Wissenschaftliche Untersuchungen

zum Neuen Testament · 2. Reihe

Herausgeber / Editor
Jörg Frey (Zürich)

Mitherausgeber / Associate Editors

Markus Bockmuehl (Oxford) · James A. Kelhoffer (Uppsala)
Hans-Josef Klauck (Chicago, IL) · Tobias Nicklas (Regensburg)
J. Ross Wagner (Durham, NC)

385
 Stephen C. Carlson

The Text of Galatians

and Its History

Mohr Siebeck
Stephen C. Carlson, born 1968; 1988 BS in Computer Engineering from Carnegie-Mellon
University; 1996 J. D. from George Mason University School of Law; 2012 PhD from Duke
University; currently a research fellow in the Institute for Religion and Critical Inquiry in the
Faculty of Theology and Philosophy of Australian Catholic University, Melbourne, and formerly
a post-doctoral researcher in Pre-Constantinian Christianity at Uppsala University.

e-ISBN PDF 978-3-16-153324-2

ISBN 978-3-16-153323-5
ISSN 0340-9570 (Wissenschaftliche Untersuchungen zum Neuen Testament, 2. Reihe)
The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliogra-
phie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de.

© 2015 by Mohr Siebeck, Tübingen, Germany. www.mohr.de

This book may not be reproduced, in whole or in part, in any form (beyond that permitted by
copyright law) without the publisher’s written permission. This applies particularly to reproduc-
tions, translations, microfilms and storage and processing in electronic systems.
The book was printed by Laupp & Göbel in Nehren on non-aging paper and bound by Buch-
binderei Nädele in Nehren.
Printed in Germany.
To my family
 Preface

This monograph is a lightly revised version of my 2012 Ph.D. dissertation at

Duke University. This project has been such a long time in the making, and
any project of this scope has so many people who deserve my thanks that I
cannot possibly name everyone. As a result, I can only acknowledge the most
formative influences upon me and my thesis.
I first encountered the heady field of New Testament textual criticism in
the academic mailing lists of the mid-1990s. Though I was but an amateur, I
got to interact with some of the best, brightest, and most technologically
savvy people in the field of Biblical historical criticism. Two of those I had
interacted with nearly twenty years ago, Mark Goodacre and Bart Ehrman,
have become my co-supervisors of this dissertation. I am profoundly grateful
to them for their knowledge, inspiration, and encouragement over these
decades. Over this time, I have kept up with the academic mailing lists, and
the one that has probably helped me the most is the Biblical Greek mailing
list, from whose participants I first heard about modern linguistic approaches
to Hellenistic Greek. I am particularly appreciative of the help I received over
the years from Carl Conrad, Edward Hobbs, Mike Aubrey, and Steve Runge.
In fact, Steve was kind enough to encourage me to call him to discuss his
work with Greek discourse grammar.
The dissertation itself derives its inspiration from an email sent to a textual
criticism mailing list about a text-critical challenge where a biologist
successfully classified various Old Norse manuscripts using off-the-shelf
software for classifying biological organisms. As someone trained in com-
puter programming myself, I became fascinated by computer-assisted meth-
ods of textual criticism using insights from systematic biology. On-line dis-
cussions with two members of the mailing list in particular, Robert B. Waltz
and Larry W. Hurtado, helped crystallize my thinking. I am grateful to them,
as well as to two biologists, Robert J. O’Hara and John Alroy, who answered
my email queries while I was learning the theory.
By this time, I had started attending sessions of New Testament textual
criticism section at the annual meetings of the Society of Biblical Literature,
and I would like to acknowledge my gratitude for all the stimulating discus-
sions I have had about textual criticism over the years, including Michael W.
Holmes, David Parker, Bill Warren, Peter Head, Tommy Wasserman, and Jan
VIII Preface

Krans among others. I have learned a lot in my discussions with them and
from reading their writings. I am especially thankful that Mike Holmes was
able to join my committee and make many useful comments on earlier drafts
of this dissertation.
I would never have been able to follow up on my idea unless I had the op-
portunity to enroll in Duke’s Ph.D. program, and I decided to make this work
a focus of my doctoral research during one of Bart Ehrman’s textual criticism
seminars open to Duke students. I learned so much from my Duke family,
including two of the members of my dissertation committee, Joel Marcus and
Douglas Campbell. They taught me much about scholarship in general and
about Judaism and Paul in particular. I am also grateful to my fellow students
at Duke and UNC, from whom I have received much intellectual stimulation,
including Hans Arneson, Maria Doerfler, Nathan Eubank, T. J. Lang, Tom
McGlothlin, Ken Olson, Jason Staples, and many others.
I would also like to express my appreciation for Jim Kelhoffer of Uppsala
University for encouraging me to submit my dissertation for publication with
Mohr Siebeck. He has been a gracious and generous host for me and my
family in my post-doc at Uppsala, and he has become a good friend to boot.
My experience in Uppsala with wonderful colleagues has been made so much
better as a result.
Of course, being able to pursue my passion at Duke and then Uppsala
meant that my family had to make many sacrifices. It is to my wife Aili and
my children, therefore, that I dedicate this book. Without their love, sacrifice,
and encouragement, this would never have been accomplished.

Uppsala and Melbourne, Summer 2014 Stephen C. Carlson

 Table of Contents

Preface .......................................................................................................VII
List of Abbreviations ................................................................................ XIII

Chapter 1: Theories of New Testament Textual Criticism ............ 1

1.1 Introduction ............................................................................................. 1

1.2 An Illustrative Textual Transmission Scenario ......................................... 9
1.3 Notions of Textual Evidence .................................................................. 13
1.4 Methods of Textual Criticism ................................................................. 18
1.4.1 Copy-Text Method ......................................................................... 18
1.4.2 Majority-Text Method .................................................................... 22
1.4.3 Stemmatics ..................................................................................... 26
1.4.4 Reasoned and Thoroughgoing Eclectic Methods ............................ 34
1.4.5 Coherence-Based Genealogical Method ......................................... 39
1.5 Proposals and Road Map....................................................................... 43

Chapter 2: Materials and Methods ..................................................... 46

2.1 Materials and Manuscripts .................................................................... 46

2.2 Computational Stemmatics: Cladistics................................................... 54
2.3 Cladistics in Action: A Simplified Example ............................................ 61
2.4 Contamination ....................................................................................... 74
2.5 A Proposed, Unoriented Stemma for Galatians ..................................... 80
X Table of Contents

Chapter 3: Orienting the Stemma of Galatians ............................... 87

3.1 Introduction ........................................................................................... 87

3.2 Examination of Variants for Inner Branch Oppositions ......................... 90
3.2.1 Inner Branch a ................................................................................ 90
3.2.2 Inner Branch b.............................................................................. 103
3.2.3 Inner Branch c .............................................................................. 117
3.2.4 Inner Branch d.............................................................................. 131
3.3 Orienting the Stemma .......................................................................... 136

Chapter 4: Examining and Establishing the Text ....................... 141

4.1 Introduction ......................................................................................... 141

4.2 Examination of Variation Units with Early, Split Support.................... 143
4.3 A Critical Text of Galatians ................................................................. 173

Chapter 5: History of Textual Variation in Galatians ................. 181

5.1 Introduction ......................................................................................... 181

5.2 The Western Branch and Prototype ..................................................... 182
5.3 The Eastern Branch and the Road to the Byzantine Text ...................... 210
5.3.1 Stage α ......................................................................................... 210
5.3.2 Stage γ .......................................................................................... 214
5.3.3 Stage ψ ......................................................................................... 220
5.3.4 Stage σ ......................................................................................... 228
5.3.5 Stage κ: The Byzantine Text and Prototype .................................. 233
5.4 Conclusions ......................................................................................... 239

Chapter 6: Contrasts and Conclusions ..................................... 241

6.1 An External History of the Text of Galatians ....................................... 241

6.2 Zuntz and Hort .................................................................................... 244
 Table of Contents XI

6.3 Contrasts with the Critical Texts of the Galatians ............................... 250
6.4 Suggestions for Further Study.............................................................. 252

Appendix A: Errata for Swanson ............................................................... 255

Appendix B: Critical Text of the Byzantine Prototype ............................... 264

Bibliography.............................................................................................. 273

Index of Ancient Sources .......................................................................... 287

Index of Modern Authors .......................................................................... 297

Index of Subjects ....................................................................................... 301

 List of Abbreviations

BDAG Walter Bauer, Frederick W. Danker, William F. Arndt, and F. Wilbur

Gingrich, eds., A Greek-English Lexicon of the New Testament and
Other Early Christian Literature (3rd ed. Chicago: The University of
Chicago Press, 2000).
BDF Friedrich Blass, Albert Debrunner, and Robert W. Funk, eds., A Greek
Grammar of the New Testament and Other Early Christian Literature
(Chicago: University of Chicago Press, 1961).
BDR Friedrich Blass, Albert Debrunner, and Friedrich Rehkopf, eds., Gram-
matik des neutestamentlichen Griechisch (14th ed.; Göttingen: Van-
denhoeck & Ruprecht, 1976).
BETZ Hans Dieter Betz, Galatians: A Commentary on Paul’s Letter to the
Churches in Galatia (Hermeneia; Philadelphia: Fortress, 1979).
BURTON Ernest de Witt Burton, A Critical and Exegetical Commentary on the
Epistle to the Galatians (ICC; Edinburgh: T. & T. Clark, 1921).
BUSCEMI Alfio Buscemi, Lettera ai Galati: Commentario esegetico (Analecta
[Studium Biblicum Franciscanum] 63; Jerusalem: Franciscan Printing
Press, 2004).
EADIE John Eadie, A Commentary on the Greek Text of the Epistle of Paul to
the Galatians (Edinburgh: T. & T. Clark, 1879).
HORT B. F. Westcott and F. J. A. Hort, Introduction to the New Testament in
the Original Greek (repr. Peabody, Mass.: Hendrickson, 1988).
LÉGASSE Simon Légasse, L’épître de Paul aux Galates (LD 9; Paris: Cerf,
2000).
LIGHTFOOT J. B. Lightfoot, St. Paul’s Epistle to the Galatians (1st ed., 1865; repr.
Peabody, Mass.: Hendrickson, 1995).
LONGENECKER Richard N. Longenecker, Galatians (WBC 41; Dallas: Word Books,
1990).
LSJ Henry George Liddell, Robert Scott, and Henry Stuart Jones, A Greek-
English Lexicon (9th ed.; Oxford: Clarendon, 1940).
MARTYN J. Louis Martyn, Galatians (AB 33A; New York: Doubleday, 1997).
MATERA Frank J. Matera, Galatians (SP 9; Collegeville, Minn.: Glazier, 1992).
METZGER Bruce M. Metzger, A Textual Commentary on the Greek New Testa-
ment (2d ed. Stuttgart: Deutsche Bibelgesellschaft, 1994).
MEYER Heinrich August Wilhelm Meyer, Critical and Exegetical Handbook to
the Epistle to the Galatians (G. H. Venables, trans.; 2d ed.; Edinburgh:
T. & T. Clark, 1904).
MUSSNER Franz Mussner, Der Galaterbrief (HTKNT 9; Freiburg: Herder, 1974).
NET Michael H. Burer, W. Hall Harris III, and Daniel B. Wallace, New
English Translation: New Testament (Dallas: Net Bible Press, 2004).
XIV List of Abbreviations

NT.VMR New Testament Virtual Manuscript Room [Muenster website located

at http://intf.uni-muenster.de/vmr/NTVMR/IndexNTVMR.php]
ROHDE Joachim Rohde, Der Brief des Paulus an die Galater (THKNT 9; rev.
ed.; Berlin: Evangelische, 1988).
ROYSE James R. Royse, Scribal Habits in Early Greek New Testament Papyri
(NTTSD 36; Leiden: Brill, 2008).
SCHLIER Heinrich Schlier, Der Brief an die Galater (KEK 7; 10th ed.; Göt-
tingen: Vandenhoeck & Ruprecht, 1949).
SIEFFERT Friedrich Sieffert, Der Brief an die Galater (KEK 7; 9th ed.; Göt-
tingen: Vandenhoeck & Ruprecht, 1899).
SMYTH Herbert Weir Smyth, Greek Grammar (Cambridge, Mass.: Harvard
University Press, 1920; Rev. ed. Gordon M. Messing, 1956).
WEISS Bernhard Weiss, Textkritik der paulinische Briefe (TU 14.3; Leipzig:
Hinrichs, 1896).
ZAHN Theodor Zahn, Der Brief des Paulus an die Galater (KNT 9; Leipzig:
Deichert, 1905).
ZIMMER¹ F. Zimmer, “Zur Textkritik des Galaterbriefes,” ZWT 24 (1881): 481–
494.
ZIMMER² F. Zimmer, “Zur Textkritik des Galaterbriefes (Cap. III.),” ZWT 25
(1882): 327–343.
ZIMMER³ F. Zimmer, “Zur Textkritik des Galaterbriefes (Cap. IV-VI),” ZWT 26
(1883): 294–308.
 Chapter 1

Theories of New Testament Textual Criticism

1.1 Introduction

Paul’s Epistle to the Galatians is one of the most important books of the New
Testament for historians and theologians. As one of the earliest Christian
documents that has survived, Galatians is historically important because it
contains the most extensive autobiographical section of any early Christian
leader. It is also theologically important because of its exposition of the role
of the Torah in Christian life in the first century, and it later became founda-
tional for the doctrine of justification by faith during the Reformation. Be-
cause of its importance, readers care about what the text of this epistle means.
The text of Galatians, however, does not exist in a single form. It exists in
a variety of forms in millions of copies. To be sure, most of these copies are
found in printed Bibles in modern language translations, but the variation in
the forms of the text is no modern phenomenon. Like other books of the New
Testament, the text of Galatians is present in hundreds of medieval
manuscripts and none of these contain the exact same form of the text. In
fact, as far back in time as the evidence can reach, the text of Galatians has
exhibited a variety of forms, at least one for each manuscript witness.
Most readers of the text do not appreciate this bewildering diversity of tex-
tual forms. They want something more convenient. They want an edition they
can read. To this end, publishers have been producing editions and transla-
tions of the text for the benefit of their readers ever since the dawn of the age
of printing. There is a social dimension as well. Readers want to share their
thoughts about the text with others, and mutual exchanges of ideas are facil-
itated by having a text in common. Otherwise, there will be disputes about
what the text says in addition to the expected disputes about what it means.
In some cases, even tiny differences over what the text says can have a
huge effect on what the text means. For example, ever since a memorable
exchange between Jerome and Augustine, exegetes have been fascinated by
what happened at Antioch as Paul described it in Gal 2:11–14. According to
the current Nestle-Aland text, this incident was triggered by the arrival of
certain people from James, who intimidated Cephas into publicly changing
his mind and separating himself from the gentiles (Gal 2:12 ὅτε δὲ ἦλθον
ὑπέστελλεν “but when they came, he withdrew”). The best and earliest manu-
2 Chapter 1. Theories of New Testament Textual Criticism

scripts, however, tell another story. A difference of a single letter in the text–
from the omicron in ἦλθον (“they came”) to the epsilon in ἦλθεν (“he came”)
– results in a markedly different understanding of the incident. Instead of
being intimidated at Antioch into changing his mind, Cephas came to Antioch
with no intention of eating with the gentiles, and this is what Paul found
objectionable. What happened at the Antioch incident and even when it
happened remains a scholarly point of contention even today.
Now, the social dimension to reading, expounding, and disputing the text
means that the nature of the text chosen for the common edition is in large
part socially constructed. This is why most forms of the text of Galatians are
now found in the vernacular. Very few people can read the Greek language
that Paul spoke. They need to access the text in a form they can manage, in
their native tongue. Nevertheless, the act of translation necessitates having a
textual base from which to translate, and this shifts the onus of coping with
the diversity of textual forms from the general readers to its translators–and
these translators cope with this diversity by relying on critical editions of the
text prepared by textual critics. 1
Like other readers of the text, New Testament textual critics also operate
within a social environment and their ideal form of the text is also socially
constructed. After all, it is not self-evident what form of the text the critical
text should represent. Should it represent the most current form of the text?
Should it represent the most popular form of the text? Should it represent the
oldest surviving form of the text? Should it represent the form of the text
when the text became canonical? Should it represent, insofar as it is possible
to determine, what Paul actually wrote to the Galatians? All of these forms of
the text are within the purview of textual criticism, but they do not all get the
same level of attention from textual critics. 2 Rather, one particular state of the
text attracts the lion’s share of the attention from textual critics, who – like
any other member of society – operate within a common framework of shared
assumptions about what is important or authoritative.
The social context within which New Testament textual criticism has been
operating is modernity. The Renaissance’s call to return to the sources (ad
fontes) led to the Protestant Reformation’s insistence on sola scriptura,
making scripture in its original languages the sole authority for purposes of

1
For an overview of scholarly editing of a broad range of texts, see D. C. Greetham,
Textual Scholarship: An Introduction (Garland Reference Library of the Humanities 1417;
corr. repr.; New York: Garland, 1994), and the collection of essays in D. C. Greetham, ed.,
Scholarly Editing: A Guide to Research (New York: Modern Language Association of
America, 1995).
2
To be sure, throughout much of the history of the text, producers of new copies of the
New Testament text have simply been content to reproduce the standard text of their time
with little to no input from textual critics, even in the age of printing. This reliance on a
“received text” is an example of a socially constructed authoritative text.
 1.1 Introduction 3

theology. 3 On the secular side, this call to return to the sources also led to the
Enlightenment’s privileging of the historical situation in which the documents
were written for purposes of historical inquiry. 4 Thus, textual critics both for
the New Testament and for classical texts have confidently proclaimed that
their goal is to restore the text of the “original.” For example, on the New
Testament side, F. J. A. Hort declared that the goal of textual criticism is
“recovering an exact copy of what was actually written on parchment or
papyrus by the author of the book or his amanuensis.” 5 On the classical side,
Paul Maas asserted, “The business of textual criticism is to produce a text as
close as possible to the original (constitutio textus).” 6 As a result, textual
critics have long taken for granted that their goal was to restore the “original
text.” 7
To be sure, there are many advantages that come from restoring the text of
an ancient work as the author wrote it. For example, the exegesis of a text
requires knowledge of the historical circumstances in which the work was
composed. 8 For this reason, the form of the text subjected to exegesis should
reflect that historical context, not a mélange of multiple historical contexts for
the author and also each scribe who changed the text. Furthermore, the histor-
ical situation of a particular author is usually more accessible to critics than
those of the scribes. Though some authors were anonymous, the scribes who

3
So Michael W. Holmes, “From ‘Original Text’ to ‘Initial Text’: The Traditional Goal
of New Testament Textual Criticism in Contemporary Discussion,” in Bart D. Ehrman and
Michael W. Holmes, eds., The Text of the New Testament in Contemporary Research:
Essays on the Status Quaestionis (2d ed.; Leiden: Brill, forthcoming), preprint pp. 5.
Holmes goes on to mention that other faith traditions such as the Roman Catholic Church
and the Eastern Orthodox Church prefer to invest authority in a standard text of a later era.
For a brief debate between two different stances about which form of the text ought to be
authoritative for theology, see David Parker, “Textual Criticism and Theology,” ExpT 118
(2007): 583–589, and John C. Poirier, “Living Text or Exquisite Corpse?” ExpT 119
(2008): 437–439.
4
Edgar Krentz, The Historical-Critical Method (Philadelphia: Fortress, 1975), 7–10,
17–19, and 35–41.
5
B. F. Westcott and F. J. A. Hort, Introduction to the New Testament in the Original
Greek (repr. Peabody, Mass.: Hendrickson, 1988) [hereinafter “HORT ”], 3.
6
Paul Maas, Textual Criticism (trans. Barbara Flower; Oxford: Clarendon Press, 1958),
1.
7
Holmes, “Original Text,” 7–11.
8
E.g., Douglas Stuart, “Exegesis,” ABD 2:682–688 at 263: “Investigating and identi-
fying the historical context of a passage is a pivotal step in the exegesis process. Recon-
struction this historical context helps provide the exegete with a potentially clearer sense of
the meaning of the passage than would otherwise be possible, by providing some of the
general knowledge that its original audience(s) relied upon to understand what was said
and/or written.”
4 Chapter 1. Theories of New Testament Textual Criticism

transmitted the work almost always were. 9 As a result, there is a greater

chance of knowing something about the author than about the unknown
scribes. 10 Similarly, historical linguists studying the state of the language at
the time the work was composed would prefer to have a text that contains the
language of that time period without later elements. Historians interested in a
time period also prefer having a witness whose text reflects that particular
time period, rather than that of another or later time period. 11 This preference
– by the consumers of the textual critic’s efforts – for a text that reflects the
time and author of an ancient work’s composition creates a powerful social
environment in favor of restoring the original text.
With this single-minded focus on the original text, scholars scoured mo-
nastic libraries for Biblical manuscripts, dug up trash heaps in hopes of find-
ing papyrus scraps of the New Testament, and combed through ecclesiastical
texts in search of early scriptural quotations. They collated the thousands of
witnesses they found, they carefully sifted through the hundreds of thousands
of collated variants, and they argued over thousands of significant points of
variation. All this effort culminated in a series of highly-regarded Nestle-
Aland critical editions of the New Testament, which was the basis for exege-
sis, translation, and historical investigation for most of the twentieth centu-
ry. 12 This success came at a price, however. It led to a stagnation of the
field. 13 By the end of the twentieth century, a dwindling number of New Tes-
tament scholars continued to argue for one or another reading as being origi-
nal, and the text of the standard Nestle-Aland critical edition effectively
remained unchanged since 1975. 14 The field entered what Eldon Jay Epp once

9
A rare exception is the scribe Ephraim, who wrote the tenth-century minuscule 1739.
For more about Ephraim, see Amy S. Anderson, The Textual Tradition of the Gospels:
Family 1 in Matthew (NTTS 32; Leiden: Brill, 2004), 22–46.
10
For a careful engagement with a frustratingly small body of material about Christian
scribes, see Kim Haines-Eitzen, Guardians of Letters: Literacy, Power, and the Trans-
mitters of Early Christian Literature (New York: Oxford University Press, 2000).
11
This is why John Knox, Chapters in a Life of Paul (New York: Abingdon-Cokesbury,
1950), privileged Paul’s letters, especially Galatians, over the later Acts of the Apostles for
the study of Pauline chronology. However, if the historian’s text of Paul’s letters and Acts
contains elements from later time period, then the basis of Knox’s methodological premise
is no longer valid.
12
For an overview of the praxis of textual criticism behind the Nestle-Aland editions,
see Kurt Aland and Barbara Aland, The Text of the New Testament: An Introduction to the
Critical Editions and to the Theory and Practice of Modern Textual Criticism (2d ed.;
trans. Erroll F. Rhodes; Grand Rapids, Mich.: Eerdmans, 1989) [hereinafter “ALAND -
ALAND”].
13
On this stagnation, see Bart D. Ehrman, The Orthodox Corruption of Scripture (2d
ed.; Oxford: Oxford University Press, 2011), 333–334.
14
The third edition of the United Bible Societies’ text was set in 1975 and adopted by
the editors of the twenty-sixth Nestle-Aland edition in 1979. The twenty-seventh edition in
 1.1 Introduction 5

dubbed the “twentieth-century interlude in New Testament textual

criticism.” 15
Against this twentieth-century backdrop of the relentless pursuit of the
original text, there were voices crying in the wilderness that there ought to be
more to textual criticism than that. Specifically, they have argued that under-
standing the history of the text – how and why the text changed over time – is
also important for understanding the history of its readers. 16 The tide began to
turn in 1993 with the publication of The Orthodox Corruption of Scripture by
Bart D. Ehrman. 17 In it, Ehrman demonstrated that the behavior of the scribes
who transmitted and modified the text had a lot to say about the ongoing
Christological controversies in the first few centuries of Christianity, and vice
versa. 18 To this end, it is necessary to look at more than what the original text
can provide. The Christian controversies revolved around the texts of their
day, not around the text of an original that had long since disappeared.
Moreover, the variants that provide a window into this past are in many cases
the very ones that textual critics have purged from the text in their quest for
the original. 19 To be sure, textual critics have documented some of these early
controversial variants in the critical apparatus below the main text, but this
documentation is often spotty and unhelpful, especially in the Nestle-Aland
hand editions, because the variants in the apparatus were chosen more for
their relevance in reconstituting the editor’s notion of the original text rather

1993 made changes only to the apparatus. See Barbara Aland, Kurt Aland, Johannes Kara-
vidopoulos, Carlo M. Martini, and Bruce M. Metzger, eds., Novum Testamentum Graece
(27th ed.; Stuttgart: German Bible Society, 1993), 2*–3*, 44*–45*. Beginning with James
in 1997, the editors of the Nestle-Aland edition have been publishing a major critical
edition in parts, with some changes to the text.
15
Eldon Jay Epp, “The Twentieth Century Interlude in New Testament Textual Criti-
cism” in Eldon J. Epp & Gordon D. Fee, eds., Studies in the Theory and Method of New
Testament Textual Criticism (Grand Rapids, Mich.: Eerdmans, 1993), 85–108. See also
Eldon Jay Epp, “A Continuing Interlude in New Testament Textual Criticism?” in Studies
in the Theory and Method, 109–123; and L. W. Hurtado, “Beyond the Interlude? Develop-
ments and Directions in New Testament Textual Criticism” in D. G. K. Taylor, ed., Studies
in the Early Text of the Gospels and Acts (Atlanta: SBL, 1999), 26–48.
16
See Ehrman, Orthodox Corruption, 331–332.
17
Ehrman was not without his precursors, the most important of whom was Eldon Jay
Epp, The Theological Tendency of Codex Bezae Cantabrigiensis in Acts (SNTSMS 3;
Cambridge: Cambridge University Press, 1966).
18
Ehrman, Orthodox Corruption, 320–330.
19
See Bart D. Ehrman, “The Text as Window: New Testament Manuscripts and the
Social History of Early Christianity” in Bart D. Ehrman and Michael W. Holmes, eds., The
Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis
(SD 46; Grand Rapids, Mich.: Eerdmans, 1995), 361–379.
6 Chapter 1. Theories of New Testament Textual Criticism

than for their relevance in the study of the transmission of the text. 20
Ehrman’s study and the others that have followed in its wake have led to a
greater emphasis upon the variants generated within the historical trans-
mission of the text as a way to study the development of early Christianity. 21
In point of fact, Epp has called for a “variant-conscious” edition of the
New Testament that would facilitate the study of textual variants within the
history of the text. 22 According to Epp’s proposal:
A variant-conscious edition, then, would display significant variants in a single running
text that, at each variation-unit, places the selected variant readings in horizontal
comparison, one below another, and then reverts to a running text until the next variation-
unit is reached. A separate apparatus would provide the attestation for various readings
displayed in the running text. 23

Epp’s proposal is no mere intellectual exercise. It is, in fact, very similar to

the approach of the Editio Critica Maior in the Catholic Epistles published by
the Institute for New Testament Textual Research in Muenster. 24 The main
difference between Epp’s proposal and Muenster’s edition is that Muenster is
much more comprehensive in printing variants under the baseline text, rather
than restricting them to those deemed “significant.” 25 While this approach is
an important advance in the study of the history of the text, it does not go far
enough. The history of individual textual variation is not the history of the
text. A text is more than its individual readings. It is a pattern of readings, in
a specific sequence and embodied in a single witness, and it is this pattern of
readings that may tell us whether the readers of the text were consistent or
scatter-brained in their attitude to Christology, apologetics, Judaism, and
women. A text is a pattern of readings, so the history of the text ought to be

20
E.g., ALAND -ALAND, 232: “Nestle-Aland 27 does not report every variant, including,
e.g., the obviously singular readings of individual manuscripts.”
21
Ehrman, Orthodox Corruption, 352–361, has an overview of recent studies.
22
Eldon Jay Epp, “It’s All about Variants: A Variant-Conscious Approach to New
Testament Textual Criticism,” HTR 100 (2007): 275–308.
23
Epp, “Variants,” 301.
24
E.g., Barbara Aland, Kurt Aland, Gerd Mink, and Klaus Wachtel, Novum Testa-
mentum Graece: Editio Critica Maior, vol. 4, pt. 1.1, James (Stuttgart: Deutsche Bibelge-
sellschaft, 1997). Curiously, though Epp cites Muenster’s major critical edition (“Variants,
286 n.38), he does not mention it as a model for his variant-conscious edition.
25
It is just as well that the Muenster edition is not limited to “significant” variants.
These are notoriously are hard to define. Though Epp suggests that they ought to relate to
the “theological, liturgical, and ethical contexts . . . in the life of the church” (“Variants,”
294), this criterion seems too subjective to apply consistently and unlikely to satisfy the
full range of issues that scholars find interesting.
 1.1 Introduction 7

concerned with the history of those patterns of readings, not disembodied

variants. 26
Another important part of Epp’s proposal is that the running, baseline text
in his variant-conscious edition is pointedly not the “original” text, but the
“earliest attainable text.” This distinction is necessary, Epp contends, because
“the nature of New Testament textual transmission virtually precludes any
ultimate identification of ‘earliest attainable’ with ‘the original.’”27 This, too,
is the approach adopted by Muenster in their concept of an “initial text,”
which is somewhere between the author’s original and the text that can be
reconstructed based on the surviving variants. 28 The humility behind this
distinction is certainly refreshing, but, while these alternative phrasings may
avoid the term “original,” they do not avoid the concept of an original,
authorial text. 29
A good illustration of this tension is evident in Epp’s article on whether
there was an apostle named Junia in Rom 16:7. 30 At the beginning of the
essay, Epp problematizes the concept of an original – “at least in this writer’s
judgment, textual critics must now speak of multivalence in the term ‘original
text’” 31 and “the search for a single original text or reading may have to be
abandoned.” 32 Yet, when he addresses Rom 16:7, Epp does not abandon the
quest for the original. Rather, he makes such categorical statements as “Paul,

26
To be sure, Epp, “Variants,” 300, is aware of this point, but he counsels against it as
“inappropriate for the Greek text” because “the issue of text types in New Testament
textual criticism is highly complex and still debated, and it is essential, rather, to present
the textual variants without invoking one theory or another.” My claim is that, if we are
serious about the history of the text, then we have to confront the issue of text types (see
Chapter 2).
27
Epp, “Variants,” 294 (emphasis original).
28
See, e.g., Gerd Mink, “Problems of a Highly Contaminated Tradition: The New
Testament–Stemmata of Variants as a Source of a Genealogy for Witnesses” in Piet van
Reenen, ed., Studies in Stemmatology II (Philadelphia: Benjamins, 2004), 13–85 at 25–27.
A more detailed overview of this trend is found in Holmes, “Original Text,” 11–25.
29
See Emanuel Tov, Textual Criticism of the Hebrew Bible (3d ed.; Minneapolis, For-
tress, 2012), 162: “Almost all scholars are involved with the evaluation of textual variants,
but often they may not be aware that this procedure actually requires the acceptance of
some idea of an original text in some form.”
30
Eldon Jay Epp, “Text-Critical, Exegetical, and Socio-Cultural Factors Affecting the
Junia/Junias Variant in Romans 16,7,” in A. Denaux, ed., New Testament Textual Criticism
and Exegesis (Festschrift J. Delobel; BETL 161; Leuven: Leuven University Press, 2002),
227–291. Epp published an expanded form of this essay in book form as Epp, Junia: The
First Woman Apostle (Minneapolis: Fortress, 2005).
31
This is in reference to his earlier article, “The Multivalence of the Term ‘Original
Text’ in New Testament Textual Criticism,” HTR 92 (1999): 245–281. The sense that is
necessary to his historical argument is the “autographic text-form,” rather than the “prede-
cessor text-form,” the “canonical text-form,” or the “interpretive text-form” (276–277).
32
Epp, “Junia/Junias Variant,” 228; Epp, Junia, 5–6.
8 Chapter 1. Theories of New Testament Textual Criticism

in Rom 16,7, requested his readers,” 33 and “The conclusion to this essay is
simple and straightforward: there was an apostle Junia.” 34 These statements
presuppose that the text he recovered was Paul’s authorial original rather than
someone else’s. Otherwise, Rom 16:7 would have no more authority for
Epp’s historical analysis than the later Paulinists whom Epp held were re-
sponsible for 1 Cor 14:34–35, Eph 4:22–24, Col 3:18, and 1 Tim 2:8–15. 35
Thus, Epp’s historical criticism about the apostle Junia requires his textual
criticism of Rom 16:7 to have produced a reading closely connected to what
Paul actually wrote. In other words, for Epp’s historical argument to work,
the “earliest attainable text” has to be Paul’s authorial text – an “original”
text, not some later form of the text – yet this is the very equation that Epp
calls into question.
The shift in focus from the “original text” to the history of the text over
the past quarter-century has been a welcome development. It has rejuvenated
the field. Nevertheless, it is important not to drive a wedge between a text and
its history. We need one to study the other. 36 History, after all, is the study of
change over time. One cannot do a history of the text – that is, how the text
changed over time – without a sense of what the text looked like before a
scribe changed it and what it looked like afterwards. Textual critics, there-
fore, need to be able to assess which forms of the text precede other forms
and, as part of this enterprise, identify a form of the text that precedes all the
other surviving forms of the text. Furthermore, it is important to understand
what relationship this archetypal text has to the original, authorial text that so
interests historical critics, linguists, theologians, and other customers of tex-
tual criticism. Whether the goal is the “original text,” the “earliest attainable
text,” the “initial text,” or even the “archetypal text,” all these proposed forms
of the text are the products of modern text-critical methods. 37 As the survey
of text-critical methods in the field of the New Testament later in this chapter
will demonstrate, all of them produce a critical text that more or less approx-
imates an authorial, original text to the extent that they rely upon a class of
textual evidence called internal evidence.

33
Epp, “Junia/Junias Variant,” 243. Also Epp, Junia, 79: “the Pauline phase in Rom
16:7.”
34
Epp, “Junia/Junias Variant,” 290 (emphasis original); Epp, Junia, 80 (emphasis origi-
nal).
35
Epp, “Junia/Junias Variant,” 291; Epp, Junia, 80–82.
36
So also Holmes, “Original Text,” 7: “In short, the two goals are best understood not
as competing but as complementary pursuits–perhaps even as two sides of a single coin. To
pursue one is not at all either to privilege it or to argue against the value of the other.”
Similarly Epp, “Variants,” 294.
37
Epp, “Variants,” 287, is explicit on this point, stating that the “earliest attainable text”
is “a text, or better, texts that represent the best that modern text-critical resources and
methods can recover.”
 1.2 An Illustrative Textual Transmission Scenario 9

Where these text-critical methods significantly differ, however, is in their

approach to the history of the text. Some impose an a priori history of the
text upon the evidence, but others attempt to reconstruct the history from the
textual evidence. If the history of the text is also important, it is important to
adopt a text-critical method that actually infers a history of the text. Con-
sequently, in order to test how well various text-critical methods are able to
infer a history of the text, a hypothetical history of the text of Galatians is
presented next and then used to evaluate the efficacy of several text-critical
methods. Based on the results of this evaluation, one of these text-critical
methods is adopted for the remainder of this study into the text of Galatians
and its history.

1.2 An Illustrative Textual Transmission Scenario

Every text-critical method for producing an edition of a text presupposes a

historical model for evaluating the transmission of the text. As Michael W.
Holmes once put it, “In short, the evaluation of individual readings depends
greatly on how the critic views them in relation to the larger picture of the
history of the text, because there is a synergistic relationship between history
and method.” 38 Like all models, the history of the text presupposed by any
particular method is necessarily a simplification of the actual, complex, and
messy transmission of the text. Thus, inherent in any model are limitations
and assumptions that affect the usefulness of the model for evaluating the
evidence for a text. In order to elucidate the limitations and assumptions of
commonly employed text-critical methods, this section presents a scenario for
the transmission of the text that is more complicated and indeed messier than
the histories of the text adopted by those methods.
FIG. 1 depicts a possible scenario for the transmission of Paul’s letter to
the Galatians from the middle of the first century down to the present day.
The purpose of this scenario is not so much to propose an actual textual
history to be defended, but rather to explore and expose the limitations and
assumptions of various methods that textual critics use. Thus, the historical
scenario presented in FIG. 1 may be more complicated than what happened in

38
Michael W. Holmes, “The Case for Reasoned Eclecticism” in David Alan Black, ed.,
Rethinking New Testament Textual Criticism (Grand Rapids, Mich.: Baker, 2002), 77–100
at 92. Also earlier, Michael W. Holmes, “Reasoned Eclecticism in New Testament Textual
Criticism,” in Bart D. Ehrman and Michael W. Holmes, eds., The Text of the New Testa-
ment in Contemporary Research: Essays on the Status Quaestionis (SD 46; Grand Rapids,
Mich.: Eerdmans, 1995), 336–360 at 350.
10 Chapter 1. Theories of New Testament Textual Criticism

some aspects but simpler in other aspects. Nevertheless, this scenario is

informed by research into letter-writing and book production in antiquity. 39

Scribe's
draft
Ω 0 Sent
copies
Retained
copy
Ω 1 Ω Ω 2 3

Letter
Collection

ω 0

ω 1 ω 2

Lost Exemplars
(Greek letters)

α β γ δ

b 1 d 1

ε a 1 c 1 c 2 d 2

d 3

Extant Witnesses
e 1 e 2 e 3 e 4 e 5 e 6 e 7 e 8 e 9 (Roman letters)

FIG. 1: A Hypothetical Textual Transmission Scenario for Galatians

At the top of FIG. 1, this scenario presumes that Paul dictated the bulk of his
letter to a scribe or secretary, up through Gal 6:10, producing a draft Ω₀. 40
Though Paul is the dominant voice in the epistle, suggested by the prevalent
use of the first person, 41 the prescript discloses the presence of additional
people who are with Paul (Gal 1:2 καὶ οἱ σὺν ἐμοὶ πάντες ἀδελφοί, “and all
the brothers with me”). This means that the dictation of the letter ought to be
envisioned as occurring in a communal setting, and it is conceivable that parts

39
See especially Harry Y. Gamble, Books and Readers in the Early Church: A History
of Early Christian Texts (New Haven, Conn.: Yale University Press, 1995); and E. Ran-
dolph Richards, Paul and First-Century Letter Writing: Secretaries, Composition and
Collection (Downer’s Grove, Ill.: InterVarsity, 2004).
40
Gamble, Books, 95–96.
41
E.g. the use of the emphatic pronoun ἐγώ in Gal 1:12; 2:19, 20; 4:12; 5:2 (Ἰδε ἐγὼ
Παὐλος λέγω ὑμῖν “See, I, Paul, tell you”), 10, 11; 6:17.
 1.2 An Illustrative Textual Transmission Scenario 11

of the dictated letter may have originated or shaped from someone in the
audience with Paul at the time. 42 In all likelihood, the dictated draft of the
letter would have been written on a temporary medium such as wax tablets or
scratch sheets of parchment. 43
After some possible corrections to the temporary copy, the scribe then
copied the letter from the temporary medium to a papyrus or parchment sheet
suitable for sending to the Galatians. 44 On this copy, marked Ω₂, Paul wrote a
handwritten note beginning with 6:11 Ἴδετε πηλίκοις ὑμῖν γράμμασιν ἔγραψα
τῇ ἐμῇ χειρί (“See what large letters I write to you in my own hand.”). 45 Like
many letter writers in antiquity, it is reasonable to posit that Paul had his
scribe copy the entire letter, either from the draft or the official letter, in order
to retain a copy for his own reference. 46 This copy is labeled Ω₁. With multi-
ple destinations in Galatia (1:2), it is also conceivable that other copies of the
letter (e.g., Ω₃) could have been made at this time or later by the Galatians
themselves as the letter carrier traveled from church to church. 47 Given the
imperfections of the process of copying a text by hand and the possibility that
Paul may have made alterations to some of the copies, it is probable that there
were already variant readings among the copies Ω₀, Ω₁, Ω₂, and Ω₃. 48
Now, many scholars believe that one or more collections of Paul’s letters
were gathered together, perhaps at the end of the first century or near the
beginning of the second century. 49 It is unclear to what extent the maker of
this collection sought out different copies of the letter. Perhaps the collector
had access to Paul’s retained copy Ω₁, perhaps one of the Galatians’ copies
Ω₂, or perhaps both. The simplest scenario is that the collector merely had
access to a single copy of Galatians, but, to complicate the scenario, we will

42
Richards, Paul, 45–46.
43
Richards, Paul, 47–58.
44
Richards, Paul, 161–165.
45
This wording of course is a text-critical reconstruction, and some of the variant
readings are discussed infra.
46
Gamble, Books, 100–101; Richards, Paul, 158–161. There are many reasons why it
was prudent to retain a copy of letters in antiquity. Many of them were lost in transmission.
47
Gamble, Books, 97. Richards, Paul, 157 n.8, points out that the mention of Paul’s
handwriting in Gal 6:11 points away from the scenario that Paul had prepared multiple sent
originals.
48
The complication of this scenario raises such questions as which copy is to be
considered “the autograph.” The first and temporary copy Ω 0? The copy Ω2 sent to the
Galatians? Or Paul’s personal copy Ω1? All of these copies are autographs in the sense of
being produced and authorized by the author, yet their very plurality makes the use of the
definite article problematic.
49
Gamble, Books, 59–63, 100; Richards, Paul, 210–223. For an overview of various
theories, see also Stanley E. Porter, “When and How was the Pauline Canon Compiled? An
Assessment of Theories” in Stanley E. Porter, The Pauline Canon (PS 1; Leiden: Brill,
2004), 95–127.
12 Chapter 1. Theories of New Testament Textual Criticism

assume that the collector had access to at least two copies, collated them, and
reconciled their variants, either by selecting one of them or by writing an
alternative reading in the margin. The form of the text of Galatians in the
letter collection at this stage in the tradition is labeled ω₀ in FIG. 1.
In this scenario, the initial letter collection was copied many times, each
time introducing scribal variants in the process. One copy, ω₁, left no direct
copies but survived long enough to become a source of corrections for
manuscript b₁. Another copy of this letter collection, however, was destined
to become the common ancestor of every surviving manuscript of the letter,
labeled ω₂ in FIG. 1. In particular, copy ω₂ itself was copied several times
and four of its copies are labeled α, β, γ, and δ in FIG. 1. All of these copies
were lost to the vagaries of time, war, and worm, but they each were the
beginning of their own branch in the history of the text. Furthermore, copy α
became the exemplar for two of its own copies. One survived as extant wit-
ness a₁; another copy ε did not survive but became the ultimate ancestor of
the majority of surviving witnesses: e₁, e₂, e₃, e₄, e₅, e₆, e₇, e₈, and e₉. Copy
β, on the other hand, was only copied once, as copy b₁. Copy γ was a little bit
more fruitful: it produced two pure descendants c₁ and c₂. As for copy δ, the
history is considerably more complicated in this scenario. First, its scribe had
access to a descendant of one of the sent copies Ω₃ and used that exemplar to
change some of the readings of its main source ω₂. The mixture does not stop
here, however. One of copy δ’s descendants, d₁, itself was corrected based on
lost exemplar γ. (Another descendant d₂ was not so corrected, however.)
Furthermore, d₁’s descendant d₃ was corrected, this time against the popular
exemplar ε.
In sum, FIG. 1 presents a possible historical scenario for the transmission
of Galatians designed to challenge the assumptions of different approaches to
studying its textual history. There were multiple authorial originals, Ω₀, Ω₁,
Ω₂, and Ω₃. There was an early recension of the text ω₀ that was based on
collating and combining multiple exemplars. To complicate matters, it is
assumed that this copy presented alternative readings or notes in places. A
further complicating assumption is that ω₀ is not the most recent common
ancestor of the surviving witnesses to the text. Rather, that ancestor is ω₂.
Furthermore, the textual history from this ancestor was not uniform. One
branch of the text beginning from ε, via α, was especially prolific and it
supplies the majority of the extant manuscripts of the text. Another part β
produced only one witness but it was corrected from an earlier edition of the
letter collection. Still another part of the tradition, starting with δ, was rife
with contamination.
 1.3 Notions of Textual Evidence 13

1.3 Notions of Textual Evidence

The transmission of a text is a historical process, and this process produces

evidence of what happened in the form of manuscripts that embody a particu-
lar sequence of words. More specifically, at each stage in the transmission, a
scribe prepares a manuscript by copying and changing the wording of one or
more exemplars. 50 Unfortunately, much of this evidence has perished. Manu-
scripts have been worn out, lost, or destroyed, and scribes rarely, if ever,
identify their exemplars in those manuscripts that have survived. Thus, we
lack most of the physical evidence produced by the process of textual trans-
mission, and almost all of the relational evidence (which manuscript was the
source for the other). For example, all that is left of the hypothetical textual
history depicted in FIG. 1 are sixteen witnesses to the text.
The surviving evidence has a concrete component and an abstract com-
ponent. The concrete component is an actual manuscript that came into being
at a particular place and time in history. The abstract component constitutes
the words represented by the physical marks on the manuscript. This dis-
tinction leads naturally to the traditional division of the text-critical evidence
into external evidence and internal evidence. 51 Specifically, external evidence
pertains to the identity of the physical witness that embodies a form of the
text, and it is used whenever a textual critic argues about a variant reading
based on which manuscript has that reading. On the other hand, internal evi-
dence relates to the form of the text as compared with another, and it is used
whenever a textual critic argues for or against a reading based on its wording.
The value of external evidence provided by a witness to the text depends
on the distance between the witness and the desired point of interest within
the transmission of the text. When the textual distance is very small, external
evidence is very powerful. For example, in FIG. 1, manuscript c₁ is the best
evidence for the form of the text produced by its scribe. Here, the textual
distance is zero. Manuscript c₁ may be fairly good evidence for the form of
the text in its lost, immediate exemplar γ, though it differs from it in an un-
known number of places. On the other hand, manuscript c₁ would be a poor
representative of the text of exemplar ε, for it is further removed from it. A

50
Ulrich Schmid, “Scribes and Variants–Sociology and Typology” in H. A. G. Hough-
ton and D. C. Parker, eds., Textual Variation: Theological and Social Tendencies? Papers
from the Fifth Birmingham Colloquium on the Textual Criticism of the New Testament
(TS3 6; Piscataway, N.J.: Gorgias, 2008), 1–23, makes the point that readers and owners of
a manuscript can be a source of textual variation when they alter the text of their manu-
scripts. Schmid’s point is well taken, and it should be understood that any references in this
study to scribal changes also include the textual alterations made by readers and owners of
the manuscript that were inscribed into a copy of its text by a scribe.
51
E.g., ALAND -ALAND, 280: “Only the reading which best satisfies the requirements of
both external and internal criteria can be original.”
14 Chapter 1. Theories of New Testament Textual Criticism

better, but not perfect, representative for the text of ε would be any of its
descendants, for example, manuscript e₁.
Unlike external evidence, internal evidence is directional. In its broadest
formulation, internal evidence assesses which of two readings better explains
the other. 52 Internal evidence is typically subdivided into two components:
intrinsic probabilities and transcriptional probabilities, one relating to the
author and the other relating to scribes, respectively. 53 More specifically,
intrinsic probabilities pertain to the fitness of the readings for the author’s
content and style, and transcriptional probabilities pertain to the likelihood
that a scribe would have changed one reading into the other. 54 Though these
two components are often distinguished from one another, they are really two
sides of the same coin, for the assessment of these probabilities assumes that
it is possible to detect differences in the textual behaviors of an author who
originated a text and of a scribe who imperfectly transmitted the text.
For instance, both authors and scribes attempt to produce a text that makes
good sense, but the principle of the harder reading (lectio difficilior potior)
assumes that authors and scribes make textual sense in different ways. In
particular, a scribe’s correction usually appears smoother on the surface while
the author’s form of the text fits better at a deeper level. 55 In other words, the
reading that appears harder to a scribe may, in essence, fit the author’s text
better than what appears to be the easier reading. As a result, both the in-
trinsic and the transcriptional probabilities have to be balanced against each
other to estimate which reading has priority over the other. By selecting the
reading that appears relatively more authorial than scribal, the use of internal
evidence attempts to establish an authorial text.
The appeal to an author’s content and style has prompted a theoretical ob-
jection, however. Jacobus Petzer points out that, with the loss of the author’s
manuscripts, the only access to the author’s content and style is at best indi-
rect, mediated through the text’s surviving witnesses. 56 This critique actually
does not go far enough. It can also be applied to transcriptional probabilities.

52
HORT , 19, formulates the principle that internal evidence “consists in dealing with
each variation independently, and adopting at once in each case out of two or more variants
that which looks most probable.”
53
HORT, 20. Interestingly, intrinsic and transcriptional probabilities are not usually dis-
tinguished in Old Testament textual criticism (Tov, Textual Criticism, 275 n.21).
54
HORT, 20.
55
So HORT, 47: “It follows that, with the exception of pure blunders, readings origina-
ting with scribes must always at the same time have combined the appearance of improve-
ment with the absence of its reality. If they had not been plausible, they would not have
existed: yet their excellence must have been either superficial or partial, and the balance of
inward and essential excellence must lie against them.”
56
J. H. Petzer, “Author’s Style and the Textual Criticism of the New Testament,” Neo-
testamentica 24 (1990): 185–197.
 1.3 Notions of Textual Evidence 15

After all, there is no direct evidence of the behavior of any particular scribe
because we usually have no access to the scribe’s exemplar. 57 At best, under-
standing the scribe’s behavior is indirect, based on the inference of scribal
tendencies in analogous circumstances. 58 Consequently, the use of both
intrinsic probability and transcriptional probability requires the textual critic
to construct behavioral models, admittedly upon incomplete evidence, of both
the author who originated the text and the various scribes who handled the
text. 59 As David C. Greetham, a theorist of textual scholarship, put it:
Being a critic means being sensitive to another person’s quirks and peculiarities; it means
that the critic must by an almost phenomenological leap, “become” the other person while
preparing the text for publication. And this is true whether the other person is the author or
one of the text’s transmitters, scribe, compositor, printer, proof-reader, or publisher’s
editor. 60

The value of internal evidence is thus sensitive to the assumption that,

textually, authors behave differently from scribes. The internal evidence is
clear-cut when the intrinsic probabilities suggest one reading as coming from
the author while the transcriptional probabilities suggest the other reading as
coming from a scribe. When the probabilities line up like this, internal
evidence can be very powerful. On the other hand, when the probabilities do
not line up so nicely, internal evidence can be very difficult to evaluate.
Sometimes, however, the behavior of scribes who transmitted the text cannot
be sharply distinguished from an author who originated the text, and so the
diagnostic value of internal evidence suffers as a result.
On the one hand, authors can behave like scribes. This can happen, for
example, when an author revises an earlier edition, creating authorial variants
in the process. In the hypothetical textual history depicted in FIG. 1, authorial
variants between originals Ω₁, Ω₂, and Ω₃ can enter the tradition of surviving
manuscripts in one of two ways. One way is through marginal readings. For

57
If we had access to the scribe’s exemplar, we would just use that and eliminate the
derivative copy as irrelevant.
58
Detailed treatments of scribal behavior are few and far between, but an impressive
recent study is James R. Royce, Scribal Habits in Early Greek New Testament Papyri
(NTTSD 36; Leiden: Brill, 2008).
59
Ehrman, Orthodox Corruption, 350–353, is one of the few who seem to be aware of
this theoretical issue, though he limits the discussion to intrinsic probabilities. Specifically,
Ehrman argues (352): “We are reconstructing an author. Doing so allows us to apply
intrinsic probabilities in evaluating which of the surviving readings go back to his pen and
which ones do not. . . . This author has verbal, stylistic, literary, and theological predilec-
tions. Recognizing them allows us to decide which readings go back to his imaginary pen
and which ones were later creations of scribes.” I argue here that we must also reconstruct
the scribes, and so both intrinsic and transcriptional probabilities are on similar epistemo-
logical footings.
60
Greetham, Textual Scholarship, 296.
16 Chapter 1. Theories of New Testament Textual Criticism

example, the compiler of the letter collection ω₀ may have had access to
multiple originals or copies of them, collated them, and included the variant
readings in the margin of the text. If these marginal readings survived into the
archetype ω₂ of surviving tradition, then subsequent copyists could have
chosen to reproduce the reading in the main text or in the margin. Wherever
these copyists differed in their choices, there would be variant readings that
also happen to be authorial readings because the variant ultimately owe their
origin to the author. Another way for authorial variants to survive in the
transmission of the text is through contamination from a document whose line
of descent is independent of that of the originals used to create the archetype.
Referring again to FIG. 1, the scribe who created exemplar δ might have had
access to an authorial original Ω₃ (or its descendant) and incorporated
readings from that source into the scribe’s copy δ. This process too would
create authorial variants.
It is unclear, however, how internal evidence can help the textual critic
adjudicate between authorial variants. On the surface, authorial variants look
like any other kind of textual variant, and, as a result, the textual critic must
employ general-purpose rules for every variant, because the textual critic
cannot tell just by looking at them which textual variant is scribal and which
happens to be authorial. Although the internal evidence rules work well for
scribal variants (since they attempt to distinguish between authorial and
scribal textual behavior), they do not work so well for authorial variants.
Consider the principle of the harder reading, for instance. An author may
change infelicitous wording in an earlier form of the work so that it both
better fits the context and also appears smoother. If the harder reading is
chosen according to the harder reading principle (even though it does not
happen to be better suited to the context), then, in effect, the textual critic
would have chosen the reading that the author created but then rejected. This
consideration, in fact, is the reason why, in fields where multiple authorial
originals (autographs) have survived, textual critics generally choose the last
autograph as their goal. 61 Unless there are variants involving datable histori-
cal events, dating the autographs is a question of external evidence, not inter-
nal evidence. At any rate, no autograph of a literary work has survived from
antiquity, so the question of choosing which autograph to print and read is
moot. Even if authorial variants did exist, our methods of evaluating internal

61
G. Thomas Tanselle, Textual Criticism and Scholarly Editing (Charlottesville, Va.:
University Press of Virginia, 1990), 29: “The job of a scholarly editor, therefore, can be
stated as the exercise of critical thinking in an effort to determine the final intention of an
author with respect to a particular text.” But see James Thorpe, Principles of Textual Criti-
cism (San Marino, Calif.: The Huntington Library, 1972), 46–47, quoting editors of
English classics including Jonathan Swift’s Gulliver’s Travels for this principle, but ar-
guing that the critic should select the one that best reflects the author’s intentions even if it
is not the last.
 1.3 Notions of Textual Evidence 17

evidence are not designed to discriminate among authorial variants, but only
between authorial and scribal variants. In this sense, Kurt and Barbara
Aland’s rule “Only one reading can be original,” may not necessarily be
correct on historical grounds, 62 it is sensible to adopt on methodological
grounds.
On the other hand, scribes can behave like authors when they thoroughly
revise a text. For example, in FIG. 1, it is conceivable that the compiler of a
letter collection ω₀ could have reworked the letters in the collection so
thoroughly as to put the compiler’s own stylistic stamp on the text. This, for
example, is the view of J. C. O’Neill who holds that the current form of
Galatians had been extensively interpolated. 63 Now, O’Neill does not have
external evidence for the interpolations, so he had to presume a certain
authorial consistency of personality for the real Paul and assign all passages
that do not fit his notion of Paul to a later editor. 64 Despite O’Neill’s claim
that he is merely working within a text-critical tradition that employs
conjecture to restore the text, 65 O’Neill’s attempt to disentangle an original
author Paul from a later editor does not involve the use of internal evidence
as explicated here, because he is not comparing the behavior of an author
with that of a scribe. Rather, what O’Neill is trying to do is distinguishing
two different strands of authorship within a text. Separating different authors
properly belongs to the domain of source criticism, not textual criticism.
What the evidence and methods of textual criticism can provide is a text that
is reasonably free of later, scribal changes to the text. So, in the case of a
thorough revision of the text by a later individual, textual criticism in effect
treats both the author and the reviser as co-authors and delivers up a joint-
authored text for the source-critical customer. 66 This complication comes at a
cost, however; with many cooks the flavor of the broth may become muddled,
and the intrinsic probabilities as to the authorial style may be harder to
discern as a result and therefore less decisive.

62
ALAND -ALAND, 280. Georg Luck, “Textual Criticism Today,” AJP 102 (1981): 164–
194 at 175–177, discusses the possibility of authorial variants in ancient Greek literature.
63
J. C. O’Neill, The Recovery of Paul’s Letter to the Galatians (London: SPCK, 1972):
“The only explanation must be that Paul’s epistle was edited for publication some time
after it had been in use for preaching and teaching. At that stage the editor felt free to add
material, perhaps on the grounds that its antiquity guaranteed its authenticity.”
64
E.g. O’Neill, 9: “The great difficulty about accepting all Galatians as written by Paul
is that Paul can scarcely have adopted the attitude towards Judaism that sometimes appears
in Galatians” and 73: “the epistle which I conjecture Paul wrote is reasonably clear and
consistent.”
65
O’Neill, 84.
66
E.g., Petzer, “Author’s Style,” 192: “It is then a redacted text which is being dealt
with–a text in which a number of authors have directly participated, each changing, adding
to or omitting from the original text, becoming a co-author of that text.”
18 Chapter 1. Theories of New Testament Textual Criticism

Different methods of textual criticism balance the use of external and

internal evidence differently. Some methods rely mostly on external evidence
and hardly consider internal evidence at all, while other methods rely on
internal evidence almost exclusively. Still other methods give approximately
equal consideration to external and internal evidence. The following sections
describe a spectrum of different textual methods starting from those which
favor external evidence and ending with those which favor internal evidence.

1.4 Methods of Textual Criticism

1.4.1 Copy-Text Method

The copy-text method is the most basic method of textual criticism. Its practi-
tioner simply obtains a copy of a text (the “copy-text”) and uses it to produce
another copy, possibly with emendations to the text. 67 This method is princi-
pally the way that texts have been reproduced throughout much of its history
of transmission, not only during the manuscript age but also during the age of
printing. 68 Indeed, when the New Testament was first published by Erasmus
in 1516, he used a local Greek minuscule manuscript for his copy-text of
Paul’s letters. 69 This method has continued to be used in Biblical textual criti-
cism well into the modern day. For example, Philip Buttmann’s nineteenth-
century edition of the New Testament for the Teubner series of classical texts
used Codex Vaticanus as its copy-text. 70 As another example, the Biblia
Hebraica Stuttgartensia (BHS), which is today’s standard scholarly text for

67
In the editing of English literature, copy-text editing has acquired a specialized sense
of using a copy-text merely for a work’s accidentals (e.g. spelling and punctuation). See
W. W. Greg, “The Rationale of Copy-Text,” Studies in Bibliography 3 (1950): 19–36. As
Greg points out, in Biblical and classical textual criticism, such issues as spelling are rou-
tinely normalized (20). Indeed, Westcott and Hort’s Introduction is fairly unusual in its
detailed treatment of orthography and punctuation (HORT 302–322, App, 143–173).
68
Sebastiano Timpanaro, The Genesis of Lachmann’s Method (trans. Glenn W. Most;
Chicago: University of Chicago Press, 2005), 45–47.
69
According to Andrew J. Brown, Opera omnia Desiderii Erasmi Roterodami:
recognita et adnotatione critica instructa notisque illustrata (ordo 6; tome 3; Amsterdam:
Elsevier, 2004), 4–6, the printer’s copy was an eleventh-century manuscript, Basel, Öffent-
liche Bibliothek, A. N. III. 11 (Gregory-Aland 2817; Wettstein-Tischendorf 7p), which was
corrected with readings from a twelfth-century manuscript, Basel, Öffentliche Bibliothek,
A. N. IV. 4 (Gregory-Aland 2815; Wettstein-Tischendorf 2p).
70
Philip Buttmann, Novum Testamentum Graece: ad fidem potissimum codicis Vaticani
B recensuit, varias lectiones codicis B, Textus Recepti, editionum Griesbachii Lachmanni
Tischendorfii integras adiecit (3d rev. ed.; Leipzig: Teubner, 1865), as its title indicates.
 1.4 Methods of Textual Criticism 19

the Hebrew Bible, uses the eleventh-century Codex Leningradensis as its

copy-text, even to the point of printing the codex’s errors in the main text. 71
The historical model presupposed by the copy-text method is conceptually
simple. As illustrated in FIG. 2, the copy-text method assumes that the work
being edited was produced by an author in an original form, labeled Ω, and
eventually transmitted to an extant witness to the text, labeled e₁, by a chain
of transmission, denoted by a long arrow, involving an unknown number of
scribes who imperfectly copied their respective exemplars. The method thus
assumes that the witness e₁ is not the same manuscript as the original Ω, but
its descendant. 72 Given the inaccuracy of reproduction in ancient copies, the
witness e₁ is assumed to differ from the original Ω at various places in the
text.

e 1

FIG. 2: A Copy-Text Model

In accordance with the copy-text method, the copy-text has a presumptive

authority. That is, the text of the copy-text is the basis of the textual critic’s
edition unless it can be affirmatively shown to the critic’s satisfaction that
another reading should be adopted at various places in the text. 73 In other
words, the burden of proof is placed on any emendation of the copy-text to
reflect another reading. Depending on the nature of the emendations to the

71
Karl Elliger and Wilhelm Rudolph, Biblia Hebraica Stuttgartensia (4th rev. ed.;
Stuttgart: Deutsche Bibelgesellschaft, 1990), xii: “We have thought it best to reproduce the
text of the latest hand of L [scil. Codex Leningradensis] with close fidelity. We have ac-
cordingly refrained from «removing obvious scribal errors;» how we have dealt with
doubtful examples may be seen in the critical apparatus to Is 2,15, note a.” At the cited
place in the text, BHS has a footnote stating “sic” followed by the correct reading of many
other manuscripts in the apparatus. Thus, the user of this edition must consult the apparatus
to avoid the errors of the copy-text.
72
Indeed, the assumption that any surviving manuscript witness to the text is not the
original is non-controversial for any literary text from antiquity. For documentary texts,
such as those found at Oxyrhynchus, however, each letter, contract, receipt, and the like
are probably the original copies. In this context, the challenge becomes one of papyrology:
restoring the degraded physical remains of the manuscript witness to a readable state.
73
E.g. Tanselle, Textual Criticism, 303: “what underlies this conception of copy-text is
the idea of presumptive authority, a text to be relied on when one finds no basis for prefer-
ring one variant over another.”
20 Chapter 1. Theories of New Testament Textual Criticism

copy-text, the resulting text may become closer to or further from the original
Ω. Despite its simplicity, the copy-text method is not a mechanical procedure.
The practitioner must employ critical judgment in deciding how to emend the
copy-text and also in choosing which witness to adopt as the copy-text.
A practitioner of the copy-text method could simply decide to make no
emendations of the copy-text at all. In fact, this decision is entirely appro-
priate for producing a diplomatic edition of a manuscript, whose sole purpose
is to present the text of that manuscript. 74 Although diplomatic editions are
prized by scholars interested in the state of a living text in a particular witness
at a particular point in history, the particularity of such an edition makes the
text less accessible or useful to readers. In terms of accessibility, the witness
chosen as the copy-text may employ orthography or punctuation unfamiliar to
the readers and even contain obvious scribal errors that reduce the intelli-
gibility of the text. In terms of usefulness, the text of a diplomatic edition
may not be relevant to the point in the history of the text that interests the
reader. For example, the historical critic is interested in how the text was read
in the historical situation at the time of its composition (i.e., original Ω), but
the diplomatic edition does not purport to convey such a text, only the form
of the work at a particular point in its history (e.g., witness e₁).
If the practitioner of the copy-text method is not content with merely
producing a diplomatic edition of the text, the critic must examine the copy-
text for places to emend. The emendation of the copy-text can be assisted by
the readings of other manuscripts (emendatio ope codicum) or by the critic’s
own ingenuity (emendatio ope ingenii). 75 When two manuscripts have dif-
ferent readings at the same place in the text, it is all but assured that at least
one of them differs from the original. 76 Even without another manuscript as a
guide, sometimes the wording of the copy-text can be so unintelligible that
the critic becomes convinced that an error in the transmission of the text must
have occurred. When no other witness suggests a viable reading to account
for the error, the critic has to resort to conjecture in order to divine the
reading for the critical edition. Thus, the use of emendation assumes that it is
possible to detect readings in the copy-text that did not occur in the original
form of the text. These readings are called “errors.”

74
In fact, diplomatic editions of the witnesses to the text of Galatians, along with fac-
similes and photographs of manuscripts, have been vitally important in this study of the
text of Galatians in lieu of access to the manuscripts themselves. These are discussed in
greater detail in Chapter 2.
75
Timpanaro, Genesis, 46.
76
E.g., HORT, 3: “Where there is variation, there must be error in at least all variants
but one; and the primary work of textual criticism is merely to discriminate the erroneous
variants from the true.” Only the slim possibility of authorial variants complicates the
issue.
 1.4 Methods of Textual Criticism 21

According to the copy-text method, the only external evidence with any
authority is the copy-text itself, so it is necessary to rely on internal evidence
when deciding to emend the copy-text. Due to the presumption of the authori-
ty conferred upon the copy-text, moreover, the reading of the copy-text must
appear to be clearly erroneous in terms of the internal evidence when com-
pared with another reading in order to be emended. Unfortunately, internal
evidence is not always clear and decisive, and the textual critic may have to
struggle in balancing the intrinsic and transcriptional probabilities. The pre-
sumptive authority of the copy-text avoids much of this struggle by removing
the discretion of the textual critic in deciding such difficult cases and forcing
reliance on the wording of the copy-text.
Since the copy-text is emended only when its wording appears to be
clearly secondary according to internal evidence, the result of the copy-text
method depends on the witness that the practitioner chooses for the copy-text.
Different choices for the copy-text result in different critical texts. In FIG. 2,
extant witness e₁ was chosen as the copy-text at a certain textual distance
from the original Ω, and emending e₁ may result in a critical text somewhat
closer textually to the original Ω than witness e₁ itself. On the other hand, if
another witness, say c₁, had been chosen as the copy-text, its distance from
the original Ω may be much smaller than that of e₁, and the choice of the
witness c₁ may result in an emended text that is closer to the original Ω than
what could be attained by starting with witness e₁ as the copy-text.
The sensitivity of the copy-text method to the choice of the copy-text
means that a textual critic must use judgment in choosing the copy-text, but it
is unclear at the outset which witness that ought to be. Because emendations
of the copy-text are permitted only when it appears to be clearly erroneous, it
stands to reason that the critic ought to choose the witness closest to the
original, but the copy-text method provides no basis for determining which
witness this is. Paleographers may be able to determine the oldest manuscript
based on its handwriting, but the manuscript that is closest in time to the
original is not necessarily the closest in text. Another approach would be to
choose the “best text,” the witness with the highest number of best readings
according to internal evidence, but a single witness does not necessarily con-
tain all or even most of these best readings. 77 Consequently, investing a single
witness with the authority of a copy-text means that many apparently earlier
readings in other witnesses would not be able to surpass the presumption of

77
So A. E. Housman, M. Manilii Astronomicon, Liber Primus (London: Grant Richards,
1903), xxiii: “Chance and the common course of nature will not bring it to pass that the
readings of a MS are right wherever they are possible and impossible wherever they are
wrong.” In characteristic fashion, Housman denounces practitioners of the best-text meth-
od as follows (xxxii): “This method answers the purpose for which it was devised: it saves
lazy editors from working and stupid editors from thinking.”
22 Chapter 1. Theories of New Testament Textual Criticism

correctness conferred upon the copy-text. In other words, the copy-text meth-
od’s focus on a single manuscript for its locus of textual authority means that
best readings outside the copy-text cannot be used. This makes the copy-text
method unsuitable for a textual tradition such as that of the New Testament
where there are hundreds or thousands of witnesses to each book. 78
The copy-text method is also unsuitable for the emphasis on the history of
the text. As seen in FIG. 2, the copy-text method hypothesizes a textual
history only for the witness that embodies the copy-text, and this hypothetical
textual history is trivially true: any witness is the eventual descendant of an
original. The copy-text method does not consider the history of any other
witness in its historical model, nor is the copy-text method capable of relating
one witness historically to another. For these reasons, the copy-text method
has limited use within New Testament textual criticism.

1.4.2 Majority-Text Method

As its name suggests, the majority-text method determines the text by se-
lecting the most attested variant among the extant witnesses. This method is
fairly rare within textual criticism, but it has actually been used for the New
Testament. In particular, two editions of the New Testament have been
produced by proponents of this method: one by Zane C. Hodges and Arthur
L. Farstad, 79 and another by Maurice A. Robinson and William G. Pierpont.80
Although these editors describe their editorial theory in different terms, their
editorial practice is almost the same. They did not actually collate all the
surviving witnesses and tally up the support for each textual variant; rather,
they adopted the “Koine” text based on a large number of Byzantine
manuscripts as collated and constructed by a team of scholars under Hermann

78
Though there are hundreds if not thousands of copies of New Testament books, many
important non-Biblical texts have survived in a single manuscript copy. For example, the
Epistle to Diognetus was only preserved in a late medieval manuscript, Codex
Argentoratensis Graecus ix, until it was destroyed in 1870 during the Franco-Prussian War
(Michael W. Holmes, The Apostolic Fathers: Greek Texts and English Translations [3d
ed.; Grand Rapids, Mich.: Baker, 2007] 689–690.) When a work is attested by a single
witness, these more sophisticated methods for editing the work essentially reduce to the
copy-text method.
79
Zane C. Hodges and Arthur L. Farstad, The Greek New Testament According to the
Majority Text (2d ed.; Nashville: Nelson, 1985).
80
Maurice A. Robinson and William G. Pierpont, The New Testament in the Original
Greek: Byzantine Textform (Southborough, Mass.: Chilton, 2005). There is an earlier edi-
tion of this text, Maurice A. Robinson and William G. Pierpont, The New Testament in the
Original Greek According to the Byzantine / Majority Textform (Atlanta: Original Word,
1991), but it contains no accidentals (e.g., punctuation, accents, breathings, subscripts, up-
per case) other than verse numbering; as a result, it is virtually unreadable. Citations to the
Robinson and Pierpont edition are hereinafter to the 2005 edition.
 1.4 Methods of Textual Criticism 23

Freiherr von Soden for his early twentieth-century edition of the New
Testament. 81 Thus, both editions are highly reliant on von Soden’s research
into the Byzantine Koine text. Many additional manuscripts have been
discovered or collated in the century since von Soden’s work, so it is possible
that it no longer represents the actual majority reading. 82

a b c c d d d e e e e e e e e e
1 1 1 2 1 2 3 1 2 3 4 5 6 7 8 9

FIG. 3: A Majority-Text Model

The historical model presupposed by the majority-text method is depicted in

FIG. 3. All surviving witnesses, a₁ through e₉, are viewed as descendants
ultimately of an original Ω, through a chain of one or more ancestors (not
shown in FIG. 3). Since they all contribute equally to the reconstruction of Ω,
the historical model must necessarily assume that they are independent of
each other, or else the mathematics of the majority-text method cannot work
properly. Indeed, Robinson acknowledges that “The genealogical indepen-
dence of most of the existing MSS points back to the earliest times.” 83 This
assumption is highly problematic, however. There are thousands of copies of
the Greek New Testament. If each copy can claim an independent line of
descent from the original, then the original must have been copied thousands
of times. Such a historical possibility did not come into existence until the
age of printing, where an exemplar could be reproduced thousands of time
mechanically. Even for smaller numbers of copies, this assumption is dubious
economically. With an imperial grant from Constantine, Eusebius had trouble

81
Hermann Freiherr von Soden, Die Schriften des neuen Testaments in ihrer ältesten
erreichbaren Textgestalt hergestellt auf Grund ihrer Textgeschichte (Göttingen: Vanden-
hoeck und Ruprecht, 1911). On the use of von Soden’s apparatus to establish their texts,
see Hodges and Farstad, Greek New Testament, xxi-xxiii; and Robinson and Pierpont, New
Testament, x.
82
Another complication to using von Soden’s edition is that his apparatus is error-
prone, as recognized by Hodges and Farstad, Greek New Testament, xxii-xxiii: “Particular-
ly problematical to the editors of this edition was the extent to which his examination of K
materials appeared to lack consistency.” Robinson and Pierpont, New Testament, x, seem
more sanguine, confining the confirmed errors to “relatively few instances.”
83
Robinson and Pierpont, New Testament, 569, see also 560. The reason for “most” ra-
ther than “all” appears to be that he considers known families of manuscripts, e.g., family 1
and family 13, to be reducible to a single witness.
24 Chapter 1. Theories of New Testament Textual Criticism

producing a requested batch of fifty Bibles, 84 so producing more copies with

fewer resources in the first century would have been even more cost-
prohibitive.
In accordance with majority-text historical model, the majority reading of
the surviving witnesses dictates the result. In this way, the majority-text
method overcomes the idiosyncrasies of a single copy-text by allowing every
manuscript to contribute to what is, in effect, a virtual copy-text. Unlike the
copy-text method, however, the majority-text is not examined for emenda-
tion. 85 As a result, internal evidence plays no theoretical role in the majority-
text method unless the manuscripts are deeply split, and even then a majority-
critic may avoid the use of internal evidence by choosing a leading subgroup
of the majority text to establish the text. In fact, both sets of editors follow the
Byzantine subgroup Kx when the Byzantine text is divided with no clear
majority. 86 It is only when there is no clear majority or plurality reading that
internal evidence becomes decisive. 87 Thus, even though internal evidence is
designed to distinguish between authorial readings and scribal readings,
internal evidence plays only a marginal role in the construction of a majority-
text.
According to Hodges and Farstad, the crucial premise behind a majority-
text edition is that:
Any reading overwhelmingly attested by the manuscript tradition is more likely to be
original than its rival(s). This observation arises from the very nature of manuscript trans-
mission. In any tradition where there are not major disruptions in the transmissional histo-
ry, the individual reading which has the earliest beginning is the one most likely to survive
in a majority of documents. And the earliest reading of all is the original one. 88

Robinson likewise appeals to a principle of normality, as articulated by

Fredson Bowers, to hold “that a text reflected in the overwhelmingly majority
of MSS is more likely to have a chronological origin preceding that of any text
which might be found in a small minority.” 89 However reasonable this prem-

84
See Theodore Skeat, “The Codex Sinaticus, the Codex Vaticanus, and Constantine,”
JTS 50 (1999): 583–625.
85
Robinson and Pierpont, New Testament, 554: “The quality of the preserved evidence
for the text of the NT precludes conjectural emendation.”
86
Hodges and Farstad, Greek New Testament, xxii; Robinson and Pierpont, New Testa-
ment, x.
87
Hodges and Farstad, Greek New Testament, xxii: “Where Kx itself is sharply divided
within an Mpt reading [i.e., a reading that enjoys substantial but not majority support], the
rival variations were weighed both in terms of their distribution within the majority
tradition as a whole and with regard to intrinsic and transcriptional probabilities.”
88
Hodges and Farstad, Greek New Testament, xi-xii.
89
Robinson and Pierpont, New Testament, 540. Unfortunately, Robinson’s reliance on
Bowers’s is misplaced for Bowers was discussing the mechanics of printing books. Here is
 1.4 Methods of Textual Criticism 25

ise may appear at first sight, it is important to point out that Hodges and Far-
stad’s explanation contains a key proviso: “where there are not major disrup-
tions in the transmissional history.” 90 The reason for this proviso is that with
major disruptions in the transmissional history, some branches of the textual
tradition ceased to be copied. This scenario is illustrated in FIG. 1, where the
branches β, γ, and δ have simply stopped producing descendants after only a
few copies, while branch ε was much more prolific, producing more survi-
ving copies than all the other branches combined. Because the readings of the
descendants of ε overwhelm numerically all other surviving readings, what a
majority-text reconstruction produces is a form of the text that resembles ε
more than another point in the textual tradition. 91 Hodges and Farstad were
aware of this potential problem with the majority-text method and so they
proposed that: “Final decisions about readings ought to be made on the basis
of a reconstruction of their history in the manuscript tradition.” 92 What they
mean by this proposal is that the textual critic ought to determine a genealogy
of the manuscripts and produce a stemma showing their history. This is in
fact a different text-critical method, known as stemmatics, and the topic of
the following section.
One complication of the majority-text method is that there are portions of
the New Testament where the premise simply breaks down. Instead of there
being a dominant majority text, the textual witnesses are fractured into multi-
ple distinct groups of manuscripts. The two chief examples are the Pericope
Adulterae (John 7:53–8:11 in most manuscripts) and the book of Revelation.
For the story of the adulteress, the textual situation is fragmented. The text is

the text of the passage that Robinson quoted as it is found in Fredson Bowers, Biblio-
graphy and Textual Criticism (Oxford: Clarendon, 1964), 75 (italics added):
If one collates 20 copies of a book and finds in sheet B that only 1 copy shows the uncorrected
state of the type as against 19 showing the corrected, ‘normality’ makes it highly probable that
the correction in B was made an earlier point in time in the machining of this forme than the cor-
rection of a forme in C that shows 19 with uncorrected type and only 1 with corrected. Despite
the fact that statistics can lie, the mathematical odds are excellent that this sampling of 20
copies can be extrapolated in accord with normality.
The portions italicized in the quotation are those which Robinson edited out with ellipses
and brackets in his block quotation. The technical terminology, e.g. “machining of this
forme,” clearly shows, however, that Bowers comments are directed to printed book
production, one of his major research interests.
90
Robinson is aware of this proviso, but he has argued extensively against the idea that
there were in fact major disruptions in the transmissional history, by minimizing the effects
of the Diocletian persecution and the Islamic conquest (Robinson and Pierpont, New
Testament, 572–576). CF. ALAND -ALAND 294–295, who discuss major disturbances in the
New Testament text.
91
The classic argument against a majority reading based on genealogical principles is
HORT 57–59.
92
Hodges and Farstad, Greek New Testament, xii.
26 Chapter 1. Theories of New Testament Textual Criticism

split among seven different groups, three of which – known as μ5, μ6, and μ7 –
are the largest and of nearly equal size. 93 As for Revelation, the textual
tradition is split into nearly two equal size groups, known as Andreas and
Q. 94 On the one hand, Robinson and Pierpont’s approach to John 7:53–8:11 is
to print the text of two of the largest subgroups, μ5 and μ6, separately. 95 Their
approach to Revelation is to favor the Q group unless a significant number of
manuscripts have defected from it to the Andreas group. 96 On the other hand,
Hodges and Farstad abandon the majority-text method entirely and offer their
own stemma for establishing their text of these problematic portions of the
New Testament. 97
As can be seen in the preceding description, the majority-text method
makes extensive use of external evidence but almost no use of internal evi-
dence, even in comparison to the copy-text method. Since internal evidence
distinguishes between authorial readings and scribal readings, the diminished
role for internal evidence means that scribal readings are not purged from the
majority-text. After all, in the majority-text method, individual readings are
not decided by determining which one looks more authorial and which one
looks more scribal, but by counting the number of witnesses in favor of them.
As a result, a majority-text is not suitable to scholars interested in an author-
ial text. The majority-text method also does not establish a history of the text,
because it imposes an a priori genealogy that assumes that surviving manu-
script can trace independent lineages from the origin. This genealogy, how-
ever, is implausible. Accordingly, the majority-text method is also unsuitable
for those scholars interested in the history of the text.

1.4.3 Stemmatics
Stemmatics is a method that explicitly reconstructs the history of the text in
order to reconstruct the original. The history of the text reconstructed by this
method is usually depicted graphically in the form of a family tree of manu-
scripts or a stemma codicum. The invention of stemmatics is typically at-
tributed to the early nineteenth-century textual critic Karl Lachmann, but
there were important precursors who anticipated significant features of his
text-critical theory. 98 Perhaps the clearest and most succinct explication of the

93
Robinson and Pierpont, New Testament, xi; Hodges and Farstad, Greek New Testa-
ment, xxiv.
94
Robinson and Pierpont, New Testament, xii.
95
Robinson and Pierpont, New Testament, xi. The group μ 7 is disregarded, however,
because it is associated with a “redactional” group in the Byzantine text.
96
Robinson and Pierpont, New Testament, xiii.
97
Hodges and Farstad, Greek New Testament, xxv and xxxiii.
98
Timpanaro, Genesis, 115–118, points out that nearly every aspect of Lachmann’s
method has antecedents in earlier textual critics.
 1.4 Methods of Textual Criticism 27

method is that of Paul Maas in the middle of the twentieth-century. 99

Although stemmatics has been popular in the textual criticism of classical
Latin and Greek texts, it has hardly been applied to the New Testament. A
notable exception is that of Hodges and Farstad who applied stemmatics
based on manuscript groups for John 7:53–8:11 and Revelation. 100
According to Maas’s description of the method, stemmatics has three
phases. 101 The first phase, known as recensio, organizes the witnesses to the
text into a stemma that represents the historical development of the text from
the most recent common ancestor, called the “archetype.” This phase recon-
structs the history of the text. The second phase, known as examinatio, exam-
ines the readings of the branches of the family tree closest to the archetype
and selects which reading, if any, is “original” (i.e., authorial). This phase
reconstructs the text of the archetype. The third phase, divinatio, tries to
repair the text of the archetype when no reading appears to be authorial. This
phase emends the archetype to be closer to the original (or “autograph”).
In contrast with the copy-text and majority-text methods, the stemmatic
method does not assume a history of the text a priori. Rather, it provides a
technique for reconstructing the history of the text based on the patterns of
readings found in the extant manuscripts. In particular, it identifies which
manuscripts, based on their readings, are more closely related to each other
than they are to any other manuscript. For example, in FIG. 1, manuscript a₁
is more closely related to manuscript e₁ than to manuscript c₁, because a₁ and
e₁ share a common ancestor α that is not an ancestor of b₁. When the scribe
of α copies the manuscript from its exemplar ω₂ (the common ancestor of a₁,
c₁, and e₁), a number of textual differences, called “errors,” are introduced.
When α itself is copied by later scribes, these errors of α’s scribe are propa-
gated to its descendants a₁ and ε, and ultimately to e₁. Thus, the presence of
these errors in a₁ and e₁, which are missing in b₁, is what indicates that a₁
and e₁ descend from a common ancestor (α) that is not an ancestor of c₁. In
other words, the common errors of a₁ and e₁ are evidence of their related-
ness. 102 Accordingly, stemmatics uses a common-error criterion to diagnose
groups of related manuscripts.

99
Maas, Textual Criticism. Maas’s treatise in the Teubner series has been superseded in
some respects by Martin L. West, Textual Criticism and Editorial Technique Applicable to
Greek and Latin Texts (Stuttgart: Teubner, 1973).
100
Hodges and Farstad, Greek New Testament, xxiii-xli. Even here, the stemmatic
method was not applied at the level of individual manuscripts but to groups.
101
Maas, Textual Criticism, 1. Some textual critics refer to another phase, selectio, sit-
ting between recensio and examinatio, but Maas considers selectio an aspect of examinatio
(18).
102
Common errors are also called “indicative errors” or Leitfehler (Maas, Textual Criti-
cism, 42).
28 Chapter 1. Theories of New Testament Textual Criticism

Crucial to the success of the method, stemmatics groups manuscripts based

on their agreement in error, but not on their overall similarity – how often
two manuscripts agree on their readings even in original readings. 103 For
instance, an overall similarity measure cannot determine if a₁ and e₁ are more
closely related to each other than c₁, because it can be misled by the different
number of changes between α and its parent and children. If, on the one hand,
the number of changes between ω₂ and α is much larger than the number of
changes between α and e₁, then a₁ would appear more overall similar to e₁
than to c₁. If, on the other hand, the number of changes between ω₂ and α is
much smaller than the number of changes between α and e₁, then a₁ would
appear more overall similar to c₁ than to e₁. Even though the overall similar-
ity measure gives different answers, the genealogy is the same. These ex-
amples show that the overall similarity measure is not useful for diagnosing
the genealogy of the manuscripts because it is too sensitive to how the
changes are distributed for a given genealogy. Under certain distributions of
changes, the overall similarity measure can be misled into giving results that
go against the actual genealogy of the manuscripts. As a result, a quantitative
approach using an overall similarity measure cannot actually identify groups
having a common ancestry, though it may approximate such groups under
appropriate circumstances.
There is an obvious chicken-and-egg problem with the common error
principle, however. On the one hand, reconstructing the text (that is, deciding
which reading is true and which is error) requires reconstructing the history
of the text. On the other hand, reconstructing the history of text requires
grouping manuscripts by their common errors, which requires reconstructing
the text. So which comes first, the reconstructed text or the reconstructed
history of the text? An elegant escape out of this circularity has been
articulated by Dom J. Froger in 1968. 104 This French textual critic recognized
that there are two aspects to a stemma: one, its shape and linkage pattern
(l’enchaînement), and, two, its orientation (l’orientation). Froger realized that
it is only necessary to identify which textual variant is an error for the second
aspect, the orientation of the stemma. For the first aspect, the stemma’s
shape, it is only necessary to work with textual variants, without knowing
which one is erroneous. 105 Thus, his solution is to split the reconstruction of

103
Overall similarity is the approach of the “quantitative method” by Ernest C. Colwell
and Ernest W. Tune, “Method in Establishing Quantitative Relationships Between Text-
Types of New Testament Manuscripts” in Ernest C. Colwell, Studies in Methodology in
Textual Criticism of the New Testament (NTTS 9; Grand Rapids, Mich.: Eerdmans, 1969),
56–62.
104
J. Froger, La critique des textes et son automatisation (INS 7 ; Paris: Dunod, 1968).
105
Froger, La critique, 77, attributes this insight to Dom Quentin, but the discussion in
Henri Quentin, Essais de critique textuelle (ecdotique) (Paris: Picard, 1926), 44–50, though
on point, is less lucid.
 1.4 Methods of Textual Criticism 29

the text into two steps. The first step is to determine the shape and linkage
pattern of the stemma based on the distribution of textual variants, not errors.
By looking at variants, which do not presume originality, the determination of
the directionality of the variants is deferred to the later step, allowing the
textual critic to produce an unoriented stemma. The second step is the
orientation of this unoriented stemma, and this step is accomplished by using
internal evidence to determine which portion of the unoriented stemma is
closest to the archetype. 106
There is another benefit to Froger’s two-step solution for reconstructing
the history of the text. Froger had the prescience to realize that his first step
can be automated by a computer. 107 The possibility of computerization is
feasible because no subjective judgments as to the priority or direction of
textual changes need to be made in the first step; those judgments are
deferred to the second step. The automation of the first step is crucial because
constructing a stemma according to a common-error principle becomes
exponentially more difficult with each additional manuscript in the stemma.
A dozen or so manuscripts can be examined by hand over a couple of weeks,
but textual traditions on the order of the New Testaments (hundreds and
thousands of manuscripts) cannot be reconstructed by hand within the
lifetime of the most ardent textual critic. With modern computer programs,
based on related work by computational biologists, 108 this step can be
performed much more rapidly. For example, Peter M. W. Robinson took
several months to derive a stemma from almost fifty manuscripts of an Old
Norse saga, but it took a biologist only a few minutes to come up with a
substantially similar result using an off-the-shelf computer program. 109 As a
result, one of the greatest challenges of applying stemmatics to a tradition as
large as the New Testament can be handled by computer technology.
A serious challenge to the common-error principle, however, is contamina-
tion, which occurs when readings from one branch of the tradition are incor-

106
Froger, La critique, 77–78. See also Humphrey Palmer, The Logic of Gospel Criti-
cism: An Account of the Methods and Arguments Used by Textual, Documentary, Source,
and Form Critics of the New Testament (London: Macmillan, 1968), 78; and West, Textual
criticism, 70–72.
107
Froger, La critique, 217–266.
108
E.g., the cladistics program PAUP. On the operation of this program, see generally
Matthew Spencer, Klaus Wachtel, and Christopher J. Howe, “The Greek Vorlage of the
Syra Harclensis: A Comparative Study on Method in Exploring Textual Genealogy,” TC:
A Journal of Biblical Textual Criticism [http://purl.org/TC] 7 (2002).
109
Robert J. O’Hara and Peter M. W. Robinson. “Computer-Assisted Methods of
Stemmatic Analysis,” The Canterbury Tales Project: Occasional Papers Vol. 1 [http://
www.canterburytalesproject.org/pubs/op1-cladistics.pdf] (1993): 53–74 at 62.
30 Chapter 1. Theories of New Testament Textual Criticism

porated into an independent branch. 110 Contamination confuses the common-

error principle. For example, in FIG. 1, manuscript d₁ is a copy of exemplar
δ, corrected against exemplar γ. The text of d₁ is thus a mixture of the two
exemplars γ and δ. As a result of this mixture, d₁ contains errors in common
with d₂ that are not found in c₁ and c₂, suggesting a common ancestor δ, but
d₁ also contains common errors with c₁ and c₂ that are not found in d₂,
suggesting a different common ancestor γ for d₁. To make matters even more
difficult, the process of mixture may have removed some of γ’s errors and
some of δ’s errors from d₁. Thus d₁ would have only an incomplete coverage
of the errors of its exemplars. 111
What makes contamination particularly pernicious is that it can distort the
shape of the stemma, causing some groups of manuscripts to be regarded
more highly than they should and other groups less highly than they should.
FIG. 4 depicts a stemma for the textual tradition illustrated in FIG. 1 that has
been distorted by contamination.
One distortion of the stemma is due to the contamination of exemplar δ by
an independent line of transmission from one of the sent copies Ω₃ as shown
in FIG. 1. If enough of the contaminated readings appear to be authorial
based on internal evidence, this would have the effect of promoting exemplar
δ to a more prominent place in the stemma and demoting δ’s sisters α, β, and
γ. As shown in FIG. 4, this distortion is manifested as a binary split from the
archetype ω into two branches δ and ψ. The branch ψ represents the common
ancestor of exemplars α, β, and γ in FIG. 1. Branch ψ excludes d₁ and d₂
from its descendants, but the most recent common ancestor of a₁, b₁, and c₁
in FIG. 1 is also the ancestor of d₁. What this distortion does is that it places
the readings of δ on equal basis as ψ, the combination of α, β, and γ, whereas
in the actual textual transmission scenario, these three exemplars should
outweigh δ. In this example, the distortion of the stemma may not be too
serious. After all, δ has access to authorial readings via Ω₃ that α, β, and γ do
not, and promoting δ in the stemma at the expense of the others allows those
readings to be considered by the textual critic in the next phase of the
stemmatic method. Similarly, the access by the scribe of b₁ to another
independent stream from an authorial original (via a lost exemplar of a letter
collection ω₁ in FIG. 1) tends to promote b₁ at the expense of the descendants
of α and γ, subordinating their readings in the guise of branch χ.

110
See generally Michael W. Holmes, “Working with an Open-Textual Tradition: Chal-
lenges in Theory and Practice” in Klaus Wachtel and Michael W. Holmes, eds., The Textu-
al History of the Greek New Testament: Changing Views in Contemporary Research (TCS
8; Atlanta: Society of Biblical Literature, 2011), 65–78.
111
Maas, Textual Criticism, 7–8; West, Textual Criticism, 37–47.
 1.4 Methods of Textual Criticism 31

χ δ

ψ b
1 d 1 d 2

α γ

ε a 1 c 1 c 2

d 3

e e e e e e e e e
1 2 3 4 5 6 7 8 9

FIG. 4: A Stemmatic Model Distorted by Contamination

Another distortion of the stemma due to contamination could occur in not

being able to recognize that d₁ in FIG. 1 is a mixture of γ and δ. A textual
critic, misled by this contamination, might get the direction of mixture wrong
and conclude erroneously that γ is a mixture of χ and d₁ instead, as shown in
FIG. 4. 112 This distortion has the effect of giving less weight to γ in the re-
construction of text than it should otherwise have. For example, assume that
witness b₁ is fragmentary (like many New Testament papyri) and does not
contribute any evidence to a variation unit. In FIG. 1, the joint testimony of
c₁, d₁, and d₂ ought to outweigh the testimony of α (determined by the joint
testimony of a₁ and ε). However, in FIG. 4, the agreement of c₁ with d₁ and
d₂ would be due to contamination from d₁, so c₁ (as well as c₂) has no weight
to contribute to the decision.
Contamination is such a difficult problem that textual critics abandon the
attempt to reconstruct the history of the text when the scale of contamination
becomes too great. 113 In the face of such “open traditions,” classical textual

112
See Holmes, “Working,” 71–73, and West, Textual Criticism, 34–36, for another
illustration of this issue.
113
Maas, Textual Criticism, 49, was fairly pessimistic: “No specific has yet been
discovered against contamination.”
32 Chapter 1. Theories of New Testament Textual Criticism

critics instead identify a set of witnesses that appear to have the best readings
and use internal evidence on a reading-by-reading basis to select the readings
for their critical text. 114 By adopting this approach, the classical textual critic
in effect abandons the stemma and begins to resemble the New Testament
“eclectics” as described in the following sections.
Once a stemma is obtained, the textual critic proceeds to the next phrase,
examinatio, to reconstruct the state of the archetype ω. The general procedure
is to go from the bottom of the stemma to the top. The lower areas of the
stemma can be reconstructed by just the patterns of readings in the neighbor-
ing areas of the stemma, 115 but at the highest level, at the archetype ω, it is
necessary to use internal evidence. 116 For example, in FIG. 4, the text of ex-
emplar ε can be reconstructed by the texts of its many descendants e₁ through
e₉, including d₃. In this case, a simple majority vote suffices to determine the
text of ε. As for the text of exemplar α, it can be determined as follows. If α’s
children, a₁ and ε, agree on a reading, then α has that reading. If, however, a₁
and ε do not agree, then α’s closest relatives on the stemma, e.g., b₁, c₁, and
c₂, are used to break the tie. If all relatives of these disagree, then the text of
α is doubtful at that place. 117 By a similar procedure, the texts of exemplars γ,
δ, and ψ can determined. As for exemplar χ, the contamination of γ by wit-
ness d₁ (according to this misled stemma) means that the testimony of γ is not
available to reconstruct the text of χ whenever γ agrees with d₁.
After reconstructing the text of the two main hyparchetypes ψ and δ, the
textual critic is ready to reconstruct the text of the archetype ω. If ψ and δ
agree on a reading, then that reading becomes the text of ω. If ψ and δ do not
agree, then the textual critic must decide which reading if any is original. If
one of the readings based on internal evidence appears to be a scribal error
derivable from the other reading, then the other reading is adopted as the
reading of the archetype. If both readings appear to be errors but neither read-
ing can be plausibly derived from the other, then the reading of the archetype
may have to be restored by conjecture. In rare instances, if both readings
appear authorial and neither reading can be plausibly derived from the other,
then the textual critic must consider the possibility of authorial variants. 118
Throughout this process of examination, the textual critic is attentive to
anomalies in the text that are not only unexplainable by the author’s style or

114
This is the recommended procedure of West, Textual Criticism, 43. For example,
Maurice Bénevot, The Tradition of Manuscripts: A Study in the Transmission of St. Cypri-
an’s Treatises (Oxford: University Press, 1961; repr. Westport, Conn.: Greenwood Press,
1979), dealt with contamination in the text of Cyprian by sorting the manuscripts into three
opposing groups and using internal evidence to adjudicate among their differences.
115
Maas, Textual Criticism, 4–6.
116
Maas, Textual Criticism, 12–14.
117
Maas, Textual Criticism, 4.
118
Maas, Textual Criticism, 6, 17–18.
 1.4 Methods of Textual Criticism 33

larger context but also appear to be the result of scribal corruption. This
examination is performed even where there are no variant readings in the
tradition. In detecting anomalies, the textual critic looks at the same intrinsic
probabilities that are used in internal evidence. In determining if an anomaly
looks like a scribal error, the textual critic looks at the same transcriptional
probabilities that are used in internal evidence. Thus, the detection of such
anomalies is an application of internal evidence. If the anomaly is sufficiently
harsh and appears to be the result of scribal corruption, the textual critic can
propose to remove the anomaly by conjectural emendation. 119 This is the
third phase of stemmatics, divinatio, which seeks to correct errors in the
archetype ω, thereby bringing the reconstructed text into closer alignment
with the assumed autograph Ω. The practice of divining satisfying conjectures
is very difficult, more art than science, and much of the effort in classical
stemmatics has gone into making and evaluating conjectures. 120 If the criteria
of internal evidence have been used well during this process, then the textual
critic can be reasonably confident that the emended text of the archetype
approximates an authorial text.
In sum, stemmatics is a text-critical method that is eminently suited for
both kinds of customers of textual criticism. The use of internal evidence to
reconstruct and to correct the text of the archetype means that stemmatics can
deliver an authorial text that is suitable for readers of the text interested in
what the author wrote. As for those customers interested in the history of the
text, stemmatics not only hypothesizes an explicit history of the text in the
form of a stemma, but it is also able to reconstruct the text at many other
stages in the transmission. Thus, the stemma enables the textual historian to
trace the development of the text through the ages and ascertain how both
individual variants and also patterns of variation can be informative about the
readers, users, and copiers of the text. The major disadvantage to the
stemmatics, however, is handling the presence of contamination, which
threatens to make the reconstruction of the history of the text impossible.
Only if this problem can be overcome or at least controlled, then stemmatics
may well become a suitable text-critical method for the textual traditions of
the New Testament. Indeed, New Testament textual critics generally
recognize the benefits of a stemmatic approach, but they contend that the
massive size of the New Testament textual tradition and the pervasiveness of
contamination preclude any rigorous use of genealogy. 121

119
Maas, Textual Criticism, 11.
120
In fact, the bulk of Maas’s tiny book is devoted to examples of conjectural emenda-
tion (Textual Criticism, 24–41), as is the textbook of Robert Reneham, Greek Textual Crit-
icism: A Reader (Cambridge, Mass.: Harvard University Press, 1969).
121
E.g., Ernest C. Colwell, “Genealogical Method: Its Achievements and its Limita-
tions,” in Colwell, Studies, 63–83.
34 Chapter 1. Theories of New Testament Textual Criticism

1.4.4 Reasoned and Thoroughgoing Eclectic Methods

In its broadest sense, eclecticism is a method for constructing a critical text
by examining the witnesses at each individual variation unit and selecting the
reading for the critical text based on the textual evidence at that individual
variation unit. This variant-by-variant approach of eclecticism allows for the
considerable use of internal evidence, and it stands in sharp contrast to
documentary approaches, 122 where the reconstructed text is determined by the
contents of a specifically chosen copy-text or a majoritarian construct of the
manuscripts. Stemmatics, by contrast, can be considered to be partially
eclectic in that its examinatio phase employs internal evidence on a variant-
by-variant basis at a limited number of variation units at the base of the
tradition’s genealogical stemma as determined by its recensio phase.
Although practitioners of eclecticism for New Testament textual criticism
are in general agreement that internal evidence must play an important role in
evaluation of various readings, they disagree as to the relative role of external
evidence. On one the hand, proponents of “reasoned” eclecticism argue for a
balanced use of external evidence. As Michael W. Holmes articulated their
fundamental principle, “the variant most likely to be original is the one that
best accounts for the origin of all competing variants in terms of both external
and internal evidence.” 123 On the other hand, the “thoroughgoing” eclectics
argue that external evidence should hardly play any role in their craft,
championing instead the value of internal evidence. 124
Reasoned eclecticism has been the dominant method of editing the text of
the New Testament throughout the twentieth century. 125 It has been the
approach of the committee of textual critics who edited the United Bible
Societies’ Greek New Testament in 1966. 126 A thoroughly revised edition of
the UBS text according to eclectic principles was published in 1975, which
then became the basis of the twenty-sixth Nestle-Aland edition in 1979. 127 As
a result, the standard text of the New Testament being used now by scholars
is an edition based on an eclectic method.

122
Holmes, “Reasoned Eclecticism in N.T.,” 338; Michael W. Holmes, “Reasoned
Eclecticism and the Text of Romans,” in Sven K. Soderlund and N. T. Wright, eds., Ro-
mans and the People of God: Essays in Honor of Gordon D. Fee on the Occasion of His
65th Birthday (Grand Rapids, Mich.: Eerdmans, 1999), 187–202, at 187–188 n.2; Holmes,
“Case,” 79.
123
Holmes, “Case,” 79.
124
J. Keith Elliott, “Thoroughgoing Eclecticism in New Testament Textual Criticism”
in Ehrman and Holmes, Text, 321–335.
125
Holmes, “Reasoned Eclecticism,” 336.
126
Bruce M. Metzger, The Text of the New Testament, its Transmission, Corruption,
and Restoration (3d ed.; New York: Oxford University Press, 1992), 145–146.
127
Metzger, Text, 280.
 1.4 Methods of Textual Criticism 35

Reasoned eclecticism begins with a study of the individual readings in the

tradition, collated from a number of sources, especially Greek witnesses such
as the papyri, uncials, and minuscules, but also from versional and patristic
sources. 128 These variants are studied in terms of their internal evidence –
which ones appear to be authorial based on intrinsic probabilities and which
appear to be scribal on transcriptional probabilities. Kurt Aland recommends
that the critic should draw up a local stemma for the variation unit showing
the development of its variant readings. 129 The study of the variants in terms
of the internal evidence, however, does not end the inquiry for the reasoned
eclectic, because, in the words of Hort’s oft-quoted dictum, “knowledge of
documents should precede final judgement upon readings.” 130 As a result,
eclectics study the witnesses carefully to determine how well they transmit
their texts and how they relate to other manuscripts in various groups. 131 For
example, Bruce M. Metzger lists various criteria for evaluating external
evidence, including the age of the witness, the distribution of the witnesses
geographically, the genealogical connections among the witnesses, but the
most important of which is the age of the text-type to which the witness
belongs. 132 Thus in Metzger’s eclectic approach, text-types play an important
role.
The concept of a text-type has been difficult to define. Ernest Cadman
Colwell defined it as “the largest identifiable group of related New Testament
manuscripts,” and distinguished it from a “family” as the “smallest . . . group
whose inter-relationships are so close that its stemma or family tree can be
established with exactness.” 133 Thus, built into the very notion of a text-type
is the need to work with groupings of related manuscripts beyond the
capability of contemporary stemmatics to discover. In the editing of the UBS
text, three text-types have been identified in Paul: an Alexandrian text-type,
divided into a proto-Alexandrian group (P46  B) and a later Alexandrian
group (A C Ψ 33 1739); 134 a Western text-type (D F G and Old Latin
versions); 135 and a Byzantine text-type (L and most minuscules). 136

128
Earnest Cadman Colwell, “Hort Redivivus: A Plea and a Program” in Colwell,
Studies, 148–171 at 160. Zuntz, Text, 12, refers to this step with the stemmatic term
examinatio. See also, Holmes, “Case,” 80, n.11.
129
Aland-Aland, 281, Rule 8, also known as the local-genealogical principle.
130
Hort 31 (original in small caps). Quoted, e.g., by Colwell, “Hort,” 160; Gordon D.
Fee, “Rigorous or Reasoned Eclecticism–Which?” in Epp and Fee, Studies, 124–140;
Holmes, “Case,” 77.
131
Colwell, “Hort,” 161–164.
132
Metzger, Text, 209.
133
Ernest C. Colwell, “Method in Establishing the Nature of Text-Types of New
Testament Manuscripts,” in Colwell, Studies, 45–55 at 45–46.
134
Metzger, Text, 216.
135
Metzger, Text, 214.
36 Chapter 1. Theories of New Testament Textual Criticism

Now, the identification of text-types does not end the inquiry for the rea-
soned eclectic. The text-type must also be fit into a historical framework of
the textual tradition. 137 In the case of these three text-types for the text of the
Pauline epistles, the Byzantine text-type is considered newer and derivative
of the Alexandrian and Western text-types. 138 Among the two earlier text-
types in Paul, the Alexandrian is generally considered more reliable than the
Western, except when the Alexandrian Codex Vaticanus (B) joins with the
Western witnesses. 139 This judgment about the value of the Western text,
especially in combination with B, goes back to Westcott and Hort. 140 Günther
Zuntz, however, in his study of the Pauline epistles is more sanguine about
the Western text; he holds that “Western witnesses joined by P46 or B or 1739
are more often right than wrong.” 141 Zuntz based his conclusions mainly on
his examination of 1 Corinthians and Hebrews, so it is a desideratum to see if
his conclusions hold up in other parts of Paul, such as Galatians. 142
FIG. 5 depicts a historical model of the transmission of the text illustrated
in FIG. 1 using text-types, in general alignment with Hort’s theory of the text.
Like the stemmatic model, this historical model includes the distinction
between an autograph Ω and an archetype ω in order to permit the use of
conjectural emendation. 143 Similar manuscripts are grouped together into
text-types. In FIG. 5, the text-types are denoted by cloud figures, and there
are three of them: one contains a₁ and b₁; another contains c₁, c₂, d₁, and d₂;
and a third text-type contains d₃ and e₁ through e₉. No genealogical
relationships are depicted within the text-type clouds because those
relationships are generally not known. The diagram indicates that the third
text-type on the bottom is derivative of the other two, and so its witnesses
carry less weight in the reconstruction of the text than the other two. 144 On
the other hand, the top two text-types have more weight than the third,

136
Metzger, Text, 213.
137
Colwell, “Hort,” 164–169. Also Holmes, “Reasoned Eclecticism,” 350: “Eclecticism
does not work in a vacuum; it functions only in conjunction with a view of the history of
the transmission of the text.”
138
Metzger, Text, 212.
139
Metzger, Text, 218.
140
HORT, 240–241.
141
Günther Zuntz, The Text of the Epistles: A Disquisition upon the Corpus Paulinum.
The Schweich Lectures of the British Academy 1946 (London: Oxford University Press,
1953), 158.
142
See also Michael W. Holmes, “The Text of Epistles Sixty Years After: An Assess-
ment of Günther Zuntz’s Contribution to Text-Critical Methodology and History” in
Transmission and Reception: New Testament Text-Critical and Exegetical Studies (J. W.
Childers and D. C. Parker, eds.; TS3 4; Piscataway, N.J.: Gorgias, 2006), 89–113.
143
The term “archetype” is not used in eclecticism, however, and the use of conjectural
emendation is almost entirely absent, Acts 16:12 being the notable exception.
144
This third text-type is analogous to Hort’s view of the Byzantine text.
 1.4 Methods of Textual Criticism 37

derivative text-type because they represent the earliest forms of the text. 145
These top two text-types are not necessary weighted equally, however,
because the critic’s knowledge of their documents may indicate that the
members of one text-type tend to be more prone to scribal interference in the
text than the other. 146
Ω
ω

c c
a b
1 2
1
1
d 1 d2

e e e e d
1 2
3 4 3

e e e e e
5
6
7
8
9

FIG. 5: A Hortian Eclectic Model using Text-Types

Compared with the previously discussed text-critical methods (copy-text,

majority-text, and even stemmatics), text-type eclecticism relies more heavily
on internal evidence in reconstructing the text. The lack of genealogical
information within the text-types means that no particular witness should
necessarily dominate in furnishing readings for the critical text. Rather, any
member of an early text-type – especially one prized in terms of external
evidence – can potentially provide a good reading for the critical text, even
against the testimony of its fellow text-type members. Internal evidence is
used, in conjunction with this external evidence, to identify such good
readings in various individual members of the earliest text-types.
This greater use of internal evidence in text-type eclecticism helps to cope
with contamination in a way that stemmatics cannot handle well. As seen in
FIG. 1, both witnesses b₁ and d₁ have independent access to authorial
readings apart from the copy in the letter collection ω₂. With stemmatics, b₁’s
location on the stemma of FIG. 4 means that the readings in b₁ are always
outweighed by the agreement of δ and ψ. In the light of the mixture of
readings from ω₁, however, some of these b₁ readings may actually be
authorial, but the stemma does not permit these authorial readings in b₁ via
ω₁ to be selected. With the text-type eclecticism of FIG. 5, on the other hand,

145
These may be analogized to the Western and Alexandrian text-types.
146
The so-called “Western” text-type in Paul has this reputation.
38 Chapter 1. Theories of New Testament Textual Criticism

b₁ is present in one of the two earliest text-types, and, as a result, makes its
readings available to be selected based on internal evidence.
This use of text-types characterizes the approaches of Metzger and
Colwell, but other eclectics organize the external evidence somewhat
differently. Aland and his colleagues, for example, classify the textual
witnesses into five categories based on their readings in a number of test
passages. 147 Zuntz, on the other hand, likens the transmission of the text to a
stream with two major branches and places the different manuscripts at
various points on the stream. 148 All of these textual critics attempt to integrate
both internal and external evidence in their work, but there is another set of
eclectics who eschew external evidence altogether.
Indeed, some “radical” or “thoroughgoing” eclectics have argued that
external evidence, including text-types, ought to be abandoned altogether or a
least subordinated in favor of using internal evidence at every turn. As one of
its practitioners, J. Keith Elliott, put it:
Thoroughgoing eclecticism is the method that allows internal considerations for a read-
ing’s originality to be given priority over documentary considerations. The thoroughgoing
eclectic critic feels able to select freely from among the available fund of variants and
choose the one that best fits her or his internal criteria. This critic is skeptical about the
high claims made for the reliability of some MSS or about arguments favoring a particular
group of MSS. For him or her no MS or group contains the monopoly of readings. 149

This means that any manuscript can potentially supply a reading that a
thoroughgoing eclectic may consider original if that reading passes the tests
of internal criticism in the view of the critic.
Accordingly, depicted in FIG. 6 is the historical model presupposed by
thoroughgoing eclecticism. Rather than grouping the witnesses into text-
types, there is a single fund of witnesses, denoted by the large cloud, from
which any variant can be selected for the critical text. The stated assumption
of this model is the tenacity of the tradition, that any original reading must
have survived in at least one of the extant witnesses, rendering conjectural
emendation as unnecessary. 150 This ban on conjectural emendation puts the
thoroughgoing method in a curious place, however. On the one hand, internal
evidence is apparently so powerful that external evidence is hardly relevant,
as long as even one manuscript attests to the variant. On the other hand,

147
ALAND -ALAND 317–337. The five categories are three groups (I-III) of Alexandrian
manuscripts with various degrees of mixture with the Byzantine text, a so-called Western
category (IV), and the Byzantine text (V).
148
Zuntz, Text, 278.
149
Elliott, “Thoroughgoing Eclecticism,” 321.
150
Elliott, “Thoroughgoing Eclecticism,” 322: “On a positive note the thoroughgoing
method of textual criticism assumes that the original reading has been preserved some-
where among the extant MSS and that conjectural emendations are unnecessary.”
 1.4 Methods of Textual Criticism 39

internal evidence is so poor that it cannot supply a reading that happens to be

unattested no matter how well it explains the state of the evidence. It seems
better, thus, to resolve this paradox over the capacity of internal evidence by
giving up on the assumption of tenacity.

d d
b
2

a c c d
1
1
1 1 2 3

e e
e
7

e
9

e e
3

e
5

e e
1 2
8
4 6

FIG. 6: A Thoroughgoing Eclectic Model with No Distinct Text-Types

Compared with stemmatics, eclecticism uses internal evidence to a greater

degree and external evidence to a lesser degree. The greater use of internal
evidence helps to overcome the problem of contamination, because the
eclectic textual critic is empowered to examine for authorial readings (based
on internal evidence) a greater number of witnesses than what stemmatics
would allow according to its stemma. This greater reliance on internal
evidence, however, comes with a cost. Internal evidence is only decisive to
the extent it can distinguish between the textual behavior of an author and
that of a scribe. Given two variant readings, when one looks authorial while
the other reading looks scribal, internal evidence can be very helpful.
Nevertheless, not every variant unit is so clear. Sometimes the semantic
difference between two variants (e.g. for word order variations) is so subtle
that it may seem fruitless to figure out how one is more scribal or authorial
than the other. In this case, internal evidence may hardly seem decisive, and
the less decisive the internal evidence, the more important external evidence
must become.
Although the quality of the New Testament text produced by eclecticism
has had a well-deserved good reputation, it has come at the expense of having
a clear history of the text. Among users of text-types, there is a rough sense
of which text-types are older and which are newer, but there is no detailed
sense of the history of the text within the text-types. This means that scholars
interested in the original context of a New Testament work have been well-
served by eclecticism, but those who wish to understand the history of the
text and its readers have been disadvantaged.

1.4.5 Coherence-Based Genealogical Method

The Coherence-Based Genealogical Method (CBGM) is the newest of the
text-critical methods reviewed in this survey. It has been developed by Gerd
40 Chapter 1. Theories of New Testament Textual Criticism

Mink of the Institute for New Testament Textual Research in Muenster and
described in a number of publications beginning in 1993. 151 The CBGM has
been applied to the text of the New Testament, specifically to the Catholic
Epistles of the new Editio Critica Maior. 152
The purpose of the CBGM is to deal with the circular reasoning that Mink
identifies as inherent in contemporary eclecticism: witnesses are good be-
cause of their good readings, but readings are good because of their good wit-
nesses. Mink doubts that this circular reasoning can entirely be avoided but
he proposes that it can be controlled through his method. In particular, he
sees the CBGM as a way to refine the quality of a critical text in accordance
with his genealogical principles, especially that of genealogical coherence. 153
It is important to understand what the CBGM is and what it is not. As
Klaus Wachtel explains it:
The CBGM does not provide a means of automating the reconstruction of the initial text,
nor the ‘royal way’ to it. Furthermore, the CBGM often does not make textual decisions
more secure. It may cast new doubt on cases that seemed to be settled. It is a way to
analyse the structure of a manuscript tradition and to integrate methodically our growing
knowledge about it. 154

Thus, the CBGM is not an automated method for generating a stemma of the
New Testament or otherwise to derive the textual history of the New Testa-
ment. Rather, it enables more precision for weighing the external evidence in
support of various readings in the course of deciding the critical text. Accord-
ingly, its practitioners see this method as an advance over the use of text-type
and categories that dominated the twentieth century.
The CBGM works as follows. First, it begins with an estimate of the initial
text that is intended to be relatively close to the form of the text from which
the textual tradition of a New Testament book has originated. 155 Then, at a
large number variation units, a local genealogy of the variant readings is
drawn up, showing the evolution of the variants without regard to the wit-
nesses that transmit those readings. Similar in practice to Aland’s local ge-
nealogical principle, the practitioner of the CBGM uses internal evidence to

151
Gerd Mink, “Eine umfassende Genealogie der neutestamentliche Überlieferung,”
NTS 39 (1993): 481–499; and idem, “Problems,” 13–85.
152
E.g., Barbara Aland, Kurt Aland, Gerd Mink, and Klaus Wachtel, Novum
Testamentum Graece: Editio Critica Maior, vol. 4.1, James (Stuttgart: Deutsche Bibelge-
sellschaft, 1997); vol. 4.2, The Letters of Peter (2000); vol. 4.3, The First Letter of John
(2003); vol. 4.4, The Second and Third Letter of John, The Letter of Jude (2005).
153
Mink, “Problems,” 25.
154
Klaus Wachtel, “Towards a Redefinition of External Criteria: The Role of Coherence
in Assessing the Origin of Variants” in Textual Variation: Theological and Social Tenden-
cies (H. A. G. Houghton and D. C. Parker, eds.; TS3 6; Piscataway, N.J.: Gorgias, 2008),
109–127 at 127.
155
The first iteration of this “initial text” seems to have been the Nestle-Aland 27th ed.
 1.4 Methods of Textual Criticism 41

decide what the likely pattern of textual flow from reading to reading appears
to be. 156 At this point, it is not necessary to perform this analysis for all
variation units, just for the ones whose local genealogy is fairly clear and
uncontroversial. A local genealogy can be represented by a local stemma, and
it indicates which variants are to be considered prior or posterior to other
variants. 157
Next, the local genealogies are considered for each witness to identify
which other witness is a potential ancestor. The best potential ancestors are
those which have a high degree of agreement in readings and, of the variant
units where they do not agree, the potential ancestor has a high proportion of
the variants that are prior to the readings found in the witness under
consideration. Thus, a “substemma” can be established for each witness that
indicates which other witnesses are the best potential ancestors for it. 158
These various substemmata can be aggregated to produce a global stemma,
which shows the textual flow of the tradition. Now, Mink is careful to warn
the reader that his global stemmata are not the same as the classical stemmata
in the Lachmannian or Maasian sense. In particular, Mink eschews the use of
hyparchetypes (hypothetical ancestors) and allows for most witnesses to have
multiple ancestors (to account for contamination). Another major difference
is that, in classical stemmatics the archetype usually has two, sometimes three
immediate descendants, while in the CBGM there can be a dozen or more
witnesses whose best potential ancestor is the initial text. 159
To illustrate how a CBGM global stemma differs from a classical stemma
codicum, a mocked-up global stemma for the transmission scenario is shown
in FIG. 7.

156
This use of local genealogies makes the CBGM incompatible with a purely
documentary approach, such as the majority-text method.
157
Mink, “Problems,” 35–36.
158
Mink, “Problems,” 38–46.
159
Mink, “Problems,” 46–49.
42 Chapter 1. Theories of New Testament Textual Criticism

a 1 b 1 c 1 d 2

d 1

c 2

e 1

d 3

e
e
2

e
3

e
4

e
5

e
6

e
7

e
8
9

FIG. 7: A Global Stemma Showing Textual Flows According to the CBGM

In contrast to the classical stemma of FIG. 4, there are no hypothetical

exemplars in the diagram except for the initial text labeled Α (German
Ausgangstext). Instead of these hypothetical exemplars, the global stemma
shows multiple textual pathways for most witnesses. 160 For example,
manuscript d₁ is shown has having three potential ancestors, A, d₂, and c₁.
The link from A to d₁ indicates that d₁ has access to the readings of ω₂ in
FIG. 1 that are not found in d₂. Yet d₁ and d₂ also share some readings in
common against the rest of the tradition, and this fact is indicated by the
arrow from d₂ to d₁. As for the contamination from γ (the ancestor of c₁ and
c₂ in FIG. 1), this mixture is shown by an arrow from c₁ to d₁. Another
difference between a CBGM text-flow diagram and a classical stemma lies in
the number of children for ancestral documents in the stemma. Ancestors in
classical stemmata have a tendency to have only two descendants, 161 while
some ancestors in a CBGM text-flow diagrams can have numerous children.
For example, while the classical stemma often has a binary split at the highest
level (e.g. ψ and δ in FIG. 4), the initial text Α in FIG. 7 has a direct link to
seven witnesses.

160
The use of hypothetical ancestors without contamination has been shown to be
mathematically equivalent to the use of contamination without hypothetical ancestors. See
John Alroy, “Continuous Track Analysis: A New Phylogenetic and Biogeographic Meth-
od,” Systematic Biology 44 (1995): 152–172.
161
An issue made famous by Joseph Bédier, “La tradition manuscrite du Lai de
l’ombre: Réflexions sur l’art d’éditer les anciens textes,” Romania 54 (1928): 161–196.
 1.5 Proposals and Road Map 43

Once the global stemma has been produced, the variation units of the
initial text can be revisited, either to revise the local stemma, or even to
modify the reading of the initial based on the quality of the “coherence” of
the respective variant readings. Basically, a reading exhibits good coherence
if it is found only among witnesses that are closely related genealogically,
and poor coherence if it is found among witnesses that are distantly related.
In other words, a reading with poor coherence is likely to have originated
multiple times in the transmission of the text, and for that reason may be
considered transcriptionally a later reading. On the other hand, a reading
should belong to the initial text if it has good coherence among those
witnesses whose best potential ancestor is the initial text. If two variants
appear to be good candidates for the original text based on coherence, then
the reading is considered on internal grounds (e.g. the lectio difficilior) to be
original is chosen. After the initial text has been revised based on
genealogical coherence, this refinement process can be repeated over and
over until the repeatedly revised initial text stabilizes. 162
To its credit, the CGBM appears to be more rigorous about evaluating
external evidence than the use of text-types and categories currently
employed in reasoned eclecticism. It also applies the notion of internal
evidence when the external evidence appears balanced in terms of coherence.
The benefit of this rigor is that the “initial text” produced by the CBGM may
reflect the authorial text better, and this benefits those who are interested in
the text as it was composed by an author within its historical context.
Historians of the text, however, are less helped by the CBGM. The CBGM
does not reconstruct the history of the text, but instead a diagram of textual
flows. These flows are highly abstract and they do not correspond to the
textual state of any lost exemplar other than the initial text.

1.5 Proposals and Road Map

There are two goals now in New Testament textual criticism: reconstructing a
serviceable authorial text for those interested in the historical context of its
composition and reconstructing the history of the text for those interested in
the change of the text over time and space. Based on the foregoing survey of
five text-critical methods, two of the methods, the copy-text method and the
majority-text method, are plainly unsuitable to the task. Neither method at-
tempts to discover the history of the text; rather, they impose an a priori his-
tory upon the textual evidence. Moreover, both of these unsuitable methods
generally avoid the use of internal evidence to construct their texts; thus, their

162
Mink, “Problems,” 38–46.
44 Chapter 1. Theories of New Testament Textual Criticism

constructed text contains a combination of authorial and scribal readings that

do not interest most users of the text.
Stemmatics may appear ideal for the twin goals of contemporary New
Testament textual criticism. It explicitly generates a history of the text in the
form of a stemma according to the theoretically sound common-error criteri-
on, and it brings to bear internal evidence to identify and reconstruct the
archetype. Moreover, it also uses internal evidence in correcting the
archetype by conjectural emending. This use of internal evidence means that
the emended archetype of stemmatics is a serviceable authorial text. The
main disadvantage of stemmatics is that textual critics have not found a way
to handle contamination. By contrast, the eclectic and coherence-based
genealogical methods are designed to work in contaminated textual traditions
by emphasizing internal evidence at the expense of external evidence. To the
extent that their use of internal evidence permits, these methods produce an
authorial text suitable for historians of the time period in which the text
originated. 163 Yet this product comes with a price: these methods have largely
foregone a detailed reconstruction of the history of the text.
Accordingly, this study of the text of Galatians and its history proposes to
take a hybrid stemmatic-eclectic approach to the textual evidence. Rather
than finding a general solution to the contamination problem, this study pro-
poses both a stemmatic solution to contamination and an eclectic solution to
contamination. Specifically, on the stemmatic side, only that level of contam-
ination that can distort the stemma will be addressed in the construction of
the stemma (see section 2.3), and the remaining amount of contamination will
be handled in accordance with eclecticism, by a more extensive use of
internal evidence than stemmatics normally permits. With this hybrid
approach, this study can reap the benefits of stemmatics for the history of
Galatians and reap the benefits of eclecticism for a serviceable authorial text.
Chapter 2 sets forth the materials and methods of this study. For the
materials, Chapter 2 surveys ninety-two textual witnesses used in this study
of Galatians, detailing their characteristics and how they were collated. For
the methods, Chapter 2 describes a computer-implemented method for
producing an unoriented stemma of the witnesses to the text of Galatians. The
method itself is based on “cladistic” techniques developed by computational
biologists for finding family trees of organisms and adopted by some textual
critics. Accordingly, this chapter explains the theoretical basis of cladistics,
its past use by textual critics, how the method operates, and how it handles
contamination. Chapter 2 concludes with a presentation and description of an
unoriented stemma of Galatians.

163
Of course, the distinction between the original text and the initial or critical text re-
cognizes that the results of their methods are necessarily provisional and limited by the
value of the available textual evidence and methods.
 1.5 Proposals and Road Map 45

Chapter 3 is directed toward orienting the stemma. To this end, thirty-six

variants on four early branches of the stemma are examined in terms of their
internal evidence. In particular, this chapter sets forth how internal evidence
is evaluated in accordance with the basic methodological approach of rea-
soned eclecticism. The result of this evaluation is to determine which variants
are authorial based on internal evidence and then to orient the stemma by
chosen as the base of the stemma the portion with the most authorial
readings.
Chapter 4 examines the resulting text near the base of the stemma and
resolves textual variants where the earliest witnesses are divided among
themselves. Again internal evidence is strongly relied upon to establish the
text. After resolving these issues, the chapter concludes with a critical text of
Galatians.
Chapter 5 traces the history of the text. Due to the size and complexity of
the textual transmission, this chapter concentrates on the Western Branch,
and the rise of the Byzantine text within the Eastern branch. For each branch,
its new readings are examined to ascertain the nature of the transmission
process for the branch. Specifically, the nature of the textual transmission
will be examined to determine if it is relatively strict or periphrastic, and
whether there is a discernible theological or other tendency to its changes.
Chapter 6 concludes the dissertation with a brief outline of the history of
the text of Galatians, and contrasts the history of the text depicted in the
stemma to that of other scholars with special attention to Hort and Zuntz.
 Chapter 2

Materials and Methods

2.1 Materials and Manuscripts

For most of the past century and a half, a researcher on the text of Galatians
and its history had limited options for grasping the state of the manuscript
evidence. The physical manuscripts were scattered across the globe, and most
of them were not published at all. Because of the limited accessibility of the
manuscripts themselves, researchers had to consult microfilms of them in
specialized centers, 1 or they had to rely on published collations and editions.
Tischendorf’s great eighth critical edition of the nineteenth century had been
helpful to many critics in the age of Westcott and Hort, garnering praise for
its range of witnesses, breadth of variation units, and reputation for
accuracy. 2 Its usefulness, however, has diminished with every new important
manuscript discovery. 3 For the Pauline corpus, these include the early third-
century Bodmer papyrus P46 and codex 1739. The Nestle-Aland handbooks,
especially its 26th and 27th editions, present a wealth of up-to-date and
accurate textual evidence, but these editions are not exhaustive and remain
less comprehensive than Tischendorf.
Over the past decade especially, there have been on-going projects to im-
prove the evidentiary basis for the practice of textual criticism. The Inter-
national Greek New Testament Project (IGNTP) is producing a volume on the
Gospel of John. 4 For the Catholic epistles, the Muenster Institute has pro-

1
Such as the Institute for New Testament Textual Research (Institut für Neutestament-
liche Textforschung, INTF) in Muenster.
2
Konstantin Tischendorf, Novum Testamentum Graece: ad antiquissimos testes denuo
recensuit apparatum criticum omni studio perfectum apposuit commentationem isagogi-
cam (2 vols.; 8th ed., critica maior; Leipzig: Giesecke & Devrient, 1869).
3
Another inconvenience of the edition has been a later change in the enumeration of the
(mainly miniscule) witnesses by Caspar René Gregory. As a result, scholars have to work
with conversion tables between Tischendorf’s system (largely inherited from Wettstein)
and Gregory’s. See generally, Kurt Aland and Barbara Aland, The Text of the New Testa-
ment: An Introduction to the Critical Editions and to the Theory and Practice of Modern
Textual Criticism (2d ed.; Erroll F. Rhodes, trans.; Grand Rapids, Mich.: Eerdmans, 1989),
38–41.
4
Unfortunately the earlier products for Matthew and Mark under the editorship of S. C.
E. Legg are poorly regarded (Aland and Aland, Text, 23–24).
 2.1 Materials and Manuscripts 47

duced a Critica Maior with over a hundred fully-collated witnesses, 5 and they
are beginning to work on the Acts of the Apostles. For Paul, scores of
manuscripts of the Hauptbriefe have been collated, transcribed, and published
by Rueben J. Swanson. 6 These publications promise to make more compre-
hensive and accurate compilations of the textual evidence available to scholar
than ever before.
Accordingly, the evidentiary basis for this project begins with Swanson’s
collations of manuscripts of Galatians, because of its comprehensive cover-
age of the earliest substantial witnesses to the text, including P46, , A, B, C,
D, F, and G. 7 Swanson’s manuscripts were further supplemented with some
additional Greek witnesses (for example, 1780 because of its presence at
Duke University and 1881 because it is consistently cited in the 27th Nestle-
Aland edition) and versional witnesses. Nevertheless, to the extent reasonably
feasible, Swanson’s collations of the most important witnesses have been
double-checked against available images of the manuscripts in question, with
transcriptions from the Muenster Institute available on its New Testament
Virtual Manuscript Room website (NT.VMR), and from other publications.
As a result, this study has double-checked Swanson’s edition of Galatian for
eleven manuscripts, some of which are the most important for understanding
its text.
The following witnesses have been collated for this study:
P46. The text of P46 used for this study was originally obtained from
Swanson’s transcription and then double-checked against high-resolution im-
ages of Gal 1:1–6:8 available at the University of Michigan website APIS and
the transcriptions of P46 at the Muenster Institute website NT.VMR. Based
on the images and transcriptions, I propose a number of corrections (or dif-
fering interpretation of P46) to Swanson’s text. These are documented in
Appendix A. 8

5
E.g., Barbara Aland, Kurt Aland, Gerd Mink, and Klaus Wachtel, Novum Testa-
mentum Graece: Editio Critica Maior, vol. 4, pt. 1.1, James (Stuttgart: Deutsche Bibelge-
sellschaft, 1997); idem, Novum Testamentum Graece: Editio Critica Maior, vol. 4.2, The
Letters of Peter (Stuttgart: Deutsche Bibelgesellschaft, 2000); idem, Novum Testamentum
Graece: Editio Critica Maior, vol. 4.3, The First Letter of John (Stuttgart: Deutsche Bibel-
gesellschaft, 2003); idem, Novum Testamentum Graece: Editio Critica Maior, vol. 4.4: The
Second and Third Letter of John, The Letter of Jude (Stuttgart: Deutsche Bibelgesellschaft,
2005).
6
Swanson has since retired and died in 2009. His materials have been entrusted with
Kent D. Clarke at Trinity Western University, who is currently at work on the volume for
Revelation; see http://twu.ca/academics/faculty/profiles/clarke-kent.html. It is unclear
when the remaining volumes for the Pauline corpus will come out.
7
A witness is considered “substantial” for this study if it can supply a reading for at
least 250 variation units. As a result, fragmentary manuscripts, including P51, H, I, 0176,
and 0261 were not used for this study.
8
The following are the most substantive corrections to Swanson in P46:
48 Chapter 2. Materials and Methods

/01. 9 Codex Sinaiticus. The text was obtained from Swanson and it was
also double-checked against the images and transcriptions found online at
http://codexsinaiticus.org/. As Swanson did not discriminate the first and
second correctors, 10 the transcriptions at the Nestle-Aland 28th edition
website were consulted to determine which correction belongs to which
corrector. Based on the images and transcriptions, I differ from Swanson at a
number of places, particularly in the assignment of the correctors. These
differences are documented in Appendix A. 11
A/02. Codex Alexandrinus. The text obtained from Swanson was double-
checked against the black-and-white photographs from Thompson and the
NA28 transcription. 12 Based on my inspection, there is one substantive
change to be made to Swanson’s collation. 13 At Gal 3:29, Swanson 45 reads

At Gal 1:14, Swanson 9 has περισσότερον ὡς, but the image shows περισσοτερο̇ν̇ ως,
with deletion dots above the ending ο̇ν̇ and with ως written inline afterward. The ὡς
accordingly represents a corrected ending, not a new word; so also ROYSE 228; Philip W.
Comfort and David P. Barrett, The Text of the Earliest New Testament Greek Manuscripts:
New and Complete Transcriptions with Photographs (2d ed.; Wheaton, Ill.: Tyndale
House, 2001), 313 n.d. Therefore, this variant is coded as the P46* reading περισσότερον
and P46¹ reading περισσότερως.
At Gal 1:17, Swanson 10 confusingly writes οὐδὲ ···λθον, which might suggest partial
support for ἀνῆλθον or ἀπῆλθον. It is clear from the image, however, that P46 only has
enough space for η to read ἦλθον rather than either variant.
At Gal 5:17, Swanson 73 reverses the indications for the original hand of P46 and its
corrector. According to the image, however, P46* actually reads τὸ δὲ πνεῦμα and P46C
reads τὸ δὲ τὸ πνεῦμα; cf. Comfort and Barrett, Text, 313 n.a; ROYSE 220, 223, 239, 241,
and 244 assigns the supralinearly written το to a third hand of P46.
9
This study prefers the traditional Wettstein-Tischendorf capital letters for some of the
uncial (majuscule) manuscripts due to their compactness, though the recent trend is to use
only the numeric sigla for them; see D. C. Parker, An Introduction to the New Testament
Manuscripts and their Texts (Cambridge: Cambridge University Press, 2008), 36–37.
10
The various correctors have been collapsed to a single document state labeled  C.
11
The following are the most substantive differences:
At Gal 2:13, Swanson 23 reads συναπήχθη (“got carried away”) for , which is for both
* and  c, but, according to the image and transcription  c (actually ²) corrected the verb
to read συνυπήχθη (“withdrew together”).
At Gal 5:3, Swanson 65 reads ἀνθρώπων for both * and  c; according to the image
and transcription, however, both read ἀνθρώπῳ.
At Gal 5:18, Swanson 73 seems to have misread the letters τιε added above εστε as a
command to replace the word ἐστε with τις, as if  C has οὐκ τις. Actually, as apparent in
the image and the transcription, the correction is to insert the letters τιε after the ε in εστε
so that  C (i.e., ²) should read οὐκέτι ἐστὲ.
12
E. M. Thompson, Facsimile of Codex Alexandrinus: New Testament and Clementine
Epistles (vol. 4; London: Trustees of the British Museum, 1879).
13
At Gal 4:21, the NA28 transcription states that the θέλοντες originally read λέγοντες
(a singular reading that is perhaps as a scribal assimilation to the preceding λέγετε), but
this is not apparent to me–or presumably to Swanson 57–from the photograph.
 2.1 Materials and Manuscripts 49

ὑμεῖς for A, but the first letter is not evident from the photograph and
assigned to part of a lacuna in the NA28 transcription. Therefore, A should
not be recorded as support for the reading ὑμεῖς. 14
B/03. Codex Vaticanus. The text was obtained from Swanson, and double-
checked against the color photographs of Carlo M. Martini with reference to
the NA28 transcription. 15 Based on my inspection of these color photographs
of B, Swanson generally fails to record the mostly orthographic letter
deletions by the medieval corrector throughout the text (these are documented
in Appendix A), 16 but otherwise has two substantive errors. 17
C/04. Codex Ephraemi Syri Rescriptus. The text was initially obtained
from Swanson and then double-checked against Tischendorf’s 1843 edition
and the corrections recommended by R. W. Lyon. 18 Tischendorf’s edition of
C is laid out so that the base text of C for Galatians is found on pp. 253–258
but his notes about corrections are located in an appendix on pp. 351–352.
Swanson seems to have largely ignored Tischendorf’s appendix with the cor-
rector notes, as well as Lyon’s own list of corrections. Thus, various changes,

14
This is the case even despite the fact that A reads ἐστέ later in the sentence, for 1241 s
reads both ἡμεῖς and ἐστέ (spelled as ἐσται).
15
Carlo M. Martini, Novum Testamentum e codice vaticano graeco 1209 (Codex B):
Tertia vice photoypice expressum ( Codices e vaticanis selecti 30; Vatican City: Biblio-
theca apostolica vaticana, 1968).
16
Some corrections listed in the NA28 transcript of B are not apparent to me in the col-
or photographs, which could reflect a limitation of inspecting the photographs rather than
the manuscript directly. These corrections include: Gal 1:11 εστι[ν], 2:8 ενεργησε[ν], 5:7
πειθεισθ[ε]<αι>, 5:19 εστι[ν] τα. Moreover, at Gal 1:11, where the first hand of B reads τὸ
εὐαγγέλιον τὸ εὐαγγέλιον τὸ εὐαγγέλιον (three times), Swanson’s layout implies that the
second and third occurrences were deleted by a later corrector, while the color photographs
indicate that the first and second occurrences were deleted.
17
At Gal 2:4, Swanson 16 reads καταδουλώσουσιν for B with no correction, but,
according to the color photographs and the NA28 transcript, the medieval corrector of B
changed the word to read καταδουλώσωσιν.
At Gal 2:16, Swanson 26 reads ὅτι ἐξ ἔργων νόμου δικαιωθήσεται πᾶσα σάρξ for B,
omitting οὐ (“not”) between νόμου and δικαιωθήσεται. According to the color photographs
and the NA28, however, Swanson apparently misread the ligature in νόμου at the end of
the line. The ligature consists of the letters μου, and Swanson seems to have taken that lig-
ature as merely the letter mu (μ), and so considered the following letters ου at the begin-
ning of the next to be the ending of νόμου instead of the separate word οὐ. At any rate, the
medieval corrector of B added a smooth breathing to οὐ–not noted in Swanson–indicating
that, at least in the mind of the corrector, those letters do not belong to νόμου.
18
Konstantin Tischendorf, Codex Ephraemi Syri Rescriptus sive fragmenta Novi
Testamenti e codice graeco parisiensi celeberrimo quinti ut videtur post Christum seculi
(Leipzig: Tauchnitz, 1843); R. W. Lyon, “A Re-Examination of Codex Ephraemi Re-
scriptus,” NTS 5 (1959): 260–272.
50 Chapter 2. Materials and Methods

detailed in Appendix A, were made to Swanson’s transcription of C for this

study. 19
D/06. Codex Claromontanus. The text was obtained from Swanson and
double-checked against Tischendorf’s edition of 1852. 20 There appears to be
one substantive error in Swanson’s transcription of D*. 21 Swanson is much
less accurate and precise with the correctors of D, however. Tischendorf
discriminates between various correctors, especially the seventh-century D**
(i.e. D¹) and the ninth-century D*** (i.e. D²). 22 Swanson, on the other hand,
uses the siglum Dc for most of the corrections to D and, on rare occasions,
refers to individual correctors. Appendix A lists my changes to Swanson’s
handling of D’s correctors, including those cases where either D¹ or D² differs
from Swanson’s Dc. Nevertheless, because of the confusion over which cor-
rection belongs to which correction, this study uses only D² as the corrected
form of D.
F/010. Codex Augiensis. The text for this analysis was obtained from
Swanson and double-checked against Scrivener’s 1859 edition. 23 It appears
that while Scrivener had access to the manuscript, Swanson looked at photo-
graphs, which explains why Swanson occasionally missed or misread some
corrections. These and other errors are detailed in Appendix A. 24
G/012. Codex Boernernianus. The text for this analysis was originally ob-
tained from Swanson and double-checked against the photographic facsimile

19
In particular: Gal 2:5 Cvid διαμένῃ (Tischendorf 351 appendix); 2:7 C εἰδότες; 3:10 C
ἐμμένει πᾶσιν (Lyon); 3:14 C γένηται; 3:16 C τῷ Ἀβραὰμ (Lyon); 3:19 C*vid ἀγγέλου, CC
ἀγγέλων; 3:21 C*vid δύναμος [sic] ζῳοποιῆσαι (Lyon contra Tischendorf 352 appendix
δυνάμενος ζῳοποιεῖν); 4:1 C* πάντων, CC πάντων ὤν; 4:14 C ἐδέξασθέ (Lyon); 4:26 CC
πάντον [sic]; 5:2 C ὠφειλήσει [sic]; 5:10 C* ἐγὼ δὲ πέποιθα, CC ἐγὼ πέποιθα; 5:10 C ἐὰν
(Lyon); 5:11 C* εἴ τι, CC ἔτι; 5:17 C* ἃ θέλητε, CC ἃ ἂν θέλητε; 5:23 C ἔστι (Lyon); 6:14
C* κἀγὼ κόσμῳ, CC κἀγὼ τῷ κόσμῳ; and 6:16 C* στοίχουσιν, CC στοιχήσουσιν.
20
Konstantin Tischendorf, Codex Claromontanus sive epistulae Pauli omnes graece et
latine ex codice parisiensi celeberrimo nomine Claromontani plerumque dicto sexti ut vi-
detur post Christum saeculi (Leipzig: Brockhaus, 1852).
21
At Gal 6:8, Swanson 81 reads εἰς τοῦ π̅ν̅ς̅, but according to Tischendorf 287 (line 7),
D* actually reads ἐκ τοῦ π̅ν̅ς̅.
22
There is also a sixteenth-century corrector, which Tischendorf cites as Dnov and NA27
sometimes cites as DC.
23
Frederick Henry Scrivener, An Exact Transcript of the Codex Augiensis, a Græco-
Latin Manuscript of S. Paul’s Epistles, Deposited in the Library of Trinity College, Cam-
bridge, to Which is Added a Full Collation of Fifty Manuscripts Containing Various Por-
tions of the Greek New Testament in the Libraries of Cambridge, Parham, Leicester,
Oxford, Lambeth, the British Museum, &c. (Cambridge: Deighton, Bell, 1859).
24
Two of Scrivener’s errors can also be identified. At Gal 2:5, Scrivener 141 appears to
be in error for printing ειζαμεν instead of ειξαμεν; similarly, at 4:27, Scrivener 150 prints
Ρηζον instead of Ρηξον.
 2.1 Materials and Manuscripts 51

by Reichardt. 25 The manuscript has since suffered extensive damage during

World War II, so the facsimile remains among the clearest sources for the
Greek and Latin text of this manuscript in Paul. Suggested errata for Swan-
son’s transcription of G are detailed in Appendix A.
33. “The Queen of the Cursives.” The text of 33 for this study has been
taken from Swanson 1999 and double-checked against photographs of the
manuscript available at the New Testament Virtual Manuscript Room
(NT.VMR), hosted at a Muenster Institute website. Almost all of the Gala-
tians portion of this manuscript is clearly legible on these photographs, but
some letters are not, especially at the beginnings or ends of lines, but Swan-
son hardly indicates difficulty reading them. 26 I have not listed such illegible
readings in Appendix A, due to their extent, but other, more substantive
corrections to Swanson’s transcription are noted. 27
1424. The text of this manuscript was obtained from Swanson 1999 and
checked against the exquisite color photographs published online at the
website for The Center for the Study of New Testament Manuscripts. While
Swanson’s transcription of the base text was generally accurate, he did err
when corrections were involved. All the proposed changes for 1424 are de-
tailed in Appendix A. 28
1735. The text of this manuscript for this study has been taken from Swan-
son and double-checked against images of the manuscript available at the
New Testament Virtual Manuscript Room (NT.VMR), hosted at a Muenster
Institute website. Suggested errata for Swanson’s transcription of 1735 are
detailed in Appendix A. 29

25
Alexander Reichardt, Der Codex Bœnernianus: Der Briefe des Aposteks Paulus (Msc.
Dresd. A 145h) (Leipzig: Hiersemann, 1909).
26
For example, on fol. 59v, line 50 (third up from bottom), Swanson records the letters
····εἰς, indicating that 33 read ἔρεις instead of ἔρις at Gal 5:20. Unfortunately, the
photograph only appears to have two round black blotches at the extant beginning of the
line where this word is found. Other editors of Galatians, in fact, do not decide what 33
read here; for example, 33’s reading is not listed at all in Tischendorf or in Nestle-Aland.
For purposes of this study, however, I am reluctantly supporting Swanson’s reading of
ἔρεις for 33vid here at Gal 5:20, because the round shape of the blotch better supports an
epsilon-iota ligature (εἰ) instead of a bare iota (ι).
27
Specifically, the substantive corrections include: Gal 1:4 ἐξελεῖται (not ἐξέληται);
3:11 νόμου (not νόμῳ); and 3:14 γένηται (not γενήσεται).
28
The substantive issues include: Gal 2:20, 1424 C reads κ(αὶ) π αραδόντος, 1424*vid
seems to have omitted καί; Gal 3:10, 1424* ὁ κρεμάμενος in erasure after πᾶς (cf. 3:13, but
Swanson did not identify erased text); Gal 3:18, both 1424* and 1424c omit ἡ after erasure
(of unidentifiable content); Gal 3:22, 1424 C ἐκ πίστεως ι̅υ̅ χ̅υ̅ (not ἐκ πίστεως χ̅ν̅) ; Gal 4:5,
1424*vid reads υἱοθεσίαν λάβωμεν (not υἱοθεσίαν ἀπολάβωμεν); and Gal 6:8, 1424* αυτου,
1424C ἑαυτοῦ.
29
Only orthographic errors have been noticed, one involving a corrected reading.
52 Chapter 2. Materials and Methods

1780. Durham, N. C.: Duke University, Greek 1. The text of this manu-
script was directly collated from the codex itself on April 21, 2011.
1874. The text of this manuscript for this study has been taken from Swan-
son and double-checked against images of the manuscript available at the
New Testament Virtual Manuscript Room (NT.VMR), hosted at a Muenster
Institute website. Suggested errata for Swanson’s transcription of 1874 are
detailed in Appendix A. 30
1881. The text of this manuscript was taken from a copy of the microfilm
photographs of the manuscript obtained from the Library of Congress.
2138. The text of this manuscript was taken from a collation published by
Barbara Aland and Andreas Juckel. 31
2423. Durham, N.C.: Duke University, Greek 3. The text of this manu-
script was taken from a collation by Kenneth W. Clark and checked against
the actual manuscript on April 22, 2011. 32 No errors were discovered in
Clark’s collation.
2892. The text for this manuscript was taken from its online photographs
hosted by the Center for the Study of New Testament Manuscripts. 33
Latin versions. The Latin versions used in this study include the following
witnesses, taken from the cited published sources: b (Beuron 89), 34 d (Beuron
75), 35 and f (Beuron 78). 36 The text of the Vulgate was obtained from the
edition of Wordsworth and White. 37
Syriac versions. The text of the Peshitta and the Harklensis Syriac ver-
sions was taken from the edition of Aland and Juckel. 38
Marcion. The reconstructed text of Marcion’s edition of Galatians was
taken from the study by Ulrich Schmid. 39 At places, Schmid is more cautious

30
The only substantive error that might affect the outcome of this study is that, at Gal
3:28, 1874 reads ἄρσεν ἢ θῆλυ, not ἄρσεν καὶ θῆλυ.
31
Barbara Aland and Andreas Juckel, Das Neue Testament in syrischer Überlieferung
(vol. 2.2; Berlin: de Gruyter, 1995), 507–514.
32
Kenneth W. Clark, Eight American Praxapostoloi (Chicago: University of Chicago
Press, 1941), 160–165.
33
Online at http://www.csntm.org/.
34
Hermann Josef Frede, Eine neuer Paulustext und Kommentar. II: Die Texte (Vetus
Latina 8; Freiburg: Herder, 1974), 217–233.
35
Tischendorf, Codex Claromontanus, 2:255–289, 568–571.
36
Scrivener, Codex Augiensis, 137–155, 277–278.
37
John Wordsworth and Henry Julian White, Nouum Testamentum Domini Nostri Iesu
Christi Latine secundum editionem Sancti Hieronymi (pt. 2; fasc. 4; Oxford: Clarendon,
1934), 367–405
38
Aland and Juckel, Syrischer Überlieferung, 181–247.
39
Ulrich Schmid, Marcion und sein Apostolos: Rekonstruktion und historische Einord-
nung der marcionitischen Paulusbriefausgabe (ANTT 25; Berlin: de Gruyter, 1995), 314–
319.
 2.1 Materials and Manuscripts 53

than the editors of the Nestle-Aland text, and so he sometimes declines to

consider Marcion a witness for a particular textual variant. 40
Chrysostom. This siglum refers to the text of John Chrysostom’s
commentary on Galatians, not to his quotations of the epistle in his other
works. The text for the study was taken from a tenth-century manuscript of
his commentary in Codex Coxianus. 41 At times, the text discussed in the
commentary differs from that presented in a lemma, so the text adopted for
this study is determined on a case-by-case basis but generally favoring that of
the commentary. One of the issues with working with manuscripts of patristic
commentaries rather than with a critical edition is a manuscript text may be
contaminated with Byzantine readings. In this case, it was felt that the tenth-
century date of the manuscript allowed it to be early enough to avoid much
contamination from the Byzantine text, and in any event that the method
described in this chapter for detecting contamination should work. According
to this study, it turns out that Chrysostom’s text as found in Codex Coxianus
is pre-Byzantine and does not have a significant amount of Byzantine con-
tamination.
The text from the following manuscripts have also been taken from Swan-
son’s transcription, though without double-checking: K/018, L/020, P/025,
Ψ/044, 056, 075, 1, 6, 69, 88, 104, 131, 205, 209, 226, 323, 330, 440, 460,
489, 517, 547, 614, 618, 796, 910, 927, 945, 999, 1175, 1241S, 1242, 1243,
1245, 1270, 1315, 1319, 1352, 1448, 1505, 1573, 1611, 1646, 1734, 1738,
1739, 1827, 1836, 1837, 1854, 1874, 1891, 1982, 2125, 2147, 2344, 2400,
2412, 2464, 2495, and 2815. Based on double-checking Swanson’s
transcriptions in other manuscripts, it can be expected that there would be
about 3–6 errors in each of these manuscripts, usually involving corrections
and singular readings. Nevertheless, it seems unlikely that correcting Swan-
son’s transcriptions will materially change the results because no corrections
aside from those of DC are used in this study, singular readings do not affect
the shape of the stemma, and most of Byzantine manuscripts usually differ
from one another in over ten variation units.
The text from the following manuscripts have also been taken from Clark’s
collations: 223, 876, 1022, 1799, 1960, and 2401. These have not been
double-checked, but Clark’s accuracy for 2423 was so impressive as to sug-
gest that any errors in the rest of his work should be minor or at least fewer
than Swanson’s.

40
One example is Gal 1:8, which is discussed in greater detail in Chapter 3.
41
Wesley L. Hemphill, Codex Coxianus of the Homilies of Chrysostom on Ephesians
and his Commentary on Galatians (Ph.D. diss., University of Pennsylvania; Norwood,
Mass.: Norwood, 1916), published a collation of Codex Coxianus against Migne. Thus, all
the citations to this commentary are to J.-P. Migne, Joannis Chrysostomi opera omnia
quae exstant (PG 61; Paris, 1862), modified by any differences noted by Hemphill.
54 Chapter 2. Materials and Methods

In all, 92 witnesses (counting D* and DC as distinct witnesses) were col-

lated for this study at 1624 variation units. Generally, orthographic variants
are not considered unless they produce a grammatically and syntactically
coherent form (interchange between ι and ει, between ε and αι, 42 προλημφθῇ/
προληφθῇ), because these can easily be removed by scribal conjecture and
are therefore not amenable to stemmatic analysis. Moreover, nonsense spell-
ings corrected by the first hand are also ignored. These collations provide the
evidentiary basis for the computational analysis used in this study as
described in the following sections.

2.2 Computational Stemmatics: Cladistics

As discussed in the previous chapter, the approach that seems the most theo-
retically promising for deducing a usable history of the text is stemmatics, as
long as some allowance for contamination can be made. None of the other
methods are geared to producing an account of the textual history in the form
of a genealogy of documents. The copy-text method assumes a trivially true
but unusable history of the text for just a single witness, while the majority-
text method imposes an artificial history of the text upon a much larger set of
witnesses. The eclectic and coherence-based genealogical methods,
moreover, are designed to cope with contamination but at the expense of
losing precision over the history of the text. For eclecticism, the questions of
textual history and the genealogical relationships between manuscripts are
sacrificed in favor of reconstructing the original text based on internal
evidence. For the coherence-based genealogical method, the textual history is
scrapped in favor of a highly abstract diagram of textual flows that does not
convey the textual state of any lost exemplar other than the initial text. Only
stemmatics promises to present a history of the text in the form of a
genealogy with both extant manuscripts and their hypothetical ancestors.
Ever since the times of Karl Lachmann, textual critics have understood the
importance of the common-error principle to identify genealogically related

42
For example, the following ε/αι variants from the UBS text produced grammatically
and syntactically coherent forms and were therefore considered as appropriate variants for
this study: 3:3 οὕτως ἀνόητοί ἐστε, ἐναρξάμενοι πνεύματι νῦν σαρκί ἐπτιτελεῖσθαι (“Are
you, who got started in the spirit, so senseless as to get finished now in the flesh?”); 3:7
γινώσκεται ἄρα ὅτι (“then it is known that”); 4:9 οἷς πάλιν ἄνωθεν δουλεύειν θέλεται (“to
which it is wanted to enslave all over again”); 4:10–11 ἡμέρας παρατηρεῖσθαι ... φοβοῦμαι
ὑμᾶς (“I fear you keep days ….”); 4:12 γίνεσθαι ... δέομαι ὑμῶν (“I beg of you to be-
come”); 4:18 καλὸν δὲ ζηλοῦσθε ἐν καλῷ πάντοτε (“but you are always being sought well
for good”); 5:10 ὅτι οὐδὲν ἄλλο φρονήσεται (“that nothing else is considered”); 5:18 οὐκ
ἐσται ὑπὸ νόμου (“it will not be under the law”); and 6:15 ἀλλὰ κενὴ κτίσις (“but an empty
creation”).
 2.2 Computational Stemmatics: Cladistics 55

groups, but they have had difficulty applying this principle to the messy
details of actual documents. In particular, accounting for contamination had
been a major theme of their research. Within the field of New Testament tex-
tual criticism, most of the intellectual effort in wrestling with the Lach-
mannian common-error principle has occurred, not in conjunction with Paul,
but over the question of the so-called “Caesarean Text” in Mark. According-
ly, the following historical overview of stemmatic analysis in New Testament
textual criticism must necessarily shift gears and discuss the text of Mark, but
it is to be stressed that the difficulties and concepts found in relation to Mark
are also applicable to Paul.
The concept of a Caesarean text of Mark did not exist in the seminal theo-
ry of the New Testament text propounded by Westcott and Hort in 1881. 43
Many of the most important members assigned to the Caesarean text – P45,
W, and Θ – were yet to be discovered and, among the four forms of the text
they identified (neutral, Western, Alexandrian, and Syrian), 44 many of the
other Caesarean members – namely, 565, family 1, family 13, 22, 28, and 157
– were merely considered to support Western readings. 45 In their theory of the
text, Westcott and Hort “emphatically” endorsed the role of genealogy: “All
trustworthy restoration of corrupted texts is founded on the study of their his-
tory, that is, of the relations of descent or affinity which connect the several
documents.” 46 This genealogy is discovered by applying the principle as they
formulated it that, apart from accidental coincidence, “identity of reading im-
plies identity of origin.” 47 This formulation is actually incomplete and
potentially misleading, because Westcott and Hort properly disregarded
agreements in original readings and used scribal errors and other forms of
alterations in identifying their textual groups. A more careful formulation of
the genealogical maxim was the “old rule” quoted by Kirsopp Lake: “com-
munity of error implies unity of origin.” 48
As famously observed by E. C. Colwell, Westcott and Hort did not actual-
ly employ the genealogical method to its fullest extent; they did not prepare
any stemma of the New Testament textual tradition. 49 The reason for this, of

43
B. F. Westcott and F. J. A. Hort, Introduction to the New Testament in the Original
Greek (repr. Peabody, Mass.: Hendrickson, 1988). References to this work use the abbre-
viation HORT.
44
HORT 178.
45
HORT 165.
46
HORT 40 (small capitals original).
47
HORT 40.
48
Kirsopp Lake, Codex 1 of the Gospels and its Allies (TS 7; Cambridge: Cambridge
University Press, 1902), xxiii.
49
E. C. Colwell, “Genealogical Method: Its Achievements and its Limitations” in idem,
Studies in Methodology in Textual Criticism of the New Testament (NTTS 9; Grand Rap-
ids, Mich.: Eerdmans, 1969), 65–66.
56 Chapter 2. Materials and Methods

course, was mixture, which, if not taken into account, threatens to invert the
genealogical relations and distort the stemma. Accordingly, Westcott and
Hort approximated the history of the text by relying on a few, well-chosen
“best documentary representatives” for the different forms of text that they
found. 50 Among their four forms of text, they considered the neutral text, best
exemplified in Codex Sinaiticus () and especially Codex Vaticanus (B), as
the purest form of the original text, while the Syrian (now called
“Byzantine”) text was viewed as the latest, being a recension of the previous
three forms. 51
About twenty years later, the first major modification of this genealogical
principle was proposed by Kirsopp Lake. 52 In his edition of Codex 1 and its
allies, Lake modified the old text-critical rule that “community of error
implies unity of origin” to account for mixture from Byzantine manuscripts.
Specifically, Lake argued that genealogical relations for late manuscripts
“can be deduced, in the absence of direct information, by studying the
variations from the standard text which they share in common.” 53 By
“standard text,” Lake means the Byzantine text. In practical terms, however,
Lake used the Textus Receptus as the standard of comparison when he edited
the text of four members of family 1, recognizing that the unusual readings of
the Textus Receptus have to be noted. 54
After working out a stemma for family 1, Lake proceeded to consider the
family’s connections with other manuscripts. Even though this phase of the
investigation evaluated readings in early manuscripts, Lake continued to ex-
clude any reading that agrees with the Byzantine text, worried that Byzantine
contamination could still have affected the archetype of family 1. Among the
remaining non-Byzantine readings, Lake also excluded those readings which
were found in almost all of the witnesses because such readings were
probably original and therefore not indicative of “the special affinities of the
text of fam1.” 55 Then, Lake compared the uncommon, non-Byzantine readings
in family 1 with other leading witnesses and found that while family 1 seems
to have “a more definite connection with the Old Syriac,” there is a “close
connection between fam1 and fam13 22 28 565 700.” 56 Lake concluded that
these manuscripts belong to a “larger family [that] seems to represent a local
text or local texts which were current in a comparatively limited region in the
East.” 57 Thus, it is evident that Lake had used a modification of the common-

50
HORT 59.
51
HORT 210 and 132–135.
52
Lake, Codex 1.
53
Lake, Codex 1, xxiii.
54
Lake, Codex 1, xxiii.
55
Lake, Codex 1, xlviii.
56
Lake, Codex 1, l.
57
Lake, Codex 1, liv.
 2.2 Computational Stemmatics: Cladistics 57

error principle to establish genealogical relationships, but only after

excluding any possible agreements in Byzantine readings.
Lake’s justification for filtering out Byzantine readings may seem reason-
able when working out the relationships within family 1, because a major
source of accidental coincidences in readings is independent assimilation to
the current standard text. The standard text was probably memorized by
monks who had spent a greater part of their lives studying, reciting, and copy-
ing the Biblical texts. In fact, his wife and collaborator Silva Lake reported:
“In collating aloud in Greek monastic libraries I have found that the librarian
almost always breaks in with a correction when I have read from a ms. a
phrase which differs from that which he is accustomed to hear.” 58 Kirsopp
Lake’s reasoning for excluding Byzantine readings when assessing the affini-
ties of family 1 with non-Byzantine texts, however, makes less sense because
many of those texts were not contaminated by Byzantine mixture so the risk
of over-counting accidental coincidences should be lower. One wonders
whether there was a more pragmatic basis underlying Lake’s decision: manu-
scripts were routinely collated against the Textus Receptus, and it actually
requires additional work to include rather than exclude Byzantine readings. In
other words, the analysis may have been made to fit the form in which the ev-
idence was collected. Regardless of the demerits of Lake’s first step, Lake’s
second step properly excluded original readings from being used to erect
groups, and Lake thus arguably found a limited community of error implying
a common origin for his set of manuscripts.
Lake’s alliance of Markan manuscripts was expanded over the ensuing
decades as new ones were discovered and published. For example, Lake and
Blake added Θ, the Koridethi Codex, to this alliance in 1923, and considered
it to be its leading member. 59 In 1924, Burnett Hillman Streeter accepted their
findings of a family Θ and connected this family to the text of Origen when
he lived in Caesarea. 60 At this point, Streeter changed Lake’s method in a
manner that would prove disastrous. Streeter kept Lake’s problematic first
step in filtering out all agreements with the Textus Receptus, but, instead of
looking for agreements in error, he calculated rates of agreements with Alex-
andrian and Western representatives and considered any manuscript that fell
somewhere in between to be “Caesarean.” 61 The problem with Streeter’s

58
Silva Lake, Family Π and the Codex Alexandrinus: The Text According to Mark (SD
5; London: Christophers, 1936; repr. Salt Lake City: University of Utah Press, 1965), 17
n.1.
59
Kirsopp Lake and Robert P. Blake, “The Text of the Gospels and the Koridethi Co-
dex,’ HTR 16 (1923): 267–286.
60
B. H. Streeter, The Four Gospels: A Study of Origins Treating of the Manuscript Tra-
dition, Sources, Authorship, & Dates (1st ed., 1924; 2d ed., London: Macmillan, 1926),
91–102.
61
See Streeter, Four Gospels, 81–84.
58 Chapter 2. Materials and Methods

method is that it now groups anything not clearly Byzantine, Alexandrian, or

Western into a catch-all category without any regard for how their readings
are patterned. In effect, Streeter treated the Caesarean text as a wellspring of
variants that could turn up in any manuscript, and Streeter designated so
many collateral members of other text-types as secondary or tertiary witness-
es to the Caesarean text that almost any non-standard variant could be classi-
fied as Caesarean. For example, Streeter thought that the Purple Uncials (N,
O, Σ, and Φ) were secondary Caesarean witnesses, 62 and in the years to come
Streeter continued to find additional members, such as 157, 1071, the uncial
U, and, for that matter, most of von Soden’s I-text. 63
Streeter’s approach is also deeply flawed because it too easily elides the
distinction between a manuscript, its text, and its readings. A text is not a
hodgepodge of variant readings; it is a pattern of sequential readings em-
bodied in a document. Though Streeter identified a collection of interesting
readings, it was questionable whether a substantial majority of these variants
were ever embodied together in the text of a real manuscript that actually
existed in history. In other words, it was unclear whether Streeter’s Caesarean
text corresponded to historical reality or whether it was merely an artifact of
his defective method.
In his second edition of The Four Gospels, Streeter responded to these crit-
icisms with a new appendix arguing that Codex W from Mark 5:31 – 16:8
was the purest authority for his Caesarean text: “I found a total of 260 read-
ings in which W differs from the T.R. In 37 of these the readings of W is
supported by no other MS. or version. … Ignoring these, there remain 223
readings. Of these 189, i.e. all but 85%, are, I found, supported by at least
one, and usually by several, of the six above-named authorities for fam. Θ.” 64
Streeter’s argument may sound impressive, but some of the agreements
against the Textus Receptus are in original readings and are therefore not
indicative of genealogical affinity below the level of the original. Of the
remaining agreements, Streeter was satisfied by any agreement with even one
member of his Caesarean gaggle of witnesses. Yet this procedure presupposes
that his Caesarean set of texts was already a true community of error having a
common origin, not an artificial assortment of variants. Thus, Streeter’s
response is a form of circular reasoning.
Starting in the 1950s, Ernest Cadman Colwell began publishing a series of
papers calling for a methodological refocus in the practice of New Testament

62
Streeter, Four Gospels, 575–577.
63
B. H. Streeter, “Codices 157, 1071 and the Caesarean Text” in Robert P. Casey, Silva
Lake, and Agnes K. Lake, eds., Quantulacumque (Kirsopp Lake FS; London: Christophers,
1937), 149–150.
64
Streeter, Four Gospels, 599.
 2.2 Computational Stemmatics: Cladistics 59

textual criticism. 65 Most importantly, Colwell established that a text-type is a

group of documents, not a collection of various readings. 66 Colwell also crit-
icized the practice of only counting agreements among deviations from an
external standard, namely, the Textus Receptus. 67 As a result, Colwell in a
very influential paper co-authored with Ernest Tune proposed a “quantitative
method,” in which all pairs of manuscripts were to be compared against each
other directly, rather than indirectly through the Textus Receptus. 68 Colwell
and Tune applied their ideas to a sample of 18 manuscripts and their cor-
rectors and found that, for their sample, a minimum of 70% total agreement
in a discriminated set of variants with a 10% gap corresponded well to the
distribution of known text-types. Colwell and Tune expressed hope that their
method could be extended to greater numbers of manuscripts and that their
70% agreement and 10% gap criteria would carry over generally. 69
This quantitative approach differs from stemmatics in the kinds of agree-
ments that are counted. Lachmannian stemmatics counts agreements in error,
but the Colwell-Tune approach counts all agreements in the variation units,
whether their readings are original or secondary. As a result, the quantitative
approach cannot actually identify groups having a common ancestry,
although it may approximate such groups under appropriate circumstances. In
its favor, however, the quantitative approach outlined by Colwell and Tune is
easy to calculate by hand and yields a rough, first-order approximation of
which manuscripts are probably related and which is going to be worth
studying in more detail. If the goal is the history of the text, however, the
quantitative method was not ultimately designed for that purpose.
At approximately the same time that Colwell and Tune were developing
their quantitative approach, systematic biologists also began investigating nu-
merical techniques for the classification of biological organisms. 70 One of the
first attempts in this area by Robert Sokal and Peter Sneath, called “numerical
taxonomy” or “phenetics,” also calculated percentages in overall agreements.

65
These are republished in E. C. Colwell, Studies in Methodology.
66
E. C. Colwell, “Method in Grouping New Testament Manuscripts,” in Studies in
Methodology, 9.
67
Colwell, “Method,” 4 n.4, credits Harold Murphy, “Eusebius’ New Testament Text in
the Demonstratio Evangelica,” JBL 73 (1954): 167–168, for this insight.
68
Colwell and Tune, “Method in Establishing Quantitative Relationships Between
Text-Types of New Testament Manuscripts” in Studies in Methodology, 56–62.
69
Colwell and Tune, “Quantitative Relationships,” 59.
70
For excellent textbooks on cladistics, see Joseph Felsenstein, Inferring Phylogenies
(Sunderland, Mass.: Sinauer, 2004); and Ian J. Kitching, Peter L. Forey, Christopher J.
Humphries, and David M. Williams, Cladistics: The Theory and Practice of Parsimony
Analysis (The Systematics Association Publication 11; Oxford: Oxford University Press,
2000). The historical overview presented here is based on Felsenstein, Inferring Phylo-
genies, 123–138.
60 Chapter 2. Materials and Methods

Though independent of Colwell and Tune, Sokal and Sneath’s numerical

taxonomy operates on a similar theoretical basis. Its proponents, however,
merely conceived of their method as an aid to classification, not a technique
for inferring genealogical relationships. 71
In 1963 and 1964, two population geneticists, Anthony Edwards and Luca
Cavalli-Sforza worked out a family tree of human populations based on blood
group gene frequencies using the principle of “parsimony,” which is that the
tree with the smallest number of changes would correspond to the organisms’
“phylogeny” or historical pedigree. 72 Because scientists had good access to
mainframe computers in the 1960s, they began to develop algorithms for
calculating the number of changes on a set of possible trees and choosing the
optimal tree. 73 By 1970, one of these computational biologists, James S.
Farris, noticed that a German entomologist named Willi Hennig had provided
the philosophical justification for using parsimony as a metric for evaluating
different phylogenies as far back as 1950. 74 Hennig’s justification or
“auxiliary principle” is that “the presence of apomorphous characters in dif-
ferent species ‘is always reason for suspecting kinship . . ., and that their
origin by convergence should not be assumed a priori.’” 75 In other words,
Hennig reinvented Lachmann’s common error principle articulated a century
earlier. Since computational biologists had figured out how to implement this
criterion in computer software, this means that much of the laborious effort in
classical stemmatics can be performed by a computer.
Indeed, researchers began noticing in the 1970s deep analogies between
cladistics in biology and stemmatics in philology. In fact, one article
published in 1977 urged biologists to use cladistics for classifying organisms
based on shared, derived characteristics precisely because of the success in
textual criticism of the common error principle in stemmatics for recognizing
kinship. 76 By 1987, cladistics had been applied to manuscript traditions, 77 yet

71
Felsenstein, Inferring Phylogenies, 123–124.
72
Felsenstein, Inferring Phylogenies, 125–128.
73
Felsenstein, Inferring Phylogenies, 129–130.
74
Felsenstein, Inferring Phylogenies, 137.
75
Willi Hennig, Phylogenetic Systematics (English trans, D. Dwight Davis & Rainer
Zangerl; Urbana, Ill.: University of Illinois Press, 1966, repr. 1999), 121. “Apomorphous”
is a technical term referring to a derived character in biology; it corresponds to a scribal
error in textual criticism.
76
Nelson I. Platnick & H. Don Cameron, “Cladistic Methods in Textual, Linguistic, and
Phylogenetic Analysis,” Systematic Zoology 26 (1977): 380–385. Also H. Don Cameron,
“The Upside-Down Cladogram: Problems in Manuscript Affiliation” in Henry M. Hoenigs-
wald & Linda F. Wiener, Biological Metaphor and Cladistic Classification: An Inter-
disciplinary Perspective (Philadelphia: University of Pennsylvania Press, 1987), 227–242.
77
Arthur R. Lee, III, “Numerical Taxonomy Revisited: John Griffith, Cladistic Analysis
and St. Augustine’s Quaestiones in Heptateuchum,” Studia Patristica 20 (1989): 24–32,
orig. deliv. (Oxford: 10th International Conference of Patristic Studies, 1987).
 2.3 Cladistics in Action: A Simplified Example 61

it was four years later that cladistics managed to pass a significant test. Peter
Robinson, who had painstakingly constructed a stemma of an Old Norse
narrative based, in part, on direct external evidence, posted an on-line
challenge in 1991 to reconstruct his stemma solely by mathematical
techniques. One of the entrants was a biologist who ran the data through a
cladistics computer program and successfully found all of Robinson’s textual
groups. 78 Robinson later used cladistics in the editing of the Canterbury
Tales. 79 In fact, cladistics was even applied to the text of the New Testament,
specifically manuscripts of the Epistle of James. 80 As a result, cladistics has
proven to be an appropriate method for implementing stemmatics with a
computer, both for theoretical and practical reasons.

2.3 Cladistics in Action: A Simplified Example

Cladistics works on the principle of “maximum parsimony.” It searches

through many different possible stemmata for a set of witnesses, calculating
the number of changes each stemma requires, and choosing a stemma with
the minimum number of changes. Every change is a hypothesis that a certain
event happened in history (e.g., that such and such a scribe changed this word
to another), and, following Occam’s razor, the maximally parsimonious ex-
planation is the one that requires the fewest separate hypotheses of historical
change. In other words, where two manuscripts agree on a non-original
reading, it is more parsimonious of an explanation that the textual variant
arose once rather than twice in history. Of course, the principle of maximum
parsimony does not mean that the pattern of textual variation is such that no
coincidental alterations of the text ever happened. They did, multiple times in
fact. Rather, what it means is that the textual history that minimizes the
number of historical hypothesis is prima facie superior to one that requires
additional hypotheses of accidental coincidences.
The method of constructing a stemma using cladistics can be illustrated us-
ing a simplified example of four witnesses over a single chapter of Galatians.
The method illustrated here follows Dom Froger’s approach to stemmatics,

78
Peter M.W. Robinson & Robert J. O’Hara, “Report on the Textual Criticism Chal-
lenge 1991,” Bryn Mawr Classical Review 3 (1992): 331–337.
79
Robert J. O’Hara & Peter M.W. Robinson, "Computer-Assisted Methods of Stem-
matic Analysis," Occasional Papers of the Canterbury Tales Project 1 (1993): 53–74.
80
Spencer et al., “Greek Vorlage.” The authors however states that their use of cladis-
tics “makes no attempt to identify original readings and does not require genealogical
information based on philological work, although it can use such information if available.”
This study precisely develops how to do incorporate such work.
62 Chapter 2. Materials and Methods

mentioned in the first chapter. 81 In order to avoid the circularity of having to

identify beforehand which reading is authorial and which is a scribal error,
the method splits the reconstruction of the stemma into two steps. First, the
shape and linkage pattern (l’enchaînement) of the stemma is determined
based on the pattern of textual changes, not errors. The distinction between
textual changes and errors is that the former are local but the latter are global.
In other words, textual changes are those differences between a copy and its
immediate exemplar, while errors are those differences between that copy and
its first ancestor. For example, a scribe can correct an error in an exemplar,
but this correction still counts as a textual change even though is what the
author wrote. Froger’s proposal is to minimize the textual changes, not the
error, so it is unnecessary at this step to know whether a reading is true or an
error. By minimizing the textual changes, a maximally parsimonious un-
oriented stemma can be found without having to determine the directionality
of the changes.
The second step is the orientation (l’orientation) of this unoriented stem-
ma, and this step is accomplished by using internal evidence to determine
which branch of the unoriented stemma is closest to the archetype. As
discussed earlier, internal evidence allows the textual critic to identify those
readings that appears more authorial than scribal. Solely for the purposes of
this simplified example, the Nestle-Aland text is used to identify which
readings are considered authorial.
Accordingly, the preliminary step in this simplified example is to collate
the witness. When P46, , A, and B are collated over the first chapter of
Galatians, a total of thirty-four variation units are found, as set forth in the
following Table 1: 82

81
J. Froger, La critique des textes et son automatisation (INS 7 ; Paris: Dunod, 1968),
77–78.
82
In this table, the first column contains the citation of the textual variant, the second
column holds the text of the Nestle-Aland 27th edition, with the relevant portion of the text
marked in bold to indicate the variation unit. In each of the remaining columns for the
witnesses, the equal sign (=) indicates that the respective witness agrees with the reading of
the Nestle-Aland text. If the witness disagrees with the Nestle-Aland text, the replacement
text is given in the column, with omissions being indicated by three en-dashes (–––).
Where a witness is lacunose (as with P46 at 1:8b–9) or otherwise cannot attest to any of
the readings due to an overriding textual variant (e.g., 1:8c dominating 1:8d and 1:11c
dominating 1:11d), then that lack of attestation is indicated by a question mark.
 2.3 Cladistics in Action: A Simplified Example 63

Table 1: Variation Units in Galatians 1

Gal NA Text P46  A B

1:1 τοῦ ἐγείραντος = αὐτῶν 83 = =
αὐτὸν ἐκ νεκρῶν
1:3 θεοῦ πατρὸς ἡμῶν καὶ κυρίου = = καὶ κυρίου
καὶ κυρίου Ἰησοῦ ἡμῶν 84 ἡμῶν
Χριστοῦ
1:4a τοῦ δόντος αὐτόν 85 = = =
ἑαυτόν
1:4b ὑπὲρ τῶν περί 86 περί περί =
ἁμαρτιῶν
1:4c ὑπὲρ τῶν ––– 87 = = =
ἁμαρτιῶν
1:4d κατὰ τὸ θέλημα = ––– 88 = =
τοῦ θεοῦ
1:6a Θαυμάζω ὅτι οὕτω 89 = = =
οὕτως ταχέως
1:6b ἀπὸ τοῦ ἡμᾶς 90 = = =
καλέσαντος ὑμᾶς
1:6c ἐν χάριτι Χριστοῦ ––– 91 = = =
1:7 καὶ θέλοντες = ––– 92 = =
μεταστρέψαι
1:8a ἀλλὰ καὶ ἐὰν = = = κἄν 93
ἡμεῖς
1:8b εὐαγγελίζηται ? εὐαγγελίσηται εὐαγγελίσηται =
ὑμῖν
1:8c εὐαγγελίζηται ? ––– 94 = =
ὑμῖν
1:8d εὐαγγελίζηται ? ? = ὑμῖν εὐαγγ.
ὑμῖν
1:9 ὡς προειρήκαμεν ? προείρηκα 95 = =

83
Although this variant by the original hand of  is apparently orthographic since
omicron and omega merged by the fourth century, this textual change potentially has a
different sense: “who raised from their dead.”
84
See Chapter 3 for a more detailed discussion of this transposition.
85
Also shared by 33 and 056*. If αυτον has the rough breathing, it is an orthographic
variant of ἑαυτόν. If it has the smooth breathing, the referent of the participle would have
to be God the Father instead of Jesus Christ.
86
See Chapter 4 for a more detailed discussion of this substitution.
87
This is one of 28 singular omissions of the article cataloged by ROYSE 273.
88
By *, then corrected to include τό.
89
This is probably a mere orthographic change.
90
By P46*, then corrected to ὑμᾶς.
91
See Chapter 3 for a more detailed discussion of this omission.
92
This is an omission by the original hand of . 2464* has the same mistake.
93
WEISS 63 holds that this crasis is due to a scribal error in B also apparent in Gal 5:17.
94
This and the following variant are discussed in greater detail in Chapter 3.
64 Chapter 2. Materials and Methods

Gal NA Text P46  A B

1:11a Γνωρίζω γὰρ ὑμῖν δέ 96 = δέ δέ
1:11b τὸ εὐαγγέλιον = = = ter 97
1:11c τὸ εὐαγγελισθὲν ὅθεν 98 = = =
ὑπ’ ἐμοῦ
1:11d τὸ εὐαγγελισθὲν ? = εὐηγγελισθὲν =
ὑπ’ ἐμοῦ
1:12 οὔτε ἐδίδαχθην = οὐδέ 99 οὐδέ =
1:14a και προέκοπτον ἐπροέκοπτον 100 = = =
ἐν τῷ Ἰουδαϊσμῷ
1:14b περισσοτέρως περισσοτέρον 101 = = =
1:15a ὅτε δὲ εὐδόκησεν = = ηὐδόκησεν =
[ὁ θεός]
1:15b ὅτε δὲ εὐδόκησεν ––– 102 = = –––
[ὁ θεός]
1:15c καὶ καλέσας διὰ ––– 103 = = =
τῆς χάριτος αὐτοῦ
1:16 ἵνα εὐαγγελίζωμαι εὐαγγελίσωμαι 104 = = =
αὐτόν
1:17a οὐδὲ ἀνῆλθον εἰς ἦλθον 105 = = ἀπῆλθον 106
Ἱεροσόλυμα
1:17b ἀλλὰ ἀπῆλθον εἰς ἀπῆλθα 107 = = =
Ἀραβίαν
1:18a Ἔπειτα μετὰ ἔτη = τρία ἔτη 108 τρία ἔτη =
τρία

95
This textual variant by the original hand of  appears to be an assimilation to the
number of the following λέγω and has the effect of aggrandizing Paul’s importance.
96
This substitution is discussed in greater detail in Chapter 3.
97
That is, repeated three times as τὸ εὐαγγέλιον τὸ εὐαγγέλιον τὸ εὐαγγέλιον, apparent-
ly out of confusion with the immediately following phrase τὸ εὐαγγελισθὲν ὑπ’ ἐμοῦ.
98
A nonsense reading; see ROYSE 258 n.328, for more detail.
99
Discussed in greater detail in Chapter 3.
100
Another example of a double augment in a singular reading by P46 is found at Eph
2:2 (ROYSE 305).
101
By P46*; see also ROYSE 226 n.136, for editors misreading the correction.
102
Discussed in greater detail in Chapter 3.
103
This omission is apparently occasioned by a leap from μητρός μου to αὐτοῦ; for
more detail see ROYSE 288 n.514. This omission is also attested in 1739 and 6 but,
unfortunately, not discussed by Günther Zuntz, The Text of the Epistles: A Disquisition
upon the Corpus Paulinum (The Schweich Lectures of the British Academy 1946; London:
Oxford University Press, 1953).
104
The substitution for the more usual subjunctive is also attested in D*, 330, and 2400.
105
The singular omission of prefixes is common in P46 (ROYSE 326).
106
The change in the verbal prefix is discussed in greater detail in Chapter 3.
107
A singular reading in P46; other instances of first aorist endings with second aorist
stems occur at Heb 9:11, 11:15, and 12:9 (ROYSE 303).
108
This transposition is discussed in greater detail in Chapter 3.
 2.3 Cladistics in Action: A Simplified Example 65

Gal NA Text P46  A B

1:18b καὶ ἐπέμεινα πρὸς ἔμεινα 109 = = =
αὐτὸν
1:19 οὐκ εἶδον οὐχ 110 = = οὐχ
1:21a ἦλθον εἰς τα ? = κλήματα 111 =
κλίματα
1:21b τῆς Συρίας καὶ ? ––– 112 = =
τῆς Κιλικίας
1:22 ταῖς ἐκκλησίαις ? = = ἐκκλησίας
τῆς Ἰουδαίας

Out of these thirty-four variation units, all but five involve singular read-
ings. 113 The distribution of singular readings is not even, and the bulk of the
singular readings comes from P46 with 13; there are 7 singulars each for 
and B, and just four for A. 114 It is possible for a variation unit to have more
than one singular reading; for example, the variation unit at 1:17a has two
singular readings, with P46 reading ἦλθον and B reading ἀπῆλθον against the
ἀνῆλθον of  and A. Of the five non-singular readings (1:3, 1:12, 1:15b,
1:18a, and 1:19), they all oppose the combination of P46 and B against the
combination of  and A.
The next step in the method is to apply the pattern of textual variation to
possible unoriented stemmata for the witnesses and count how many textual
changes each unoriented stemma implies. Typically, implementations of cla-
distics do not permit one witness to be the direct ancestor or descendant of
another and, by a simplifying assumption, they limit internal nodes to three
branches. Thus, there would only be three unoriented stemmata to be consid-
ered. 115 This constraint is justified in zoology because systematic biologists
are dealing with animals, for whom the biological species concept forbids
interbreeding, but for manuscripts it seems to better to permit one distinct
manuscript to be the ancestor of another. Accordingly, this study follows an
extension of cladistics described by biologist John Alroy and termed

109
This is another example of P46’s penchant to omit verbal prefixes.
110
Discussed in greater detail in Chapter 4.
111
This itacism produces a reading on the verge of nonsense (“I went to the branches of
Syria of Cilicia.”).
112
The lack of the article would tend to indicate that Syria and Cilicia are taken
together; see Daniel B. Wallace, Granville Sharp’s Canon and Its Kin: Semantics and Sig-
nificance (SBG 14; New York: Peter Lang. 2009), 167 (“discrete items which are united”)
and 170 (“Geographic terms were easily the largest type of nouns to belong in the discrete
group.”).
113
Many of these readings, however, are attested in witnesses not part of this simplified
example.
114
In 1:8d both word orders are considered singular for the data set. When more
witnesses are added, neither reading remains singular.
115
They are unoriented stemmata (17), (18), and (19), shown below.
66 Chapter 2. Materials and Methods

“Continuous Track Analysis.” 116 Alroy reformulated the parsimony objective

of minimizing the number of changes to be one of minimizing the number of
introduced variants, which he called “continuous tracks.” Because any extant
witness can be linked to any other, Continuous Track Analysis straight-
forwardly handles extant ancestors, multiple children, and even
contamination.
(1) P46 ℵ A B
(2) P46 ℵ B A
(3) P46 A ℵ B
(4) P46 A B ℵ
(5) P46 B ℵ A
(6) P46 B A ℵ
(7) ℵ P46 A B
(8) ℵ P46 B A
(9) ℵ A P46 B
(10) ℵ B P46 A
(11) A P46 ℵ B
(12) B P46 ℵ A

FIG. 8: Twelve Linear Unoriented Stemmata of Four Manuscripts

Thus, when direct ancestry of witnesses is permitted, there are actually twen-
ty possible unoriented stemmata for any four witnesses. These unoriented
stemmata can be classified depending on how many branches they have and
how hypothetical intermediaries they include. Perhaps the simplest con-
figuration is a straight linear stemma with no more than two branches at each
node or and hypothetical intermediaries. According to combinatory theory,
there are 4! = 24 possible linear orderings of four distinct elements. Since the
stemmata are unoriented, however, each ordering is equivalent to the reversal
of the ordering for 4! / 2 = 24 / 2 = 12 possible orderings. These twelve linear
stemmata are shown in FIG. 8.
It is important to keep in mind that these stemmata are unoriented. In other
words, the starting point of the tradition, when the stemma is oriented, can be
at any node where there is a witness or at any branch between two nodes.
Thus, for unoriented stemma (1), the order P46––A–B is equivalent to the
unoriented order B–A––P46 at this stage in the method.

116
John Alroy, “Continuous Track Analysis: A New Phylogenetic and Biogeographic
Method,” Systematic Biology 44 (1995): 152–172.
 2.3 Cladistics in Action: A Simplified Example 67

Another configuration is with one manuscript in the middle with three

branches to other manuscripts respectively, but with no hypothetical inter-
mediaries. This configuration can be arranged in the shape of a “T” and there
four such unoriented stemmata, one for each manuscript in the middle, as in
shown in FIG. 9.

ℵ P46 B P46 ℵ B

(13) A (14) A

ℵ A B ℵ B P46

(15) P46 (16) A

FIG. 9: Four Unoriented Stemmata in a T-Configuration

Finally, if hypothetical intermediaries are permitted, there are another sixteen

possibilities, four of which are shown in FIG. 10. 117

P46 A P46 ℵ
(17) • • (18) • •
ℵ B A B

P46 A P46 A
(19) • • (20) •
B ℵ B ℵ

FIG. 10: Four Unoriented Stemmata with Hypothetical Intermediaries

All of these unoriented stemmata represent possible histories of the text, but
some fit the textual evidence better than others. Accordingly, the textual vari-
ants are then used to evaluate the possible unoriented stemmata in terms of
the number of textual changes they imply. In cladistics, the number of
changes is referred to as a “cost.” This evaluation is performed on a variation-
unit-by-variation-unit basis to count the cost, i.e., the number of textual
changes implied by each variation unit. The total cost for an unoriented
stemma is determined by adding up all the individual costs of the variation
units. In accordance with the maximum parsimony principle, the goal is to
choose the unoriented stemma with the minimum cost.

117
The other twelve possibilities are in a Y-configuration, with the hypothetical inter-
mediary at the middle of the Y and one of the branches with two manuscripts in a row. As
consideration of these do not affect the final result, they are omitted from this discussion to
simplify the description.
68 Chapter 2. Materials and Methods

For example, the first variation unit in this example is a singular reading at
Gal 1:1 where  reads αὐτῶν while the other witnesses (P46, A, and B) all
read αὐτόν. For unoriented stemma (1), with the pattern P46––A–B, there
are two textual changes: one between P46 and  (since P46 and  have dif-
ferent readings, αὐτόν and αὐτῶν, respectively), and another textual change
between  and A. The same cost of 2 also applies for unoriented stemmata
(2), (3), (5), (11), and (12), because in all these configurations  is an internal
node with two branches. For unoriented stemma (14), the cost goes up to 3
because now there must be a textual change between  and P46, between 
and A, and between  and B. Here,  is an internal node with three branches,
and all three branches connected to  incur a cost because  has a reading
different from its neighbors on the unoriented stemma. When  is located on
the edge (or “leaf”) of a stemma, its cost is only 1. Concerning unoriented
stemma (4) P46–A–B–, for example, there is only one textual change,
between B and ; the rest of the branches incur no cost because they all have
the same reading for this variation unit. Indeed, the cost for a singular reading
on a leaf node is always 1, and this is why unoriented stemmata (6), (7), (8),
(9), (10), (13), (15), (16), (17), (18), (19), and (20) would all evaluate to a
cost of 1.
As another example, variation unit 1:3 has two non-singular readings. P46
and B have one word order (καὶ κυρίου ἡμῶν) while  and A have another
(ἡμῶν καὶ κυρίου). For unoriented stemma (1), with the pattern P46––A–B,
there are two textual changes: one between P46 and  and another between A
and B. There is no textual change between  and A, because they have the
same reading. For unoriented stemma (2), P46––B–A, the cost goes up to 3,
because there is an additional textual change in this configuration, here,
between  and B. A different cost of 1 is the result for unoriented stemma (5),
P46–B––A, because the only textual change is between B and ; the
branches between P46 and B and also between  and A do not have any cost.
All the branching stemmata (13) – (20) have a cost of 2, except for one
configuration: unoriented stemma (19) where there is only textual change
between the P46–B side and the -A side.
The costs incurred by each variation unit for each unoriented stemmata can
be presented in the form of a table, as follows:

Table 2: Cost Table for Twenty Unoriented Stemmata

Unoriented Stemma
Gal 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
1:1 2 2 2 1 2 1 1 1 1 1 2 2 1 3 1 1 1 1 1 1
1:3 2 3 2 3 1 1 3 2 1 2 3 1 2 2 2 2 2 2 1 2
1:4a 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:4b 1 2 1 2 2 2 1 2 1 2 1 1 1 1 1 3 1 1 1 1
 2.3 Cladistics in Action: A Simplified Example 69

Unoriented Stemma
Gal 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
1:4c 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:4d 2 2 2 1 2 1 1 1 1 1 2 2 1 3 1 1 1 1 1 1
1:6a 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:6b 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:6c 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:7 2 2 2 1 2 1 1 1 1 1 2 2 1 3 1 1 1 1 1 1
1:8a 1 2 1 2 2 2 1 2 1 2 1 1 1 1 1 3 1 1 1 1
1:8b 1 2 1 2 1 1 1 2 1 2 2 1 3 1 1 2 1 1 1 1
1:8c 1 1 2 1 2 1 1 1 1 1 2 2 3 2 1 1 1 1 1 1
1:8d 1 1 1 1 1 1 1 1 1 1 1 1 2 2 1 1 1 1 1 1
1:9 1 1 2 1 2 1 1 1 1 1 2 2 3 2 1 1 1 1 1 1
1:11a 2 2 2 1 2 1 1 1 1 1 2 2 1 3 1 1 1 1 1 1
1:11b 1 2 1 2 2 2 1 2 1 2 1 1 1 1 1 3 1 1 1 1
1:11c 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:11d 2 1 1 1 1 2 2 1 2 1 1 1 3 1 2 1 1 1 1 1
1:12 2 3 2 3 1 1 3 2 1 2 3 1 2 2 2 2 2 2 1 2
1:14a 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:14b 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:15a 2 1 2 2 1 2 2 1 2 1 1 1 1 1 3 1 1 1 1 1
1:15b 2 3 2 3 1 1 3 2 1 2 3 1 2 2 2 2 2 2 1 2
1:15c 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:16 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:17a 2 3 2 3 2 2 3 3 2 3 3 3 3 2 2 3 2 2 2 2
1:17b 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:18a 2 3 2 3 1 1 3 2 1 2 3 1 2 2 2 2 2 2 1 2
1:18b 1 1 1 1 1 1 2 2 2 2 2 2 3 1 1 1 1 1 1 1
1:19 2 3 2 3 1 1 3 2 1 2 3 1 2 2 2 2 2 2 1 2
1:21a 2 2 2 1 2 1 1 1 1 1 2 2 1 3 1 1 1 1 1 1
1:21b 2 2 2 1 2 1 1 1 1 1 2 2 1 3 1 1 1 1 1 1
1:22 1 2 1 2 2 2 1 2 1 2 1 1 1 1 1 3 1 1 1 1
Total 48 57 49 52 47 41 60 58 49 58 67 56 74 55 43 50 40 40 35 40

As shown in Table 2, the various possible unoriented stemmata vary greatly

in the number of textual changes they imply. More specifically, they range
from a low of 35 for unoriented stemma (19) to a high of 74 for unoriented
70 Chapter 2. Materials and Methods

stemma (13). According to the maximum parsimony principle, the unoriented

stemma with the lowest cost is the one that best explains the distribution of
the textual variation and is therefore selected as the hypothetical history of
the textual transmission. Since the most parsimonious unoriented stemma is
unoriented stemma (19) with a low cost of 35 textual changes, it is
accordingly selected as the proposed textual history that best describes the
state of the external textual evidence.
The final step is to orient the stemma. The step of orienting the stemma
looks for the portion of the stemma, a node or a branch, that appears to be
closest to the text of the work as the author wrote it. External evidence does
not determine this, and the previous operations up to this point have so far
involved only external evidence. Rather, in order to orient the stemma it is
necessary to use internal evidence, because internal evidence can identify
which reading more likely owes its origin to the author of the work and which
reading more likely owes its origin to a copyist. In other words, it is
necessary to decide, based on internal evidence, which textual variant is more
likely to be authorial than the others in the variation unit. A detailed exam-
ination of the internal evidence follows in Chapter 3, so for purposes of this
simplified example it is assumed that the reading of the Nestle-Aland text is
the reading adjudged to be the earliest form of the text even though this edi-
tion does not always depend on internal evidence in establishing its text.
The following table itemizes the readings collated for this simplified
example, with the scribal variants in italics (defined for this example as those
readings that do not agree with the Nestle-Aland text). Between the columns
for the witnesses are the authorial reading counts for the intermediary nodes
(IN) between P46 and B and between  and A, as well as the middle branch
(MB).

Table 3: Authorial Readings in Galatians 1 per Nestle-Aland

Gal P46 IN B MB  IN A
1:1 αὐτὸν 1 αὐτὸν 1 αὐτῶν 1 αὐτὸν
1:3 καὶ κυρίου ἡμῶν 0 καὶ κυρίου ἡμῶν 1 ἡμῶν καὶ κυρίου 1 ἡμῶν καὶ κυρίου
1:4a αὐτόν 1 ἑαυτόν 1 ἑαυτόν 1 ἑαυτόν
1:4b περί 0 ὑπὲρ 0 περί 0 περί
1:4c ––– 1 τῶν 1 τῶν 1 τῶν
1:4d τό 1 τό 1 ––– 1 τό
1:6a οὕτω 1 οὕτως 1 οὕτως 1 οὕτως
1:6b ἡμᾶς 1 ὑμᾶς 1 ὑμᾶς 1 ὑμᾶς
1:6c ––– 1 Χριστοῦ 1 Χριστοῦ 1 Χριστοῦ
1:7 καὶ θέλοντες 1 καὶ θέλοντες 1 ––– 1 καὶ θέλοντες
1:8a καὶ ἐάν 1 κἄν 1 καὶ ἐάν 1 καὶ ἐάν
1:8b ? 1 εὐαγγελίζηται 1 εὐαγγελίσηται 0 εὐαγγελίσηται
1:8c ? 1 ὑμῖν 1 ––– 1 ὑμῖν
1:8d ? 0 ὑμῖν εὐαγγ. 1 ? 1 εὐαγγ. ὑμῖν
1:9 ? 1 προειρήκαμεν 1 προείρηκα 1 προειρήκαμεν
 2.3 Cladistics in Action: A Simplified Example 71

Gal P46 IN B MB  IN A
1:11a δέ 0 δέ 0 γάρ 0 δέ
1:11b τὸ εὐαγγέλιον 1 ter 1 τὸ εὐαγγέλιον 1 τὸ εὐαγγέλιον
1:11c ὅθεν 1 τὸ εὐαγγελισθέν 1 τὸ εὐαγγελισθέν 1 τὸ εὐηγγελισθέν
1:11d ? 1 εὐαγγελισθέν 1 εὐαγγελισθέν 1 εὐηγγελισθέν
1:12 οὔτε 1 οὔτε 1 οὐδέ 0 οὐδέ
1:14a ἐπροέκοπτον 1 προέκοπτον 1 προέκοπτον 1 προέκοπτον
1:14b περισσοτέρον 1 περισσοτέρως 1 περισσοτέρως 1 περισσοτέρως
1:15a εὐδόκησεν 1 εὐδόκησεν 1 εὐδόκησεν 1 ηὐδόκησεν
1:15b ––– 0 ––– 1 ὁ θεός 1 ὁ θεός
1:15c ––– 1 καὶ . . . αὐτοῦ 1 καὶ. . . αὐτοῦ 1 καὶ. . . αὐτοῦ
1:16 εὐαγγελίσωμαι 1 εὐαγγελίζωμαι 1 εὐαγγελίζωμαι 1 εὐαγγελίζωμαι
1:17a ἦλθον 0 ἀπῆλθον 1 ἀνῆλθον 1 ἀνῆλθον
1:17b ἀπῆλθα 1 ἀπῆλθον 1 ἀπῆλθον 1 ἀπῆλθον
1:18a ἔτη τρία 1 ἔτη τρία 1 τρία ἔτη 0 τρία ἔτη
1:18b ἔμεινα 1 ἐπέμεινα 1 ἐπέμεινα 1 ἐπέμεινα
1:19 οὐχ 0 οὐχ 1 οὐκ 1 οὐκ
1:21a ? 1 κλίματα 1 κλίματα 1 κλήματα
1:21b ? 1 τῆς 1 ––– 1 τῆς
1:22 ? 0 ἐκκλησίας 1 ἐκκλησίαις 1 ἐκκλησίαις
Total 8 26 25 32 23 29 26

As can be seen from Table 3, different nodes and branches have different
authorial reading counts, and these counts can be mapped onto the unoriented
stemma, as shown in FIG. 11:

P46 A
8 26

32
26 • • 29
25
23
B ℵ

FIG. 11: Unoriented Stemma (19) with Counts of Authorial Readings

P46 has the least amount of authorial readings, only 8, and this reflects the
fact that it is lacunose in a number of places and has many singular readings
in the variation units that are extant for this papyrus. Somewhat better is 
with 23 authorial readings, and B with 25 authorial readings, but A has the
most with 26 authorial readings. None of these extant witnesses, however,
contain as many authorial readings as the intermediary nodes and middle
branch. The intermediary node between P46 and B has 26 authorial readings,
while the one between  and A has 29 authorial readings. Even more authori-
al readings, moreover, can be assigned to the middle branch between P46–B
on one side and -A on the other. Indeed, a total of 32 authorial readings can
72 Chapter 2. Materials and Methods

be assigned to the middle branch. There are only two variation units whose
authorial readings according to the Nestle-Aland edition are not found on the
middle branch, and in both cases the singular reading is the one read by the
Nestle-Aland main text. One is at 1:4b where B reads ὑπέρ, while P46, , and
A all read περί instead. 118 The other is at 1:11a where  reads γάρ, while P46,
A, and B all read δέ instead. 119 At any rate, the authorial reading counts
indicate that the base of the stemma – where the textual tradition of Galatians
begins to diverge – should be located in the middle, between the places where
P46–B and –A join the stemma, as shown in FIG. 12.

Ω 34

32
ω
26 29
• •
8 25 26 23
P46 B A ℵ

FIG. 12: Oriented Stemma (19) with Counts of Authorial Readings

On this stemma, the starting point of the extant textual tradition is the arche-
type ω, located between the P46–B group and the -A group. Because the
archetype text ω differs from the original text as designated by the Nestle-
Aland edition in two places, there are also two changes between the earliest
form of the text Ω and the archetype ω. These differences suggest that either
the identification of the earliest form of the text was erroneous (indeed, it will
be argued in Chapter 3 that the authorial reading of 1:11a is actually δέ) or
that a scribal change was so common as to have been committed multiple
times independently (as argued in Chapter 4 for the variation unit at 1:4b).
This simplified example shows how cladistics can be used to construct a
plausible stemma of the textual tradition of four witnesses to the text of the
first chapter of Galatians. Of course, as the number of witnesses and the
number of variation units expand, it quickly becomes impossible to be done
by hand, so it is necessary to implement the cladistic method in computer
software. There are many existing programs that can perform cladistic
analysis and some have even been applied to manuscript traditions. 120

118
Naturally, in the full dataset the numbers work out differently. See Chapter 4 for a
more detailed discussion of this variation unit.
119
In Chapter 3, it will in fact be argued that δέ should be established for the critical
text.
120
E.g., Spencer et al., “Greek Vorlage.”
 2.3 Cladistics in Action: A Simplified Example 73

It should be acknowledged, however, that after a certain point not even

computers can search every possible stemma within a reasonable amount of
time, because the total of number of possible unoriented stemmata quickly
becomes too great to search exhaustively. For example, the number of
possible unoriented, strictly bifurcating genealogical trees with no mixture is
given by the double-factorial formula (2n − 5)!!. 121 For three witnesses, there
is (2×3 − 5)!! = (6−5)!! = 1 possible unoriented strictly bifurcating tree. 122
For four witnesses, there are (2×4 − 5)!! = 3!! = 3 such trees, which
corresponds to unoriented stemmata (17), (18), and (19) in FIG. 10. 123 For
five witnesses, the number climbs to 15. For six, the number is 105. For nine,
the number exceeds two million. By 15 witnesses the number exceeds a
trillion, and they increase astronomically from there. 124 With 94 witnesses,
the number of such stemmata with no mixture is 1.93 × 10¹⁶⁶. This number is
unfathomably immense. By way of comparison, the number of atoms in the
universe has been estimated to be a mere 10⁸⁰, less than the square root of the
number of such stemmata. 125
As a result, computer programs that implement cladistics must find their
results by a heuristic search. 126 Heuristic searches are designed to start with a
relatively good estimate of the result and attempt to improve it with minor
modifications to the best current estimate of the result. The initial estimate is
usually determined by adding one witness at a time to a stemma. 127 This ini-
tial estimate is then rearranged by switching branches around to see if a better
result can be found. 128 After these possibilities are exhausted, then all the
variants are randomly reweighted, and various rearrangements of the stemma
are retested. 129 By use of such heuristic searches, computer programs can find
acceptable results within a reasonable amount of time (e.g., less than a month
of constant computer time).
Therefore, it must be kept in mind that the result of a cladistic analysis of
the data by any computer program does not necessarily represent the actual

121
Felsenstein, Inferring Phylogenies, 24–25. The expression (2n – 3)!! is equal to (2n –
3)! / 2n–1(n–1)!, where the factorial operator (!) on a number n indicates the product of all
the numbers from n down to 1.
122
If direct links between witnesses are permitted, the number rises to 7.
123
With direct links and the permissibility of more than three branches per node, the
number rises 20.
124
A table of these numbers is found on Felsenstein, Inferring Phylogenies, 24.
125
E.g., John D. Barrow, The Constants of Nature: The Numbers that Encode the Deep-
est Secrets of the Universe (New York: Vintage, 2004), 118.
126
See generally Felsenstein, Inferring Phylogenies, 37–53.
127
This is called a “greedy algorithm” (Felsenstein, Inferring Phylogenies, 38).
128
Felsenstein, Inferring Phylogenies, 38–51, details several ways to rearrange phylo-
genetic trees.
129
Felsenstein, Inferring Phylogenies, 51–52.
74 Chapter 2. Materials and Methods

genealogical history. First, the history can be messy and there is no guarantee
that the actual process that generated this history did so with the least number
of changes. From an epistemological perspective, however, the stemma with
the least number of changes explains the textual evidence with fewer aux-
iliary hypotheses than any of its competitors. Thus, even if a computer pro-
gram found the maximally parsimonious stemma, that stemma is only pre-
sumed, in the absence of countervailing evidence, to be the stemma that
represents the history of the text. Second, once exhaustive searches become
impossible due to the astronomical numbers of different stemmata, the result
of a computer program can only be thought of as the “best found” stemma,
not the “best possible” stemma. It is always conceivable that with additional
computer time a better result might be found, though it may take months or
years to find one.
With the practical impossibility of finding the ideal genealogical tree,
practitioners of cladistics have become more interested in grasping how well
the trees that they find are supported by the evidence. In particular, they are
interested in understanding what parts of the tree are strongly supported by
the evidence and what parts are poorly supported by the evidence. For this
reason, computational biologists have devised various tests to gauge the
amount of uncertainty in their results. 130 A common test for this purpose is
the “bootstrap test,” which works by randomly resampling the variants in the
original dataset to produce a series of related datasets known as “bootstrap
replicates.” From these bootstrap replicates, a set of genealogical trees are
generated and compared against the tree under test. If a group of witnesses in
the tree under test is supported by 70% of the trees among the bootstrap
replicates, then that group is considered to be well supported. On the other
hand, groups in the tree under test that are supported by fewer than 70% of
the bootstrap replicates are considered to be poorly supported.
The heuristic search and bootstrap testing principles just described are
standard features found in available, off-the-shelf cladistics software
packages. What they do not do, however, is account for contamination. As a
result, it became necessary for this study of the text of Galatians and its his-
tory to implement these principles in special-purpose software in a way that
takes contamination into account, as described in the following section.

2.4 Contamination

Standard cladistics computer packages make several assumptions that are out-
of-place for manuscript filiation. For example, most computer packages
assume a strictly dichotomous view of descent, in which each ancestor has

130
See generally, Felsenstein, Inferring Phylogenies, 335–363.
 2.4 Contamination 75

exactly two immediate descendants. 131 Moreover, they also assume that every
ancestor is lost, or, in other words, that no extant organism is the ancestor of
any other organism being studied. The most serious assumption, however, is
that lineages only diverge, never converge. All of these assumptions are prob-
lematic when applied to manuscripts: manuscripts may have more than two
copies made from them; some extant manuscripts are known to be exemplars
of other manuscripts, and contamination is sufficiently common that it cannot
be ignored. In fact, mixture is the major problem when studying the text of
Paul, and off-the-shelf computer packages produce strange results, such as
the Western text being a late and highly derived member of the Byzantine
text.
Nevertheless, various approaches have been attempted among computa-
tional biologists in adapting cladistics to handle mixture, hybridization, hori-
zontal gene transfer, and other forms of what they generically call “reticula-
tion.” Some of these ideas are unfeasible, unfortunately. For example, some
methods first generate a family tree assuming no mixture and attempt to add
links to account for reticulation. 132 The problem with this approach is that
contamination distorts the tree before the links are added, even reversing the
direction of development in some of the branches. Merely adding reticulating
links without fixing the reversed branches will not correct the distortion.
Another approach is something called “reduced median networks,” 133 which
is too restrictive in the input it accepts, being limited to binary characters
with no missing data, 134 and unrealistic in its output – in one case, generating
8517 hypothetical manuscripts based on an original set of 82 manuscripts. 135
A different approach proposed by Allan W. Dickerman is that of “hyper-
trees,” which are trees with additional, “reticulating” links added to connect
different lineages. 136 Under Dickerman’s hypertree idea, the additional link
represents a set of possible non-reticulating trees, and only one of those trees

131
This recalls Bédier’s complaint against the bifid stemma; cf. Ernest C. Colwell,
“Genealogical Method: Its Achievements and its Limitations,” in idem, Studies in Method-
ology in Textual Criticism of the New Testament (NTTS 9; Grand Rapids, Mich.: Eerd-
mans, 1969), 63–83 at 78–79.
132
So P. Legendre & V. Makarenkov, “Reconstruction of Biogeographic and Evolu-
tionary Networks Using Reticulograms,” Systematic Biology 51 (2002): 199–216; Gareth
Nelson, “Reticulation in Cladograms” in Advances in Cladistics 2 (Norman I. Platnick &
V. A. Funk, eds.; New York: Columbia University Press, 1983), 105–111.
133
So Matthew Spencer, Klaus Wachtel, and Christopher J. Howe, “Representing Mul-
tiple Pathways of Textual Flow in the Greek Manuscripts of the Letter of James Using
Reduced Median Networks,” Computers and the Humanities 38 (2004): 1–14.
134
Spencer et al., “Multiple Pathways,” 11.
135
Spencer et al., “Multiple Pathways,” 5.
136
Allan W. Dickerman, “Generalizing Phylogenetic Parsimony from the Tree to the
Forest,” Systematic Biology 47 (1998): 414–426.
76 Chapter 2. Materials and Methods

in the set is applied at each variation unit, but the tree is chosen to minimize
the number of changes in the hypertree. Because each reticulating link will
reduce the score for the hypertree, a stopping point in adding links is not
immediately apparent, so Dickerman recommended imposing a cost for every
additional link until no more can be added. The major problem with
Dickerman’s approach, however, is that the calculation of the states of the
hypothetical ancestors is very expensive, and, in fact, computationally
tractable only for binary characters – those with exactly two states. This is a
significant limitation given the number of manuscripts, variation units, and
variant readings of interest to textual critics.
The approach of this study builds on Dickerman’s hypertree concept. Like
Dickerman’s hypertrees, a slight penalty is imposed for each case of mixture,
on a “pay as you go” system. In order to add a reticulating link, a link must
first be removed somewhere else in the stemma, either by taking out some
other reticulating link or by forcing some ancestors to have more than two
descendants. Nevertheless, Dickerman’s suggestion that there should be a
cost for each case of mixture is a good one, and a “reticulation cost” is
assigned to each case of contamination to keep the mixture down to a man-
ageable amount, that is, to account for the mixture that could distort the
stemma.
Setting an appropriate reticulation cost, however, is theoretically challen-
ging. On the one hand, the cost should not be set so high as to be unable to
find even the most significant cases of contamination. On the other hand, the
cost should not be so low so as to eliminate any coincidental variation from
the stemma by attributing them to contamination. Coincidence and
contamination are two competing hypotheses for non-stemmatic agreements
in error, but the goal is to produce a workable stemma, not to account for
every apparent agreement in error in the textual history. As a result, con-
structing a stemma does not need to generate reticulating links for all possible
cases of mixture, only to generate those links whose absence would distort
the shape of the stemma. Low levels of mixture show up in the results as
another form of accidental coincidence, and they can be identified by looking
at the final set of changes made to each manuscript. Based on my earlier work
on the Caesarean text in Mark, I have discovered that the most workable way
to estimate an appropriate cost is by setting it to the average number of
introduced variants per manuscript in the stemma. 137 For manuscripts of
Mark, this formula for the reticulation costs produced a stemma that coheres
well with other researchers’ views of the history of the Markan text. 138

137
Stephen C. Carlson, “The Origin(s) of the ‘Caesarean’ Text” (paper presented at the
annual meeting of the Society of Biblical Literature, San Antonio, 20 Nov. 2004).
138
For example, it was able to detect that 118 is a mixed Byzantine member of family 1,
long known by Lake and reconfirmed by Amy S. Anderson, The Textual Tradition of the
 2.4 Contamination 77

It is also necessary to avoid the problem that befell Streeter. When there is
a lot of mixture in the witnesses and most of the variation units have only two
variant readings, there is a tendency for the readings to sort themselves into
two different ancestors, which become the putative ancestors of a large
number of mixed manuscripts. This set of affairs is illustrated in a portion of
a stemma depicted in FIG. 13, in which hypothetical ancestors α and β are
parents of mixed manuscripts Α, Β, Γ, Δ, Ε, Ζ, and Η where one of the
readings would be assigned to ancestor α and the other reading to ancestor β.
In a data set involving the New Testament witnesses, ancestor α could end up
with all the Byzantine readings, and ancestor β would be assigned the non-
Byzantine variants present in only two of the manuscripts Α, Β, Γ, Δ, Ε, Ζ,
and Η.

Α Β Γ ∆ Ε Ζ Η

FIG. 13: The “Streeter Problem” of Contamination

This configuration is not unlike Streeter’s 1924 vision of the Caesarean text
and subject to the same criticisms. A text is not merely a collection of diverse
readings but a sequence of readings embodied in a manuscript. Thus, the evi-
dentiary basis for a hypothetical text must be the texts of its descendants. Al-
though unmixed descendants provide the requisite sequential evidence for the
reconstruction, mixed descendants do not – they can only substantiate the
readings that the other parent did not supply. If, however, all of the descend-
ants of a hypothetical ancestor are mixed, there is no evidence of a connected
sequence of readings in the descendants left for the reconstruction. In fact, the
problem is still evident with a single unmixed child and multiple mixed
children, because any two of the mixed children can out-vote the attested text
of the unmixed child. For this reason, a constraint is imposed that rejects any
stemma in which the number of mixed children of a hypothetical ancestor
exceeds the number of non-mixed children. Extant manuscripts are not
subject to this constraint since we have the direct evidence for their text.

Gospels: Family 1 in Matthew (NTTS 32; Leiden: Brill, 2004), 101–102. It also found
support for Larry Hurtado’s observation that Θ and 565 seem to be more closely related to
Codex Bezae (D) than to W. See Larry W. Hurtado, Text-Critical Methodology and the
Pre-Caesarean Text: Codex W in the Gospel of Mark (SD 43; Grand Rapids, Mich.:
Eerdmans, 1981), 30–43, summarized at 44 and 86.
78 Chapter 2. Materials and Methods

This constraint upon mixture is consistent with the way contamination is

handled by classical textual critics working with open traditions. For exam-
ple, in his study of the text of Cyprian, Maurice Bénevot discussed the
importance of finding opposing pairs, triples, or larger sets of manuscripts,
which he defined as “when, though each varies often from the resultant text,
their variants agree together only very rarely.” 139 Michael P. Weitzman, ap-
proving of Bénevot’s approach, called such opposing manuscripts poles, and
explained that a stemma can be constructed even in the presence of con-
tamination by assuming that the archetype has only two poles, an assumption
he called “far less restrictive than that of total absence of contamination.” 140
Like the maximum parsimony principle that chooses the stemma with the
least amount of coincidental error, the “bipolarity assumption, reasoning from
consistent pairs to parent groups, chooses whichever possibility posits the
least contamination.” 141 The mixture constraint of this study finds the oppos-
ing poles of Weitzman’s bipolarity assumption within a maximum parsimony
evaluation framework.
Accordingly, these principles have been incorporated into a computer pro-
gram developed for this study to search for unoriented stemma that minimizes
the number of textual changes (the maximum parsimony principle), while at
the same detecting some degree of mixture at a level that threatens to distort
the shape of the stemma. 142

139
Maurice Bénevot, The Tradition of Manuscripts: A Study in the Transmission of St.
Cyprian’s Treatises (Oxford: University Press, 1961; repr. Westport, Conn.: Greenwood
Press, 1979), 148.
140
Michael P. Weitzman, “The Analysis of Open Traditions,” Studies in Bibliography
38 (1985): 82–120 at 102.
141
Weitzman, “Open Traditions,” 104–105.
142
The program was written in the C programming language and consists of 6548 lines
of code in 21 source files.
 2.4 Contamination 79

FIG. 14: Best-Found Unoriented Stemma of 94 MSS of Galatians

80 Chapter 2. Materials and Methods

2.5 A Proposed, Unoriented Stemma for Galatians

In this study, ninety-four witnesses have been collated over 1624 variation
units. Of these variation units, 827 consist of singular readings, and the re-
maining 797 variation units have at least two witnesses agreeing against all
the other witnesses. These collated data were inputted into a computer pro-
gram written by the author of this study in accordance with the principles de-
scribed in this chapter. After inputting the collated data and running the com-
puter program for about two weeks, the best found stemma with the minimal
cost turned out to be one with a cost of 4032 and is depicted in FIG. 14. This
stemma has a reticulation cost parameter of 25, which means that, for mixture
to be found, the contaminating source must contribute at least 25 variants.
It must be kept in mind that this “best-found” stemma, like the result of
any cladistic analysis of the data, does not necessarily represent the actual
genealogical history of these manuscripts of Galatians. Indeed, the actual
history is almost certainly messier. But it does represent the stemma best
supported by the textual evidence of the ninety-two witnesses supplied to the
program. From an epistemological perspective, the best-stemma explains the
textual evidence with fewer auxiliary hypotheses than any of its competitors
about the sources of the textual variants. In other words, the “best-found”
stemma may not be the “best” in terms of correspondence to the actual his-
tory of the text, but it is better than its competitors on the basis of the same
textual evidence. As such, the “best-found” stemma constitutes a prudent
working hypothesis for the history of the text of Galatians and a starting point
for further stemmatic analysis.
This unoriented stemma has a general linear progression, from which the
manuscripts and groups diverge. On one side are the so-called “Western”
witnesses, consisting of the Greek-Latin diglots D*, DC, F, and G, as well as
the Old Latin and Vulgate witnesses b, d, f, and vg. Some of these Western
witnesses are mixed, and estimates of their mixture percentage are put next to
their names. Not surprisingly, the corrected text of Codex Claromontanus
(DC) has been determined to be a mixed witness of D*, but it was drastically
corrected to a Byzantine text, such that nearly two-thirds of the readings that
differ between its parent D* agree with the Byzantine text. The Vulgate was
also found to be mixed, with 80% Old Latin and 20% proto-Alexandrian
sources. This result generally coheres with the conclusions of scholarship that
“Jerome reacted against the predominance of the Western type of text, and
deliberately sought to orientate the Latin more with the Alexandrian type of
text.” 143 Latin manuscript f is also a mixture, being 91% Vulgate and 9% in
agreement with F, the Greek text on the other side of the page.

143
Bruce M. Metzger, The Early Versions of the New Testament: Their Origin, Trans-
mission, and Limitations (Oxford: Clarendon Press, 1977), 359.
 2.5 A Proposed, Unoriented Stemma for Galatians 81

Closest to the Western witness is the text of Marcion (labeled “Marc”), fol-
lowed by a cluster including P46 and B. Within this cluster, the Syriac
version known as the Peshitta appears to be 20% contaminated by B. In this
regard, the Peshitta may be compared with the Vulgate – both are sanctioned
translations that seem to incorporate some older readings into their text base.
Next to P46 and B on the main line of the stemma is a collection of
witnesses headed by  (labeled “01”) and 33. Both these manuscripts are
considered “Alexandrian” and their location on the stemma next to P46 and B
shows that their text is very important. The Alexandrian manuscript 33 has an
ancestor that contributes about 55% of the readings to a mixed manuscript
1175. According to the stemma, 1175 itself contributes about 44% (four-
ninths) of the contaminated readings to the ancestor of a cluster consisting of
1319 and 1573.
At this point on the stemma, the text diverges into three directions. One of
them is a source of about 20% of the Vulgate readings. Another direction is a
group of what is generally known as “secondary Alexandrian” manuscripts
with A, C, P, and 1241S plus various others whose text has been conformed
to the Byzantine standard in varying percentages. In particular, this group
includes manuscript 88 (39% Alexandrian, 61% Byzantine), 104 (56% Alex-
andrian, 44% Byzantine), 1837 (49% Alexandrian, 51% Byzantine), and 2464
(48% Alexandrian, 52% Byzantine). The result that these later manuscripts
appear to be Byzantinized indicate that the Alexandrian text did not so much
cease to be copied but that its copies were steadily and progressively cor-
rected to the dominating Byzantine text. Due to the implementation of Weitz-
man’s bipolarity assumption for contamination as described above, it should
be recognized that the failure to detect mixture for 1241S only means that this
manuscript is a secondary Alexandrian pole; it is possible that it too was
mixed with Byzantine readings but its status as a pole means that it is merely
the least Byzantine contaminated among these later secondary Alexandrian
witness. 144
The third direction of divergence is toward the Byzantine text. Next in this
direction is a group of related manuscripts that include 1739 and 1881, plus
the partially Byzantinized 6. The leading member of this group was exten-
sively studied by Günther Zuntz in his examination of the text of Paul. 145
Zuntz considered 1739 to be an excellent manuscript, on par with P46, B, and
, but, according to the best-found stemma, 1739 is still very good but it is a
sister to the common ancestor of the secondary Alexandrians A and C. Next

144
The fact of that three of these mixed Alexandrian-Byzantines–88, 104, and 1837–all
seem to be contaminated from the same place in the Byzantine text near 1836, which has
some Alexandrian readings too, suggests that the other “pole” in accordance with Weitz-
man’s bipolarity assumption is probably 1836.
145
Zuntz, Text, 68–83.
82 Chapter 2. Materials and Methods

in the direction toward the Byzantine text is the Athos Codex Ψ, followed by
a group of “Syrian” witnesses including the text of John Chrysostom, the
Peshitta, the Harklean, and its Greek allies 1505, 1611, and 2495. 146 On one
side of the stemma are the bulk of the witnesses in this study and they
correspond to the Byzantine text.
Up to this point, the discussion of the results of the computer-analysis of
the text of Galatians has revolved around the best-found stemma and the con-
tamination that has been detected within the textual tradition. The next step is
to subject the stemma to a bootstrap test to see which groups identified in the
best-found stemma are strongly supported by the textual evidence and which
groups are poorly supported. The reason for this step is that not every branch
in the best-found stemma is supported by substantial evidence and those
branches with insubstantial support should not be presented as if they are one
par with those with substantial support. Accordingly, a bootstrap test with
1000 replicates was run against the best-found stemma and the groups with at
least 70% support in the replicates were determined. This particular level of
support – 70% – is to some extent conventional, but it is a common con-
vention in cladistics.
Based on a comparison of the best-found stemma with the result of the
bootstrap analysis, it appears that the groups of related textual witnesses iden-
tified in the best-found stemma have generally held up with two major
exceptions. The first exception is the relationship between  (labeled 01) and
33. According to the best-found stemma,  and 33 appear to be more closely
related to each than to any other witness, except for 1175 and 1319–1573,
which were contaminated by a text close to 33. According to the bootstrap
analysis, however, this relationship between  and 33 dissolves. Further scru-
tiny of the evidence initially supporting  and 33 indicates that this outcome
of the bootstrap is reasonable. Just four textual variants initially supported a
close relationship between  and 33, which were deemed insufficient by the
bootstrap test. Now, the bootstrap test works only with the numeric level of
support at particular places in the stemma, and it did not find these four
variants to be sufficient to warrant the conclusion of a close textual relation-
ship between these two witnesses. The numerical bootstrap test, of course,
knows nothing about the Greek language, but further inspection of these
variants shows the wisdom of this result of the bootstrap test. These variants
are: Gal 3:3 ἐπιτελεῖσθαι for ἐπιτελεῖσθε, 4:18 ζηλοῦσθε for ζηλοῦσθαι, 6:9
θερίσωμεν for θερίσομεν, and 6:10 ἔχωμεν for ἔχομεν. 147 All of these variants
feature orthographic or phonetic variants occasioned by mergers in Greek

146
These three Greek manuscripts had already been identified to be related to the
Harklensis Syriac version by Aland and Juckel, Syrischer Überlieferung, 507–514. They
also included 2138, but its text appears to be Byzantine in Galatians.
147
All of these variants are discussed in greater detail in Chapter 4.
 2.5 A Proposed, Unoriented Stemma for Galatians 83

phonology: by the second century, the diphthong αι came to have the same
sound as ε, 148 and the length distinction between ο and ω was lost. 149 As a
result, these four changes could well be coincidental and thus not strong
evidence of a close genetic relationship. In other words, the mathematical
bootstrap test independently agrees with the philology of the variants, even
though the test knows nothing about the philology.
The other place in the stemma where the groups hypothesized by the best-
found stemma dissolved in the light of the bootstrap test is found within the
Byzantine text. According to the best-found stemma, there are many groups
that are only diagnosed by just a few textual variants. The bootstrap test,
however, concludes that the evidence supporting these Byzantine subgroups
is numerically insufficient. For example, the best-found stemma holds that
there is a Byzantine subgroup consisting of 056, 6 (mixed), 876, 1352, 1448,
1646, and 1735, united by their agreement in error at Gal 6:12. This lone
agreement in error is numerically insufficient to support this group and
further inspection of the variant (an essentially orthographic change of
διώκονται for διώκωνται) confirms this conclusion as well.
As a result of the bootstrap analysis, much of the early internal structure of
the Byzantine text disappears, leading to twenty separate lines of descent
from the common ancestor of the Byzantine text within this dataset. This
result has both positive and negative aspects. On the positive side, this means
that the text of the common ancestor of the Byzantine text – the Byzantine
prototype – can be reconstructed fairly precisely merely on the basis of exter-
nal evidence on majority-text principles. Each of the twenty separate lines of
descent would contribute a vote to the establishment to the text of Byzantine
text. The reconstructed Byzantine prototype is presented in Appendix B. Most
of the Byzantine manuscripts differ from this prototype in about a dozen or so
places, but, as it turns out, one of these witnesses – the eleventh century
manuscript 1854 – has a text that is almost identical to that of the Byzantine
prototype, differing in just four places (the addition of με after καλέσας in
1:15, the addition of the article before θεόν in 4:9, the spelling διώκωνται in
6:12, and the omission of Ἰησοῦ before Χριστοῦ in 6:17). Nevertheless, the
Byzantine prototype must be centuries older than 1854; the prototype is, after
all, the imputed ancestor of the ninth century uncials K and L as well. What
this shows is that the transmission of the Byzantine text was sufficiently
controlled, at least from the ninth to the eleventh centuries, that 1854 can
accurately reflect the text of a few centuries earlier.
This apparent control over the transmission of the Byzantine text relates to
the negative side of the bootstrap results. There does not appear to be enough

148
W. Sidney Allen, Vox Graeca: The Pronunciation of Classical Greek (3d ed.; Cam-
bridge: Cambridge University Press, 1987), 79.
149
Allen, Vox Graeca, 94.
84 Chapter 2. Materials and Methods

evidence sufficiently patterned within the text of Galatians to resolve the

early history of the Byzantine text subsequent to the Byzantine prototype. It
is possible that this problem may disappear when additional textual evidence
is brought to bear on the Byzantine text, for example, the text of another letter
in the Pauline Corpus such as 2 Corinthians. Nevertheless, the textual critic
ought to keep the possibility open that the transmission of the Byzantine text
may have not have operated in accordance with the assumption of in-
dependent transmission common to both stemmatics and cladistics. As Silva
Lake discovered when she was collating manuscripts of Mark, those re-
sponsible for copying the text may have been so familiar with the standard,
ecclesiastical text that they were able to detect deviations from that standard
text in their exemplars and correct for them. 150 The implication of this scribal
familiarity with the standard, Byzantine text is that scribes were able to
obliterate the agreements in error that are necessary to establish stemmatic
subgroups. 151 Though the stemmatic software has some provisions for de-
tecting contamination, it also appears that the level of contamination within
the controlled transmission of the Byzantine text was so low (less than the re-
ticulation cost of 25 variants) that it does not register on the stemma. In other
words, the Byzantine text appears to be a highly complex interconnected
network of texts that neither stemmatics nor cladistics can resolve with any
precision. Fortunately, the results of the bootstrap test suggest that this
weakness in the ability of stemmatics to reconstruct the history of the text
appears to be limited for the carefully controlled Byzantine text. Outside of
the Byzantine text (with the exception of  and 33), the bootstrap test
indicates the discovered manuscript groups are well supported by the textual
evidence.
Indeed, the unoriented stemma of selected witnesses to the text of
Galatians is comparable in many respects with Hort’s influential theory of the
text. 152 According to Hort, there are four kinds of text in Paul, and, with the
prevalence of mixture, these textual groups are overlapping to a large extent.
Referring to FIG. 15, ovals representing the textual groups identified by Hort
are overlaid upon the unoriented stemma of Galatians, showing that Hort’s
theory of the text and the unoriented stemma are roughly compatible.

150
Perhaps this familiarity comes from liturgical use, as with lectionary texts, but no
lectionary texts were collated for this study.
151
This practice could also account for the convergence of the secondary Alexandrian
branch of the text toward the Byzantine.
152
B. F. Westcott and F. J. A. Hort, Introduction to the New Testament in the Original
Greek (repr. Peabody, Mass.: Hendrickson, 1988). As before, references to this work are
cited by the abbreviation HORT .
 2.5 A Proposed, Unoriented Stemma for Galatians 85

Chrys-hark-1611

Byz
Marc 33-×1175 Syrian •
(P46)-B (1739)
Neutral
• • • • • •

•
Western
ℵ Alexandrian
Ψ
• A-C-P-1241S

D* F-G d-b-×vg
FIG. 15: Unoriented Stemma of Galatians Compared with Hort's Textual Groups

In particular, Hort identifies a “Western” text, preserved mainly in the Greek-

Latin diglots D and G. 153 Famously, Hort identifies two “Neutral” witnesses,
 and B, and holds that they are primarily but not entirely free from influ-
ences from other textual groups. 154 Specifically, Hort has pointed out that in
B’s text of Paul, “there is an unquestionable intermingling of readings de-
rived from a Western text nearly related to that of G3.” 155 This intermingling
is indicated in FIG. 15 by the overlapping of the Western and Neutral ovals.
Now, Hort has interpreted this intermingling in terms of contamination, but
the unoriented stemma suggests that this commonality may be due to genea-
logy: some of the B-Western agreements could instead go back to a common
ancestor of B and the Western manuscripts. 156 As for , it too is not entirely
free from what Hort has termed “Alexandrian corrections,” 157 and these are
indicated in FIG. 15 by the overlapping of the Neutral and Alexandrian ovals.
According to Hort, the “Alexandrian text” is not pure, but contains mixtures

153
HORT 149. Hort believes that F was a copy of G and discounted out, but, according
to this study as explained in the following chapter, F is seen as an independent but inferior
copy of one of G’s close ancestors.
154
HORT 150–151.
155
HORT 150. “G3” is Hort’s symbol for this study’s “G” (Gregory-Aland 012).
156
As suggested by Zuntz, Text, 153.
157
HORT 153.
86 Chapter 2. Materials and Methods

of other texts, the most dominant one being the “Syrian” text. 158 The un-
oriented stemma is thus generally compatible with Hort’s theory of the text.
The same groups that Hort identified can be located on the stemma and the
transitional areas, where one group blurs into another, are also there. The
main difference is that Hort usually appeals to mixture to account for these
transitional texts, while the stemma implies that the reason for them is
genealogical.
Hort’s “Neutral” text is of particular interest because it is the text that Hort
has identified, based on internal evidence, to contain the least amount of scri-
bal corruptions. 159 The importance of a text with the least scribal corruption is
that it should resemble an authorial text the most and can therefore be chosen
as the base of an oriented stemma. Thus, if Hort’s judgment is correct as to
the relative purity of the Neutral text based on the internal evidence of
readings, then it appears that the base of the stemma should be located some-
where in the area within the oval for the Neutral text. In other words, if the
stemma were to be oriented based on Hort’s judgment of the text, then the
starting point of the transmission of the text should be somewhere near the in-
ternal points where P46–B or  and 33 join the stemma. This point is not far
from the oldest portions of the stemma, especially considering that Hort did
not have the benefit of the later discovered P46. The two oldest witnesses to
the text of Galatians are P46 and Marcion, so the oldest parts of the stemma
would be near the internal points where Marc(ion) and P46–B join the
stemma.
This proposed unoriented stemma for Galatians is an important but
beginning step in this study of the text of Galatians and its history. The next
steps are to orient the stemma based on internal evidence (Chapter 3),
examine the text at the base of the stemma (Chapter 4), and then to
investigate the history of the text and its variation (Chapter 5).

158
HORT 151–155.
159
HORT 126–130.
 Chapter 3

Orienting the Stemma of Galatians

3.1 Introduction

The next step in this study of the text of Galatians and its history is to orient
the unoriented stemma that was generated in Chapter 2. At this stage in the
investigation, the stemma is unoriented, which means that it does not indicate
any particular starting point where the textual tradition of Galatians began. It
does show however which witnesses and which groups of witnesses are more
closely related and which are more distantly related. The goal of this chapter
is to determine which portion of the unoriented stemma should be the starting
point (or “base”) of the stemma. The rule for identifying the base is that it is
the part of the stemma that has the most authorial readings, as determined by
internal evidence.
Owing to the sheer scale of the textual tradition, particularly within the
Byzantine text, it is expedient to concentrate on the portions of the unoriented
stemma that are the most promising for the base of the stemma. This study
proposes three independent but converging approaches for pinpointing these
promising areas of the stemma for further examination: Hort’s theory of the
Neutral text, the oldest branches in the stemma, and the part of the stemma
closest to the Nestle-Aland critical text.
On the following page, FIG. 16 depicts the shape and linkage pattern for a
stemma of selected witnesses of Galatians that was generated according to the
cladistic method described in the previous chapter, with four internal branch-
es labeled for future reference. According to Hort’s theory of the text (de-
scribed in some detail in the previous chapter), the two most important, “neu-
tral” witnesses are  and B. In other words, Hort’s neutral text corresponds to
the area of the stemma between  and B, specifically inner branch a and inner
branch b. Thus, if Hort’s theory of the text still has some validity, then the
textual changes along the inner branches a and b ought to be scrutinized.
Another approach is to use the oldest surviving portions of the stemma.
The two oldest substantial witnesses of Galatians are P46 and Marcion. Their
age indicates that the portions of the stemma that ought to be scrutinized are
inner branch d and inner branch b.
A third approach is to identify the portion of the stemma that is the closest
to the Nestle-Aland critical text. The justification for this approach is that the
88 Chapter 3. Orienting the Stemma of Galatians

Nestle-Aland critical text constitutes the locus of intense text-critical activity

over the past century. If this text approximates the authorial text of Galatians
to a large extent, then the part of the stemma closest to its text would be very
promising. As it turns out, the place in the stemma closest to the Nestle-Aland
critical text is the node at which  and 33 join the stemma, flanked by the
Eastern branch on one side and by the Western branch and P46–B cluster on
the other side. This criterion points to inner branch d and inner branch b as
the most fruitful places to scrutinize for the base of the stemma.

Chrys-hark-1611

Byz
Marc 33-×1175 •
P46-B 1739
Eastern
b Branch
d a c
• • • • • •
Western
Branch
• ℵ Ψ
• A-C-P-1241S

D* F-G d-b-×vg
FIG. 16: Unoriented Stemma of Selected Witnesses to Galatians

These three different approaches toward finding the base of the stemma –
Hort’s “Neutral” text, the oldest portion of the stemma, and the point of the
stemma closest to the Nestle-Aland text – all converge on an area of the un-
oriented stemma somewhere between  and B, comprising inner branches a,
b, c, and d. This area, then, is the location on the stemma that is provisionally
the most important for constructing a critical text of Galatians. Accordingly,
the construction of an eclectic, critical text for Galatian in this study should
focus on the variants that change along the course of inner branches a, b, c,
and d. Variants that arise outside of this area are not ignored but considered
in the following chapters on examining the text and studying the history of
the text.
Accordingly, each of the inner branches are considered in turn as follows.
First, the witnesses on the opposing sides of the inner branch are used to
 3.1 Introduction 89

identify the variants that change between the opposing sides. For example,
inner branch a defines an opposition between the Western branch plus the
P46–B cluster, on one hand, and the Eastern branch plus  and 33 on the
other hand. Each textual variant over which the two sides of the inner branch
are opposed is then analyzed, and the analysis of each variant has a common
plan. First, the variation unit is underlined and presented within a relevant
context. Then, a table of the readings attested for the variation unit is pre-
sented, so that the external evidence can be seen at a glance. This table of
external evidence has subdivisions for the Western and Eastern branches, as
well as columns for their closest relatives near the preliminary base of the
stemma. Specifically, these “basic” witnesses include, on the one hand, P46
and B, which are closer to the Western branch than the Eastern, and, on the
other hand, , 33, and the non-Byzantine portion of 1175, which are closer to
the Eastern branch than the Western.
In addition to the manuscript and versional witnesses, nine editions of the
text of Galatians (including this study) are cited to give a sense of the scholar-
ly support for the readings. 1 These editions are generally chosen for their
editorial independence and historical importance. 2 These editions and their
respective sigla, in reverse chronological order, are: Carlson 2015 (SCC),
Holmes 2011 (SBL), Robinson-Pierpont 2d ed. 2005 (RP), Nestle-Aland 27th
ed. 1993 (NA), von Soden 1911 (S), B. Weiss 1902 (BW), Westcott-Hort 1881
(WH), Tregelles 1869 (T), and Lachmann 1850 (L). Where these editions pre-
sent more than one reading, the reading in the main text is indicated by a
superscripted T, and the reading in the margin is indicated by a superscripted
M. For bracketed readings in the critical editions, the full reading within the
bracket is designated as the main text reading, while the omission of the
bracketed reading is designated as the marginal reading.
Following the table of external evidence is an exegesis of each variant
reading under consideration. This exegesis looks at both the traditional se-
mantic and syntactic meaning, and – in a new contribution to the field of New
Testament textual criticism – the pragmatic considerations related to the
placement of emphasis on particular constituents of the sentence. 3 Although
the CBGM was not chosen as the text-critical method for this study, because
it does not infer a history of the text, its notion of “coherence” was found use-
ful and occasionally applied, albeit in stemmatic terms. After the exegesis of

1
The judgments of various commentators of Galatians, however, are cited in the dis-
cussion of the variants, where appropriate.
2
Perhaps the only striking omission on this list is Tischendorf’s great edition, which is
rightly more famous for the quality of its apparatus than the quality of its text.
3
In general, the principles discussed by Stephen H. Levinsohn, Discourse Features of
New Testament Greek: A Coursebook on the Information Structure of New Testament
Greek (2d ed.; Dallas: SIL International, 2000), have been found to be particularly helpful
for assessing the prominence of sentence constituents.
90 Chapter 3. Orienting the Stemma of Galatians

the variants, the internal evidence is then evaluated in accordance with

reasoned eclecticism, taking into account the relative contributions of the
intrinsic probabilities and the transcriptional probabilities. For purposes of
this study, the criteria and limitations of the internal evidence, especially the
harder reading, have been followed as they are outlined by Bruce M.
Metzger, 4 with one important exception that has recently been subject to
particular scrutiny.
This exception involves the criterion of the shorter reading. According to
Metzger, the shorter reading should be preferred unless it can be shown that it
is the result of a mechanical scribal error (e.g. a skip of the eye from one set
of letters to another similar set) or if the longer reading appears superfluous,
harsh, or unorthodox. 5 Recent scholarship, however, has called this text-
critical canon into question. For example, James R. Royse has found that
omissions, especially of little words, were very common in his study of the
scribal habits among the six earliest and most substantial papyri. 6 He argues
that in many cases the longer reading ought to be preferred unless the longer
reading appears late, appears to be a harmonization, or appears to be a
grammatical improvement. 7 Due to this conflict over the status of the
criterion of the shorter (or longer) reading, this study is agnostic about the
value of mere length as diagnostic of a scribal error. As a result, decisions
about which variant to prefer had to rely on criteria other than the length of
the reading.

3.2 Examination of Variants for Inner Branch Oppositions

3.2.1 Inner Branch a

The following six variants are those where P46–B and the Western branch
oppose , 33, and the Eastern branch along inner branch a. In Hort’s theory
of the text formulated prior to the discovery of P46, this configuration of the
witnesses is due to contamination of B with Western readings. 8 In Zuntz’s
theory of the text, by contrast, Western agreements with B, especially where
supported by P46, are ancient, pre-Western readings. 9 Accordingly an exam-
ination of these readings based on internal evidence is not only helpful for

4
METZGER 12*–14*.
5
METZGER 13*. My characterization of “unorthodox” refers to Metzger’s more precise
but longer formulation as “contrary to pious belief, liturgical usage, or ascetical practice.”
6
James R. Royse, Scribal Habits in Early Greek New Testament Papyri (NTTSD 36;
Leiden: Brill, 2008), 703–736. This work is cited hereinafter by the abbreviation ROYSE.
7
ROYSE 735.
8
HORT 150–151, 166, 171, 224, 240–241.
9
E.g., ZUNTZ 158.
 3.2 Examination of Variants for Inner Branch Oppositions 91

orienting the stemma but it is also useful for adjudicating between Hort and
Zuntz on their theories of the text. 10

Gal 1:3 χάρις ὑμῖν καὶ εἰρήνη ἀπὸ θεοῦ πατρὸς καὶ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ
Var. Western Eastern EDS.
11
καὶ κυρίου ἡμῶν DFG P46 B 1739 1611 SCC SBL RP
d ×vg Byz M T
WH T L
καὶ κυρίου Chrys 12 WH
M

ἡμῶν καὶ κυρίου b  33 A P 1241S Ψ 13 T M

NA S BW WH T
? Marc C

This variation unit exhibits the divide between the Western branch and P46–
B on the one hand and the Eastern branch with  and 33 on the other hand,
though the Eastern branch itself is divided among the readings. The Western
branch, together with P46 and B, as well as half the Eastern branch read,
“grace to you and peace from God the Father and our Lord Jesus Christ,”
while the other half of the Eastern branch plus  and 33 read, “from God our
Father and the Lord Jesus Christ.” The former reading, “our Lord Jesus
Christ,” puts more prominence on the common lordship of Jesus Christ that
Paul shares with the Galatians, while the latter reading focuses more on the
common fatherhood of God. In other words, the emphasis of the former is
Christological, while the latter is theological.
The intrinsic probabilities relate to Paul’s usual expression as well as the
reading’s function within the context of the entire letter. To be sure, the “God
our Father” reading is identical to most other Pauline salutations, 14 but there
is little evidence that the greeting formula had actually become stereotyped
by the time Galatians was written. 15 The earliest letter, 1 Thessalonians,
shows a different greeting formula (1:1, merely “grace to you and peace”).
Even with a late dating of Galatians, the only undisputed letters with the

10
Hort’s leading example for Western contamination in B is Col 1:12, where B reads
καλέσαντι καὶ ἰκανώσαντι against D* F G καλέσαντι and P46  A C 1739 Byz (HORT
240). Hort did not know of P46, of course, but Zuntz points out that P46’s failure to con-
firm B’s reading shows that it can help assess B’s quality (ZUNTZ 41). In other words,
P46’s agreement with B on an apparent Western reading means that the reading is pre-
Western. (Hort’s other example is 2 Thess 3:4, at which P46 is unfortunately lacking.)
11
This reading also enjoys the support of P51vid, which was too fragmentary to be ana-
lyzed for the stemma.
12
Also attested by the Byzantine manuscripts: 547 910 999 1319 1424 1448 1799 1874
1891 2147 2423.
13
Also attested by the Byzantine witnesses 056 131 876 1573 1836 1837 2125 2424.
14
This is the reason given by those who favor reading ἡμῶν with God: BURTON 11;
BUSCEMI 4; METZGER 520: “the apostle’s stereotyped formula”; MATERA 32.
15
NET 859, argues that “in an early letter such as this one his [scil. Paul’s] regular style
was yet to be established,” but this argument assumes that Galatians is an early letter.
92 Chapter 3. Orienting the Stemma of Galatians

“God our Father” formula that can confidently be assigned to predate

Galatians are 1 Corinthians and possibly 2 Corinthians. 16 The rest of the
letters with the formula are too late to establish the existence of the pattern at
the time Galatians was composed.
Bernhard Weiss asserts that Paul never uses the pronoun ἡμῶν with an
anarthrous κύριος. 17 Although Weiss does not list the instances behind his
observation, it nonetheless holds for dozens of instances throughout both the
undisputed and disputed Pauline epistles, without a single counter-example in
the Nestle-Aland text or corresponding apparatus. 18 Somewhat complicating
the force of his observation, however, is that Paul also routinely omits the
article with κύριος when it is the second element of a compound joined by
καί with θεός (with or without the article) as the first element, especially in
the greeting formula. 19 Thus, there is a question of deciding which tendency
in Paul’s writing should dominate the other when ἡμῶν is used with κύριος as
the second element of a compound with θεός.
Contextually, the Christological focus of the placement of the pronoun as
“our Lord Jesus Christ” better fits the Christological theme of the letter. 20
Though strictly unparalleled in Paul, the placement of the pronoun ἡμῶν with
κυρίου instead of πατρός anticipates the importance of Christ in the argu-
mentation of the letter (e.g. 2:16). Galatians famously lacks a thanksgiving
section, which often foreshadows themes to be developed later in other
epistles, so the function of the typical Pauline thanksgiving is handled instead
by the overloaded introductory section (vv.1–5). 21 The placement of ἡμῶν
with κυρίου thereby anticipates the Christological emphasis of the letter.
Moreover, scribal transcriptional probabilities support the “our Lord Jesus
Christ” reading. The “God our Father” reading is likely to be scribal assimila-

16
For those who support an early date for Galatians, then even 1 and 2 Corinthians
would not predate Galatians. Second Thessalonians also shows “God our Father,” though B
D P 33 1739 omit the pronoun.
17
WEISS 99: “aus begreiflichem Grunde, n ir g en d s bei Paul κυρ. ημων ohne Artikel
steht.”
18
Specifically, Rom 1:4, 4:24, 5:1, 11, 21, 6:23, 7:25, 8:39; 15:6, 30, 16:18, 20; 1 Cor
1:2, 7, 8, 9, 10, 5:4, 9:1, 15:31, 57; 2 Cor 1:3, 8:9; Gal 6:14, 18; 1 Thess 1:3, 2:19, 3:11,
13, 5:9, 23, 28; cf. Eph 1:3, 17, 3:11, 5:20, 6:24; 2 Thess 1:8, 12, 2:1, 14, 16, 3:18; 1 Tim
1:2, 12, 14, 6:3, 14; 2 Tim 1:8.
19
Rom 1:3; 1 Cor 1:3; 2 Cor 1:2; Gal 1:3, Phil 1:2; 1 Thess 1:1; Phlm 3; cf. Eph 1:2,
6:23; 2 Thess 1:1, 2, 12.
20
On the other hand, BETZ 41 n.47 claims that having the pronoun with God the Father,
“establishes the proper hierarchical order of the Father, God, and his Son, the ‘Lord Jesus
Christ,’” but his argument is obscure, for the order is established by the word order, not the
pronoun.
21
For a recent discussion, see Robert E. Van Voorst, “Why is There No Thanksgiving
Period in Galatians? An Assessment of an Exegetical Commonplace.” JBL 129 (2010):
153–172.
 3.2 Examination of Variants for Inner Branch Oppositions 93

tion to Rom 1:3, 1 Cor 1:3, 2 Cor 1:2, etc., 22 and there are similar cases of
such scribal assimilation in 1 Thess 1:1. 23 Indeed, there is evidence for this
very assimilation within the Byzantine text. The bulk of the Byzantines fa-
vors the “Our Lord Jesus Christ” reading, but there are some Byzantine man-
uscripts that conform the wording to the “stereotyped formula” preferred by
the UBS committee. 24 The transcriptional probabilities in the other direction
are harder to discern, however. The Committee’s contention that the pronoun
was moved to “Lord Jesus Christ” out of piety 25 is too undeveloped to be
probative (both readings are pious) and is uncorroborated by any evidence for
an analogous change in the other letters of the Pauline corpus. 26
For these reasons, both the intrinsic and the transcriptional probabilities
favor the unusual Christological placement of the pronoun in the greeting. 27

Gal 1:17 οὐδὲ ἀπῆλθον εἰς Ἱεροσόλυμα πρὸς τοὺς πρὸ ἐμοῦ ἀποστόλους, ἀλλὰ
ἀπῆλθον εἰς Ἀραβίαν καὶ πάλιν ὑπέστρεψα εἰς Δαμασκόν.
Var. Western Eastern EDS.
ἀπῆλθον 28 D F G B SCC BW TM L
ἦλθον (d b ×vg veni) P46
ἀνῆλθον  33 A C P 1241S 1739 Ψ Chrys SBL NA RP S WH TT
1611 Byz
? Marc

All three of these readings go back to the earliest period, and thus the external
evidence is not decisive on its own. The reading ἀνῆλθον enjoys the support
of the Eastern branch plus  and 33, but the reading ἀπῆλθον is also ancient,
being supported by B and the Greek-Latin diglots D, F, and G. The oldest
witness, P46, has another reading ἦλθον due to the scribe’s tendency to

22
Philip W. Comfort, Early Manuscripts & Modern Translations of the New Testament
(Wheaton, Ill.: Tyndale House, 1990), 149; EADIE 9; NET 859; ZAHN 31 n.7.
23
NET 859 and 869.
24
For example, 056 131 876 1573 1836 and 2125.
25
METZGER 520: “The apostle’s stereotyped formula was altered by copyists who, ap-
parently in the interest of Christian piety, transferred the possessive pronoun so it would be
more closely associated with ‘Lord Jesus Christ.’” So also MATERA 39.
26
NET 859. WEISS 99 proposes a two-stage corruption in which the pronoun had ac-
cidentally dropped out from the standard formula and then restored to the wrong place.
Indeed, many Byzantine manuscripts have lost the pronoun (547 910 999 1319 1424 1448
1799 1874 1891 2147 2423), but this occurred within a tradition in which the
Christological reading predominated.
27
R. V. G. Tasker, The Greek New Testament Being the Text Translated in the New
English Bible 1961 (Oxford: Oxford University Press, 1964), 428, decides in favor of the
Christological placement based on external attestation.
28
This reading is also supported by P51, which was too fragmentary to be analyzed for
the stemma.
94 Chapter 3. Orienting the Stemma of Galatians

simplify prefixed words, 29 so either ἀνῆλθον or ἀπῆλθον could have stood in

P46’s exemplar. 30
Turning to the internal evidence, the meanings of the verb phrases ἀπ-
ῆλθον εἰς Ἱεροσόλυμα (“leave for Jerusalem”) and ἀνῆλθον εἰς Ἱεροσόλυμα
(“went up to Jerusalem”) are almost the same, but the emphases are different.
On the one hand, ἀνῆλθον with its prefix ἀνα- (“up”) is commonly used for
the hilltop city of Jerusalem (e.g. 2:1 ἀνέβην εἰς Ἱεροσόλυμα). 31 On the other
hand, ἀπῆλθον keeps the topic of Damascus in mind (“nor did I leave
[Damascus] for Jerusalem”), minimizing the importance of Jerusalem and
consequently that of the Jerusalem apostles.
The specter of scribal harmonization haunts both ἀνῆλθον and ἀπῆλθον. 32
Scribes would have been at least as prone – if not more so – to use the more
common ἀνῆλθον with a Jerusalem destination (also in context at v.18 ἀνῆλ-
θον εἰς Ἱεροσόλυμα, cf. 2:1), 33 as they would have been to assimilate the verb
to the following clause ἀλλὰ ἀπῆλθον εἰς Ἀραβίαν. 34 Intrinsically, either
reason would apply to Paul, but the implicit minimization of Jerusalem with
the ἀπῆλθον reading fits Paul’s purposes better. If the objection is that Paul
would not have repeated the verb, 35 there are examples to the contrary, such
as Rom 5:23 καυχώμεθα ... ἀλλὰ καυχώμεθα and 8:15 ἐλάβετε ... ἀλλὰ ἐλά-
βετε. 36
On balance, it seems that it was the apparent unusualness of the expression
ἀπῆλθον εἰς Ἱεροσόλυμα that induced scribes to modify the verb to the more
typical ἀνῆλθον despite its fitness to the context.

29
ROYSE 326, citing P46’s singular and near singular readings in Rom 8:17 (πάσχομεν),
12:16 (συναγόμενοι), Gal 1:17 (ἦλθον), 1:18 (ἔμεινα), 2:13 (ἀπήχθη); Eph 2:2 (ἐργοῦντος);
Col 3:16 (οἰκείτω); Heb 1:6 (ἀγάγῃ), 6:11 (δείκνυσθαι), 7:27 (ἅπαξ), 10:25 (συναγωγήν),
12:4 (ἀγωνιζόμενοι), 13:11 (καίεται).
30
ROYSE 309.
31
BURTON 55; LÉGASSE 99 n.5; LIGHTFOOT 83; LONGENECKER 33. These intrinsic
analyses do not go beyond noting the relative popularity of the collocation ἀνῆλθον εἰς
Ἰεροσόλυμα.
32
BUSCEMI 106.
33
EADIE 47; MEYER 11; ZAHN 66 n.74. So also WEISS 35, who argued that ἀπῆλθον
was assimilated by the scribes to the technical usage of ἀνῆλθον in v.18 due to its seeming-
ly tiresome repetition.
34
BETZ 73 n.173; BURTON 55; LÉGASSE 99 n.5; LIGHTFOOT 83; LONGENECKER 33;
MUSSNER 91 n.61; SIEFFERT 65–66 n.**; ZIMMER¹ 488. In fact. there is evidence of just
this scribal harmonization to ἀπῆλθον among several minuscules (88 1799 1836 2344
2464). BETZ 73 n.173 notes that “the synonymous ἀναβαίνω is more common,” which Paul
uses in Gal 2:1, 2; so also ZIMMER¹ 488.
35
So LIGHTFOOT 83: “the repetition makes the sentence run awkwardly”.
36
EADIE 47, citing Rom 8:15. LIGHTFOOT 83 concedes that Rom 8:15 and 1 Cor 2:13 (ἃ
καὶ λαλοῦμεν οὐκ ἐν διδακτοῖς ἀνθρωπίνης σοφίας λόγοις ἀλλ’ ἐν διδακτοῖς πνεύματος)
are contrary examples.
 3.2 Examination of Variants for Inner Branch Oppositions 95

Gal 1:18 Ἔπειτα μετὰ ἔτη τρία ἀνῆλθον εἰς Ἱεροσόλυμα ἱστορῆσαι Κηφᾶν καὶ
ἐπέμεινα πρὸς αὐτὸν ἡμέρας δεκαπέντε,
Var. Western Eastern EDS.
ἔτη τρία D F G P46 B Ψ Chrys 1611 Byz SCC SBL NA RP BW T L
d b ×vg
τρία ἔτη  33 A P 1241S 1739 S WH
? Marc C

This is a good example of the split of the P46–B and the Western branch
against  33 and the Eastern branch, though the Byzantines and Ψ join the
Western branch. Internal evidence is necessary to resolve this.
This word order variant may seem, at first glance, to have hardly any effect
on the meaning – both would be translated “after three years.” In noun-
adjective pairings, however, Greek prefers to put the adjective first only if it
is more informative or contrastive than the noun. 37 Thus, the order μετὰ ἔτη
τρία suggests a contrast later in the verse with ἡμέρας δεκαπέντε (“for fifteen
days”); the units of time are fronted to emphasize that while Paul spent years
away from Jerusalem, he spent only days with Cephas in the city. 38 On the
other hand, the word order μετὰ τρία ἔτη invites a correspondence with Gal
2:1 διὰ δεκατεσσάρων ἐτῶν (“after fourteen years”), and the point would then
have to be on enumerating the different time intervals.
Paul’s usage elsewhere confirms this behavior in the placement of the
number for emphasis. 39 For example, Paul places the number before the noun
ἔτη to emphasize the amount of time, e.g., Gal 2:1 διὰ δεκατεσσάρων ἐτῶν
(“after fourteen years”) to highlight his independence from Jerusalem, and
3:17 μετὰ τετρακόσια καὶ τριάκοντα ἔτη (“after four hundred and thirty
years”) to underscore the length of time between Abraham and Moses. In 2
Cor 12:2 πρὸ ἐτῶν δεκατεσσάρων (“fourteen years ago”), however, he places
the number afterwards when the length of time is not important. 40

37
Stéphanie Bakker, “Adjective Ordering in Herodotus: A Pragmatic Explanation” in
Rutger Allan and Michel Buijs, eds., The Language of Literature: Linguistic Approaches to
Classical Texts (ASCP 13; Leiden: Brill, 2007), 188–210 at 195.
38
WEISS 130. Indeed, it is noteworthy just how well Weiss was able to anticipate the
modern pragmatic analysis more than a century ago: “dass τρια eine sehr kleine Zahl ist,
die nicht durch Voranstellung betont werden konnte, so wenig wie das δεκαπεντε im
Folgenden.” Cf. ZAHN 69 n.77, who takes the number to be more important because Paul
had not enumerated anything yet.
39
Pace ZIMMER¹ 488–489, who claims that Paul always places small numbers after the
noun. Zimmer’s observation is confounded by the pragmatics of the word order. Small
numbers are usually less important than the noun and would therefore go after the noun,
but they are not always unimportant in appropriate contexts. As argued here, Gal 1:18 is
such an appropriate context.
40
There is also case of a non-numeric adjective before the noun, cf. Rom 15:23 ἀπὸ
πολλῶν ἐτῶν (“for many years”)
96 Chapter 3. Orienting the Stemma of Galatians

Intrinsically, the word order of μετὰ ἔτη τρία fits the passage better. It is
more amenable to the argument of the letter to emphasize his independence
from Jerusalem, than to count the time intervals away. Thus, the proper point
of contrast is the years apart from Jerusalem compared with the days in Jeru-
salem, not a three year delay compared with a fourteen year delay in going to
Jerusalem. Transcriptionally, the reason for the year first order could be
scribal assimilation to the order of 2:1. 41 Thus, the internal evidence decisive-
ly favors the order μετὰ ἔτη τρία of the Nestle-Aland text.

Gal 2:20 19 ἐγὼ γὰρ διὰ νόμου νόμῳ ἀπέθανον, ἵνα θεῷ ζήσω. Χριστῷ συνεσταύρω-
μαι· 20 ζῶ δὲ οὐκέτι ἐγώ, ζῇ δὲ ἐν ἐμοὶ Χριστός· ὃ δὲ νῦν ζῶ ἐν σαρκί, ἐν
πίστει ζῶ τῇ τοῦ θεοῦ καὶ Χριστοῦ τοῦ ἀγαπήσαντός με καὶ παραδόντος
ἑαυτὸν ὑπὲρ ἐμοῦ.
Var. Western Eastern EDS.
τοῦ θεοῦ καὶ Χριστοῦ D* F G d P46 B T
SCC BW T L
τοῦ υἱοῦ τοῦ θεοῦ ×vg  33 A C 1241S 1739 Ψ Chrys SBL NA RP
1611 Byz S WH T
M

τοῦ Χριστοῦ καὶ θεοῦ b

τοῦ θεοῦ 330
? Marc P

The first two readings are old and could conceivably go back to the most re-
cent common ancestor of the text. 42 Though the shortest reading is attested in
330, this manuscript is so far removed from the ancestor that selecting its
reading is tantamount to making a conjectural emendation. 43 As the two old-
est readings are evenly split between the Western and Eastern branches with
their respective allies, internal evidence must decide this variation unit.
The Nestle-Aland reading, “by faith of/in the Son of God,” raises two
major exegetical issues. The first is whether the genitive τοῦ υἱοῦ τοῦ θεοῦ is

41
WEISS 130.
42
HORT App.121 and BURTON 139, however, characterized this as a Western cor-
ruption of B, but this was before the discovery of P46.
43
In fact, J. C. O’Neill, The Recovery of Paul’s Letter to the Galatians. London: SPCK,
1972), 45, has proposed this very conjectural emendation. Edmond Farahian, Le “je”
paulinien: Étude pour mieux comprendre Gal. 2, 19–21 (Analecta gregoriana 253; Rome:
Editrice Pontificia Università Gregoriana, 1988), 237, seems sympathetic to the conjecture
but ultimately decides against it.
What is good about this conjecture is that an original τοῦ θεοῦ can easily account for
both early extant readings as two independent attempts at fixing a problematic reading.
What is bad about it is that the phrase, “God, who … gave himself up for me,” is anoma-
lous for Paul, much more so than either of the extant readings. The conjecture solves a
transcriptional problem at the cost of introducting a major intrinsic problem. O’Neill’s
argument that “[t]he verb παραδίδωμι is used with God as the implied subject in Rom.
4.25” does not take into account the object of this verb; it is Jesus whom God delivered up
in Rom 4:25, but, in his emendation of Gal 2:20, God delivered up himself.
 3.2 Examination of Variants for Inner Branch Oppositions 97

subjective, “faith(fulness) of the Son of God,” 44 or objective, “faith in the Son

of God.” 45 Proponents of the view that the genitive in Gal 2:20 is objective
usually argue that this is controlled by a similar genitive in v.16, which they
maintain to be objective. 46 On the other hand, proponents of the subjective
genitive here argue that the context of v.20, rather than v.16, indicates that
the genitive is subjective. The following participial phrase “who loved me
and gave himself over for me” epexegetically restates what the Son of God’s
faithfulness is, 47 and Paul’s statement that he is no longer living but Christ
living in him (v.20a) suggests a Christological rather than anthropological
focus for the term “faith.” 48 Moreover, the articulation of v.20 supports the
subjective genitive reading. Generally, when the noun lacks the article but its
modifier has the article, the modifier is necessary for the correct identification
of the noun. 49 This suggests that this Son-of-God faithfulness implicitly
contrasts with Paul’s own faithfulness inherent in his life in the flesh (ὃ δὲ
νῦν ζῶ ἐν σαρκί). Just as Paul’s own faithfulness is subjective, so too is the
Son of God’s faithfulness.
The second issue raised by the Nestle-Aland reading is to account for why
the text introduces the term “Son of God” at all. 50 The context does not
trigger the use of this term; rather, the more commonly Pauline term “Christ”

44
So Richard B. Hays, The Faith of Jesus Christ: The Narrative Substructure of
Galatians 3:1–4:11 (2d ed.; Grand Rapids, Mich.: Eerdmans, 2002), 153–155; MATERA
96.
45
So BURTON 139; BUSCEMI 224; James D. G. Dunn, “Once More, ΠΙΣΤΙΣ ΧΡΙΣΤΟΥ”
in Hays, Faith, 249–271, at 262; Gordon D. Fee, Pauline Christology: An Exegetical-
Theological Study (Peabody, Mass.: Hendrickson, 2007), 223–226; Thomas H. Tobin,
Paul’s Rhetoric in its Contexts: The Argument of Romans (Peabody, Mass.: Hendrickson,
2004), 132–133.
46
BUSCEMI 224; Tobin, Paul’s Rhetoric 133. Oddly, LONGENECKER 94 reads the
genitives of v.16 as subjective but then understands the genitive in v.20 to be objective due
to the dative article τῇ, yet the article is dative merely because πίστει is dative.
47
Campbell, Deliverance, 847; Hays, Faith, 291. But Dunn, “Once More,” 262 sees the
genitive participle phrases as specifying the content of the faith in the Son of God.
48
Campbell, Deliverance, 847–848. Hays, Faith, 154–155, sums it up as follows: “The
whole context portrays Christ as the active agent and Paul as the instrument through which
and/or for whom Christ’s activity comes to expression.” Hays, Faith, 154, also points to a
syntactical analogy with Rom 5:15 ἐν χάριτι τῇ τοῦ ἑνὸς ἀνθρώπου Ἰησοῦ Χριστοῦ (“in the
grace of one person Jesus Christ”).
49
See generally, Stéphanie Bakker, The Noun Phrase in Ancient Greek: A Functional
Analysis of the Order and Articulation of NP Constituents in Herodotus (ASCP 15; Leiden:
Brill, 2009), 271–282 and 289–290, especially at 290: “The use of the NaX [scil. noun-
article-modifier] pattern, by contrast, presupposes that the modifier is essential for the
identification of a referent: without the help of the modifier, the addressee would identify
the wrong referent or no referent at all.”
50
Noted but not elaborated by Campbell, Deliverance, 645: “‘The Son of God’ is not a
name for Paul (and only Gal. 2:20 uses it).”
98 Chapter 3. Orienting the Stemma of Galatians

would be more appropriate in the immediate context. 51 Moreover, the phrase

“faith in/of the Son of God” is unique to Paul, 52 and the fully articulated
postpositive genitive phrase τοῦ υἱοῦ τοῦ θεοῦ is found only here in this exact
form among the undisputed Paulines. 53 Consequently, commentators have
struggled to account for the phrasing of the Nestle-Aland reading of Gal
2:20. 54
As for the “God and Christ” reading, there are also two exegetical issues.
The first is whether the lack of the article before Christ as the second member
of a compound noun phrase means that the text is equating God and Christ, 55
thereby making this reading one of the most explicit statements of Christ’s
divinity in the Pauline corpus. 56 Though this construction admits this inter-

51
So BURTON 139, who concedes: “What particular phase of the meaning of this title as
applied to Jesus is here in mind, or why it is chosen instead of Χριστός or Χριστός Ἰησοῦς,
which have been used in this passage thus far, there is nothing in the context to indicate.”
Burton goes on to argue that perhaps Paul used this phrase because he was thinking of
Christ’s love.
52
Cf. Hermas 93.5 οἱ ἀπόστολοι καὶ οἱ διδάσκαλοι . . . κοιμηθέντες ἐν δυνάμει καὶ
πίστει τοῦ υἱοῦ τοῦ θεοῦ “the apostles and teachers . . . fallen asleep in the son of God’s
power and faith.” If the genitive τοῦ υἱοῦ τοῦ θεοῦ construes with both power and faith,
then since the first (power) is Christological, so should the second (faith).
53
But cf. Eph 4:13 τῆς ἐπιγνώσεως τοῦ υἱοῦ τοῦ θεοῦ “of the knowledge of the Son of
God” (however τοῦ υἱοῦ is omitted by F G b). Among the undisputed letters, Paul’s
favorite way by far of expressing divine filiation is ὁ υἱὸς αὐτοῦ (Rom 1:3, 9, 5:10, 8:29; 1
Cor 1:19; Gal 1:16, 4:4, 6; 1 Thess 1:10), though with some variations: τὸν ἑαυτοῦ υἱοῦ
(Rom 8:3), τοῦ ἰδίου υἱοῦ (Rom 8:32), absolute ὁ υἱός (1 Cor 15:28), anarthrous υἱοῦ θεοῦ
(Rom 1:4), and the pre-posed genitive ὁ τοῦ θεοῦ γὰρ υἱός (2 Cor 1:19).
54
Some argue that Paul was incorporating a traditional formulation, but Gabriella
Berényi, “Gal 2,20: a Pre-Pauline or a Pauline Text?” Bib 64 (1984):490–537 at 522–523,
counters that Gal 2:20 is not a traditional saying because the use of παραδοῦναι with the
reflexive pronoun is found only in the New Testament at Gal 2:20 and Eph 4:19, 5:2, 25.
She goes on to argue that, despite the fact that the use of the phrase “Son of God” is “very
rare” in Paul (528 n.62), this “very solemn title” was chosen to enhance the paradox of
Christ’s delivering himself up for Paul and lend “an audacious, unusual character” to the
entire statement (528–529, also 530, 536).
55
This article-substantive-καί-substantive construction is also known as the “Granville
Sharp construction”; see especially Daniel B. Wallace, Granville Sharp’s Canon and Its
Kin: Semantics and Significance (SBG 14; New York: Peter Lang, 2009), 7 n.21. Wallace
argues strongly that this construction with personal, singular, and non-proper referents
means that the construction as a whole refers to a single person, but grammarians of
Classical Greek prefer to think of this construction as merely having the effect of “a single
notion” (SMYTH 291 § 1143) or being “depicted as one whole” (Bakker, Noun Phrase,
177).
56
So NET 860: “The construction ‘of God and Christ’ appears to be motivated as a
more explicit affirmation of the deity of Christ (following as it apparently does the
Granville Sharp rule). Although Paul certainly has an elevated Christology, explicit ‘God-
talk’ with reference to Jesus does not normal appear until the later books.”
 3.2 Examination of Variants for Inner Branch Oppositions 99

pretive possibility, 57 it is more likely that Paul’s use of “Christ” as a proper

name makes the term sufficiently definite for Paul that the article is
unnecessary. 58 This means that, if the reading was original to Paul, it would
refer to two separate persons, 59 not to a single God-Christ person. 60
Second, as with the other reading, the genitive phrase τοῦ θεοῦ καὶ Χρι-
στοῦ can be either subjective, “faith of God and Christ,” or objective, “faith
in God and Christ.” Though this issue for this specific reading has hardly
been discussed in the literature, 61 the objective genitive seems difficult to
sustain. Against the objective reading, Paul does not otherwise describe two
faiths for a Christian, one in God and another in Christ. 62 On the other hand,
the subjective reading of the genitive is appropriate because Paul would be
talking about God’s faithfulness (cf. Rom 3:3 τὴν πίστιν τοῦ θεοῦ), 63 as well
as Christ’s. Both persons are mentioned in the nearby context, especially v.19
(“that I live to God; I am crucified in Christ”). Moreover, as explained above,

57
See generally Nigel Turner, A Grammar of New Testament Greek: Syntax (vol. 3;
Edinburgh: T. & T. Clark, 1963), 181: “The art. may be carried over from the first noun to
the other(s), especially if they are regarded as a unified whole and the gender and number
are the same. . . . [However,] the repetition of the art. was not strictly necessary to ensure
that the items be considered separately.” See also Bakker, Noun Phrase, 177–178: “the
different entities are depicted as one whole.”
58
So Wallace, Granville Sharp’s Canon, 237, esp. n.7: “Since they [scil. proper names]
do not require an article to be definite, one cannot conclude that the article ‘carries over’ to
the proper name in the sense of referential identity.”
59
John Bligh, Galatians: A Discussion of St Paul’s Epistle (Householder Commentaries
1; London: St Paul Publications, 1969), 215, argues in favor of two persons because Gal
2:20–21 corresponds to 1:4–5 where Jesus Christ gave himself according to the will of God
the Father. Note also the dual reference to God the Father and the Lord Jesus Christ in 1:3.
60
Pace Bart D. Ehrman, The Orthodox Corruption of Scripture: The Effect of Early
Christological Controversies on the Text of the New Testament (New York: Oxford Uni-
versity Press, 1993), 86–87; NET 860; perhaps also BETZ 125 n.104. Wallace, Granville
Sharp’s Canon, 237 abandons sub silentio the prior position in NET 860, of which he was
an editor. His later, more-considered position is the correct one, because his statement of
the Granville-Sharp rule explicitly excludes proper names from its scope (Daniel B.
Wallace, Greek Grammar Beyond the Basics: An Exegetical Syntax of the New Testament
[Grand Rapids, Mich.: Zondervan, 1996], 271–272), and he concedes that Christ is a pro-
per name in the epistles (Wallace, Granville Sharp’s Canon, 252 n.52).
61
After the author’s dissertation had been defended, this issue has now been taken up in
response to it by Jermo van Nes, “‘Faith(fulness) of the Son of God’? Galatians 2:20b
Reconsidered,” NovT 55 (2013): 127–139, who accepts the author’s conclusion that τοῦ
θεοῦ καὶ Χριστοῦ ought to be read in the text.
62
Van Nes’s counter-argument that “Rom 4:23 and Gal 2:16 clearly refer to having
faith in God and Christ respectively” (“Faith(fulness),” 136) begs the pistou Christou
question without argument. As the vast debate on the question shows, it is hardly “clear.”
63
Howard Eshbaugh, “Textual Variants and Theology: A Study of the Galatians Text of
Papyrus 46.” JSNT 3 (1979): 60–72 at 67.
100 Chapter 3. Orienting the Stemma of Galatians

the articulation of ἐν πίστει ... τῇ τοῦ θεοῦ καὶ Χριστοῦ implicitly contrasts
Paul’s own faithfulness (subjective) with that of God and Christ.
Weighing the intrinsic probabilities between the first two readings is
complicated by the lack of exact parallels for both wordings in Paul. Both
“faith of God and Christ” and “faith of the Son of God” are found only here
in Paul, 64 and both genitive phrases after “faith” are uniquely formulated in
the undisputed letters of Paul, though analogous formulations are attested for
both τοῦ θεοῦ καὶ Χριστοῦ 65 and τοῦ υἱοῦ τοῦ θεοῦ. 66 Metzger’s denial that
“Paul nowhere else expressly speaks of God as the object of a Christian’s
faith” 67 not only begs the question that this genitive is objective but is also
too narrowly drawn to be fully cogent. For example, Paul expressly talks of
the Thessalonians’ faith in God at 1 Thess 1:8 (ἡ πίστις ὑμῶν ἡ πρὸς τὸν
θεόν). As another example, in Rom 4:23–24, Paul refers to his contemporary
Christians as “those believing in the one who raised our Lord Jesus from the
dead” (τοῖς πιστεύουσιν ἐπὶ τὸν ἐγείραντα Ἰησοῦν τὸν κύριον ἐκ νεκρῶν).
Even within Galatians itself, God is the object of the faith of Abraham, who
functions as a model for gentile Christians (Gal 3:6, 8–9; see also Rom 4:3,
11–12). 68
Moreover, the Christology of the reading τοῦ θεοῦ καὶ Χριστοῦ is not too
high for Paul for him to have written the phrase, 69 because the lack of the
article before Christ does not necessarily equate God and Christ in this
construction. Indeed, having both God and Christ in focus fits the context
very well, with the two of them together being mentioned in both v.19 and

64
Noted by Eshbaugh, “Textual Variants,” 67.
65
Though Raymond E. Brown, Jesus, God and Man: Modern Biblical Reflections
(Milwaukee: Bruce,1967), 11, argues that “[t]he phrase tou theou kai Christou is never
found elsewhere in the Pauline writings, and so is suspect,” various fuller phraseologies are
found throughout the Pauline corpus (Rom 1:7; 1 Cor 1:3; 2 Cor 1:2; Gal 1:3; Phil 1:2; 1
Thess 1:2; Phlm 3; cf. Eph 1:2, 6:23; 2 Thess 1:2; 1 Tim 1:2, 5:21; 2 Tim 1:2, 4:1; Tit 1:4,
2:13).
66
Farahian, Le “je” paulinien, 234 n.10 properly calls into question Metzger’s appeal
to “the customary Pauline expression τοῦ υἱοῦ τοῦ θεοῦ” (METZGER 524, followed by
MATERA 96). Metzger’s claim works only at a conceptual level–as stated above, Paul
prefers other formulations to express divine filiation.
67
METZGER 524. Also BETZ 125 n.104; BURTON 139 (“The apostle never speaks of
God expressly as the object of a Christian’s faith.”); BUSCEMI 196 (“mai altrove in Paulo
Dio è oggetto della fede”); Ehrman, Orthodox Corruption, 86; LONGENECKER 94; Pedro
Mendoza Magallón, «Estar crucificado juntamente con Cristo»: el nuevo status del cre-
yente en Cristo. Estudio exegético-teológico de Gal 2,15–21 y Rom 6,5–11 (Tesi Grego-
riana Serie Teologica 122; Rome: Editrice Pontificia Università Gregoriana, 2005), 94
n.185; Kazimeirz Romaniuk, L’amour du Père et du Fils dans la sotériologie de saint Paul
(Analecta biblica 15; Rome: Pontifical Biblical Institute, 1961), 31.
68
See Campbell, Deliverance, 873.
69
Pace Ehrman, Orthodox Corruption, 86–87; NET 860; perhaps BETZ 125 n.104.
 3.2 Examination of Variants for Inner Branch Oppositions 101

v.21, 70 while there is little intrinsic reason for Paul to write “Son of God.” 71
Furthermore, the τοῦ θεοῦ καὶ Χριστοῦ reading lends itself better to a sub-
jective genitive reading, which also fits the context very well. 72
Transcriptionally, the τοῦ υἱοῦ τοῦ θεοῦ reading is unobjectionable to
scribes, 73 so textual critics have found it necessary to posit a two-stage error
scenario in which τοῦ υἱοῦ was omitted by accident to read the very difficult
τοῦ θεοῦ (“faith in/of God who … gave himself up for me”), and then was
later corrected by adding καὶ Χριστοῦ. 74 The transcriptional probabilities
indeed go in the other direction. An anti-Patripassionist scribe could well
have misread the τοῦ θεοῦ καὶ Χριστοῦ as a single person, which would
suggest the passibility of God in this context, and fixed the reading to avoid
that implication. 75
Accordingly, the intrinsic probabilities suggest that “God and Christ” fits
the context better than “Son of God,” and the transcriptional probabilities also
favor the “God and Christ” reading. For these reasons, the “God and Christ”
reading, τοῦ θεοῦ καὶ Χριστοῦ, ought to be adopted for a critical text of
Galatians.

70
WEISS 9: “Aber das του θεου και χριστου (Lchm. Treg. txt. nach BDFG d. e. g.) ist
nicht nur offenbar die schwierigere Lesart, sondern wird auch durch v. 21 bestätigt wo
deutlich vorausgesetzt ist, dass vorher von beiden geredet war.”
71
Pace Romaniuk, L’amour, 32, who argues that the clause “I am crucified with Christ”
indicates that Paul only had the Son in view. Contradicting this argument is the
immediately preceding clause in v.19, ἵνα θεῷ ζήσω, “so that I may live to God,” which
also puts the Father in view. At any rate, Romaniuk does not explain why Paul wrote “Son
of God” instead of merely “Christ.”
72
So Norbert Baumert, Der weg des Trauens: Übersetzung und Auslegung des Briefes
an die Galater und des Briefes an die Philipper (Würzburg: Echter, 2009), 50.
73
Peter M. Head, “Galatians 2.20: ‘I live by faith in God and Christ ...’,” Evangelical
Textual Criticism (March 15, 2006), online: http://evangelicaltextualcriticism.blogspot.
com/2006/03/galatians-220-i-live-by-faith-in-god.html; cf. Ehrman, Orthodox Corruption,
86; LIGHTFOOT 119.
74
METZGER 524; MEYER 54. Also Ehrman, Orthodox Corruption, 86.
75
WEISS 1896:9 calls “God and Christ” the apparently more difficult reading (“offenbar
die schwierigere Lesart”) but does not explain why. On the other hand, Eshbaugh, “Textual
Variants,” 67 suggests the “Son of God” reading “shows a higher, more formalized
Christology.” Fee, Pauline Christology, 511–512 deduces the Christology of Gal 2:20 from
Paul’s use of Son of God: “the very fact that in this case he identifies Christ as ‘the Son of
God’ suggests that what overwhelms Paul about such love is not simply Christ’s death on
his behalf [but] that the preexistent, and therefore divine, Son of God is the one who by
incarnation as well as crucifixion ‘died for me.’ Contrarily, Brown, Jesus, 11 sees the “Son
of God” as “the less developed reading from a theological viewpoint,” apparently if the
phrase “God and Christ” is understood as equating the two persons.
102 Chapter 3. Orienting the Stemma of Galatians

Gal 5:17b . . . ἵνα μὴ ἃ ἐὰν θέλητε ταῦτα ποιῆτε.

Var. Western Eastern EDS.
ἃ ἐάν BC  ×1175 A 1241S SCC SBL NA S BW WH T L
T

ἃ ἄν CC Ψ 1611 Byz RP
ὃ ἄν D* F G
ἄν B*
ἅ C* 1739 Chrys L
M

? Marc d b ×vg P46 33 P

Not noted in the NA27 apparatus, this variation unit is fragmented along the
basic branches of the textual transmission of Galatians. Nevertheless, B and
the Western branch agree with each other by reading ἄν, against the rest of
the tradition which reads ἐάν when it is attested. The only witness to omit the
relative pronoun ἅ is B*, but this omission is probably due to parablepsis
from the alpha of ἅ to the alpha of (ἐ)άν (and B was corrected to read ἃ
ἐάν); 76 thus, B*’s ἄν actually fails to provide substantial support for either ἄν
or ἐάν when the relative pronoun is included. With B* out of the picture, the
two basic readings are ἃ ἐάν of the Eastern branch plus  (and BC) and ὃ ἄν
of the Western branch. The external evidence thus favors the reading ἃ ἐάν.
Originally a combination of εἰ and ἄν, the particle ἐάν was used in class-
ical Greek as a conditional marker with the meaning “if,” but by Hellenistic
times it came to be frequently used after relative pronouns. 77 Since Galatians
was written in Hellenistic Greek, this use of ἐάν after a relative pronoun is
intrinsically suitable, but classically trained scribes may have been motivated
to change the reading to a more classical ἄν. 78 Furthermore, the singular
relative pronoun of the Greek-Latin diglots is intrinsically unsuitable due to
the number mismatch with its resumptive pronoun ταῦτα. Thus, the internal
evidence too favors the reading ἐάν.

Gal 6:2 Ἀλλήλων τὰ βάρη βαστάζετε καὶ οὕτως ἀναπληρώσετε τὸν νόμον τοῦ
Χριστοῦ.
Var. Western Eastern EDS.
ἀναπληρώσετε Marc F G (P46 ἀπο-) B SCC SBL NA
d b ×vg M T
BW T L
ἀναπληρώσατε D  33 A C 1739 Ψ Chrys RP S WH
1241S 1611 Byz T M
T L
? P

The internal evidence must be determinative, because the external evidence is

split between the Western branch and its allies (P46 and B) and the Eastern

76
WEISS 63.
77
BDAG “ἐάν,” 267–268; BDF 57 § 107; LSJ “ἐάν,” 465.
78
Cf. the discussion in ZIMMER³ 301–305.
 3.2 Examination of Variants for Inner Branch Oppositions 103

branch and its allies ( 33). 79 The difference between the two readings lies in
the inflection of the verb. The Western branch’s verb ἀναπληρώσετε is in
future indicative: “Bear each other’s burdens and thus you will fulfill the law
of Christ,” while the Eastern branch has an aorist imperative ἀναπληρώσατε:
“Bear each other’s burdens and thus fulfill the law of Christ.” The future
indicative is a better fit to the context because it gives a promised conse-
quence of the earlier imperative, 80 while the aorist imperative appears to be a
scribal harmonization to the mood of the preceding verb. 81 Harmonization,
however, is not sufficent by itself to settle this reading because it is con-
ceivable that the harmonization could go in the other direction, for the pre-
ceding βαστάζετε, being a present imperative, ends with the same letters as
the future: −ετε. 82

3.2.2 Inner Branch b

The following fifteen variation units occur along inner branch b where P46
and B are in agreement against the readings of their nearest stemmatic rela-
tives, i.e., the Western branch, , and 33:

Gal 1:12 οὐδὲ γὰρ ἐγὼ παρὰ ἀνθρώπου παρέλαβον αὐτὸ οὐδὲ ἐδιδάχθην, ἀλλὰ δι’
ἀποκαλύψεως Ἰησοῦ Χριστοῦ.
Var. Western Eastern EDS.
οὐδέ D* F G  33 ×1175 A P 1241S 1739 Ψ Chrys M
SCC WH T L
οὔτε P46 B 1611 Byz SBL NA RP
T
S BW WH
? Marc d b ×vg C

The οὔτε reading is only found in the P46–B cluster, and the Harklean
(including the Greek 1611) and Byzantine groups. 83 Ordinarily, the external
evidence strongly favors οὐδέ, being found in both the Western and Eastern
branches along with  and 33. The lack of genealogical coherence for the
οὔτε readings suggests that it arose on separate occasions and is therefore
secondary, unless it appears that it is much more likely that scribes would

79
METZGER 530 favors the future indicative reading of P46, B, and G because of its
“early and diverse external diversification,” but this is in tension with the more usual low
regard for B-Western agreements by rational eclectics.
80
LONGENECKER 275.
81
BUSCEMI 568; MATERA 214; METZGER 530; MEYER 319; MUSSNER 399 n.23;
ROHDE 260; SIEFFERT 333 n.**; WEISS 49.
82
BURTON 330 has a different argument against the future form: “The fut. is probably
due to the natural tendency to convert the second imperative into a promissory apodosis,”
but without any evidence of this “natural tendency.”
83
So ZAHN 52 n.52. BURTON 40, to the contrary, weighs the external evidence against
οὐδέ because it is Western, but this does not account for the reading in  33 A P 1241S
1739 and Ψ.
104 Chapter 3. Orienting the Stemma of Galatians

have changed οὔτε into οὐδέ than the reverse. This assessment, then, requires
an examination of the internal evidence.
The distinction between οὐδέ and οὔτε in this context is fairly subtle.
When these negative conjunctions follow a negative particle (as here, the οὐ
in the adverbial οὐδὲ γάρ), οὐδέ adds another negative thought while οὔτε
subdivides the earlier negative thought. 84 Thus, the reading οὐδὲ ἐδιδάχθην
implies that Paul was also denying that he been taught the gospel by a human
at an occasion separate from his receiving of the gospel (οὐδὲ . . . παρέλα-
βον). 85 After all, Paul is about to admit that he spent a fortnight in Jerusalem
with Cephas three years after his call (v.18), and this denial of being taught
with οὐδέ to indicate a separate negative thought allows him preemptively to
deny being taught the gospel even then. As for the reading οὔτε ἐδιδάχθην,
this usage is rare in Greek because οὔτε is usually found paired with at least
one other οὔτε, 86 so the construction οὐ(δέ) . . . οὔτε . . . tends to suggests
that the second negative element is an afterthought. 87 This reading would
suggest that Paul re-characterized his initial reception of the gospel as not
involving teaching and implies no denial of being taught at another occasion.
Intrinsically, the separate denial of the οὐδὲ ἐδιδάχθην reading fits Paul’s
defensiveness about his apostleship better. For example, in 1:1 Paul carefully
laid out that he was an apostle not from any group of people nor by anybody
(Παῦλος ἀπόστολος οὐκ απ’ ἀνθρώπων οὐδὲ δι’ ἀνθρώπου). 88 Transcrip-
tionally, it has been argued that οὔτε could be the harder reading, since it is
not usual construction in ancient Greek, 89 so a scribe, not noticing that the οὐ
in οὐδέ substituted for an initial οὔτε, could have assimilated it to the
preceding οὐδέ. 90 On the other hand, dissimilation can also occur, and these

84
BDF 231 § 445(2); George Benedict Winer, A Grammar of the New Testament Dic-
tion: Intended as an Introduction to the Critical Study of the Greek New Testament
(Edward Masson, tr.; 4th ed.; Edinburgh: T. &. T. Clark, 1863), 509.
85
ZAHN 52.
86
SMYTH 661 § 2941. Elsewhere in Paul, οὔτε is always paired with at least one other
οὔτε: Rom 8:38–39 (10 times); 1 Cor 3:7, 6:9–10 (7 times), 8:8, 11:11; Gal 5:6, 6:15; 1
Thess 2:5–6 (4 times). ZIMMER¹ 487 concludes from this that an unpaired οὔτε is non-
Pauline, but the rarity of the expression makes such a conclusion precarious.
87
J. D. Denniston, The Greek Particles (2d ed.; Oxford: Clarendon Press, 1954), 509–
510; Winer, Grammar, 513.
88
Cf. LONGENECKER 23, who connects 1:12 with 1:1 though he does not seen any
difference in meaning between οὐδέ and οὔτε here (20 n.b).
89
BURTON 40 (“unusual”); LIGHTFOOT 80 (“less regular”).
90
WEISS 61: “Die Emendatoren übersahen, dass sich dasselbe [οὔτε] an das in ουδε
liegende ου anschliesst, wenn sie nicht einfach nach diesem conformirten.” So also MA-
TERA 52–53. Within the Byzantine tradition, there are some examples of assimilating οὔτε
to οὐδέ here (69 205–209 2147).
 3.2 Examination of Variants for Inner Branch Oppositions 105

conjunctions are often interchanged among the manuscripts, 91 due to their

similarity of sound 92 or perhaps because a scribe did not realize that οὐδὲ
ἐδιδάχθην referred to a separate incident at which Paul could have learned
about the gospel. At any rate, the internal evidence does not seem decisive
enough to overturn the excellent external evidence for οὐδέ.

Gal 1:15 Ὅτε δὲ εὐδόκησεν ὁ θεὸς ὁ ἀφορίσας με ἐκ κοιλίας μητρός μου καὶ
καλέσας διὰ τῆς χάριτος αὐτοῦ 16 ἀποκαλύψαι τὸν υἱὸν αὐτοῦ ἐν ἐμοί, . . .
Var. Western Eastern EDS.
ὁ θεός D d  33 ×1175 A P 1241S 1739 Ψ T
SCC NA RP
Chrys Byz T T T
S WH T L
− F G b ×vg P46 B 1611 M
SBL NA BW
M M M
WH T L
? Marc C

Although the shorter reading without ὁ θεός (“God”) is classified here as a

P46–B cluster reading, it is also found in some members of the Western
branch (F G b ×vg), so it could have also been classified as belonging to the
basic Western/Eastern opposition, with the longer reading of D and d being a
secondary development. The longer reading is also found in the Eastern
branch, and the basic witnesses  and 33. Though textual critics have viewed
the balance of the external evidence as tilting in favor of the inclusion of
“God,” 93 it is necessary to consider which reading the internal evidence
supports.
With the shorter reading, the participial clause ὁ ἀφορίσας με . . . καὶ καλέ-
σας . . . is the subject of the verb εὐδόκησεν: “the one who set me apart . . .
and called [me] . . . was pleased.” The unnamed agent is clearly God, as evi-
dent in the complementary infinitive clause ἀποκαλύψαι τὸν υἱὸν αὐτοῦ (“to
reveal his son”), and the longer reading makes this explicit. 94 With the longer
reading, ὁ θεός is the explicit subject of the verb, while the participial clause

91
BDAG “οὔτε,” 740. For example, the scribe of P75 created a singular reading of οὐδὲ
... οὔτε from an original οὔτε ... οὔτε at John 8:19 (ROYSE 680 n.344). Examples of
changes from οὐδέ to οὔτε include Rom 8:7 (L), 1 Cor 3:2 (L al), 5:1 (88), 11:14 (1175),
15:50 (33); Gal 2:3 (F G), 3:28 (Ψ 323 796 943). Examples of changes from οὔτε to οὐδέ
include 1 Cor 3:7 (A C), 1 Cor 6:9–10 (D).
92
The devoicing of the stop in οὐδέ to οὔτε could have been triggered by the preceding
word αὐτό. Such devoicing of consonants is also a characteristic of the Greek speech in
Egypt at this time, see generally Francis T. Gignac, A Grammar of the Greek Papyri of the
Roman and Byzantine Periods (vol. 1: Phonology; Milan Goliardica, 1976).
93
So METZGER 521 (majority of the committee); ZAHN 60. Conceded by BUSCEMI 106.
The longer reading is also found in Irenaeus, Adv. haer. 5.12.5: Cum autem placuit Deo,
qui me ex utero matris meae segregavit (W. Wigan Harvey, Sancti Irenaei episcopi Lugdu-
nensis: libros quinque adversus haereses [2 vols.; Cambridge: Cambridge University
Press, 1857], 2:354).
94
BETZ 69–70; BURTON 51.
106 Chapter 3. Orienting the Stemma of Galatians

is in apposition functioning like a relative clause: “God, who set me apart . . .

and called [me] . . ., was pleased to reveal his son in me.” In either case, Gal
1:15 states that God is the one who set apart and called Paul.
Intrinsically, it is true that Paul often omits θεός with substantivized parti-
ciples that refer to God. 95 But not always so. In many cases, Paul explicitly
mentions God when there is a contrast with another entity. 96 For Gal 1:15, the
contrast was already signaled in v.10, where Paul rhetorically asks if he is
trying to appeal to people or to God (ἄρτι γὰρ ἀνθρώπους πείθω ἢ τὸν θεόν;).
Verse 15, then, answers the question – it was God who was pleased (εὐδόκη-
σεν ὁ θεός), and the appositional participial clauses elaborate that answer by
stating actions that can be done by God, not Paul’s contemporaries.
Transcriptionally, the shorter reading has usually been valued, except in
cases where accidental omission by skipping between similar letters is appar-
ent. 97 In this case, the omission of ὁ θεός occasioned by skipping between the
two masculine singular nominative articles (ὁ) is certainly feasible. 98 Royse,
on the other hand, has argued that, for early variants the longer reading ought
to be preferred unless it appears to be a harmonization or a grammatical im-
provement. 99 Neither exception is applicable to Gal 1:15. The longer reading
does not appear to be a harmonization, 100 and adding an explicit subject does

95
LIGHTFOOT 82. For example, Gal 1:6 ἀπὸ τοῦ καλέσαντος (“from the one who
called”), 2:8 ὁ γὰρ ἐνεργήσας Πέτρῳ (“the one who worked in Peter”–perhaps the holy
spirit, however), 3:5 ὁ οὖν ἐπιχορηγῶν ... καὶ ἐνεργῶν (“the one who supplies . . . and
works”), 5:8 οὐκ ἐκ τοῦ καλοῦντος ὑμᾶς (“not from the one calling you:); Rom 8:11 ὁ
ἐγείρας Χριστὸν ἐκ νεκρῶν (“the one who raised Christ from the dead”); Phil 1:6 ὁ ἐναρ-
ξάμενος ἐν ὑμῖν ἔργον ἀγαθὸν (“the one who began a good work in you”); and 1 Thess
5:24 πιστὸς ὁ καλῶν ὑμᾶς (“faithful is one who calls you”).
96
See Rom 8:33 θεὸς ὁ δικαιῶν (“It is God who justifies” – in contrast with the one
who accuses), 9:22 θέλων ὁ θεὸς ἐνδείξασθαι (“God wanting to demonstrate” – in contrast
with the mere human of v.20); 1 Cor 3:7 ἀλλ’ ὁ αὐξάνων θεός (“but the one who grows is
God”–in contrast with Paul who planted and Apollos who watered); 2 Cor 4:6 ὁ θεὸς ὁ
εἰπών (“God, the one who said” – in contrast with the god of this age, v.4); and Phil 2:13
θεὸς γὰρ ὁ ἐνεργῶν ἐν ὑμῖν (“for it is God who works in you” – in contrast with Paul).
97
METZGER 13*: “In general the shorter reading is to be preferred, except where (a)
Parablepsis arising from homoeoarcton or homoeoteleuton may have occurred (i.e., where
the eye of the copyist may have inadvertently passed from one word to another having a
similar sequence of letters) . . .”
98
Pace WEISS 83, who claims that there is no apparent accidental reason for the om-
ission: “da hier an einen blossen Schreibfehler nicht zu denken ist.” LÉGASSE 91 n.3 raises
the possibility of an intentional omission to avoid duplication with the participle ὁ ἀφο-
ρίσας or by assimilation to the various participial parallels where Paul does not name God
explicitly.
99
ROYSE 735.
100
The possible parallel to 1 Cor 1:21 εὐδόκησεν ὁ θεός is too remote and contextually
too dissimilar to plausibly prompt a harmonization, though it corroborates that the ex-
pression belongs to Paul’s thinking and style.
 3.2 Examination of Variants for Inner Branch Oppositions 107

not improve the grammar. 101 As a result, the shorter-reading canon is not
applicable to this particular case.
Furthermore, the lack of genealogical coherence plays an important role in
assessing the transcriptional probabilities. The witnesses for the shorter and
longer readings are not close genealogical relatives, so the issue becomes de-
ciding which is more likely to account for this genealogical incoherence. Is it
more likely for two unrelated groups to add the same exact words in the same
exact place coincidentally (i.e., D and , 33, Eastern witnesses adding ὁ θεός
after εὐδόκησεν), or for two unrelated groups to omit the same words inde-
pendently (i.e., F–G and P46–B)? Since it is generally easier to agree ac-
cidentally on what to omit rather than on what to add in its exact wording, the
balance of probabilities favors coincidental omission over addition.
In summary, the longer reading is intrinsically suitable for Paul, and the
omission could have been occasioned by the repetition of the masculine
singular article (ὁ). As a result, the internal evidence favors of the longer
reading, which corroborates, not overturns, the views of textual critics as to
the weight of the external evidence.

Gal 1:19 ἕτερον δὲ τῶν ἀποστόλων οὐκ εἶδον εἰ μὴ Ἰάκωβον τὸν ἀδελφὸν τοῦ
κυρίου.
Var. Western Eastern EDS.
οὐκ εἶδον  33 A P 1739 Ψ Chrys SCC SBL NA RP
×1175 1241S 1611 Byz S BW WH T L
οὐχ εἶδον P46 B
εἶδον οὐδένα D F G
d (b) ×vg
? Marc C

This variation unit is actually a three-way split among the Western branch,
the P46–B cluster, and the Eastern branch with its  and 33 allies.
All three readings deny that Paul saw any apostle other than Cephas and
James the brother of the Lord, but their emphases vary. The least emphatic is
the simple negative οὐκ εἶδον of the Eastern branch and its allies. As for P46–
B, its form of negative οὐχ may be the result of confusion over whether εἶδον
begins with an aspirate, 102 or an elided form of οὐχί (i.e. οὐχ’ εἶδον), 103 which
is a strengthened form of the negative and thus serves to heighten Paul’s de-

101
Contrary to many critics who feel that adding an explicit subject for a verb is a rea-
son to reject the longer reading: BETZ 70 n.132; BURTON 51–52; BUSCEMI 106; LIGHT -
FOOT 82; LONGENECKER 26 n.a; M ATERA 59; METZGER 521–522 (minority view of Bruce
Metzger and Allen Wikgren); MEYER 11; MUSSNER 81; NET 859–860; SIEFFERT 60 n.**.
102
So WEISS 60. BDF 10 § 14 gives various examples of οὐχ with the verb ἰδεῖν. If this
is the case, then the P46–B reading is merely an orthographic variant of οὐκ εἶδον.
103
Cf. the reading οὐχ Ἰουδαϊκῶς at Gal 2:14 in * A C 33 1175 1241S for οὐχὶ
Ἰουδαϊκῶς in  C B Ψ.
108 Chapter 3. Orienting the Stemma of Galatians

nial of not seeing any other apostle. As for the Western branch reading, the
use of the negative pronoun οὐδένα (“no one”) also serves to strengthen
Paul’s denial about not seeing any other apostle. 104
Accordingly, the reading that best explains the origin of the others is the
simple negative οὐκ εἶδον, which the Western branch and the P46–B
strengthened in different ways.

Gal 2:13 12 . . . [Κηφᾶς] ὑπέστελλεν καὶ ἀφώριζεν ἑαυτὸν φοβούμενος τοὺς ἐκ

περιτομῆς. 13 καὶ συνυπεκρίθησαν αὐτῷ καὶ οἱ λοιποὶ Ἰουδαῖοι, ὥστε καὶ
Βαρναβᾶς συναπήχθη αὐτῶν τῇ ὑποκρίσει.
Var. Western Eastern EDS.
καί DFG  33 ACP Ψ Chrys 1611 T
SCC SBL NA RP S WH
T

db ×1175 1241S Byz TL

− ×vg P46 B 1739 M
NA BW WH
M

? Marc

The omission of the adverbial καί (“too”) is found in a few unrelated

witnesses, vg, P46–B, and 1739. Thus, the genealogical coherence of the
external evidence is on the side of the inclusion of καί.
Without the adverb, the first clause of Gal 2:13 states that “the rest of the
Jews pretended to agree with him [i.e., Cephas].” 105 The reference to the “rest
of the Jews” is to be compared with the reference to those of the circumcision
party (τοὺς ἐκ περιτομῆς) at the end of v.12, suggesting two sets of Jews: one
who advocated circumcision for gentiles (v.12) and another who did not
(v.13). At any rate, the first clause of v.13 has the latter group pretending to
go along with the former group and Cephas. Paul’s suggestion of pretense
implies that he thought they knew better but masked their true feelings. If the
adverbial καί is read in v.13, then it would emphasize a parallelism between
the rest of the Jews and the circumcision party plus Cephas. 106 This
parallelism thus implies that both the circumcision party and Cephas also
recognized that withdrawing table fellowship from gentile Christians was
“not walking rightly in the truth of the gospel” (v.14).

104
Cf. BETZ 77 n.201, who considers this an “attempt[] to clarify the text.”
105
J. H. Moulton and G. Milligan, Vocabulary of the Greek Testament (London: Hodder
& Stoughton, 1930; repr. Peabody, Mass.: Hendrickson, 1997), “συνυποκρίνομαι,” 615–
616, recommend the translation “pretended to agree with” based on Polybius, Hist. 3.92.5:
Φάβιος . . . συνυπεκρίνετο τοῖς προθύμως καὶ φιλοκινδύνως διακειμένοις. BDAG “συν-
υποκρίνομαι,” 976–977, has “to join in playing a part or pretending, join in pretense/
hypocrisy.”
106
See Levinsohn, Discourse Features, 99–101, on the two functions of the non-
conjunctive καί, non-contiguous parallelism (“too, also”) and confirmation (“even”). The
adverbial καί before Βαρναβᾶς, on the other hand, indicates confirmation: “so that even
Barnabas got caught up in their pretense.”
 3.2 Examination of Variants for Inner Branch Oppositions 109

Intrinsically, the inclusion of the adverbial καί fits Paul’s context better,
for it underscores Paul’s stance that there is no excuse for their behavior be-
cause everyone recognizes what “the truth of the gospel” is. When they do
not behave properly, even if out of fear (v.13), it is just a pretense on their
part. Transcriptionally, though the addition of καί is possibly an assimilation
to the following adverbial καί before Βαρναβᾶς, 107 it is nonetheless difficult
to explain why isolated scribes should independently insert the same adver-
bial καί before οἱ λοιποὶ Ἰουδαῖοι. 108 Admittedly, the internal evidence for
including καί does not seem particularly decisive, but it agrees with the
genealogical coherence of the external evidence in its favor.

Gal 2:16 εἰδότες δὲ ὅτι οὐ δικαιοῦται ἄνθρωπος ἐξ ἔργων νόμου διὰ πίστεως Ἰησοῦ
Χριστοῦ, καὶ ἡμεῖς εἰς Χριστὸν Ἰησοῦν ἐπιστεύσαμεν,
Var. Western Eastern EDS.
Χριστὸν Ἰησοῦν D F G d b ×vg  A C P Ψ Byz SCC SBL NA RP
T
S WH T L
Ἰησοῦν Χριστόν P46 B 33 ×1175 1241S 1739 1611 BW WH
M

Ἰησοῦν Chrys
? Marc

Here, the opposition is between P46–B and the Western branch, while the
Eastern branch is divided at every level. Internal evidence must decide this
reading. Intrinsically, Paul has a tendency to prefer the order Christ Jesus
except in genitives, 109 and transcriptionally, the order of Ἰησοῦν Χριστόν
appears to be an assimilation to the order of Ἰησοῦ Χριστοῦ in the preceding
clause. 110 These considerations together suggest that Χριστὸν Ἰησοῦν is the
authorial reading.

Gal 3:7 6 Καθὼς Ἀβραὰμ ἐπίστευσεν τῷ θεῷ, καὶ ἐλογίσθη αὐτῷ εἰς δικαιοσύνη·
7 γινώσκετε ἄρα ὅτι οἱ ἐκ πίστεως, οὗτοι υἱοί εἰσιν Ἀβραάμ.
Var. Western Eastern EDS.
υἱοί εἰσιν P46 B  P 1241S Ψ Chrys SCC SBL NA
T T
BW WH T L
εἰσιν υἱοί D F G d b ×vg 33 ×1175 A C 1739 1611 Byz M M
RP S T L
? Marc

This is another opposition of P46–B against the Western branch, while the
Eastern branch is split. Internal evidence, then, must decide the variant.

107
So WEISS 113.
108
BURTON 109; LONGENECKER 63 n.d. BUSCEMI 182 suggests that it might have been
omitted because it seemed superfluous, especially with an adverbial καί in the following
clause. ZAHN 115 n.46 puts forward the views that the adverbial καί implies that Cephas
too was a hypocrite.
109
BURTON 122–122.
110
B also transposes Χριστοῦ Ἰησοῦ of the preceding clause.
110 Chapter 3. Orienting the Stemma of Galatians

The transposition of εἰσιν and υἱοί hardly affects the sense, but the em-
phasis is different. Both readings mean “these are Abraham’s sons.” On the
one hand, the reading with the discontinuous noun phrase οὗτοι υἱοί εἰσιν
Ἀβραάμ features the rhetorical figure of hyperbaton, 111 which serves to em-
phasize the first element, “sons.” 112 On the other hand, the word order of
οὗτοι εἰσιν υἱοὶ Ἀβραάμ keeps the noun phrase υἱοὶ Ἀβραάμ together and
does not emphasize “sons.” In light of the context, the emphasis on “sons” is
a better fit intrinsically, because it anticipates Paul’s point about the Galatians
being sons in Christ Jesus (3:26). 113
Transcriptionally, though the hyperbaton of οὗτοι υἱοί εἰσιν Ἀβραάμ lends
a more sophisticated air to the clause, οὗτοί εἰσιν υἱοὶ Ἀβραάμ seems to be
the easier reading because it is the unmarked word order and it keeps the
noun phrase together. 114 Furthermore, the transposition of οὗτοι υἱοί εἰσιν to
οὗτοί εἰσιν υἱοὶ could be accidental. Specifically, the υἱοί of the οὗτοι υἱοί
εἰσιν reading is susceptible to dropping out by parablepsis occasioned by the
common ending in -οι, yielding οὗτοί εἰσιν, which can be fixed by adding the
omitted υἱοὶ after the εἰσιν. Thus, both the intrinsic and transcriptional pro-
babilities favor the word order υἱοί εἰσιν.

Gal 3:21a ὁ οὖν νόμος κατὰ τῶν ἐπαγγελιῶν τοῦ θεοῦ; μὴ γένοιτο. . . .
Var. Western Eastern EDS.
τοῦ θεοῦ D (F G −τοῦ) b  33 AC 1739 Ψ Chrys T
SCC SBL NA RP S
×vg ×1175 1241S 1611 Byz T
WH T L
T

− d P46 B M
NA BW WH L
M M

? Marc P

The external evidence for the omission of τοῦ θεοῦ (“of God”) is limited to
the Old Latin d and to the common ancestor of P46 and B. Since the Old
Latin d is distantly related to P46–B and its closest relatives include the
phrase, its agreement with P46–B is most likely coincidental. In other words,
the inclusion of this phrase is strongly supported by the genealogical co-
herence of the external evidence and ought to be adopted for the critical text
of Galatians, unless internal evidence can show that P46–B has managed to
preserve the earliest form of the text.
The noun phrase τῶν ἐπαγγελιῶν τοῦ θεοῦ (“the promises of God”) is a
fully articulated construction in which the head noun precedes the modifier.
According to recent research on this construction, the first article indicates

111
Levinsohn, Discourse Features, 57–60 under the term “discontinuous constituent”;
SMYTH 679 § 3028 as “hyperbaton.”
112
BUSCEMI 256; SIEFFERT 174.
113
BUSCEMI 249; MATERA 118.
114
LONGENECKER 108 n.b.; SIEFFERT 173 n.**: “jene Stellung ist, als die regel-
mässigere, Korrektur”; ZIMMER² 329.
 3.2 Examination of Variants for Inner Branch Oppositions 111

that the referent of the noun can be identified from the noun alone, while the
second article means the modifier either confirms the identification of the
noun or specifies a subset of the entities identified by the noun. 115 For the
particular noun phrase in Gal 3:21, the modifier phrase τοῦ θεοῦ should not
be strictly necessary to identify what τῶν ἐπαγγελιῶν refers to. And this is the
case. The head noun refers back to the promises (αἱ ἐπαγγελίαι) that were
spoken to Abraham in v.16. Because of this, the modifier is not essential to
understanding the flow of Paul’s argument.
Nevertheless, the modifier phrase τοῦ θεοῦ has a function of its own. It is
confirmatory. It allows Paul’s recipients to be sure that the identification with
the promises spoken to Abraham in v.16 is correct. Although the passive ἐρ-
ρέθησαν (“were said”) in v.16 allowed Paul to omit the actual agent of the
speaking (i.e. God), those familiar with Genesis would know that it was God
who said those promises to Abraham about his offspring (e.g. Gen 13:14–17,
17:3–8; cf. 24:7–8). Mentioning in v.21 that the promises were God’s allows
the recipients to confirm that it was the promises in these passages that Paul
had in mind. The specification of τοῦ θεοῦ also fosters the cohesion of Paul’s
argument. In v.19 Paul states that the Law (i.e. the Torah) was ordained by
angels through the hand of a mediator (i.e., Moses). In v.20, Paul makes a
distinction between the mediator and God. Then v.21 launches into the
rhetorical question: “So is the Law against the promises [of God]?” The
modifier phrase τοῦ θεοῦ is what provides the connecting link between the
mention of God in v.20 and the promises spoken in v.16. Without the
modifier phrase “of God,” the connection between the promises of v.16 and
God in v.20 would not be so clear.
As for the intrinsic probabilities, the modifier phrase τοῦ θεοῦ is admitted-
ly not necessary for the sense of the argument, for the argument can be made
without it. 116 Yet this is to be expected for a post-nominal modifier whose
head noun has an article – it not strictly necessary to identify what the noun
refers to. The mere fact that a phrase appears superfluous is not sufficient to
hold that an author did not write it, because all writing has material not strict-
ly necessary to convey the propositional meaning of the content. 117 What the

115
Bakker, Noun Phrase, 289–290.
116
Indeed, some critics hold that the terseness is in accordance with Paul’s style, who
often refers to God’s promises without a modifier: BURTON 193; LONGENECKER 143;
METZGER 524.
117
Paul Maas, Textual Criticism (trans. Barbara Flower; Oxford: Clarendon, 1958), 14–
15: “[Interpolations] are particularly dangerous, as it is often very difficult to prove that a
text based on them has been deformed (whereas scribal blunders normally produce obvious
nonsense); and in texts where such an interpolation has been demonstrated much becomes
suspect simply because it appears to be superfluous. And it is so easy simply to cut out
what could easily be dispensed with! But there is undoubtedly superfluous (or at least not
demonstrably indispensable) matter in every original. So very thorny problems arise.”
112 Chapter 3. Orienting the Stemma of Galatians

modifier phrase does do, however, is make the argument cohesive and clear.
It connects the unity of God in v.20 to his promises in v.16. Since the
modifier phrase τοῦ θεοῦ is eminently suitable for its context, the intrinsic
probabilities lean toward its inclusion.
The primary argument for omitting the words τοῦ θεοῦ is the text-critical
canon of preferring the short reading. 118 For example, Bruce Metzger sug-
gests that “the words τοῦ θεοῦ may be a natural addition made by copyists
who recalled such passages as Ro 4.20 or 2 Cor 1.20.” 119 Yet the parallels at
Rom 4:20 (εἰς δὲ τὴν ἐπαγγελίαν τοῦ θεοῦ) and 2 Cor 1:20 (ὅσαι γὰρ ἐπαγγε-
λίαι θεοῦ) are physically remote and literarily inexact. In fact, these parallels
are conceptual and within the orbit of Paul’s thinking, a fact that bolsters the
intrinsic probabilities for the longer reading. 120 Another problem with the
shorter reading canon is that, in the earliest period of transmission (to which
the common ancestor of P46 and B certainly belongs), there was a tendency
to make small omissions, often accidentally. 121 In fact, the coherence of the
external evidence suggests that accidental omission occurred once, if not
twice. The phrase is missing in the Old Latin d, when its closest relatives
include it. The phrase is included in such diverse and distantly related texts as
the Western branch, the basic  and 33, and the Eastern branch, so that
multiple independent additions of the same words seems progressively
unlikely. 122
Therefore, with the strong intrinsic probabilities for inclusion and the weak
transcriptional probabilities against it, the internal evidence agrees with the
coherence of the external evidence that Gal 3:21 should read τῶν ἐπαγγελιῶν
τοῦ θεοῦ.

118
E.g. METZGER 13*: “In general the shorter reading is to be preferred,” with ex-
ceptions for accidental omissions and material that looks superfluous, harsh, or impious to
scribes. WEISS 90 sees no reason why τοῦ θεοῦ should have dropped out and argues that it
was added to emphasize even more the impossibility of a conflict between the law and the
promise, but SIEFFERT 218 n.* counters this suggestion by arguing that “of God” was de-
leted to avoid the idea that promises are opposed to the Law as if the reason was that the
Law did not also come from God.
119
METZGER 525; so also BUSCEMI 317; LÉGASSE 261 n.6; ZAHN 178 n. 43. MUSSNER
250 n.31 also suggests an influence from Gal 3:18 δι’ ἐπαγγελίας κεχάρισται ὁ θεός,
though here is the promise is singular, not plural.
120
BETZ 173 n.95.
121
ROYSE 703–736. For a fuller treatment of this canon, see the above discussion about
this at Gal 1:15.
122
LONGENECKER 143; MEYER 129; NET 860–861. More reluctantly, BURTON 193;
BUSCEMI 317; METZGER 526. MEYER 129 also suggests the omission of τοῦ θεοῦ was a
harmonization to v.16, which lacks τοῦ θεοῦ.
 3.2 Examination of Variants for Inner Branch Oppositions 113

Gal 3:21b ὄντως ἐκ νόμου ἂν ἦν ἡ δικαιοσύνη

Var. Western Eastern EDS.
ἐκ νόμου D F G d b ×vg  33 A C 1241S 1739 Ψ Chrys SCC SBL NA RP
1611 Byz S BW WH T L
ἐν νόμῳ P46 B
? Marc P

Instead of ἐκ νόμου (“of the Law”), P46 and B read the phrase ἐν νόμῳ (“by
the Law”) against every other witness. Transcriptionally, this appears to be a
scribal harmonization in reminiscence of the same prepositional phrase with
the cognate verb in Gal 3:11 ὅτι δὲ ἐν νόμῳ οὐδεὶς δικαιοῦνται παρὰ τῷ
θεῷ, 123 though there is an intervening example of ἐκ νόμου at v.18 εἰ γὰρ ἐκ
νόμου ἡ κληρονομία, albeit with “inheritance” rather than with “righteous-
ness.” This indicates that the ἐν νόμῳ reading is secondary.

Gal 3:24 ὥστε ὁ νόμος παιδαγωγὸς ἡμῶν γέγονεν εἰς Χριστόν, ἵνα ἐκ πίστεως
δικαιωθῶμεν.
Var. Western Eastern EDS.
γέγονεν D F G  33 A C 1241S 1739 Ψ Chrys SCC SBL NA RP
1611 Byz S BW WH T L
ἐγένετο P46 B
? Marc d b ×vg P

This is a reading exclusively shared between P46 and B. The change from the
perfect tense γέγονεν to the more common aorist tense ἐγένετο appears to
foster a temporal sense for εἰς Χριστόν (“until Christ”), rather than as a goal
(“about Christ”). 124 There is a similar change in B from the perfect γέγονα to
the aorist ἐγενόμην in 1 Cor 13:11. 125 Transcriptionally, then, the change to
the aorist is the easier reading and therefore secondary.

Gal 4:23 ἀλλ’ ὁ μὲν ἐκ τῆς παιδίσκης κατὰ σάρκα γεγέννηται, ὁ δὲ ἐκ τῆς ἐλευθέρας
δι’ ἐπαγγελίας.
Var. Western Eastern EDS.
μέν DFGdb  33 ×1175 A C 1241S 1739 Ψ SCC SBL NA RP
Chrys 1611 Byz T
S WH T L
T

− ×vg P46 B M M M
BW WH T L
? Marc 126 P

123
BURTON 194; WEISS 56.
124
So BUSCEMI 318. Cf. BETZ 178.
125
WEISS 43. Unfortunately, P46 is lacunose in 1 Cor 13:11, so it is not clear whether
this change had happened in the common ancestor of P46 and B.
126
There is no Latin word corresponding to μέν in Tertullian, Marc. 5.4.8, sed qui ex
ancilla carnaliter natus est, qui vero ex libera per repromissionem, while other Old Latin
witnesses read quidem for μέν. On the one hand, John J. Clabeaux, A Lost Edition of the
Letters of Paul: A Reassessment of the Text of the Pauline Corpus Attested by Marcion
114 Chapter 3. Orienting the Stemma of Galatians

The particle μέν has the discourse function of marking its clause as less im-
portant than the corresponding δέ clause. 127 Intrinsically, this reading fits
Paul’s argument well because it highlights the positive member of the
allegory. Transcriptionally, the omission of this little word, even if it was ac-
cidental, 128 has the effect of foregrounding the first clause, perhaps to
strengthen the force of ἀλλά. 129 On balance, therefore, the internal evidence
weakly favors the inclusion of μέν.

Gal 5:26 μὴ γινώμεθα κενόδοξοι, ἀλλήλους προκαλούμενοι, ἀλλήλοις φθονοῦντες.

Var. Western Eastern EDS.
ἀλλήλοις D F  33 AC 1739 Ψ Chrys SCC SBL NA RP S
×1175 1611 Byz T T M
BW WH T L
ἀλλήλους G P46 B P 1241S Byzpt M M T
WH T L
? Marc d b ×vg

The external evidence is sharply divided. This variation unit features an

agreement between P46 and B against their closest relatives, though the ac-
cusative ἀλλήλους later acquired many scattered supporters, especially in the
Byzantine text. 130 Intrinsically, φθονέω takes a dative object, though the ac-
cusative is occasionally attested, especially in later Greek. 131 The trans-
criptional evidence is more clear, as the accusative form is an ostensible har-
monization to the preceding accusative ἀλλήλους. 132

(CBQMS 21; Washington, D.C.: The Catholic Biblical Association of America, 1989), 86,
takes this lack in Tertullian as evidence that Marcion’s Greek text also lacked the particle.
On the other hand, Schmid, Marcion, 20, discounts the evidentiary value of Tertullian for
this and other cases of “textual trivia” due to Tertullian’s tendency for omission.
127
Levinsohn, Discourse Features, 170.
128
BURTON 253.
129
Clabeaux, Lost Edition, 86, strongly argues in favor of the μέν-less reading as
original on the grounds of its attestation in P46–B, the Vulgate, and (he contends) Marcion,
and on the grounds that the shorter reading is also the harder reading due to a “horror
asyndeti.” Clabeaux’s argument about avoiding asyndeton overlooks the presence of the
conjunction ἀλλά. For an assessment of Clabeaux’s text-critical judgment, see Gordon D.
Fee, review of John J. Clabeaux, A Lost Edition of the Letters of Paul: A Reassessment of
the Text of the Pauline Corpus Attested by Marcion, CBQ 53 (1991): 320.
130
Byzantine manuscripts with the accusative ἀλλήλους include: 1 6 69 104 131 209
330 460 489 517 614 618 796 910 927 999 1242 1243 1245 1315–1573 1319 1424 1646
1734 1738 1827 1837 1881 1891 2125 2147 2400 2464.
131
BDAG “φθονέω,” 1054; BDF 84 § 152(1). Conceded by LIGHTFOOT 214, who
nonetheless opts for the accusative on the basis of external attestation and its lack of
grammaticality as the harder reading (cf. BUSCEMI 529).
132
So BURTON 324; MEYER 318; WEISS 27–28.
 3.2 Examination of Variants for Inner Branch Oppositions 115

Gal 6:4 τὸ δὲ ἔργον ἑαυτοῦ δοκιμαζέτω ἕκαστος, . . . 5 ἕκαστος γὰρ τὸ ἴδιον

φορτίον βαστάσει.
Var. Western Eastern EDS.
ἕκαστος D F G d b ×vg  33 A C 1241S 1739 Ψ SCC SBL NA RP
Chrys 1611 Byz T
S BW WH T L
− P46 B WH
M

? Marc P

The omission of the pronoun ἔκαστος (“each”), found only in P46 B, has the
effect of making the commandment τὸ ἔργον ἑαυτοῦ δοκιμαζέτω (“let him
examine his own work”) apply to the person in the previous verse who is self-
deceived about his importance, rather than to each Galatian. Despite the
superficial suitability of this reading, the reference to ἕκαστος in the fol-
lowing γάρ-clause shows that Paul had shifted the topic to each person. 133 As
a result, the omission is probably inadvertent. 134 Thus, the internal evidence
favors the inclusion of ἕκαστος.

Gal 6:11 Ἴδετε πηλίκοις ὑμῖν γράμμασιν ἔγραψα τῇ ἐμῇ χειρί.

Var. Western Eastern EDS.
πηλίκοις D F G d b ×vg  A C 1241S 1739 Ψ SCC SBL NA RP
Chrys 1611 Byz T
S BW WH T L
ἡλίκοις P46 B 33 WH
M

? Marc P

These two related correlative pronouns differ by a single letter. Instead of the
interrogative correlative pronoun πηλίκοις used for an indirect question (“See
how large letters I have written to you in my own hand”), P46 and B have an
exclamatory correlative ἡλίκοις (“See what large letters . . .”). 135 The latter is
more classical and may thus represent an Atticism, 136 but this change could
also be due to an inadvertent omission of a letter. 137 In either case, the inter-
nal evidence, especially the transcriptional probabilities, are against the read-
ing of P46, B, and 33.

133
LONGENECKER 277.
134
BURTON 332. WEISS 92 thinks the omission occurred out of pure negligence (“auf
reiner Nachlässigkeit”).
135
See BDF 159–160 § 304.
136
BUSCEMI 597.
137
WEISS 27 suggests that this omission is a near-haplography of ΠΗ.
116 Chapter 3. Orienting the Stemma of Galatians

Gal 6:12 μόνον ἵνα τῷ σταυρῷ τοῦ Χριστοῦ μὴ διώκωνται.

Var. Western Eastern EDS.
Χριστοῦ DFGdb  33 AC 1739 Ψ Chrys SCC SBL NA RP S
×vg 1241S 1611 Byz T
WH T L
Χριστοῦ Ἰησοῦ P46 B ×1175 BW WH
M

? Marc P

The internal evidence is against the longer reading. Intrinsically, the genitive
Χριστοῦ Ἰησοῦ is somewhat less common in Paul than Ἰησοῦ Χριστοῦ. 138
Transcriptionally, the full phrase Χριστοῦ Ἰησοῦ appears to be an assimila-
tion to v.14 ἐν τῷ σταυρῷ τοῦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ. 139 The longer
reading also has a semantic effect that suits the later theological interests of
the proto-Orthodox; specifically, the addition of “Jesus” to the “cross of
Christ” underscores that the human Jesus really did die on the cross as Christ,
and is therefore anti-Separationist in effect. 140 For these reasons, the longer
reading appears secondary.

Gal 6:13 οὐδὲ γὰρ οἱ περιτεμνόμενοι αὐτοὶ νόμον φυλάσσουσιν ἀλλὰ θέλουσιν ὑμᾶς
περιτέμνεσθαι, ἵνα ἐν τῇ ὑμετέρᾳ σαρκὶ καυχήσωνται.
Var. Western Eastern EDS.
περιτεμνόμενοι Marc D ×vg  33 A C 1739 Chrys SCC SBL NA
1241S 1611 Byzpt 141 T T M
S WH T L
περιτετμημένοι F G d b P46 B ×1175 Ψ Byzpt 142 M M T
RP BW WH T L
? P

Though the Western and Eastern branches are split, P46 and B stand in
opposition to  and 33 as to the tense of the participle. The present participle
περιτεμνόμενοι (“for not even those who are getting circumcised themselves
keep the Law”) is notoriously obscure. 143 Does it refer to the circumcision
party, 144 or does it refer to converted proselytes? 145 By contrast, the perfect
participle περιτετμημένοι (“for not even those who have been circumcised
themselves keep the Law”) is less ambiguous as it refers to those already

138
BURTON 358–359.
139
Pace WEISS 88, who claims the addition lacks an analogy.
140
See Bart D. Ehrman, The Orthodox Corruption of Scripture: The Effect of Early
Christological Controversies on the Text of the New Testament (New York: Oxford
University Press, 1993), 154, 175 nn.159 and 162.
141
K 075 131 323 460 547 876 910 945 1242 1243 1315 1424 1448 1836 1874 1982
2125 2147.
142
L 056 1 ×6 69 205 209 223 226 330 440 489 517 614 618 796 927 999 1022 ×1175
1270 1319 1352 1573 1646 1734 1735 1738 1780 1799 1827 1854 1891 1960 2138 2400
2401 2412 2423 2815 2892.
143
BETZ 316–317.
144
So LONGENECKER 292 as meaning “those who belong to the circumcision.”
145
So LIGHTFOOT 223.
 3.2 Examination of Variants for Inner Branch Oppositions 117

circumcised, i.e., Judaizers (or proselytes to Judaism) who were pressuring

the Galatian gentiles to get circumcised. 146 For this reason, the present parti-
ciple is the harder reading, and hence it is favored by the internal evidence. 147

3.2.3 Inner Branch c

The inner branch c oppositions refer to those textual variants where the
Eastern branch agrees among its members against , 33, P46, B, and the
Western branch. For this inquiry, the Eastern branch includes two sub-
branches: one with A, C, P, and 1241S; and another with 1739, Ψ, the
Harklean group, and the Byzantine text. There are eight such oppositions, as
follows:

Gal 1:11 Γνωρίζω δὲ ὑμῖν, ἀδελφοί, τὸ εὐαγγέλιον τὸ εὐαγγελισθὲν ὑπ’ ἐμοῦ ὅτι οὐκ
ἔστιν κατὰ ἄνθρωπον·
Var. Western Eastern EDS.
δέ b P46 *,² A 1241S 1739 Ψ 1611 Byz M
SCC RP WH L
γάρ D* F G d ×vg B ¹ 33 ×1175 SBL NA S
T
BW WH T
− P Chrys
? Marc C

The external evidence is equally divided, with the Eastern branch reading δέ,
and the Western branch reading γάρ. Also divided is each of the early groups.
B disagrees with P46, and * disagrees with 33. 148 With such an even balance
of the external evidence, this variation unit must be decided by internal evi-
dence.
Some critics have tried to resolve the issue by appealing to general tran-
scriptional probabilities. They claim that it is more likely, as a general matter,
for scribes to change the blander δέ to γάρ than the reverse. 149 There is very
little evidence, however, to support this claimed tendency. 150 Within the
textual transmission of Galatians, there are only sporadic changes between δέ

146
BURTON 353; LIGHTFOOT 223.
147
WEISS 52, however, considers the present participle to be a harmonization to the
present infinitive περιτέμνεσθαι later in the verse.
148
Moisés Silva, Interpreting Galatians (2d ed.; Grand Rapids, Mich.: Baker, 2001),
44–45, argues that the first corrector of ¹ may have been contemporaneous with the
original hand (*) and therefore of equal weight. So also ZIMMER¹ 485.
149
NET 493 n.11; Zuntz, Text, 203–204.
150
In Moisés Silva’s study collating ten manuscripts against the Nestle-Aland text in
Galatians and Philippians, he concludes that γάρ was changed to δέ in five verses (Gal
1:11, 4:25c, 5:17b, 6:5, and Phil 1:19), while δέ was changed to γάρ at Gal 3:20 and 4:18.
Moisés Silva, “Text and Language in the Pauline Corpus: With Special Reference to the
Use of Conjunctions in Galatians,” Neot 24 (1990): 273–281 at 277; see also Silva,
Interpreting, 46–47.
118 Chapter 3. Orienting the Stemma of Galatians

and γάρ in either direction. 151 Among the six early papyri studied by Royse,
there are very few singular readings involving a change between δέ and γάρ,
specifically, just two changes in each direction. 152 As a result, the evaluation
of the internal evidence must carefully consider the particular context of Gal
1:11.
Apart from the conjunction, there are three features within v.11 that
suggest that this verse begins a major section in the letter. 153 First, there is the
disclosure formula (or meta-comment), γνωρίζω ὑμῖν (“I am letting you
know”), which calls attention to what follows. Second, there is the attention-
getting vocative, ἀδελφοί (“fellow Christians”). Third, there is the prolepsis
of the subordinate clause’s subject, τὸ εὐαγγέλιον τὸ εὐαγγελισθὲν ὑπ’ ἐμοῦ
(“the gospel preached by me”), which is raised into the independent clause in
order to establish it as the topic of the succeeding discourse. Elsewhere in
Paul, a similar set of features also signals the beginning of a new major
section, for example, 1 Cor 15:1 Γνωρίζω δὲ ὑμῖν, ἀδελφοί, τὸ εὐαγγέλιον ὃ
εὐηγγελισάμην ὑμῖν (“I am letting you know, fellow Christians, about the
gospel which we preached to you”) and 2 Cor 8:1 Γνωρίζομεν δὲ ὑμῖν,
ἀδελφοί, τὴν χάριν τοῦ θεοῦ τὴν δεδομένην ἐν ταῖς ἐκκλησίαις τῆς Μακε-
δονίας (“We are letting you know, fellow Christians, about God’s grace that
has been given among the churches of Macedonia”). 154 Thus, all these
features combine to indicate that a new section of the letter begins here,
having the theme of Paul’s gospel. 155

151
From δέ to γάρ: 3:8 (1881 pesh Chrys), 3:20 (33), 3:23 (489–927), 4:13 (pesh), 4:18
(33 104), 5:19 (pesh). From γάρ to δέ: 2:6 (pesh), 3:18 (pesh), 4:24 (pesh), 4:25b, 5:5 (88
1175 1245), 5:13 (F G b 1243 2138 Chrys hark–1611–1505–2495), 5:17 (2138 1611–
1505–2495), 5:18 (Eastern branch), 6:5 (Ψ 1 33 440 1315).
152
For P45, from δέ to γάρ at Luke 11:11 (ROYSE 171). For P46, from γάρ to δέ at Rom
12:20 (ROYSE 320). For P47, none (ROYSE 385, only change in conjunction is from ἵνα to
ὅτι at Rev 14:13). For P66, from δέ to γάρ at John 9:28 (ROYSE 528). For P72, none
(ROYSE 598, only change in conjunction is from διότι to ὅτι at 1 Pet 1:12). For P75, from
γάρ to δέ at John 5:13 (ROYSE 680). Thus, there are two changes from δέ to γάρ, and two
changes from γάρ to δέ. Much more common among the papyri, however, is the change
from δέ to τε.
153
Stephen H. Levinsohn, “Los rasgos discursivos comparativos aplicados a la traduc-
ción de Gálatas” (2009), 21; online at http://www.recursosteologicos.org/Documents/
Galatas_traduccion_discursivo.pdf. Accessed June 1, 2010. These features are also noted
by BUSCEMI 79–81.
154
Though these examples do not strictly involve prolepsis, they do establish the object
of γνωρίζω as the topic for the succeeding discourse.
155
BUSCEMI 79; BURTON 35; LÉGASSE 75; LONGENECKER 20–22; ROHDE 49; ZAHN
53. James. D. Hester, “The Rhetorical Structure of Galatians 1:11–2:14,” JBL 103 (1984):
223–233 at 226, identifies vv.11–12 as the stasis, the basic issue of the case, before begin-
ning the narratio proper at v.13. Others, however, identify the beginning of a major section
 3.2 Examination of Variants for Inner Branch Oppositions 119

The conjunction δέ has two fundamental usages, adversative and

connective, 156 and both of these reflect a basic function to mark some kind of
a shift or mid-to-low level discontinuity in the discourse. 157 As a marker of
discontinuity, δέ is an appropriate particle to signal the beginning of a major
section of the letter. 158 Intrinsically, this is an excellent fit for the overarching
structure of the letter, but there is a loose end brought about by this reading.
If v.11 begins a new section, then v.10 ends a section 159 with its statement
that if Paul were still pleasing people, he would not be a slave of Christ. The
logic of Paul’s statement is not completely self-evident, but presumably it
would mean that as Christ’s slave, he would be pleasing Christ instead of
other people. At any rate, the discontinuity signaled by the particle δέ
suggests that the explanation for v.10 is not to be found directly in v.11.
Verse 10 is left hanging in its immediate context.
As for the conjunction γάρ, exegetes have struggled to account for its use
in v.11. The usual sense of γάρ is that it explains, clarifies, or strengthens the
previous point. 160 Weiss, for example, holds that the conjunction γάρ is indis-
pensable for clarifying the otherwise obscure statement of Paul in v.10 that he
“would not be Christ’s slave.” 161 This interpretation, however, closely cou-
ples v.11 to v.10 such that it is difficult to see v.11 as starting a new section.
Accordingly, some exegetes suggest that the γάρ in v.11 actually attaches to

at v.10, including Robert G. Hall, “The Rhetorical Outline for Galatians: A Reconsi-
deration,” JBL 106 (1987): 277–287 at 284–285; MARTYN 136; MUSSNER 62.
156
E.g., BDAG “δέ,” 213: “one of the most common Gk. particles, used to connect one
clause to another, either to express contrast or simple continuation.”
157
See especially Stephanie L. Black, Sentence Conjunctions in the Gospel of Matthew:
καί, δέ, τότε, γάρ, οὖν and Asyndeton in Narrative Discourse (JSNTSS 216; STG 9;
London: Sheffield Academic Press, 2002), 142–178. See also Randall Buth, “On
Levinsohn’s ‘Development Units’,” START 5 (October 1981), 53–56; Stephen H.
Levinsohn, “Some Constraints on Discourse Development in the Pastoral Epistles,” in
Stanley E. Porter and Jeffrey T. Reed, eds., Discourse Analysis and the New Testament:
Approaches and Results (JSNTSS 170; STG 4; Sheffield: Sheffield Academic Press,
1999), 316–333 at 320–325; Steven E. Runge, A Discourse Grammar of the Greek New
Testament: A Practical Introduction for Teaching & Exegesis (Lexham Bible Reference
Series; Bellingham, Wash.: Logos Research Systems, 2010), 42–52 at 46.
158
LONGENECKER 22. Curiously, MATERA 52 argues that δέ has the effect of
“providing a closer connection with the previous verse” than γάρ; the basis for this claim,
however, is not explained.
159
BUSCEMI 75 distinguishes between Gal 1:6–10 as the exordium and 1:11–2:21 as the
narratio.
160
BDAG “γάρ,” 189–190. So also Black, Sentence Conjunctions, 254–281; Levinsohn,
“Constraints,” 319–320; Runge, Discourse Grammar, 69–73.
161
WEISS 67. So also LÉGASSE 76; MARTYN 141: “Paul’s freedom from the need to
please his audience, his new identity as the slave of Christ, has its ground in the gospel
itself.”; MUSSNER 65 n.110; Silva, Interpreting, 48–49.
120 Chapter 3. Orienting the Stemma of Galatians

v.9 (or vv.6–9), skipping v.10 as a parenthesis, 162 but this function is more
appropriate for a resumptive δέ. Others understand the γάρ in v.11 as an in-
ferential, a function more appropriate for οὖν. 163 These unusual senses pro-
posed for γάρ in v.11 ought to be taken as a sign for what it is, namely, that
the usual explanatory sense of γάρ does not fit the larger structure of the letter
well, however appropriate this explanatory sense may seem for connecting
v.10 directly to v.11.
Putting these considerations together, the case for δέ over γάρ is compel-
ling. 164 Intrinsically, the particle δέ is an eminently suitable conjunction for
the beginning of a major section in Paul’s epistle. 165 This conjunction also
comfortably fits Paul’s style in similar examples at 1 Cor 15:1 and 2 Cor
8:1. 166 That these examples occur in other epistles makes it unlikely that the
δέ reading is scribal assimilation to such distant parallels. 167 After all, harmo-
nization is much more common to local parallels, 168 and the local context
abounds with γάρ (vv.10, 12–13). 169 Furthermore, the obscurity of v.10
causes the δέ in v.11 to become the harder reading for scribes. It is the ob-
scurity of v.10 that needed clarification, and changing the conjunction to γάρ
enabled scribes to clarify why Paul would be Christ’s slave, namely, that the
gospel preached by him did not come from anybody else. 170 Yet this
clarification is a local, superficial change – at the expense of the structure of
the letter, a global consideration scribes are less likely to appreciate. In other
words, the discontinuity between v.10 and v.11 signaled by the δέ is the

162
BURTON 35; cf. ZAHN 53.
163
MARTYN 141. So also Hester, “Rhetorical Structure,” 225 n.11: “Gar in 1:11 does
have the sense of connection but also acts subtly inferentially, i.e., it could be translated,
‘so I want you to know.’” For the instances of γάρ in vv.12–13, Hester takes these particles
with their usual strengthening or explanatory meaning.
164
LONGENECKER 20–22. So also EADIE 33; LIGHTFOOT 79–80; NET 493 n.11; Zuntz,
Text, 203–204.
165
EADIE 33; LONGENECKER 22.
166
LONGENECKER 22. So also LIGHTFOOT 80: “more after the Apostle’s manner”;
Zuntz, Text, 204: “These, however, are for once genuine parallels.”
167
Pace METZGER 521; ROHDE 50; SIEFFERT 52 n.**; ZAHN 53; ZIMMER¹ 486. So also
Silva, Interpreting, 47, who recognizes that the argument could go the other way.
168
ROYSE 737: “Further sharpening of the generally accepted principle that scribes tend
to harmonize may prove useful, for it seems evident that the major influence on out scribes
was the immediate context.”
169
MEYER 11. On the contrary, ROHDE 50 claims that the fourfold use of γάρ in vv.10–
13 inspired scribes to replace the one in v.11, but assimilation–not dissimilation–is more
common among the scribes. Similarly SIEFFERT 52 n.**; ZIMMER¹ 486.
170
Neither LIGHTFOOT 80 (“the latter [γάρ] would seem to be the obvious connecting
particle to transcribers”) nor Zuntz, Text, 204 (“The superficial appropriateness of γάρ is
here so striking”), explain their reasoning, but Silva, Interpreting, 48, no fan of the reading
δέ, helpfully lays out their implicit argument.
 3.2 Examination of Variants for Inner Branch Oppositions 121

harder reading that prompted scribes to make the connection between these
two verses clearer.

Gal 2:12 11 Ὅτε δὲ ἦλθεν Κηφᾶς εἰς Ἀντιόχειαν, κατὰ πρόσωπον αὐτῷ ἀντέστην, ὅτι
κατεγνωσμένος ἦν. 12 πρὸ τοῦ γὰρ ἐλθεῖν τινας ἀπὸ Ἰακώβου μετὰ τῶν
ἐθνῶν συνήσθιεν· ὅτε δὲ ἦλθεν, ὑπέστελλεν καὶ ἀφώριζεν ἑαυτὸν
φοβούμενος τοὺς ἐκ περιτομῆς.
Var. Western Eastern EDS.
τινας . . . ἦλθεν D* F G B  33 T
SCC T L
τινα . . . ἦλθεν d P46
τινας . . . ἦλθον b ×vg ×1175 A C 1241S 1739 Ψ SBL NA RP
Chrys 1611 Byz S BW WH T
M

? Marc P

The external evidence for ἦλθεν (“he came”) is found in all early branches of
the text except for the Eastern branch, which has the plural ἦλθον (“they
came”). Indeed, some opponents of the ἦλθεν reading have conceded that it
goes back to the archetype, if not the autograph itself. 171 Given the strength of
the external evidence, the internal evidence would have to strongly be in
favor of the ἦλθον to prevail.
The ἦλθεν reading is difficult and stylistically clumsy. 172 The subject of
ἦλθεν does not point back to the subject of πρὸ τοῦ ἐλθεῖν τινας (“before the
coming of certain people”) as its antecedent, due to the mismatch in number.
Rather, the subject of ὅτε δὲ ἦλθεν has to go further back to v.11 and refer to
Cephas, 173 as must the immediately preceding συνήσθιεν and the following
ὑπέστελλεν and ἀφώριζεν. Thus the ὅτε δὲ ἦλθεν of v.12 is resumptive of ὅτε
δὲ ἦλθεν Κηφᾶς in v.11. 174 On the other hand, both alternative readings τινα
. . . ἦλθεν (P46) and τινας . . . ἦλθον (Eastern) are easier on the surface. The

171
HORT App.121: “an unusually well attested Western reading . . . or a primitive
error”; LIGHTFOOT 112: “it may possibly be due to an error of the original amanuensis”;
ZIMMER¹ 493.
172
F. F. Bruce, The Epistle to the Galatians: A Commentary on the Greek Text (NIGTC;
Grand Rapids, Mich.: Eerdmans, 1982), 130, notes regarding the singular verb: “this would
be intolerably awkward; the verb here rather catches up the preceding ἐλθεῖν.”
173
James D. Hester, “The Use and Influence of Rhetoric in Galatians 2:1–14,” TZ 42
(1986): 386–408 at 403; Andreas Wechsler, Geschichtsbild und Apostelstreit: Eine
forschungsgeschichtliche und exegetische Studie über den antiochenischen Zwischenfall
(Gal 2,11–14) (BYNT 62; Berlin: Walter de Gruyter, 1991), 316.
174
A less likely possibility, as Origen apparently understood it, is that the singular
subject of ἦλθεν refers to the last mentioned individual, James. Origen, Contra Celsum 2.1
ὅτι Πέτρος ἔτι φοβούμενος τοὺς Ἰουδαίους, παυσάμενος τοῦ μετὰ τῶν ἐθνῶν συνεσθίειν,
ἐλθόντος Ἰακώβου πρὸς αὐτὸν ἀφώριζεν ἑαυτόν (“that Peter, still being afraid of the Jews
and having stopped from eating together with the Gentiles, when James came to him, he
separated himself”). Also Tasker, The Greek New Testament, 438, who understands the
singular verb to refer to a subsequent arrival by James.
122 Chapter 3. Orienting the Stemma of Galatians

finite verb ἦλθεν or ἦλθον refers back to the ἐλθεῖν τινα or ἐλθεῖν τινας, re-
spectively. 175 With the P46 reading, ἐλθεῖν τινα . . . ἦλθεν, the incident in-
volves the coming of a certain person from James, 176 while the plural reading,
ἐλθεῖν τινας . . . ἦλθον, refers to certain people coming from James. In either
case, the subject of the following verbs switches back to Peter, who began to
withdraw and separate himself.
Intrinsically, the ἦλθεν reading (with ἐλθεῖν τινας) is admittedly awkward,
but not impossibly so. 177 In fact, the reading ἦλθεν makes sense upon further
consideration. Paul’s account of the Antioch incident begins with a statement
that when Cephas came (ὅτε δὲ ἦλθεν Κηφᾶς) to Antioch, he confronted him.
After giving background information in v.12a that Cephas used to eat with
gentiles before the coming of people from James, Paul restarts the account by
repeating the triggering phrase ὅτε δὲ ἦλθεν in v.12. On this reading, only the
arrival of Cephas triggered the incident, which is what Paul claimed in v.11.
With the ἦλθον reading, on the other hand, there are two separate triggering
events for the Antioch incident. Paul stated in v.11 that “when Cephas came
to Antioch, I opposed him to his face,” but in v.12, Cephas had already been
there for an unspecified amount of time eating with gentiles. 178 As a result, it
is not until the coming of people from James that Cephas began to withdraw
and thus provoked Paul’s opposition. This goes against Paul’s claim in v.11
that he confronted Cephas when he came to Antioch. 179 By contrast, the
ἦλθεν reading is consistent: according to both v.11 and v.12b, the incident
happened when Cephas came.
Transcriptionally, either reading could be explained as a scribal harmoni-
zation. On the one hand, scribes could have found the singular ἦλθεν in v.12
difficult, because it appears to go back to the preceding infinitive πρὸ τοῦ γὰρ
ἐλθεῖν τινας with its plural subject. Readings τινα . . . ἦλθεν and τινας . . .
ἦλθον would thus reflect two different attempts to fix the difficulty – by

175
It is grammatically possible that ἦλθον here means “I [scil. Paul] came,” but this
reading conflicts with both v.11 (when Cephas came) and v.12 (when some from James
came).
176
MATERA 85 speculates that P46’s reading may be in reference to the agitator of Gal
5:10.
177
Pace HORT 224: “unquestionably wrong,” App.121: “It cannot in any case be
genuine”; LIGHTFOOT 112: “it can scarcely have been the word intended by St Paul”;
METZGER 523–524: “the sense of the passage seems to demand the plural ἦλθον.” None of
these explain why the reading is wrong beyond their bare assertion.
178
The imperfect συνήσθιεν suggests a customary activity over many meals (e.g. BETZ
447).
179
Hester, “Use and Influence,” 403–404: “It makes little sense for Paul to say in 2:11
that he opposed Peter when he came to Antioch if in fact the confrontation occurred after
he had been there awhile.”
 3.2 Examination of Variants for Inner Branch Oppositions 123

harmonizing the first or the second element, respectively. 180 On the other
hand, if the plural ἦλθον were original, scribes could have been motivated to
assimilate it either to the singular of the adjoining verbs συνήσθιεν, ὑπέστελ-
λεν, and ἀφώριζεν, 181 or to the phrasing of ὅτε δὲ ἦλθεν Κηφᾶς of v.11. 182
Accordingly, the intrinsic probabilities favor the reading ἦλθεν and the tran-
scriptional probabilities by themselves are not decisive enough to overcome
the intrinsic probabilities. This finding is consistent with the external
evidence. 183
Though the textual difference is merely that of a single letter, this variant
has significant ramifications for Paul’s biography. The reading with the sin-
gular ἦλθεν results in a markedly different understanding of the incident.
What triggered the incident was not the coming of people from James (that
happened earlier) but the coming of Cephas himself, just as Gal 2:11 ex-
plicitly states. Instead of being intimidated at Antioch into changing his mind,
Cephas came to Antioch with no intention of eating with the gentiles. This is
what Paul found objectionable. After all, Paul had been in Antioch for some
time after the meeting in Jerusalem before Cephas arrived (v.11). During this
time, it is reasonable to suppose that Paul with his colleague Barnabas had
been eating with the gentile Christians. All along, Paul thought he had an
understanding with Cephas, James, and John back in Jerusalem that – with
their acknowledgment of his gospel to the uncircumcision – his uncircum-
cised gentiles would be welcomed into the community (v.9). The issue of
table fellowship had already come up in Antioch when those from James
came while Cephas was still eating with the gentiles there (v.12a), so Paul
believed the meeting in Jerusalem must have resolved that issue. When
Cephas came to Antioch, however, he refused to engage in table fellowship
with Paul’s gentiles but withdrew and separated himself (v.12b). To make
matters worse, even Barnabas went along with Cephas (v.13). Paul had to
object, immediately. His apostolic mission of bringing the gospel to the
gentiles was at stake.

180
Pace Tasker, Greek New Testament, 438, who asserts that “the choice of variants is
between the singular in both places and the plural in both places, and that the reference is
to a single occasion.”
181
METZGER 523–524.
182
HORT App.121; LIGHTFOOT 112; MATERA 86; METZGER 523–524; MEYER 53;
WEISS 37.
183
David I. Yoon, “The Antioch Incident and a Textual Variant: ‘ΗΛΘΟΝ’ or
‘ΗΛΘΕΝ’ in Galatians 2:12,” ExpT 125 (2014): 423–439, cites and broadly endorses the
dissertation’s conclusion on this variant.
124 Chapter 3. Orienting the Stemma of Galatians

Gal 4:3 οὕτως καὶ ἡμεῖς, ὅτε ἦμεν νήπιοι, ὑπὸ τὰ στοιχεῖα τοῦ κόσμου ἤμεθα
δεδουλωμένοι·
Var. Western Eastern EDS.
ἤμεθα D* F G P46  33 SCC SBL NA
×1175 S BW WH T
M

ἦμεν B A C 1241S 1739 Ψ Chrys 1611 Byz T

RP T L
? Marc d b ×vg P

The evidence for the first person plural imperfect active ἦμεν is confined to
the Eastern branch plus B. Its usual allies  and 33, however, agree with the
Western branch and P46 in having the middle form ἤμεθα. In Hellenistic
Greek, the verb “to be” is in the process of changing from active forms as in
classical Greek into middle forms as in Modern Greek. 184 Thus, the middle
form ἤμεθα is not out of place for the Hellenistic Greek of Galatians. 185
The transcriptional probabilities, moreover, are against ἦμεν. First, the
active ἦμεν is the more classical form, and so the middle ἤμεθα would be the
harder reading for later, educated scribes. 186 Second, the presence of the
active ἦμεν earlier in the verse could well have triggered scribes to harmonize
the more unusual middle to the earlier active. 187 Because B is not closely
related to the Eastern branch, the genealogical coherence of their agreement
ought to be viewed as coincidental scribal improvements of the text.

Gal 4:19a τέκνα μου, οὓς πάλιν ὠδίνω ἄχρις οὐ μορφωθῇ Χριστὸς ἐν ὑμῖν·
Var. Western Eastern EDS.
τέκνα D* F G d B * 1739 SCC SBL NA
M T
BW WH T L
τεκνία b ×vg C 33 ×1175 A C 1241S Ψ Chrys 1611 Byz T M
RP S WH L
? Marc P46 P

The external evidence favors τέκνα (“children”) because it includes the West-
ern branch, B, *, and 1739, while the diminutive τεκνία (“little children”) is
mostly an Eastern branch phenomenon. 188 Intrinsically, τέκνα is in accord-

184
BDF 49 § 98; LONGENECKER 100 n.a.
185
The alternation between ἦμεν and ἤμεθα in Gal 4:3 is curious. A possible reason,
admittedly speculative, is that it helps to give the pluperfect periphrastic ἤμεθα
δεδουλωμένοι a more passive feel; compare with the non-passive Acts 20:8 οὗ ἦμεν
συνηγμένοι (“where we had gathered”).
186
BURTON 216; BUSCEMI 365; SIEFFERT 265 n.*; ZIMMER³ 294–295.
187
WEISS 38: “In Gal. 4, 3 ist das ημεθα (Tisch. WH. Treg. a. R.) in ABCEKLP dem
unmittelbar vorhergehenden ημεν conformirt.” So also LONGENECKER 100 n.a; SIEFFERT
265 n.*.
188
Those who argue that the external attestation supports τέκνα include BURTON 249;
BUSCEMI 413; and even by those against it on internal grounds: LIGHTFOOT 178;
LONGENECKER 195.
 3.2 Examination of Variants for Inner Branch Oppositions 125

ance with Pauline usage elsewhere, 189 while τεκνία is only found elsewhere in
the New Testament among the Johannine writings. 190 Transcriptionally,
scribes may have modified τέκνα to the diminutive τεκνία in order to
aggrandize Paul or to have a more superficial fit with Paul’s birthing imagery
(ὠδίνω “I am groaning in labor”), 191 perhaps under the influence of the
Johannine usage. 192 To be sure, τεκνία lends a more affectionate tone to the
verse, 193 but this does not explain why scribes would remove the power of the
expression by changing τεκνία to τέκνα. These considerations all point in
favor of reading τέκνα in Gal 4:19.

Gal 4:28 Ὑμεῖς δέ, ἀδελφοί, κατὰ Ἰσαὰκ ἐπαγγελίας τέκνα ἐστέ.
Var. Western Eastern EDS.
ὑμεῖς . . . D* F G d b P46 B 33 ×1175 1739 SCC SBL NA
M T
ἐστέ S BW WH T L
ἡμεις . . . ×vg  A C P Chrys 1611 T M
RP WH T
ἐσμέν 1241S Ψ Byz
? Marc

With support from the Western branch, P46 and B, 33, and 1739, the external
evidence favors the second person plural ὑμεῖς … ἐστέ (“you are”) over the
first person plural ἡμεῖς . . . ἐσμέν (“we are”), supported by  and the Eastern
branch. Nevertheless, the internal evidence must be checked to see if it
overturns the external evidence.
The second person plural reading is the harder reading because it might
appear to scribes that Paul has implicitly excluded himself from this
statement: “But you, fellow Christians, are children of the promise according
to Isaac.” Despite this apparent exclusion of Paul, the second plural forms are
appropriate because it strongly assures the gentile Galatians that they are

189
E.g., Rom 8:16, 17, 9:7, 8; 1 Cor 4:14, 7:14; 2 Cor 6:13, 12:14; Gal 4:28, 31; Phil
2:15; 1 Thess 2:7, 11; Phlm 10 cf. Eph 2:3, 5:1, 8, 6:1, 4; Col 3:20, 21. LIGHTFOOT 178,
however, points out that this evidence is weaker than it appears because only Eph 6:1 and
Col 3:20 (both disputed Paulines) have vocatives like Gal 4:19.
190
John 13:33; 1 John 2:1, 12, 28, 3:7, 18, 4:4, 5:21. The lack of Pauline examples of
τεκνία in comparison to τένκα is decisive for BETZ 233 n.146; SCHLIER 151 n.4.
191
ZAHN 221–222 n.20. So also ROHDE 189 n.88.
192
WEISS 14. So also LÉGASSE 334 n.1, who suggests that τεκνία was probably
influenced (“vraisemblablement influencée”) by the Johannine writings. On the contrary,
LONGENECKER 195 suggests that τεκνία is the harder reading because “an early scribe
might have wanted to conform it to Paul’s usual practice” (so also Zimmer, Exegetische
Probleme, 202), but, given the relative infrequency of a vocative τέκνα in Paul compared
with that of the vocative τεκνία in the Johannine writings, it seems more likely that scribes
would change the vocative to the more common New Testament form.
193
LIGHTFOOT 178, who states that “the diminutive, expressing both the tenderness of
the Apostle and the feebleness of his converts, is more forcible.” So also LONGENECKER
195; SIEFFERT 273.
126 Chapter 3. Orienting the Stemma of Galatians

children of Abraham. 194 Moreover, the conclusion in v.31 explicitly includes

Paul: “Therefore, fellow Christians, we are not children of the slave girl but
of the free woman.” With the textual possibility of Paul being excluded from
Isaac’s promise in v.28, 195 scribes would have been inclined to harmonize
v.29 to the local context of v.31, as well as v.26 (“the free woman, who is our
mother”). 196 As a result, the internal evidence does not overturn but rather
corroborates the external evidence in favor of the second person plural forms.

Gal 4:30 . . . οὐ γὰρ μὴ κληρονομήσῃ ὁ υἱὸς τῆς παιδίσκης μετὰ τοῦ υἱοῦ τῆς
ἐλευθέρας.
Var. Western Eastern EDS.
κληρονομήσῃ (F G −μή) AC 1739 Ψ Chrys SCC RP S L
M

1611 Byz
κληρονομήσει D (P46 −μή)  33 1241S Byzpt 197 SBL NA BW
B ×1175 WH T L
T

? Marc d b ×vg P

The aorist subjunctive κληρονομήσῃ enjoys much support from the Eastern
branch, as well as the Western Greek-Latin diglot F and G. The future indica-
tive κληρονομήσει has the support of the basic witnesses P46, B, , and 33,
as well as D and 1241S in the Western and Eastern branches, respectively. On
balance, therefore, the external evidence appears to favor κληρονομήσει, 198
but the internal evidence may tell a different story.
With the coalescence of ει and ῃ to the same sound starting in the second
century, 199 the difference between these two variant readings throughout
much of the transmission history may have merely been orthographic. Further
obliterating the distinction between these forms is that both the future indica-
tive and the aorist subjunctive denote a strong denial with οὐ μή. 200 Because
of the closeness in pronunciation and meaning of these different spellings,

194
BETZ 249 n. 105; BURTON 265; LÉGASSE 361; LONGENECKER 216.
195
Similarly, MARTYN 443 suggests that the original distinction of ethnicity between
gentile and Jew was reinterpreted by this variant reading as a Christian-Jew distinction.
196
BETZ 249 n. 105; BURTON 265; BUSCEMI 450; EADIE 371; LÉGASSE 361 n.4;
LIGHTFOOT 183; MATERA 171; METZGER 528; MUSSNER 328–329 n.60; NET 500 n.4;
ROHDE 202 n.74; SIEFFERT 293 n.*; WEISS 25; ZAHN 240 n.52. Zuntz, Text, 107 n.2, notes
that this internal argument was so strong that even textual critics before the discovery of
P46 and 1739 had already adopted the ὑμεῖς . . . ἐστέ reading despite the ordinarily suspect
B-Western agreement.
197
The future indicative is found in the following Byzantine witnesses: 075 ×6 69 ×88
205 209 330 460 489 517 618 927 ×1175 1245 1319 1646 1734 1735 1738 1780 1836
×1837 1874 2138 2400 2423 ×2464.
198
Scholars who found the external evidence decisive, mainly on the strength of P46, ,
B, and D, include: BUSCEMI 450; SIEFFERT 295 n.*.
199
Allen, Vox Graeca, 74.
200
SMYTH 626 § 2755; BDF 184 § 365.
 3.2 Examination of Variants for Inner Branch Oppositions 127

manuscripts can vary in their spelling without affecting the pronunciation or

meaning. In fact, there are many such variants between these verb forms
throughout the New Testament; among grammatically unambiguous forms,
there is only one case of the future (Matt 16:22) but numerous cases of the
aorist. 201 This evidence suggests that the usual Koine usage at the time the
New Testament was written was to employ the aorist subjunctive. 202 In other
words, this general pattern of usage suggests that the intrinsic probabilities
back the aorist subjunctive.
Transcriptionally relevant is the fact that Gal 4:30 is a quotation of Gen
21:10. There the Septuagintal text reads οὐ γὰρ κληρονομήσει, 203 without μή
and with the future indicative. 204 Agreement with the Septuagint is not neces-
sarily a sign of originality, 205 especially when there are other indications that
Paul is adapting the Old Testament text, as he has here. 206 However, scribes
are more likely to change Old Testament quotations toward the Septuagint
than away from it. 207 This tendency is most evident in P46, whose omission
of μή and use of the future have both conformed the text to the Septuagint. In
addition, dozens of Byzantine manuscripts also feature the harmonization of
the verb. Since the aorist subjunctive is the non-Septuagintal form, tran-
scriptional considerations also favor the aorist subjunctive. 208

201
BDF 184 § 365(1)-(3).
202
So also Winer, Grammar, 528. Note that BUSCEMI 451 suggests that this is a
transcriptional reason why scribes would have preferred the aorist subjunctive.
203
Alfred Rahlfs and Robert Hanhart, eds., Septuaginta (rev. ed., Stuttgart: Deutsche
Bibelgesellschaft, 2006), 1:28.
204
Note, however, the Codex Alexandrinus includes the μή.
205
Pace LIGHTFOOT 184.
206
Paul adds μή (also attested in the fifth century Codex Alexandrinus) and changes τοῦ
υἱοῦ μου Ἰσαάκ to τοῦ υἱοῦ τῆς ἐλευθέρας.
207
Bruce M. Metzger, The Text of the New Testament, its Transmission, Corruption,
and Restoration (3d ed.; New York: Oxford University Press), 197–198: “Frequently Old
Testament quotations are enlarged from the Old Testament context, or are made to
conform more closely to the Septuagint wording.” Kurt Aland and Barbara Aland, The Text
of the New Testament: An Introduction to the Critical Editions and to the Theory and
Practice of Modern Textual Criticism (2d ed.; Grand Rapids, Mich.: Eerdmans, 1989), 290:
“Quotations from the Old Testament which differ from the text of the Septuagint popular
in the Church were often corrected to agree with it.”
But note the warning of Aland and Aland, Text, 281: “Neither should the commonly
accepted rule of thumb that variants agreeing with parallel passages or with the Septuagint
in Old Testament quotations are secondary be applied in a purely mechanical way.”
208
So MEYER 214.
128 Chapter 3. Orienting the Stemma of Galatians

Gal 5:17a 16 Λέγω δέ, πνεύματι περιπατεῖτε καὶ ἐπιθυμίαν σαρκὸς οὐ μὴ τελέσητε.
17 ἡ γὰρ σὰρξ ἐπιθυμεῖ κατὰ τοῦ πνεύματος, τὸ δὲ πνεῦμα κατὰ τῆς σαρ-
κός, ταῦτα γὰρ ἀλλήλοις ἀντίκειται, ἵνα μὴ ἃ ἐὰν θέλητε ταῦτα ποιῆτε.
Var. Western Eastern EDS.
γάρ D* F G P46 B  33 SCC SBL NA
d b ×vg BW WH T L
δέ ×1175 A C P 1241S 1739 Ψ Chrys 1611 Byz RP S
? Marc

The reading δέ is confined to the Eastern branch; the other early branches and
witnesses of the text (Western, P46–B, , and 33) all bear witness to γάρ.
Though some scholars consider this external evidence to be decisive support
in favor of γάρ, 209 the internal evidence must nonetheless be considered.
This passage begins with Paul telling the Galatians to “walk in the spirit
and in no way satisfy the desire of the flesh” (5:16). Supporting this state-
ment, as indicated by the conjunctive particle γάρ, are two oppositions: “for
the flesh desires against the spirit and the spirit against the flesh” (v.17a–b).
These antimonies are then followed by the statement that “these things are
opposed to each other so that you do not do whatever you desire” (v.17c–
d). 210 The role of the conjunctive particles γάρ and δέ in v.17c is to indicate
how this clause is to relate to the previous statement. On the one hand, if the
particle is γάρ, then the purpose of the clause is to explain or clarify why they
desire against each other, 211 namely, that they are so generally opposed to
each other as to make doing what one wants impossible. 212 On the other hand,
if δέ is read, then it would mark a discontinuity between v.17c and v.17a–
b. 213 Although v.17c is not adverse to v.17a–b, there are two other possible
analyses. One possibility is that v.17c is broader in scope than v.17a–b, so the
discontinuity marked by the particle would be a shift from the specific (i.e.,
desiring against each other) to the general (i.e., being opposed to each
other). 214 Another possibility is that v.17c–d presents a consequence of Paul’s

209
So BETZ 279 n.73; BUSCEMI 527; MUSSNER 376 n.16. Perhaps also LONGENECKER
238 n.e.
210
The ἴνα clause is probably result rather than purpose (MARTYN 494), but LONG -
ENECKER 246 takes it as the purpose of the personified flesh and spirit not to let people do
whatever they want.
211
The discourse marker γάρ signals that the previous point is to be strengthened, sup-
ported, or confirmed by the current discourse unit. See, e.g., BDAG “γάρ,” 189–190;
Black, Sentence Conjunctions, 254–281; Levinsohn, “Constraints,” 319–320; Runge,
Discourse Grammar, 69–73.
212
Levinsohn, “Los rasgos,” 69.
213
See Black, Sentence Conjunctions, 142–178; Buth, “Levinsohn’s,” 53–56; Levin-
sohn, “Constraints,” 320–325; Runge, Discourse Grammar of the Greek New Testament,
46.
214
Black, Sentence Conjunctions, 165, has an example of δέ in Matt 27:16 marking a
shift in the other direction, from the general (Pilate’s custom) to the specific (Barabbas).
 3.2 Examination of Variants for Inner Branch Oppositions 129

advice in v.16; with this understanding, the δέ signals a shift resuming the
main line of discourse in v.16 and marking v.17a–b to be a parenthesis. 215
Intrinsically, both particles fit the context tolerably well. Transcriptionally,
however, the assimilation from γάρ to δέ could well have been prompted by
the panoply of δέ particles in the immediate context (vv.16, 17 [immediately
preceding], 18, and 19) and the apparent difficulty of having a consequence
(v.17d) in what should be a confirmatory statement. 216 Thus, the internal
evidence is at least consistent with, if not supportive of, the external evidence
in favor of γάρ.

Gal 5:21 19 φανερὰ δέ ἐστιν τὰ ἔργα τῆς σαρκός, ἅτινά ἐστιν . . . 21 φθόνοι, μέθαι,
κῶμοι καὶ τὰ ὅμοια τούτοις . . .
Var. Western Eastern EDS.
φθόνοι (Marc) P46 B  33 P SCC SBL NA
M M M
S WH T L
φθόνοι φόνοι D F G d b ×1175 A C 1739 Ψ Chrys T
RP S BW T
T

(×vg homicidia) 1241S 1611 Byz L

T

The external evidence is split. The Western and Eastern branches include the
word φόνοι (“murders”) after the similar-sounding φθόνοι (“envies”), while
the basic witnesses, P46–B, , and 33 lack the word. As tempting as Mar-
cion’s attestation for the lack of φόνοι may seem, it is best not to put much if
any weight upon his testimony here due to Epiphanius’s inconsistency in
quoting Marcion. 217 Then again, Irenaeus, Adv. haer. 5.11.1, quoting Gal 5:21
as including invidiae, ebrietates, comissationes (φθόνοι, μέθαι, κῶμοι), 218 is
an early witness to a form of the text without φόνοι. Though the external

215
ZAHN 262.
216
ZAHN 262: “Da das Neue, was hierzu v.17b hinzubringt, vor allem in dem
Absichtssatz liegt, dieser aber nicht eine Bestätigung, sondern eher eine Folge der Aussage
von v.16 enthält, so begreift es sich daß man das vorzüglich bezeugte γάρ teils durch δέ
(. . .) teils durch οὖν (. . .) teils durch καί (. . .) ersetzte.”
217
Though Schmid, Marcion, 318, prints the sequence φθόνοι, μέθαι, κῶμοι, derived
from Epiphanius, Pan. 42.12.3, for his reconstructed text of Marcion’s Galatians, he also
put these words in italics, indicating that they are unsecure readings. The problem is anoth-
er citation of the same passage by Epiphanius, Pan. 42.11.8, in which two of the words are
quoted in the singular instead of the plural and, more importantly, φθόνοι is replaced by
φόνοι. Thus, while Epiphanius does not provides evidence for the inclusion of both φθόνοι
and φόνοι in his text of Marcion, it is unclear which word was actually present in his
source. As a result, Schmid is skeptical that the nature of Epiphanius’s quotation is exact
enough to be useful for inferring such “textual trivia” (192).
218
Harvey, Sancti Irenaei, 2:348.
130 Chapter 3. Orienting the Stemma of Galatians

evidence is divided, the earliest evidence slightly favors the lack of the word
φόνοι, 219 but the internal evidence has to be consulted.
In this case, the usual text-critical canon of preferring the shorter reading is
hardly applicable, 220 because it could easily have arisen by accident since the
two words φθόνοι and φόνοι have the same ending in -όνοι. 221 After writing
any of the letters ονοι in φθόνοι, a scribe could have leapt to the same letters
in φόνοι and accidentally deleted the word φόνοι. 222 Conversely, the addition
of φόνοι could have arisen from harmonization to Rom 1:29 φθόνου φόνου
ἔριδος δόλου, 223 though the parallel is distant and non-verbatim, 224 and it
provides evidence that the collocation of φθόνοι and φόνοι was on Paul’s
mind. 225 Another possibility is harmonization to a wordplay popular in
antiquity between these two words, for instance, Euripides, Tro. 768–769
(εἶτα δὲ Φθόνου Φόνου τε Θανάτου). 226
Nevertheless, if Paul had been thinking of φόνοι when he wrote this list of
the deeds of the flesh, it seems odd that Paul included it next to sins of over-
indulgence in v.21 (e.g., μέθαι, κῶμαι “drunkenness, carousing”) rather than
sins of discord in v.20 (e.g., ἔχθραι, ἔρις “enmities, discord”) as he did with
φθόνου φόνου ἔριδος in Rom 1:29. 227 On the other hand, the strength of the
sin φόνοι (“murders”) is so strong that a scribe may have thought the list was
incomplete without it, and added the terrible sin. A similar motive seems to

219
Most critics, however, consider this external evidence to be strong grounds for
omission, including: BURTON 304; BUSCEMI 528; METZGER 529; SIEFFERT 325 n.*; ZAHN
265 n.98.
220
In fact, ROYSE 703–736 argues that little omissions were so common among the
papyri that in many cases the longer reading ought to be preferred unless the longer
reading appears late, appears to be a harmonization, or appears to be a grammatical
improvement (735). The first two of these grounds arguably apply to Gal 5:21.
221
Cf. The first exception in Metzger 13* against preferring the shorter reading:
“Parablepsis arising from homoeoarcton or homoeoteleuton may have occurred (i.e., where
the eye of the copyist may have inadvertently passed from one work to another having a
similar sequence of letters).”
222
Among those who include φόνοι: EADIE 419; LIGHTFOOT 212 (bracketed); MEYER
283; WEISS 125. This point is often conceded by proponents of the shorter reading,
including: BUSCEMI 528; LÉGASSE 427 n. 5; METZGER 529.
223
BETZ 284 n.119; BUSCEMI 528; METZGER 529; ROHDE 243; SIEFFERT 325 n.*.
LONGENECKER 248 n.d and ZAHN 265 n.98 also suggest Mark 7:21 or Matt 15:19 for a
source, neither of which has φόνοι but not φθόνοι, however.
224
WEISS 125.
225
LIGHTFOOT 212.
226
BETZ 284 n.119. Winer, Grammar, 659 notes that paronomasia is common in Paul.
227
MUSSNER 383 n.60 argues that φόνοι ought to have been associated with ζῆλος in
the lists, but all his examples (1 Macc 8:16, T. Sim. 4.5, 1 Clem 3.2, 4.7, 5.2) involve
φθόνοι instead of φόνοι. LÉGASSE 427 n. 5 also found it incredible that there would be
“assassins” in the Galatian communities.
 3.2 Examination of Variants for Inner Branch Oppositions 131

have been in effect for those witnesses who add μοιχεία (“adultery”) before
πορνεία (“fornication”) in v.19.
Accordingly, the internal evidence is frustratingly inconclusive, though to
some extent it points against the inclusion of φόνοι in the text. With the ex-
ternal evidence also slightly favoring the omission of φόνοι, it seems that the
least objectionable conclusion is that the earliest form of Gal 5:21 lacked this
word.

3.2.4 Inner Branch d

The Western branch diverges from its closest relatives along inner branch d
in about seventy places. Virtually all of the differences are secondary, and
they are discussed in greater detail in Chapter 5. For purposes of orienting the
stemma, however, it suffices to examine the earliest attested stage of the
Western branch, namely, the most recent common ancestor of Marcion and
the Greek-Latin diglots D, F, and G. There are seven such variation units to
consider:

Gal 2:9 καὶ γνόντες τὴν χάριν τὴν δοθεῖσάν μοι, Ἰάκωβος καὶ Κηφᾶς καὶ Ἰωάννης,
οἱ δοκοῦντες στῦλοι εἶναι,
Var. Western Eastern EDS.
Ἰάκωβος καὶ ×vg B  33 C 1241S 1739 Ψ Chrys SCC SBL NA RP S
Κηφᾶς 1611 Byz BW WH T L
Ἰάκωβος καὶ P46
Πέτρος
Ἰάκωβος A
Πέτρος καὶ Marc D F G
Ἰάκωβος db
? P

The Western branch places the name “Peter” first, followed by James and
John, while all the other witnesses place James first, and – with the exception
of P46 – use the name “Cephas” instead of Peter. Thus, there are two sub-
variants to be considered.
The internal evidence strongly favors the name “Cephas.” Intrinsically, it
is Paul’s typical usage (with the exception of Gal 2:7–8), and the Aramaic
name is the harder reading since the Greek name “Peter” was more famil-
iar. 228 In addition, the presence of the name “Peter” in vv.7–8 probably
prompted the harmonization of the name at v.9. 229
The internal evidence also favors the James-first order. It is the more
difficult reading as it gives James the pride of place instead of Peter. 230 In

228
BURTON 59, 95; BUSCEMI 126; METZGER 522.
229
ROYSE 298 n.561.
230
BUSCEMI 151; LIGHTFOOT 109; METZGER 523.
132 Chapter 3. Orienting the Stemma of Galatians

addition, the Peter-first order could be a harmonization to 1:18–19, where

Cephas/Peter is mentioned before James in a chronological context.

Gal 3:14 ἵνα εἰς τὰ ἔθνη ἡ εὐλογία τοῦ Ἀβραὰμ γένηται ἐν Χριστῷ Ἰησοῦ, ἵνα τὴν
ἐπαγγελίαν τοῦ πνεύματος λάβωμεν διὰ τῆς πίστεως.
Var. Western Eastern EDS.
ἐπαγγελίαν ×vg B 33 A C 1241S 1739 Ψ Chrys SCC SBL NA RP
1611 Byz S BW WH T L
εὐλογίαν Marc 231 D* F G d b P46 ×88 489 927
? P

The reading εὐλογίαν is found not only in the Western branch, but also in
P46, the mixed Alexandrian 88, and the Byzantine 489 and 927. The lack of
genealogical coherence for this reading suggests that it arose independently
several times. Indeed, it appears to be a harmonization to the local context.
Gal 3:14 is the first appearance of the word ἐπαγγελία (“promise”) in the
letter and the context up to this point concerns the blessing (εὐλογία) of Abra-
ham. 232 The unexpected appearance of “promise” probably induced scribes to
assimilate it to εὐλογίαν in the immediately preceding clause. 233

Gal 4:6 Ὅτι δὲ ἐστε υἱοί, ἐξαπέστειλεν ὁ θεὸς τὸ πνεῦμα τοῦ υἱοῦ αὐτοῦ εἰς τὰς
καρδίας ἡμῶν κράζον· αββα ὁ πατήρ.
Var. Western Eastern EDS.
υἱοί ×vg P46 B  33 A C P 1241S 1739 Ψ Chrys SCC SBL NA RP
1611 Byz S BW WH T L
υἱοὶ θεοῦ Marc D F G d b
?

The addition of θεοῦ is exclusively found within the Western branch.

Intrinsically, the genitive θεοῦ is certainly appropriate for the context,
because that relationship of being sons of God is on Paul’s mind back in 3:26.
Nevertheless, since the concept was on Paul’s mind, it is not necessary to
express it explicitly; it can be assumed. Transcriptionally, the genitive θεοῦ
appears to be an assimilation to 3:26 πάντες γὰρ υἱοὶ θεοῦ ἐστε. 234 Thus, the
shorter reading ought to be favored as the harder reading.

231
Marcion’s text as quoted by Tertullian, Marc.5.3.11, is fairly divergent but at any
rate supports the blessing variant: accepims igitur benedictionem spiritualem per fidem
(Schmid, Marcion, 123).
232
Clabeaux, Lost Edition, 104, further supports the ἐπαγγελίαν reading as an instance
of Paul’s style of using “bridge words” to shift topics from one point to another. For exam-
ples of such “bridge words,” Clabeaux, Lost Edition, 104 n.53, cites σοφία in 1 Cor 1:17
and τὸ ἀσθενές in 1 Cor 1:25.
233
METZGER 525.
234
METZGER 526.
 3.2 Examination of Variants for Inner Branch Oppositions 133

Gal 4:8 Ἀλλὰ τότε μὲν οὐκ εἰδότες θεὸν ἐδουλεύσατε τοῖς φύσει μὴ οὖσιν θεοῖς·
Var. Western Eastern EDS.
ἐδουλεύσατε P46 B  33 A C 1739 Ψ Chrys SCC SBL NA RP
τοῖς ... θεοῖς 1241S 1611 Byz S BW WH T L
τοῖς ... θεοῖς DFG
ἐδουλεύσατε d b ×vg
τοῖς ... θεοῖς Marc 235
δουλεύετε
? P

The reading with the verb at the end of the clause is found only in Western
witnesses. By placing the verb at the end, this word order is more highly
marked and serves to emphasize the fact that they were enslaved to false
gods. Yet, intrinsically, the point of the passage has more to do with the
Galatians’ enslavement rather than the nature of those to whom they were
enslaved. 236 Transcriptionally, the more emphatic reading seems to be super-
ficially the easier reading, as it heightens the anti-pagan force of the verse,
but it is also conceivable that the transposition could have been due to an
accidental skip of the verb, followed by an in-line correction at the end of the
clause.

Gal 5:14a ὁ γὰρ πᾶς νόμος ἐν ἑνὶ λόγῳ πεπλήρωται, ἐν τῷ· ἀγαπήσεις τὸν πλησίον
σου ὡς σεαυτόν.
Var. Western Eastern EDS.
ἐν ἑνὶ λόγῳ ×vg P46 B  33 A C 1739 Ψ Chrys SCC SBL NA RP
1241S 1611 Byz S BW WH T L
ἐν ὑμῖν ἐν ἑνὶ λόγῳ D* F G d b
ἐν ὑμῖν Marc 237
? P

The Western branch includes the phrase ἐν ὑμῖν (“among you”), so that Paul
states that the whole Law has been fulfilled among the Galatians. The internal
evidence does not provide much guidance on deciding this variant. The short-
er reading could easily be the result of a leap from one ἐν to the next, and the

235
Tertullian, Marc. 5.4.5: si ergo his, qui ‘in natura sunt’ dei, servitis. Schmid,
Marcion, 116, suggests that the omission of μή is due to a corruption of τοῖς μὴ φύσει into
τοῖς τῇ φύσει.
236
Indeed, commentators struggle to account for Paul’s attitude toward pagan gods
here; e.g. BETZ 215–215; LONGENECKER 179–180.
237
Schmid, Marcion, 130–131 and 182, citing Tertullian, Marc. 5.4.12: tota enim,
inquit, lex ‘in vobis’ adimpleta est: dileges proximum tuum tamquam te, and Epiphanius,
Pan. 42.11.8, 42.12.3 ὁ γὰρ πᾶς νόμος ὑμῖν πεπλήρωται· ἀγαπήσεις τὸν πλησίον ὡς σεαυ-
τόν, respectively. Schmid proposes that the phrase ἐν ὑμῖν is possibly a mechanical
corruption of ἐν ἑνὶ <λόγῳ>, as is clear in Marcion, which the rest of the Western branch
then conflated.
134 Chapter 3. Orienting the Stemma of Galatians

prepositional phrase ἐν ὑμῖν makes the statement more contingent to the his-
torical situation of the Galatians. On the other hand, the longer reading seems
to be motivated by its context (e.g. v.13 ὑμεῖς γὰρ ἐπ’ ἐλευθερίᾳ ἐκλήθητε,
“for you were called to freedom”) to strengthen the point by specifying that
the Law has already been fulfilled among the gentile Galatians. 238 On
balance, the shorter reading seems preferable.

Gal 5:14b ὁ γὰρ πᾶς νόμος ἐν ἑνὶ λόγῳ πεπλήρωται, ἐν τῷ· ἀγαπήσεις τὸν πλησίον
σου ὡς σεαυτόν.
Var. Western Eastern EDS.
ἐν τῷ P46 B  33 A C 1241S 1739 Ψ Chrys SCC SBL NA RP
1611 Byz S BW WH T L
− Marc 239 D* F G d b ×vg
? P

The lack of the introductory signal for the quotation from Lev 19:18 is found
only in the Western branch; all the other branches include ἐν τῷ. With the
introductory signal in apposition to the term ἐν ἑνὶ λόγῳ, the Levitical in-
junction supports Paul’s point that the Law has been fulfilled in a single
statement (ἐν ἑνὶ λόγῳ). Without the introductory signal, the scriptural
quotation is not clear, and it may appear to be a separate command.
Intrinsically, the words ἐν τῷ fit the context because they define the
“single statement” as Lev 19:18. Transcriptionally, the omission of ἐν τῷ here
is paralleled by a similar omission of the same phrase at Rom 13:8, where B,
P46, F, and G, but not D, omit the signal. On balance, the probabilities seem
to favor the inclusion of the words.

Gal 5:24 οἱ δὲ τοῦ Χριστοῦ τὴν σάρκα ἐσταύρωσαν σὺν τοῖς παθήμασιν καὶ ταῖς
ἐπιθυμίαις.
Var. Western Eastern EDS.
Χριστοῦ Marc D F G P46 P Chrys 1611 SCC SBL NA
M

d b ×vg Byz RP L
M

Χριστοῦ Ἰησοῦ B (* κυρίου χ. ι.) A C 1739 Ψ T

NA S BW
33 ×1175 1241S WH T L
T

The external evidence for adding Ἰησοῦ (“Jesus”) comprises the earliest
members of the Eastern branch, its usual allies  and 33, and B, while the
external evidence against it comprises the Western branch, P46, plus the

238
Eshbaugh, “Theological Variants,” 187.
239
Schmid, Marcion, 130–131 and 182, citing Tertullian, Marc. 5.4.12: tota enim,
inquit, lex ‘in vobis’ adimpleta est: dileges proximum tuum tamquam te, and Epiphanius,
Pan. 42.11.8, 42.12.3 ὁ γὰρ πᾶς νόμος ὑμῖν πεπλήρωται· ἀγαπήσεις τὸν πλησίον ὡς
σεαυτόν, respectively.
 3.2 Examination of Variants for Inner Branch Oppositions 135

Byzantines. This evidence is about evenly balanced, 240 so internal grounds

ought to play a large role in deciding this variation unit.
The interpretation of this phrase appears fairly straightforward. The shorter
reading οἱ τοῦ Χριστοῦ means “those who are of Christ” (or “those who
belong to Christ”), and 1 Cor 15:23 has the same expression οἱ τοῦ Χριστοῦ.
The addition of Ἰησοῦ should just mean “those who are of / belong to Christ
Jesus.” 241 Despite the intelligibility of the longer reading, it has three
different features that make it anomalous for Paul. First, nowhere else in the
undisputed letters of Paul is an article found before Χριστοῦ Ἰησοῦ, 242 though
the article τοῦ may be necessary for the genitive phrase to be substantivized
by the nominative plural article οἱ. 243 Second, Paul regularly uses the bare
Christ for the idea of belonging to Christ: Gal 3:29 εἰ δὲ ὑμεῖς Χριστοῦ; 1 Cor
1:12 ἐγὼ δὲ Χριστοῦ; 3:23 ὑμεῖς δὲ Χριστοῦ; 15:23 οἱ τοῦ Χριστοῦ; 2 Cor
10:7 Χριστοῦ εἶναι. 244 Third, Paul has a strong preference for the order Ἰησοῦ
Χριστοῦ in the genitive, 245 though there are some exceptions. 246 Because the
longer reading is not merely anomalous in Paul, but anomalous in three
different ways, the intrinsic probabilities are on the side of the shorter read-
ing.
Transcriptionally, evidence within the transmission of Galatians shows that
scribes both expanded and contracted these expressions. Sometimes they add

240
Scholars are split over how to evaluate the external evidence. Some critics prefer the
longer reading, presumably on the agreement of B, , 33, etc.: BUSCEMI 528; LÉGASSE
437 n.2; LONGENECKER 298 n.f; MATERA 204; MUSSNER 391 n.108. Indeed, BURTON 319
condemns the support for the shorter reading as “some Western authorities,” but this was
before the publication of P46.
On the other hand, Zuntz, Text, 158, holds that “Western witnesses joined by P 46 . . . are
more often right than wrong,” while NET 861 goes further and prefers the shorter reading
on “the strength of the alignment of  46 with the Western and Byzantine text-types.”
241
Pace BURTON 319, who parses the noun phrase differently as “the titular τοῦ χρι-
στοῦ, the Christ, with Ἰησοῦ in apposition.” Thus, he understands it as “they that belong to
Christ, Jesus.”
242
BURTON 319; METZGER 529. The possible exceptions to the rule occurs in the dis-
puted Paulines Ephesians and Colossians: Eph 3:1 τοῦ Χριστοῦ [Ἰησοῦ], where the omis-
sion of Ἰησοῦ is attested by * D*FG (365) pc samss against P46 ¹ AB(C)D¹Ψ 33 1739
Byz lat sy samss bo; Eph 3:11 ἐν τῷ Χριστῷ Ἰησοῦ, where P46 reads ἐν τῷ κυρίῳ Ἰησοῦ;
Col 2:6 τὸν Χριστὸν Ἰησοῦν.
243
SCHLIER 193 n.1.
244
BETZ 289 n.168; SCHLIER 193 n.1. See also the interpretation of Gal 1:6 on similar
lines above.
245
In the undisputed Paulines: Rom 1:4, 6, 7, 8; 2:16, 3:22; 5:1, 11, 15, 17, 21; 7:25;
15:6, 30; 16:25, 27; 1 Cor 1:2, 3, 7, 8, 9, 10; 6:11; 15:57; 2 Cor 1:2, 3; 8:9; 12:13; Gal 1:1,
3, 12; 2:16; 3:22; 6:14, 18; Phil 1:2, 11, 19; 2:21; 4:25; 1 Thess 1:3; 5:9, 23, 28; Phlm 3,
25.
246
In the undisputed Paulines: Rom 1:1; 15:16; 1 Cor 1:1; 2 Cor 1:1; Phil 1:1, 6, 8; 3:8,
12; Phlm 1, 9.
136 Chapter 3. Orienting the Stemma of Galatians

Jesus to Christ, 247 but sometimes they delete Jesus from Jesus Christ or Christ
Jesus. 248 Granted, the tendency to expand seems somewhat stronger than to
contract, but it is hard to see scribal tendencies in Galatians as conclusive.
Lightfoot argues that “the omission in others [scil. manuscripts] is easily ac-
counted for by the unusual order ὁ Χριστὸς Ἰησοῦς,” 249 but this does not
explain why scribes did not address the “unusual order” by transposition
rather than deletion.
Given the intrinsically anomalous nature of the phrase τοῦ Χριστοῦ Ἰησοῦ
and no compelling transcriptional reason to delete Ἰησοῦ, the preponderance
of the evidence is against the longer reading.

3.3 Orienting the Stemma

The cladistic procedure discussed in Chapter 2 produces an unoriented

stemma that expresses the relative genealogical relationships of the witnesses
without, however, identifying a starting point, or base, for the stemma. Based
on the analysis of the internal evidence of the readings in Chapter 3, the
stemma can be oriented so that the base of the stemma is identified at the
point with the greatest number of authorial readings. Table 4 sets forth the
readings for the Western branch, for the P46–B cluster, for , for 33 and
×1175, 250 and for the Eastern branch. In this table, those readings determined
to be authorial rather than scribal are indicated in plain text, while scribal
readings are in italics. Furthermore, if a branch or cluster is divided as to its
reading, that split is marked with a dagger (†). At the bottom of the table are
the counts of the authorial readings for each branch, cluster, or inter-branch
node in the stemma.

Table 4: Authorial Readings in Opposing Branches within the Stemma of Galatians

Gal Western P46 B  33 (×1175) Eastern

πατρὸς κυρίου †πατρὸς
1:3 κυρίου ἡμῶν 1 κυρίου ἡμῶν 1 0
ἡμῶν ἡμῶν ἡμῶν
1:11 γάρ 1 †δέ 1 δέ γάρ 1 δέ
1:12 οὐδέ 1 οὔτε 1 οὐδέ οὐδέ 1 οὐδέ
1:15 †[ὁ θεός] 1 − 1 ὁ θεός ὁ θεός 1 ὁ θεός

247
E.g., Gal 2:6 hark; Gal 3:24 D* F G b d f, 88, 1175 1319 1573, 1735; Gal 3:27a 209
618; Gal 3:27b 1739 6; Gal 3:29 D* F G d; Gal 4:7 6 1646, 223, 323, 489 927, hark 1611
1505 2495, 1837, 2147, 2400 pesh; Gal 6:12 P46 B, 69, 1175, 1242.
248
E.g., Gal 2:4 Marc; Gal 2:16 517 547 1827 2401 Chrys; Gal 3:22 33 1424; Gal 3:28
1241S 1245; Gal 5:6 B; Gal 6:17 P Ψ 075 1175 1854 2464 Marc.
249
LIGHTFOOT 213.
250
The mixed manuscript 1175 is considered a witness for 33’s ancestor only when it
differs from the Byzantine text.
 3.3 Orienting the Stemma 137

Gal Western P46 B  33 (×1175) Eastern

1:17 ἀπῆλθον 1 †[ἀπ]ῆλθον 1 ἀνῆλθον ἀνῆλθον 0 ἀνῆλθον
1:18 ἔτη τρία 1 ἔτη τρία 1 τρία ἔτη τρία ἔτη 0 †τρία ἔτη
1:19 εἶδον οὐδένα 0 οὐχ εἶδον 1 οὐκ εἶδον οὐκ εἶδον 1 οὐκ εἶδον
Πέτρος καὶ †Ἰάκ. καὶ Ἰάκ. καὶ Ἰάκ. καὶ Ἰάκ. καὶ
2:9 1 1 1
Ἰάκ. Κηφᾶς Κηφᾶς Κηφᾶς Κηφᾶς
2:12 ἦλθεν 1 ἦλθεν 1 ἦλθεν ἦλθεν 1 ἦλθον
2:13 καί 1 − 1 καί καί 1 καί
Χριστὸν Ἰησοῦν Ἰησοῦν Χριστὸν †Χριστὸν
2:16 1 1 1
Ἰησοῦν Χριστὸν Χριστὸν Ἰησοῦν Ἰησ.
θεοῦ θεοῦ
2:20 1 1 υἱοῦ τοῦ θεοῦ υἱοῦ τοῦ θεοῦ 0 υἱοῦ τοῦ θεοῦ
καὶ Χριστοῦ καὶ Χριστοῦ
3:7 εἰσιν υἱοί 1 υἱοί εἰσιν 1 υἱοί εἰσιν εἰσιν υἱοί 1 †υἱοί εἰσιν
3:14 εὐλογίαν 1 †ἐπαγγελίαν 1 ἐπαγγελίαν ἐπαγγελίαν 1 ἐπαγγελίαν
3:21a τοῦ θεοῦ 1 − 1 τοῦ θεοῦ τοῦ θεοῦ 1 τοῦ θεοῦ
3:21b ἐκ νόμου 1 ἐν νόμῳ 1 ἐκ νόμου ἐκ νόμου 1 ἐκ νόμου
3:24 γέγονεν 1 ἐγένετο 1 γέγονεν γέγονεν 1 γέγονεν
4:3 ἤμεθα 1 †ἤμεθα 1 ἦμεν ἦμεν 0 ἦμεν
4:6 υἱοὶ θεοῦ 1 υἱοί 1 υἱοί υἱοί 1 υἱοί
τοῖς ... θεοῖς ἐδ. τοῖς ... ἐδ. τοῖς ... ἐδ. τοῖς ... ἐδ. τοῖς ...
4:8 1 1 1
ἐδ. θεοῖς θεοῖς θεοῖς θεοῖς
4:19 τέκνα 1 τέκνα 1 τέκνα τεκνία 0 τεκνία
4:23 μέν 1 − 1 μέν μέν 1 μέν
ὑμεις . . . ὑμεις . . . ἡμεῖς . . .
4:28 1 1 ὑμεις . . . ἐστέ 0 ἡμεῖς . . . ἐσμέν
ἐστέ ἐστέ ἐσμέν
†κληρονο- κληρονο-
4:30 0 0 κληρονομήσει κληρονομήσει 0 †κληρονομήσῃ
μήσῃ μήσει
†ἐν ὑμῖν ἐν
5:14a 1 ἐν ἑνὶ λόγῳ 1 ἐν ἑνὶ λόγῳ ἐν ἑνὶ λόγῳ 1 ἐν ἑνὶ λόγῳ
ἑνὶ λόγῳ
5:14b − 1 ἐν τῷ 1 ἐν τῷ ἐν τῷ 1 ἐν τῷ
5:17a γάρ 1 γάρ 1 γάρ γάρ 1 δέ
5:17b ὃ ἄν 0 ἄν 1 ἃ ἐάν ἃ ἐάν 1 †ἃ [ἐ]άν
5:21 φθόνοι φόνοι 1 φθόνοι 1 φθόνοι φθόνοι 1 φθόνοι φόνοι
κυρίου Χριστοῦ
5:24 Χριστοῦ 1 †Χρ. [Ἰησ.] 1 0 †Χρ. [Ἰησ.]
Χρ. Ἰησ. Ἰησοῦ
5:26 †ἀλλήλοις 1 ἀλλήλους 1 ἀλλήλοις ἀλλήλοις 1 †ἀλλήλοις
ἀναπληρώ- ἀναπληρώ- ἀναπληρώ- ἀναπληρώ-
6:2 1 1 0 ἀναπληρώσατε
σετε σετε σατε σατε
6:4 ἕκαστος 1 − 1 ἕκαστος ἕκαστος 1 ἕκαστος
6:11 πηλίκοις 1 ἡλίκοις 1 πηλίκοις πηλίκοις 1 πηλίκοις
Χριστοῦ
6:12 Χριστοῦ 1 1 Χριστοῦ Χριστοῦ 1 Χριστοῦ
Ἰησοῦ
†περι- περι- περι- περι- περι-
6:13 1 1 1
πτεμνόμενοι τετμημένοι πτεμνόμενοι πτεμνόμενοι πτεμνόμενοι
Total 25 33 20 35 26 26 26 25

As can be seen from the table, these nodes in the stemma have different
counts of authorial readings. In particular, out of the thirty-six variation units,
 has the authorial reading in 26 cases, while the ancestor of 33 and ×1175
also has 26. The Eastern branch has 25 authorial readings, and the node in
between these three parts of the stemma has 26 authorial readings. As for the
138 Chapter 3. Orienting the Stemma of Galatians

Western branch, it has 25 authorial readings, while the P46–B cluster has 20
authorial readings. The node between the Western branch and the P46–B
cluster has 33 authorial readings, while the link on the stemma connecting the
Eastern and Western brands has 35 authorial readings on either side.
Consequently, these counts of authorial readings can be mapped onto the
unoriented stemma, as shown in FIG. 17:
P46-B ℵ
20 26

35 Eastern
33 • • Branch
26
25

Western
Branch 33-×1175
25 26

FIG. 17: Unoriented Stemma with Counts of Authorial Readings

The portion of the unoriented stemma with the highest authorial reading
count (35) is the branch between where the Western and Eastern branches
join the stemma. This branch separates the Western branch and its allies, P46
and B, from the Eastern branch and its allies  and 33. This branch also lies
in the area indicated by Hort as belonging to his “neutral” text. By contrast,
the node that is between the two oldest witnesses, Marcion and P46, has an
authorial reading count of 33, two less than that of the central branch. This
result suggests that, while the age of the witnesses is an important
consideration in terms of external evidence, it is not strictly determinative of
the best readings. As for the node that is the closest to the Nestle-Aland text,
that has 26 authorial readings, nine less than the central branch. This result
suggests that the Nestle-Aland text bears some room for improvement.
Accordingly, when the stemma is oriented at the central branch, the result-
ing oriented stemma appears as shown in FIG. 18, in which the difference of
one reading between the assumed autograph Ω and the stemmatic archetype ω
means that no portion of the stemma, however, has authorial readings in all
36 variation units. The reason for this is that the reading of Gal 4:30
determined to be authorial, κληρονομήσῃ, is found in only the Western and
Eastern branches. The basic witnesses, P46, B, , and 33, however, all read
κληρονομήσει in conformance with the reading of the Septuagintal text. The
lack of genealogical coherence for this variation unit suggests at least two
possibilities. One possibility is that harmonization to the LXX was so com-
mon that it affected all the basic witnesses, but somehow left the Western and
 3.3 Orienting the Stemma 139

Eastern branch unscathed. The other is that this study’s judgment about the
secondary nature of the reading that parallels the LXX is wrong in this case,
and that what Paul wrote was the future indicative κληρονομήσει, the form of
his scriptural source.

36
Ω

ω 35
33 26
• •
25 20 26 26 25
Western P46-B ℵ 33-×1175 Eastern
Branch Branch
FIG. 18: An Oriented Stemma of Galatians Based on Authorial Reading Counts

On this stemma, there are two differences between the stemmatic archetype ω
and the common ancestor of the Western branch and the cluster P46–B. One
of the differences occurs in Gal 1:19, where neither the Western branch
reading εἶδον οὐδένα nor the P46–B cluster reading οὐχ’ εἶδον were seen as
supporting the authorial οὐκ εἶδον, but rather both as independent efforts to
strengthen Paul’s denial. It is also possible, however, that the common an-
cestor of the Western branch and the P46–B cluster did not agree with either
of its descendants and that it simply read the authorial οὐκ εἶδον. (Indeed, the
P46–B cluster reading of οὐχ’ εἶδον may merely be an orthographic variant
and, despite its spelling, actually evidence of the authorial οὐκ εἶδον.) An-
other difference between the stemmatic archetype ω and the Western/P46–B
common ancestor is the textual change between ἐάν and ἄν in Gal 5:17b. This
could well be an accidental coincidence. In fact, the ἄν reading is that of the
original hand of B and it was corrected to read ἃ ἐάν. If this reading is indeed
coincidental, then there is no remaining textual evidence within Galatians to
group the Western branch and the P46–B cluster together below the level of
the archetype. Both the Western branch and the P46–B cluster would then be
independent descendants of the archetype on par with a group consisting of
the Eastern branch and its early allies of  and 33.
In other words, the data may well support the alternative stemma shown in
FIG. 19. On this stemma, the archetype ω has three lines of descent instead of
two. One of them is the Eastern branch and its early allies, and the other two
are the Western branch and the P46–B cluster. With this configuration, estab-
140 Chapter 3. Orienting the Stemma of Galatians

lishment of the text would be fairly straightforward, for the agreement of any
two of the descendants against other would constitute clear external evidence
that can only be overturned by convincing internal evidence.

Ω 36

35
ω

26
•
25 20 26 26 25
Western P46-B ℵ 33-×1175 Eastern
Branch Branch

FIG. 19: An Alternative Oriented Stemma of Galatians

Nevertheless, it seems prudent to adopt the oriented stemma of FIG. 18 for

the text of Galatians in the Corpus Paulinum, because it is less sensitive than
the alternative stemma to the presence of additional shared errors between
P46–B and the Western branch not considered in this study, for instance in
other books of the Pauline corpus, as well as being less sensitive to differing
assessments of the internal evidence within Galatians. For example, if further
analysis of the internal evidence of Gal 2:20 turns out to favor the “son of
God” reading over the “God and Christ” reading, then the stemma of FIG. 18
would still stand (albeit with changed authorial counts), while the stemma of
FIG. 19 would no longer represent the best choice of base for the stemma.
This orientation of the stemma does not automatically establish the critical
text of Galatians. In stemmatic terms, only the recensio step has been per-
formed. It does establish a tentative form of the text for the next step, which
is to examine the textual variants near the base of the stemma and select the
readings that appear best based on internal evidence. Even if the text is sup-
ported by the earliest variant carriers at the base of the stemma, the text must
be further examined for primitive errors that may require conjecture to fix.
This next step is performed in Chapter 4.
 Chapter 4

Examining and Establishing the Text

4.1 Introduction

A critical text is a scholarly construct that presents the textual critic’s under-
standing of the text and its history. That is, the critical text is designed, in the
studied opinion of the textual critic, to explain the state of the surviving tex-
tual evidence better than its alternatives. The state of the surviving textual
evidence includes both the pattern of readings distributed among the external
evidence and the generation of scribal variants as indicated by the internal
evidence. A critical text should be considered a provisional text, which repre-
sents the best explanation of the current state and understanding of the textual
evidence. As new evidence or superior interpretations of the evidence arise,
however, the critical text should be updated in light of those new develop-
ments.
This study has enabled the external evidence to be interpreted genealogi-
cally. That is, it groups the various witnesses to the text in the form of a stem-
ma based on their genealogical history as inferred from their patterns of
readings. In order to avoid the circularity between the reconstruction of a text
and the reconstruction of the history of the text, the construction of the stem-
ma is split into two stages. The first stage determines the shape and linkage
pattern of a stemma – but not its orientation – based on the pattern of textual
changes among the witnesses to the text. That stage, which uses external
evidence and hence does not need internal evidence or a reconstructed text to
produce its result, was performed in Chapter 2. The second stage, performed
in Chapter 3, oriented the stemma based on the examination of thirty-six
variants in four sets of opposing witnesses in accordance with the internal
evidence of their readings. Based on this examination, the stemma of
Galatians was oriented so that there are two basic divisions in the history of
the text. On one side, there is the Western branch and the P46–B cluster,
while on the other side there is the Eastern branch with  and 33 as its allies.
At this point, only those coherent variants where the opposing sets of wit-
nesses line up neatly were considered in Chapter 3. This chapter, by contrast,
examines those variation units where the witnesses are less coherent in their
distribution.
142 Chapter 4. Examining and Establishing the Text

In order to identify which additional variation units need to be examined to

establish the text, the bracketed readings in the Nestle-Aland 27th edition
were considered first. Bracketed and marginal readings have long been em-
ployed in editions of the New Testament in order to signal the editors’ doubts
about the wording of particular places in the text. There has been some con-
cern lately, however, that brackets have been overused, so this study adopts
the editorial philosophy of the recent SBL edition of the New Testament that
“one of an editor’s duties is to make choices, particularly in the ‘hard cases,’
so as to offer some degree of guidance to those making use of the resulting
text.” 1 As a result, one of the goals of this study is to reassess each of the
bracketed readings in the Nestle-Aland 27th edition.
There are eight such bracketed readings in Galatians, as follows: Gal 1:6
ἐν χάριτι [Χριστοῦ], 1:8 εὐαγγελίζηται [ὑμῖν], 1:15 εὐδόκησεν [ὁ θεός], 2:6
[ὁ] θεός, 2:13 [καὶ] οἱ λοιποὶ Ἰουδαῖοι, 2:16 εἰδότες [δὲ] ὅτι, 3:21 κατὰ τῶν
ἐπαγγελιῶν [τοῦ θεοῦ], 5:7 [τῇ] ἀληθείᾳ, and 5:24 οἱ δὲ τοῦ Χριστοῦ
[Ἰησοῦ]. Of these eight readings, three have been already treated in Chapter 3
(Gal 1:15, 2:13, and 5:24). The remaining five all involve places where P46
and B are split, with P46 agreeing against B with another witness and with B
agreeing against P46 with still another witness. This result suggests that P46–
B splits, where neither witness is singular, are good candidates for
examination.
In fact, there are thirty-five such supported P46–B split variation units in
Galatians. Most variants against the Nestle-Aland text can be cursorily
resolved. For example, seven appear to be harmonizations, 2 three involve
orthographic or morphological issues, 3 five appear to be the easier reading, 4
and two pertain to scribal habits of P46. 5 Four such variants were already
discussed in Chapter 3, 6 and three will be discussed in Chapter 5. 7 This
leaves eleven P46–B split variation units to be reassessed in detail for this
chapter: 1:4 περί, 1:6 [Χριστοῦ], 1:8 εὐαγγελίζηται [ἡμῖν], 2:6 [ὁ], 2:16a [δέ],
3:21 ἦν ἄν, 4:18 ζηλοῦσθε, 4:19 ἄχρις, 4:25 τὸ γὰρ Σινᾶ, 5:7 [τῇ], and 6:10
ἔχωμεν.
In addition, there are four readings where  and 33 agree with each other
against the Eastern branch. These readings are found at Gal 3:3 ἐπιτελεῖσθαι,
4:18 ζηλοῦσθε, 6:9 θερίσωμεν, and 6:10 ἔχωμεν. Two of these are also P46–

1
Michael W. Holmes, The Greek New Testament: SBL Edition (Atlanta: Society of Bib-
lical Literature, 2010), xv n.16.
2
Gal 1:16 εὐαγγελίσωμαι, 2:7 εἰδότες, 2:8 ἐνέργησεν, 2:16b Χριστοῦ Ἰησοῦ, 4:23a
γεγένηται, 5:10a δέ, and 6:7b ταῦτα.
3
Gal 2:8 κἀμοί, 5:10 φρονήσητε, and 6:12 διώκονται.
4
Gal 3:19b ἄν, 4:9 δουλεῦσαι, 4:23b τῆς, 5:12 ἀποκόψωνται, and 6:7a ἄν.
5
Gal 5:5 ἐκδεχόμεθα and 5:25 omit καί.
6
Gal 1:11 γάρ, 2:9 Πέτρῳ, 3:14 εὐλογίαν, and 5:24 Ἰησοῦ.
7
Gal 3:19a πράξεων, 4:31 ἄρα, and 5:17 ἀντίκειται ἀλλήλοις.
 4.2 Examination of Variation Units with Early, Split Support 143

B split variants (4:18 and 6:10), so there are two additional variation units to
reassess. Consequently, a total of thirteen variation units are reexamined in
detail for this chapter.

4.2 Examination of Variation Units with Early, Split Support

Gal 1:4 τοῦ δόντος ἑαυτὸν ὑπὲρ τῶν ἁμαρτιῶν ἡμῶν,
Var. Western Eastern EDS.
περί D F G P46  A Ψ 1611 Byz M
SCC RP S WH T L
ὑπέρ B 33 ×1775 P 1241S 1739 Chrys Byzpt SBL NA BW WH
T

? Marc d b ×vg C

The attestation for the two prepositions is fractured among the witnesses.
Every branch and cluster, except for the Western branch, is divided about
evenly between περί and ὑπέρ. Thus, if external evidence has any weight at
all, there is a slight presumption in favor of περί because the Western branch
is unified as to that reading.
The difference in meaning between περί and ὑπέρ in this context is at best
subtle. 8 Normally, ὑπέρ means “on behalf of” and περί means “concerning,”
but the first clause of the verse can equally be translated as “[Christ] who
gave himself for our sins” with either preposition. 9 Pauline usage evidences a
semantic overlap for these prepositions, 10 and Paul attests to this meaning
with this particular noun for both prepositions: Rom 8:3, περὶ ἁμαρτίας (“for
sin”), 11 and 1 Cor 15:3, ὑπὲρ τῶν ἁμαρτιῶν ἡμῶν (“for our sins”). Indeed,
New Testament usage is varied even within a single author for this phrase. 12

8
BETZ 42 n.55. LIGHTFOOT 73, however, endeavors a syntactic distinction: “Generally
it may be said that περὶ is things, ὑπὲρ of persons, . . . but exceptions are very numerous.”
9
When used with ἁμαρτία, BDAG offers nearly identical glosses for these prepositions:
compare “to take away, to atone for” (BDAG “περί, 1g” 798) with “in order to atone for
(the) sins or to remove them” (BDAG “ὑπέρ, A1b” 1030).
10
BURTON 13; LONGENECKER 8. Although MATERA 39 argues that “the preposition
[ὑπέρ] functions as a technical term when the Apostle speaks of Christ’s death on our be-
half,” all of his examples except for 1 Cor 15:3 (viz. Rom 5:6, 7, 8; 14:15; 1 Cor 1:13;
15:3; 2 Cor 5:14, 15; Gal 2:20; 3:13) involve the preposition ὑπέρ with a personal object,
not with the abstract noun sin. For an alleged “technical term,” such distinctions are
crucial.
11
BDAG “περί, 1g” 798, but T. C. G. Thornton, “The Meaning of καὶ περὶ ἁμαρτίας in
Romans viii. 3,” JTS 22 (1971): 515–517 proposes “on the charge of sin” or “on the
grounds of sin,” instead. See also 1 Pet 3:18 περὶ ἁμαρτιῶν ἔπαθεν; 1 John 2:2 περὶ τῶν
ἀμαρτιῶν ἡμῶν.
12
Conzelmann, 1 Corinthians: A Commentary on the First Epistle to the Corinthians
(Hermeneia; James W. Leitch, trans.; Philadelphia: Fortress, 1975), 253 n.48, compares
Heb 10:12 ὑπὲρ ἁμαρτιῶν with 10:26 περὶ ἁμαρτιῶν.
144 Chapter 4. Examining and Establishing the Text

There is little internal guidance for this variant. As for intrinsic probabili-
ties, the parallel with 1 Cor 15:3, ὃ καὶ παρέλαβον, ὅτι Χριστὸς ἀπέθανεν
ὑπὲρ τῶν ἁμαρτιῶν ἡμῶν κατὰ τὰς γραφὰς (“what I also received, that Christ
died for our sins according to the scriptures”) demonstrates that ὑπὲρ τῶν
ἁμαρτιῶν ἡμῶν occurs in a Pauline text, but the wording may only be pre-
Pauline, as Paul explicitly stated. Indeed, Gal 1:4 does not reproduce the
traditional formula of 1 Cor exactly: apart from the preposition, it differs over
whether to say ἀπέθανεν (“died”) or δόντος ἑαυτόν (“gave himself”). Thus,
the phraseology of 1 Cor 15:3 could be the traditional formulation, and Gal
1:4 the Pauline adaptation of it. This means 1 Cor 15:3 is not strong intrinsic
evidence of Paul’s stylistic preference.
The transcriptional probabilities are also hard to discern because the
interchange between περί and ὑπέρ was common among the manuscripts. 13
On the one hand, the ὑπέρ variant in Galatians could be a harmonization to
the preposition in the established form of 1 Cor 15:3. On the other hand,
Bernhard Weiss argued that a scribe may have been scandalized by the ὑπέρ
variant, thinking that the text said that Christ gave himself on behalf of sin,
and changed the preposition to περί. 14 The lack of textual variation, however,
in the preposition ὑπέρ in the 1 Cor 15:3 parallel weakens this argument. 15
Though Colwell rightly warns that scribal behavior in one passage cannot be
easily generalized to another passage, 16 this warning does not overcome the
difficulty that Weiss’s proposal about the scribal motivation behind the
change in preposition is uncorroborated, especially where it would be most
expected.

13
Breytenbach, Versöhnung: Eine Studie zur paulinischen Soteriologie (WMANT 60;
Neukirchen-Vluzn: Neukichener, 1989), 197, points to variations between these two
prepositions in the LXX and in the textual transmission of Gal 1:4 and 1 Thess 5:10. See
also LONGENECKER 8: “And this same interchangeability of prepositions appears in the
extant Koine Greek materials outside the NT.” Bortone, Greek Prepositions: From Anti-
quity to the Present (Oxford: Oxford University Press, 2010), 231, points out that περί was
being phased out, even in its non-spatial sense, during the medieval period, while ὑπέρ re-
mained in use.
14
WEISS 56. Weiss’s suggestion that the preposition was modified to περί on the analo-
gy of Rom 8:3, however, is strained and unnecessary, because the interchange between
these prepositions was already common in antiquity even without a specific Pauline model.
15
Swanson, New Testament Greek Manuscripts: 1 Corinthians (Wheaton: Ill.: Tyndale
House, 2003), 239.
16
Earnest Cadman Colwell, “Hort Redivivus: A Plea and a Program” in Studies in
Methodology in Textual Criticism of the New Testament (NTTS 9; Grand Rapids, Mich.:
Eerdmans, 1969), 148–171 at 160: “But the knowledge gained from one of these com-
pendia cannot be transferred to another, nor can the knowledge gained from one period be
generalized for another, nor can the treatment of one part of the New Testament be
generalized for the entire canon” (emphasis original).
 4.2 Examination of Variation Units with Early, Split Support 145

In sum, both the external evidence and the internal evidence tilt slightly in
favor of the preposition περί.

Gal 1:6 Θαυμάζω ὅτι οὕτως ταχέως μετατίθεσθε ἀπὸ τοῦ καλέσαντος ὑμᾶς ἐν χάριτι
Χριστοῦ εἰς ἕτερον εὐαγγέλιον,
Var. Western Eastern EDS.
T
Χριστοῦ ×vg B  33 AP 1739 Ψ Chrys SCC SBL NA RP
×1175 1611 Byz S BW WH T L
Ἰησοῦ Dd 1241S
Χριστοῦ
− F* G b P46vid 17 NA
M

? Marc 18 C

Both the Western branch and the P46–B group are split as to the addition or
omission of the genitive Χριστοῦ, while the Eastern branch and  33 are solid
support for the longer reading. P46’s apparent support for the shorter reading
may not be significant since P46 has a tendency for accidental omission. 19 As
a result, the divided support for the shorter reading in P46vid and FG does not
seem sufficient to offset the weight of the coherent support for the longer
reading in the rest of the tradition. 20
The exegesis of the shorter reading ἐν χάριτι is straightforward. The anar-
throus noun “grace” has a qualitative effect, 21 and, with the preposition, the
expression as a whole acquires an adverbial force: “with grace” or “gra-
ciously.” 22 As a result, the shorter text would have Paul say: “from the one

17
P46 has a lacuna from where the word χάριτι should be until the end of the line.
Comfort, Early Manuscripts, 150, argued that the gap may have been large enough to ac-
commodate the nomen sacrum χ̅υ̅, but this is not apparent to me from the high-resolution
photograph of the sheet, and, in any case, this reconstruction of the lacuna was later
abandoned sub silentio by Philip W. Comfort and David P. Barrett, The Text of the Earliest
New Testament Greek Manuscripts: New and Complete Transcriptions with Photographs
(2d ed.; Wheaton, Ill.: Tyndale House, 2001), 313.
18
Against the indication in UBS4 that Marcion omitted the genitive Χριστοῦ, Schmid,
Marcion, 64–65, argues well that Tertullian’s quoting style does not permit us to determine
if Marcion had included or omitted the word.
19
ROYSE 357 states that P46’s tendency for accidental omission has to be taken in
account. Note, however, that his survey shows P46’s omission of proper names to be fairly
rare among its singular or near-singular omissions (278, listing only three examples at Rom
15:17b [Ἰησοῦ]; 1 Cor 9:9 [Μωϋσέως], 15:45 [Ἀδάμ]).
20
METZGER 520 cautions that the shorter “reading is supported by only part of the
Western tradition.”
21
BURTON 21: “The absence of the article before χάριτι has the effect, and is doubtless
due to the intention, of giving the word qualitative rather than individualizing force.”;
LIGHTFOOT 76.
22
BDAG “ἐν” 11, 330. BETZ 48 prefers a locative meaning to “designate the Galatians’
present state of salvation” (so also MARTYN 109), though, if that were meant, εἰς would
have been more appropriate. Other translations render it as “by grace” (e.g. NET).
146 Chapter 4. Examining and Establishing the Text

who called you with grace (or graciously).” Though this reading does not
specify the agent of the calling, Paul is explicit elsewhere that it is God the
Father who does the calling. 23
On the other hand, the longer reading has a number of possible exegetical
options, though none of them are as straightforward as the shorter reading.
The most common option is to construe Χριστοῦ as a genitive of possession
with the immediately preceding noun χάριτι, thus “Christ’s grace.” This cre-
ates a slight complication in that the prepositional phrase ἐν χάριτι Χριστοῦ
occurs nowhere else in Paul, 24 and scholars have wrestled over the question
whether it could be instrumental (“called by/with Christ’s grace”) or locative
(“in Christ’s grace”). Neither one of these possibilities, however, is fully
satisfactory. The instrumental possibility conflicts with the possessor of the
grace in Paul’s own call later in 1:15–16, εὐδόκησεν ὁ ἀφορίσας . . . καὶ
καλέσας διὰ τῆς χάριτος αὐτοῦ ἀποκαλύψαι τὸν υἱὸν αὐτοῦ ἐν ἐμοί (“the one
who set me apart . . . and called by his grace was pleased to reveal his son in
me”). 25 Less problematic might be a locative construal if ἀπὸ τοῦ καλέσαντος
ὑμᾶς ἐν χάριτι is taken to refer to calling the Galatians to a destination state
of grace, as “from the one who called you to be in Christ’s grace,” 26 although
the preposition εἰς would have been better suited for this meaning and Paul
elsewhere uses the preposition ἐν to refer to the pre-call condition, e.g., 1 Cor
7:18, ἐν ἀκροβυστίᾳ κέκληταί τις, μὴ περιτεμνέσθω (“if anyone is called in [a
state of] uncircumcision, let him not get circumcised”). 27

23
See especially Rom 8:30, 9:12, 24; 1 Cor 1:9, 7:15, 17, 18, 20, 21, 22ab, 24; Gal
1:15; 1 Thess 2:12, 4:7, 5:24; cf. also 2 Thess 2:14. The instances of Gal 5:8 and 13 do not
specify the agent of the calling.
24
A much longer, stereotyped formulation, ἡ χάρις τοῦ κυρίου (ἡμῶν) Ἰησοῦ Χριστοῦ
(“the grace of [our] Lord Jesus Christ”), occurs in Paul’s subscriptions (2 Cor 13:13; Gal
6:18; Phil 4:23; 1 Thess 5:28; 2 Thess 3:18; Phlm 25; cf. Rom 16:20 ἡ χάρις τοῦ κυρίου
ἡμῶν Ἰησοῦ), as well as in 2 Cor 8:9 γινώσκετε γὰρ τὴν χάριν τοῦ κυρίου ἡμῶν Ἰησοῦ
Χριστοῦ (“for you know the grace of our Lord Jesus Christ”).
25
Pace LONGENECKER 15, who argues that “just as in the letter’s salutation God and
Christ are presented as completely at one in mankind’s salvation, so in these texts they are
presented interchangeably as the source of redemptive grace.”
26
See BETZ 48, “ἐν (‘in’) should not be taken as the ‘means by which’ but as a
definition of the situation before God enjoyed by those who were called” (footnotes
omitted). So also MARTYN 109.
27
Martinus C. de Boer, Galatians: A Commentary (NTL; Louisville, Ky.: Westminster
John Knox, 2011), 40 n.53, argues that Paul sometimes uses ἐν to mean “into” when used
with καλέω, but the examples do not establish the point. First Corinthians 7:15 is no
exception, for ἐν δέ εἰρήνῃ κέκληκεν ὑμᾶς ὁ θεός (“God has called you in peace”) refers to
the peacefulness of the marriage at the time when a person is called to Christ. In 1
Thessalonians 4:7 οὐ γὰρ ἐκάλεσον ἡμᾶς ὁ θεὸς ἐπὶ ἀκαθαρίᾳ ἀλλ’ ἐν ἀγιασμῷ, the
preposition ἐπί refers to destination, while the ἐν could be instrumental. Rom 5:2 lacks the
word “call” but uses stand, for which the term “into” is inappropriate.
 4.2 Examination of Variation Units with Early, Split Support 147

Another possibility is to regard Χριστοῦ as an epexegetical genitive or,

similarly, as a genitive of content: “the grace that is (or contains) Christ.” 28
Although the idea that Christ could be a grace or a benefaction has been
suggested by some scholars, 29 this notion is hard to square with how Paul is
using the preposition ἐν in this context. If Christ is equated with grace in this
exegetical possibility, then the preposition ἐν would locative as if it read ἐν
Χριστῷ, 30 yet Paul’s usual usage elsewhere would not suggest that this grace
is the destination, which the construct requires, but the state within which one
is called.
A different syntactic option with the longer reading is to construe Χριστοῦ
as the object of the preposition ἀπό with the participial phrase καλέσαντος
ὑμᾶς ἐν χάριτι in the first attributive position, such that the text would mean
“from Christ, who called you graciously.” 31 This option, however, is unlikely,
because there is no explicit corroboration elsewhere in Paul of making Christ
the subject of calling, 32 while there are a dozen places in the undisputed let-
ters of Paul where God (“the Father”) is one who calls. 33
Still another possibility is to construe the genitive Χριστοῦ as the comple-
ment of the participle καλέσαντος: “from the one who called you with grace
to be of Christ.” In this case, ἐν χάριτι becomes an adverbial prepositional
phrase describing the manner of God’s calling the Galatians: “with grace” or
“graciously.” The hyperbaton of this construction, by putting Χριστοῦ at the
end of the clause, confers additional emphasis on belonging to Christ. 34 That
Paul considers the Galatians to be “of Christ” is attested elsewhere in the let-
ter, e.g., 3:29 εἰ δὲ ὑμεῖς Χριστοῦ, “if you are of Christ,” and 5:24 οἱ δὲ τοῦ
Χριστοῦ, “those of Christ.” 35 Moreover, Rom 1:6 uses this genitive comple-
mentary construction with the adjectival cognate: ἐστε καὶ ὑμεῖς κλητοὶ
Ἰησοῦ Χριστοῦ, “you too are called to be of Jesus Christ.” This reading also

28
BETZ 48: “it refers to the grace of God which Christ represents.”
29
For notions of “grace” in its Greco-Roman context, see generally, James R. Harrison,
Paul’s Language of Grace in its Graeco-Roman Context (WUNT 2.172; Tübingen: Mohr
Siebeck, 2003); Stephan Joubert, Paul as Benefactor: Reciprocity, Strategy and Theologi-
cal Reflection in Paul’s Collection (WUNT 2.124; Tübingen: Mohr Siebeck, 2000).
30
So de Boer, Galatians, 40.
31
ZAHN 43. This interpretation is discussed as a possibility by BURTON 19.
32
So BETZ 48; BURTON 19; LIGHTFOOT 76; LONGENECKER 15; MATERA 45. ROHDE
39 grants that Rom 1:6 κλητοὶ Ἰησοῦ Χριστοῦ might be (“wohl ist”) an example if the
genitive is subjective, but, as he notes, the possessive genitive is the better understanding.
33
BURTON 20, citing Rom 4:17, 8:30, 9:11, 14, 11:29; 1 Cor 1:9, 26, 7:15, 17; Gal
1:15; Phil 3:14; 1 Thess 2:12, 4:7, and 5:24.
34
SMYTH 679 § 3028: “In prose hyperbaton is less common than in poetry, but even in
prose it is frequent, especially when it secures emphasis on an important idea by placing it
at the beginning or end of a sentence.”
35
On the justification for reading Gal 5:24 as Χριστοῦ rather than Χριστοῦ Ἰησοῦ see
the discussion below for that variation unit.
148 Chapter 4. Examining and Establishing the Text

fits the context because it contrasts being “of Christ” with that of turning to
another gospel, which is not really a gospel at all (vv.6–7).
The internal evidence does not overturn the presumption in favor of the
longer reading based on the external evidence. Stylistically, the shorter
reading is unproblemmatically Pauline because the absolute χάριτι is well
attested (Rom 11:6; 1 Cor 10:30; 2 Cor 8:7, 19; cf. Eph 2:5, 8; Col 3:16, 4:6,
2 Thess 2:16). 36 For the longer reading, on the other hand, it is also Pauline to
understand Χριστοῦ as a complement of καλέσαντος (Rom 1:6), rather than
as an otherwise unattested simple phrase ἐν χάριτι Χριστοῦ. 37
Many critics have contended that the transcriptional probabilities favor the
shorter reading here, arguing that word Χριστοῦ is a natural expansion of the
text. 38 With the dispute over the value of this criterion, occasioned by
Royse’s study of scribal habits among the six earliest and most substantial
papyri, 39 however, it is prudent not to rely on this text-critical canon.
Genealogical coherence plays a role, too. If the shorter reading were the
earlier form of the text, it is difficult to account for why scribes would have
independently hit upon the same solution of supplementing the absolute
χάριτι with Χριστοῦ to make the term more explicit. 40 Indeed, the very
naturalness of this supposed supplement is dubious in light of v.15, where it
is the grace of God – not Christ’s – by which Paul was called. Conversely, it
is conceivable that the word Χριστοῦ could have been omitted by accident
(cf. P46) or out of a misreading of the complementary genitive either with the
participle τοῦ καλέσαντος, 41 or as one of possession in such a way that would
seem to contradict v.15. 42

36
For χάριτι θεοῦ, a genitive with God rather than Christ, cf. 1 Cor 1:4 with τοῦ, 15:10;
and 2 Cor 1:2. Another example is Rom 3:24 τῇ αὐτοῦ [scil. θεοῦ] χάριτι, “by his [scil.
God’s] grace,” with God being mentioned explicitly at the end of v.23.
37
Note also in this connection that Rom 5:15 refers to Christ’s grace much more
verbosely: ἐν χάριτι τῇ τοῦ ἑνὸς ἀνθρώπου Ἰησοῦ Χριστου (“by the grace of the one
person Jesus Christ”). In this verse Paul also has ἡ χάρις τοῦ θεοῦ (“the grace of God”).
38
MARTYN 109; Tasker, Greek New Testament, 438; cf. METZGER 520, who argues that
“the absence of any genitive qualifying ἐν χάριτι . . . has the appearance of being the
original reading, which copyists supplemented”
39
ROYSE 703–736. None of Royse’s exceptions for preferring the longer reading
(lateness, harmonization, or grammatical improvement) apply here.
40
The argument of NET 859 that “it is difficult to explain how this particular reading
could have arisen from the simple χάριτι, in light of Paul’s normal idioms,” somewhat
unartfully suggests a confusion between transcriptional probabilities (“arisen from”) and
intrinsic probabilities (“Paul’s normal idioms”). It is better to point out that the context
would not suggest (e.g. as by harmonization) adding Χριστοῦ.
41
MEYER 11.
42
Indeed difficulties with the “grace of Christ” interpretation have led scholars to reject
Χριστοῦ, e.g., LÉGASSE 61 n.6; MUSSNER 55.
 4.2 Examination of Variation Units with Early, Split Support 149

The reading, therefore, that best explains the textual evidence is the longer
reading with Χριστοῦ, with the understanding that it is a genitive complement
to ὑμᾶς in a double object construction with καλέσαντος, “called you to be of
Christ.”

Gal 1:8 ἀλλὰ καὶ ἐὰν ἡμεῖς ἢ ἄγγελος ἐξ οὐρανοῦ ὑμῖν εὐαγγελίζηται παρ’ ὃ
εὐηγγελισάμεθα ὑμῖν, ἀνάθεμα ἔστω.
Var. Western Eastern EDS.
ὑμῖν εὐαγγ. B ×1175 1739 SCC S BW T
M
T T
εὐαγγ. ὑμῖν (D ὑμᾶς) d 33 A P (1241S Chrys 1611 SBL NA WH RP
×vg ἡμῖν) Byz T
T L
εὐαγγ. FGb  Ψ M
NA WH T
M M

? Marc 43 P46 C

This variation unit pertains to the presence and position of the pronoun ὑμῖν
(“to you”). 44 No reading has the unambiguous support of any branch, and the
branches are divided as to whether or where the pronoun is read. Further-
more, the lacuna in P46 makes it difficult to weigh the strength of B, but B’s
reading is shared by P51vid, so there is early support for this variant. 45 Though
the presence of the pronoun is better attested than its lack, 46 all of these
variants could potentially reflect the earliest form of the text, so the decision
will have to be made on internal grounds.
The difference in meaning between the presence and absence of the pro-
noun ὑμῖν concerns the scope of the anathema. When the pronoun is present
in the text, Paul’s anathema relates to what he “or an angel from heaven”
might preach to the Galatians, the referent of ὑμῖν. As it is written, the ana-
thema is on its literal terms geographically and historically contingent to the
situation in Galatia. Paul’s own feelings may not have been so contingent, but
the words are. For example, Paul may well have felt that if a heavenly angel
preached contrary to (or in addition to) what he preached to the Corinthians,

43
Schmid, Marcion, 74, does not view Marcion as a witness for the presence or omis-
sion of the pronoun ὑμῖν (“to you”). Tertullian, Marc. 5.2.5, cites Marcion’s text as fol-
lows: “et ideo, dices, subtextuit: licet angelus de caelo aliter evangelizaverit, anathema
sit.” Rather than strictly reconstructing a Greek text for Marcion with a term such as ἄλλως
(“otherwise”), Schmid considers the Latin aliter to be an Old Latin translation variant of
the relative clause παρ’ ὃ εὐηγγελισάμεθα ὑμῖν, which Tertullian incorporated into his
citation of Marcion. Given the looseness of Tertullian’s citation, Schmid places no weight
on the apparent omission of the pronoun ὑμῖν (“to you”), and reconstructed Marcion’s text
of Gal 1:8 as ἀλλὰ καὶ ἐὰν ἡμεῖς ἢ ἄγγελος ἐξ οὐρανοῦ εὐαγγελίσηται (ὑμῖν παρ’ ὃ
εὐηγγελισάμεθα ὑμῖν), ἀνάθεμα ἔστω (Schmid 315), in which the parenthesis indicates that
the source material is too deficient to establish its contents.
44
There is also variation in the form of this verb, which this section does not address.
45
Comfort, Early Manuscripts, 150 emphasizes the weight of P51vid and B over .
46
BURTON 26; BUSCEMI 45.
150 Chapter 4. Examining and Establishing the Text

that angel too would be anathematized, but Gal 1:8 with an explicit pronoun
ὑμῖν, “to you, the Galatians,” does not express that thought. As a result, the
absence of the pronoun makes the condemnation broader in geographic scope
and less contingent on the historical occasion of Paul’s letter to the Galatians.
As for the position of the pronoun, few exegetes have had anything at all to
say about it, except that Weiss contends that the placement of ὑμῖν before the
verb is unusual. 47 Weiss is right on this point. The most common place for
such a pronominal object is immediately following the verb. 48 This common
position is, in fact, the word order of καὶ ἐὰν ... ἄγγελος ἐξ οὐρανοῦ εὐαγγελί-
ζηται ὑμῖν παρ’ ὃ εὐηγγελισάμεθα ὑμῖν (“even if . . . an angel from heaven
might preach to you besides what we preached to you”). The subject ἄγγελος
ἐξ ουρανοῦ comes before the verb to introduce the heavenly angel as a new
discourse topic. In particular, even this hypothetical angel would be anathe-
matized if it preaches anything besides what Paul preached to the Galatians.
There are two main reasons for placing a pronominal object before the
verb. If the pronoun itself is emphatic, it is placed first in the sentence, but no
variant reading of v.8 has this placement for the pronoun. Another reason is
that the pronoun follows an element fronted before the verb for emphasis,
especially to highlight the element as the most informative part of the
sentence. 49 This is in fact the word order of the variant reading: καὶ ἐὰν ... ἄγ-
γελος ἐξ οὐρανοῦ ὑμῖν εὐαγγελίζηται παρ’ ὃ εὐηγγελισάμεθα ὑμῖν, where the
preverbal constituent is ἄγγελος ἐξ οὐρανοῦ. According to this word order,
the angel from heaven is the most important part of the sentence because it is
an unpredictable participant in Paul’s argument. 50 The rest of the conditional
clause, other than the heavenly angel, is presupposed, background material.
In other words, the predicate of the conditional clause, ὑμῖν εὐαγγελίζηται
παρ’ ὃ εὐηγγελισάμεθα ὑμῖν (“preaches to you besides what we preached to
you”) has to be already in view earlier in the context. In this case, that context

47
WEISS 96: “das ungewöhnlich gestellte ὑμῖν.”
48
Levinsohn, Discourse Features, 29: “The default position for ‘unemphatic’
pronominal constituents is immediately following the verb.”
49
Levinsohn, Discourse Features, 39: “A change in the position of pronominal
constituents often occurs when a focal constituent precedes the verb.” Note that Levinsohn,
Discourse Features, 42, calls the new information of the sentence the “focus,” and the
words that instantiate the focus are the “focal constituent.” See also BDF 248, § 472: “Any
emphasis on an element in the sentence causes that element in the sentence to be moved
forward.”
50
Levinsohn, Discourse Features, 42: “Focal constituents, in contrast, typically are the
most important piece of new information in the comment about the propositional topic. As
a result, focal constituents are often anarthrous.” Thus, the fact that the subject ἄγγελος ἐξ
οὐρανοῦ is anarthrous helps to confirm that it is “new” and therefore focal.
Another indication that the angel is new and unexpected is the adverbial καί (“even”)
before ἐάν; e.g. BURTON 26: “the καί is intensive, marking the extreme nature of the
supposition”; so also LONGENECKER 16.
 4.2 Examination of Variation Units with Early, Split Support 151

is v. 7 τινές εἰσιν οἱ ταράσσοντες ὑμᾶς καὶ θέλοντες μεταστρέψαι τὸ εὐαγγέ-

λιον τοῦ Χριστοῦ (“there are some who are disturbing you and wishing to dis-
tort the gospel of Christ”). In sum, the preverbal placement of the pronoun
adds to the contextually-dependent meaning of v.8 in two ways: first, it em-
phasizes the unexpected heavenly angel as new and important to Paul’s line
of thought, and, second, it signals that preaching besides what Paul preached
to the Galatians is a distortion of the gospel of Christ.
The intrinsic probabilities pertain to how suitable the readings are to the
author’s context and style. Of the three variant readings, the one most inte-
grated into Paul’s argumentative context is the one with the preverbal place-
ment of the pronoun. Not only is this reading historically contingent with its
inclusion of the pronoun, but it is also contextually contingent because the
context is necessary to supply what Paul presupposed by his reference to
preaching besides what he had preached. As for the post-verbal placement of
the pronoun, it too is historically contingent on what was preached to the Ga-
latians, but it is less integrated into the context. Here, it introduces a hypo-
thetical heavenly angel into the discourse, though with no special emphasis,
and predicates a behavior upon it that Paul found objectionable, but without
implying a necessary connection between preaching against Paul’s preaching
and the troublemakers of v.7. As for the reading without the pronoun, it is
less contingent on the occasion of Paul’s letter and more suitable for a
statement of a general principle. The first reading fits Paul’s context the best,
while the other, progressively more decontextualized readings suit the needs
of later readers of Paul, who are no longer part of Paul’s context.
These intrinsic grounds go hand-in-hand with the transcriptional con-
siderations. Later readers and writers of the text are more likely to prefer the
more generalized forms of the statement for their theological needs, even
though these more generalized forms are less suitable for Paul’s own context.
Some critics have argued that the variation in the placement of the pronoun
suggests that it is secondary, 51 but this argument does not explain why scribes
would want to make Paul’s anathema more specific rather than more univer-
sal. Indeed, the omission of the pronoun generalizes Paul’s anathema beyond
a specific application to the Galatians. 52 To the extent that harmonization
could have affected the placement of the pronoun, that harmonization could
go in either direction; scribes could have been tempted to assimilate either to

51
NET 493 n.6 (“the fact that it floats suggests its inauthenticity”), 859. Also mentioned
by LIGHTFOOT 77; METZGER 521; ZIMMER¹ 484.
52
BUSCEMI 45; LÉGASSE 65 n.3; METZGER 521; MATERA 47. But NET 859 argue for a
motivation of clarification without addressing that such a clarification would limit the
scope of Paul’s anathema.
152 Chapter 4. Examining and Establishing the Text

v.8b, after the verb (παρ’ ὃ εὐηγγελισάμεθα ὑμῖν) 53 or to v.9b (εἴ τις ὑμᾶς
εὐαγγελίζεται) 54 before the verb.
In conclusion, the reading with the strongest intrinsic and transcriptional
probabilities is with the preverbal placement for the pronoun: καὶ ἐὰν ... ἄγ-
γελος ἐξ οὐρανοῦ ὑμῖν εὐαγγελίζηται παρ’ ὃ εὐηγγελισάμεθα ὑμῖν. This find-
ing fits with the external evidence, and so it should be adopted for a critical
text of Galatians. 55

Gal 2:6 Ἀπὸ δὲ τῶν δοκούντων εἶναί τι, – ὁποῖοί ποτε ἦσαν οὐδέν μοι διαφέρει·
πρόσωπον ὁ θεὸς ἀνθρώπου οὐ λαμβάνει – ἐμοὶ γὰρ οἱ δοκοῦντες οὐδὲν
προσανέθεντο,
Var. Western Eastern EDS.
ὁ θεός P46  33 ×1175 A P 1241S (Ψ Chrys) 56 T
SCC NA S WH
T

θεός (D F G) 57 B C 1739 1611 Byz M

SBL NA RP
M
BW WH T L
? Marc d b ×vg

The Eastern branch is split as to the presence or absence of the article before
God, but its closest basic allies  and 33 include it. P46 and B are split, with
P46 agreeing with  and 33 on its inclusion, and B agreeing with the Western
branch on its omission. B and the Western branch, however, do not agree on
the position of the word θεός. If the external evidence leans in any direction,
it would appear to be in favor of the agreement between P46, , and 33 in
support of the article. The lack of clarity in the external evidence, however,
means that this variation unit needs to be decided by internal evidence.
The expression λαμβάνειν πρόσωπον ἀνθρώπου (literally, “to accept a
person’s appearance [or face]”) is an idiom that means to show favoritism or
partiality to somebody. 58 With or without the article before God, the state-
ment πρόσωπον [ὁ] θεὸς ἀνθρώπου οὐ λαμβάνει means “God does not show
partiality to anybody.” The nuance of πρόσωπον as “outward appearance” is
not entirely missing, because the immediate context refers to “those who
seem to be something” (τῶν δοκούντων εἶναί τι) and “whatever they once
were” (ὁποῖοί ποτε ἦσαν). Though the meaning of the statement seems clear
enough with or without the article, the article does have an effect on where
the emphasis lies within the statement.

53
WEISS 96.
54
LÉGASSE 65 n.3.
55
LONGENECKER 16 prints this reading (against the critical text), perhaps on the
strength of P51vid and B (13 n.b).
56
Ψ and Chrys have the word order πρόσωπον ἀνθρώπου ὁ θεός.
57
D F G have the word order θεὸς ἀνθρώπου πρόσωπον.
58
BDAG “πρόσωπον 1bα,” 888.
 4.2 Examination of Variation Units with Early, Split Support 153

With the article before God, the emphasis is expected to be on the first
word of the clause, 59 πρόσωπον (“appearance”). 60 After talking earlier in the
verse about those with a good reputation (i.e., those who seem to be some-
thing) and about “whatever they once were,” the emphasis on πρόσωπον
(“appearance”) suggests that it is not their appearance or reputation that is
important to God but something else. This contrast is left implied in Gal 2:6,
but in 2 Cor 5:12 a similar contrast is explicit, between outward appearances
and the heart (τοὺς ἐν προσώπῳ καυχωμένους καὶ μὴ ἐν καρδίᾳ). 61 The idiom
is found – also with an articular God – in Rom 2:11 οὐ γὰρ ἐστιν
προσοωπολημψία παρὰ τῷ θεῷ (“for there this no partiality with God”). This
statement is in a context that everyone who does good (παντὶ τῷ ἐργαζομένῳ
τὸ ἀγαθόν) will be honored – both Jews and Greeks – where the emphasis on
impartiality helps to confirm that ethnicity does not matter. Likewise in Gal
2:6 the emphasis on “appearance” serves to support Paul’s statement that
whatever those of repute used to be does not matter to him. 62
Without the article, commentators have long noticed that the emphasis
falls on God, though they differ as to the reason why. Ernest de Witt Burton,
for example, argues that “the use of θεός without the article . . . employed
with qualitative force with emphasis upon the divine attributes, especially in
contrast with man, is an established use.” 63 Stephen H. Levinsohn, as another
example, points out: “If an anarthrous substantive has a unique referent and is
activated [i.e. mentioned in context], then its referent is prominent.” 64 Lev-
insohn illustrates this point with Gal 2:19 ἐγὼ γὰρ διὰ νόμου νόμῳ ἀπέθανον,
ἵνα θεῷ ζήσω (“for through the Law I died to the Law, so that I live for
God”). Here, both νόμῳ (“Law” i.e., “Torah”) and θεῷ (“God”) are activated,

59
See BDF 248, § 472: “Any emphasis on an element in the sentence causes that
element in the sentence to be moved forward.”
60
As applied to Gal 2:6, Levinsohn, Discourse Features, 61: “If the articular reading is
followed, it is more likely that πρόσωπον ἀνθρώπου is preposed for focus. Ὁ θεός is then a
supportive constituent (sec. 3.8.1). Πρόσωπον ἀνθρώπου is discontinuous because it is the
‘face’ of a person that relates back to ‘what they actually were’ in v. 6b (Arichea & Nida
1975:35) rather than the phrase πρόσωπον ἀνθρώπου as a whole.”
61
So David M. Hay, “Paul’s Indifference to Authority,” JBL 88 (1969): 36–44 at 42.
62
Hay, “Paul’s Indifference,” 41, arguing that this statement “is meant to be a kind a
kind of OT proof text to support the preceding clause: Paul's indifference to the Jerusalem
chiefs is grounded on God’s [impartially].” Unfortunately, Hay does not address the text-
critical issue.
63
BURTON 89. So also EADIE 120; LIGHTFOOT 108; MEYER 89; WEISS 71.
64
Levinsohn, Discourse Features, 162. This principle is applied to Gal 2:6 specifically
at Levinsohn, Discourse Features, 61: “If anarthrous θεὸς is read in Gal. 2:6c, then πρόσω-
πον ἀνθρώπου precedes the negated verb because it is a point of departure (marking a shift
from ‘what they actually were’ in v. 6b). In turn, θεὸς precedes the verb because it is a
focused constituent (“the emphatic θεός”–Eadie 1869:120) that is not negated (it occurs
before the negative particle).”
154 Chapter 4. Examining and Establishing the Text

unique referents without the article, and they are both prominent because they
are part of a double contrast construction: dying to the Law versus living for
God. The word order of the Western branch, where θεός is at the beginning of
the clause, makes this word even more prominent. So, in Gal 2:6 with an
anarthrous God being prominent, the contrast is along the lines of “God does
not show partiality (but some people do).” Indeed, the fact that anarthrous
references to God lend prominence to God makes the lack of the article
especially common in other quasi-proverbial statements. 65 For example, θεός
lacks the article in Gal 6:7 Μὴ πλανᾶσθε, θεὸς οὐ μυκτηρίζεται (“Don’t be
misled, God is not mocked”), where the emphasis on God is appropriate. As
another example, θεός is anarthrous in 1 Thess 2:5 θεὸς μάρτυς (“God [is my]
witness”), another case of appropriate prominence on God in the statement.
Accordingly, with regard to the internal evidence, the presence of the
article appears to be the harder reading. Gal 2:6 looks like a proverbial state-
ment, and proverbial statements with God often omit the article before it in
order to lend more prominence to God. While it seems likely that scribes
would want to enhance references to God in their exemplars (here, by omit-
ting the article), the opposite change is harder to understand. On the other
hand, the articular ὁ θέος in v.6 fits the author’s context better because it
supports Paul’s statement that those of high repute do not matter to him. 66 For
these reasons, the critical text should include the article before θέος.

Gal 2:16 15 Ἡμεῖς φύσει Ἰουδαῖοι καὶ οὐκ ἐξ ἐθνῶν ἁμαρτωλοί· 16 εἰδότες δὲ ὅτι οὐ
δικαιοῦται ἄνθρωπος ἐξ ἔργων νόμου ἐὰν μὴ διὰ πίστεως Ἰησοῦ Χριστοῦ,
καὶ ἡμεῖς εἰς Χριστὸν Ἰησοῦν ἐπιστεύσαμεν ἵνα δικαιωθῶμεν ἐκ πίστεως
Χριστοῦ καὶ οὐκ ἐξ ἔργων νόμου, . . .
Var. Western Eastern EDS.
δέ D* F G d b ×vg B  C 1241S Chrys T
SCC SBL NA S BW WH T L
− P46 33 ×1175 A P 1739 Ψ M
NA RP
1611 Byz
? Marc

Except for the Western branch, which supports the inclusion of the conjunc-
tion δέ, the external evidence is divided in every group. Even among the
Byzantine witnesses, which by and large omit the conjunction, there are many
manuscripts that include the δέ. 67 The evidence of P46 is not definite, how-

65
BETZ 95 calls λαμβάνει a “proverbial present.”
66
BUSCEMI 150 claims that the article better conforms to Paul’s style, but does not
explain how; cf. BURTON 89: “the N. T. writers and Paul in particular rarely use θεός as
subject without the article.” So also ZIMMER¹ 491–492, who further suggests that the
article was omitted as a harmonization to the anarthrous ἀνθρώπου, but SIEFFERT 108 n.*
argues that scribes are relatively more likely to add the article than to remove it.
67
Among those collated for this study: L 075 223 330 1319 1573 1646 1734 1827 2138
2147 2400 2892.
 4.2 Examination of Variation Units with Early, Split Support 155

ever, because it has a noticeable tendency to omit conjunctions and parti-

cles. 68 These considerations suggest that the external evidence tilts, if any-
where, toward inclusion of the conjunction.
Commentators have been divided about whether Gal 2:15 continues Paul’s
discourse to Peter in v.14 or whether it opens a new section of the letter. 69
Either possibility may seem to fit with the asyndeton of v.15, which tends to
signal either a large discontinuity such as a change in topic for the argument
(cf. the asyndeton at 1:6) or a close connection to the previous statement. 70
The connection between v.14 and v.15, however, does not appear to be
especially close, because v.15 does not seem to answer the question Paul
posed to Cephas in v.14 (“If you live like a gentile, how can you compel the
gentiles to obey Jewish law?), 71 and the topic switches from the second
person singular to first person plural. 72 Therefore, it is best to treat v.15 as the
beginning of a new section of the letter. 73
If the conjunction δέ is included in the text, it signals a shift in Paul’s
argument. 74 In this case, the shift would be a movement from Paul’s and
Cephas’s natural identity as Jews to their common trust in Christ Jesus. 75
With the discontinuity marked by δέ, the instances of the first person nomina-
tive plural pronoun ἡμεῖς (“we”) in v.15 and v.16b should then be the explicit
subjects of their respective clauses. 76 In other words, v.15 is a standalone

68
ROYSE 270–298, esp. 297: “it is clear that the scribe of  46 omits conjunctions,
articles, and pronouns – whose omissions usually results at worst in asyndeton or lack of
clarity –, much more often than he omits nouns and verbs.” ROYSE 271 documents 55
examples where P46’s omission of a conjunction or particle resulted in a singular (or a
near singular “asterisked”) reading.
69
BETZ 113, especially n.6.
70
See generally Black, Sentence Conjunctions, 179–217, esp. 216–217; Levinsohn,
“Constraints,” 330–333; Levinsohn, Discourse Features, 118–119; Runge, Discourse
Grammar, 36.
71
If Gal 2:15 is an answer to that question, it would appear to say: “You cannot do
either, because you (and I) are natural Jews and they are gentile sinners (i.e., non-observers
of the Torah).” This answer seems to take for granted a strong ethnic distinction that is be-
lied elsewhere in Paul (e.g. Gal 3:28).
72
Levinsohn, “Los rasgos,” 36.
73
LONGENECKER 80–81; ZAHN 120. Indeed, BETZ 114 identifies vv.15–21 as the epis-
tle’s propositio, which sums up the preceding statement of facts (1:12–2:14) and sets up
the arguments that follow (3:1–4:31).
74
Black, Sentence Conjunctions, 142–178; Buth, “Levinsohn’s,” 53–56; Levinsohn,
“Constraints,” 320–325; Runge, Discourse Grammar of the Greek New Testament, 46.
75
BETZ 115 n.29 takes the contrast as between φύσει (“by birth”) and εἰδότες (“know-
ing”); BUSCEMI 206–207 construes the contrast as between a Jewish-Christian position and
Paul’s Christian position. SCHLIER 52 cleverly expresses the contrast as between a Jewish-
Christian and a Jewish-Christian.
76
Curiously, LONGENECKER 83 analyzes δέ “simply as a connective without contrast,”
coordinating the participial clause with the nominal appositions. Without explanation, so
156 Chapter 4. Examining and Establishing the Text

verbless copula clause: “We are Jews by nature and not sinners from the
gentiles.” 77
If the conjunction δέ is not to be included in 2:16a, the syntax is messier
and there are a couple of viable exegetical options. One option is that the
participial phrase introduced by εἰδότες begins a new sentence: “We are
naturally Jews and not sinners from the gentiles. Knowing that a person is not
made right . . ., we too believed . . . .” The option is similar to that with the
reading δέ, except that asyndeton rather than δέ separates v.15 and v.16. The
use of asyndeton is somewhat unusual here; it does not seem to signal the be-
ginning of a new section, but, on the contrary, a fairly close connection to the
preceding statement. 78 Under this reading, the adverbial καί before ἡμεῖς
(“we too”) seems to indicate that those “sinners from the gentiles” are those
who also knew that a person is not made right except through the faithfulness
of Jesus Christ. That is, they are Antiochene gentile Christians, the people
with whom Cephas did not want share a meal.
Another option is that the participial phrase with εἰδότες goes with the pre-
vious clause in v.15, and the following clause unobjectionably begins with a
conjunctive καί: “We are naturally Jews . . . And we believed . . . .” With this
option, it is unclear whether the participial phrase construes with the subject
of v.15, ἡμεῖς (“We are naturally Jews . . . who know . . .”), or the nearest
masculine nominative plural, ἁμαρτωλοί (“We are . . . not gentile sinners who
know . . .”). Neither possibility seems viable, for the first would seem to ex-
clude gentile Christians from that knowledge of how someone is made right,
while the second would seem to exclude Paul, who is naturally a Jew, from
that knowledge.
Still another option is that there is one long clause, with v.15 setting forth
a proleptic subject and resumed by καὶ ὑμεῖς in v.16b, with v.16a interposing
a participial phrase introduced by εἰδότες: “We [who are] Jews by nature and
not sinners from the gentiles, knowing that a person is not made right . . ., we

also BURTON 119. This would be an odd use of δέ, however, which ordinarily coordinates
larger discourse units.
77
LIGHTFOOT 114; MARTYN 256; SCHLIER 51; SIEFFERT 141, pointing to 2 Cor 10:7
ὅτι καθὼς αὐτὸς Χριστοῦ, 11:6 εἰ δὲ καὶ ἰδιώτης τῷ λόγῳ ἀλλ’ οὐ τῇ γνώσει, Phil 3:15
ὅσοι οὖν τέλειοι. Conversely, ZAHN 120–121 is troubled by the lack of an explicit copula,
so he takes v.15 instead as a proleptic nominal in apposition to a renewed explicit subject
καὶ ἡμεῖς; so also BETZ 113; LÉGASSE 167; LONGENECKER 81; MATERA 92 (bracketing
“[but]”); MUSSNER 133; ROHDE 100. This is an old debate, in the nineteenth century,
EADIE 161–162 cites a half-dozen scholars on each side.
78
Cf. Douglas A. Campbell, The Deliverance of God: An Apocalyptic Rereading of
Justification in Paul (Grand Rapids, Mich.: Eerdmans, 2009), 1144 n.19: “The δέ . . .
might have been prompted by a desire to distance the material in 2:16 from the Jewish
identifications of 2:15, which can be read as a separate sentence.”
 4.2 Examination of Variation Units with Early, Split Support 157

too believed . . . .” 79 The option leaves it somewhat ambiguous whether the

participle phrase is to be construed backwards or forwards, but construing it
forwards with καὶ ὑμεῖς avoids the problems of the previous option. What
makes this third option problematical, however, is that the clause contains
two explicit pronominal subjects, ἡμεῖς in v.15 and καὶ ἡμεῖς in v.16.
Ordinarily, one explicit pronominal subject is remarkable but having two of
the same is surprising. 80
Among these three options for the asyndetonic reading, the least awkward
is the first one, where v.15 is an independent clause, and v.16 follows as its
own sentence. This, in fact, is very similar to the reading with the conjunction
δέ, except that δέ makes the connection between v.15 and v.16 explicit. 81
Paul’s argument shifts from one point (about ethnicity) to another (about a
belief or trust common to Christians). The conjunction δέ fits Paul’s argu-
ment better than the asyndeton, but both variants are acceptable to the
context.
Transcriptionally, the change can happen in either direction, and, given the
distribution of the different variant readings in all but the Western branch,
changes in both directions have probably occurred. On the one hand, the
asyndeton at v.16 may have been perceived as hard, and δέ is the natural con-
junction to supply. On the other hand, scribes may have found the καὶ ἡμεῖς
after the participial clause confusing and misinterpreted the καί conjunctively
as the start of a new sentence, leaving the participial clause high and dry – a
problem they could fix by omitting the δέ. 82 Another factor is that the
omission of the conjunction may have been occasioned by the start of a
lection. 83 Finally, in the earliest period of the transmission, the omission of
particles and conjunctions was fairly frequent, as in the case of P46. 84
All in all, the internal evidence, especially the transcriptional probabilities,
is not decisive, though the conjunction appears to be the better intrinsic fit.
Coupled with the external evidence favoring δέ, the balance of the textual
evidence tilts ever so slightly to the inclusion of the conjunction.

79
E.g., LÉGASSE 167; MATERA 92 (bracketing “[but]”).
80
More idiomatic (and Pauline) would be to use καὶ αὐτοί as in Rom 8:23; cf. ZAHN
120 n.56.
81
BUSCEMI 195, however, finds the conjunction δέ abrupt and disturbing the flow of
the argument: “il δέ in fatti è inatteso e turba la linearità del dettato.”
82
MEYER 54; SIEFFERT 142 n.**.
83
SIEFFERT 142 n.**.
84
ROYSE 721.
158 Chapter 4. Examining and Establishing the Text

Gal 3:3 οὕτως ἀνόητοί ἐστε, ἐναρξάμενοι πνεύματι νῦν σαρκὶ ἐπιτελεῖσθε;
Var. Western Eastern EDS.
ἐπιτελεῖσθε d b ×vg P46 B AC 1739 Ψ SCC SBL NA RP
Chrys 1611 Byz S BW WH T L
ἐπιτελεῖσθαι D* F G  33 ×1175 1241S
? Marc P

Not noted in the NA27 apparatus, the infinitive ἐπιτελεῖσθαι is attested both
in the otherwise reliable  33 and the Greek witnesses of the Western
branch. 85 Nevertheless, the external evidence generally favors the indicative
ἐπιτελεῖσθε. Thus, the infinitive ἐπιτελεῖσθαι would need compelling internal
evidence to overturn the indicative ἐπιτελεῖσθε.
Though these variants were pronounced identically after the second cen-
tury, 86 they have different meanings and interact with the clause in different
ways. On the one hand, the indicative ἐπιτελεῖσθε creates its own separate
clause and the verse as a whole can be rendered, “Are you so senseless? After
getting started in the spirit, are you now getting finished in the flesh?” On the
other hand, the infinitive ἐπιτελεῖσθαι is a complement to the adjective
ἀνόητοι, 87 and the verse would mean: “Are you, who got started in the spirit,
so senseless as to get finished now in the flesh?”
The infinitive ἐπιτελεῖσθαι is the smoother reading because it avoids the
asyndeton of the indicative ἐπιτελεῖσθε. 88 Thus, the internal evidence favors
the harsher reading and thereby agrees with the external evidence that the in-
dicative ἐπιτελεῖσθε should be adopted as the reading of the critical text.

85
The Latin witnesses read coeperitis, which corresponds to the indicative ἐπιτελεῖσθε.
86
W. Sidney Allen, Vox Graeca: The Pronunciation of Classical Greek (3d ed.;
Cambridge: Cambridge University Press, 1987), 79.
87
The use of a complementary infinitive for ἀνόητος is attested, for example, in Plato,
Republic 336e, and Gorgias, 514e.
88
Cf. MATERA 112–113, who felt it necessary to add an infinitive complement to his
translation “to render the Greek more clearly,” as follows: “Are you so foolish [as to think]
that having begun with the Spirit you are now made perfect by the flesh?”
 4.2 Examination of Variation Units with Early, Split Support 159

Gal 3:21 . . . . εἰ γὰρ ἐδόθη νόμος ὁ δυνάμενος ζῳοποιῆσαι, ὄντως ἐκ νόμου ἂν ἦν ἡ

δικαιοσύνη·
Var. Western Eastern EDS.
ἐκ νόμου ἂν ἦν (B) 89 AC SCC SBL NA BW
1241S TL
ἐκ νόμου ἦν ἄν (P46)  33 1739 S WH
M

×1175 (Ψ* −ἦν)

ἂν ἐκ νόμου ἦν Chrys 1611 RP
Byz
ἐκ νόμου ἦν D* WH
M

ἐκ νόμου FGd
? Marc b ×vg P

The external evidence for the position of the particle ἄν is fractured across the
witnesses. Although  and 33 support the order ἦν ἄν, both the Eastern
branch and the P46–B group are split between the orders ἂν ἦν and ἦν ἄν, 90
and the Western branch cannot resolve the choice because that branch omits
at least the particle ἄν. 91 Nevertheless, it seems more likely that the particle
would be omitted if the Western branch exemplar had read ἦν ἄν rather than
ἂν ἦν, because the former merely requires a simple skip from the first nu to
the second. These considerations suggest that the order ἦν ἄν has the
strongest external evidence in its favor, but internal evidence is what has to
decide which variant belongs to the critical text.
The Greek particle ἄν does not have an English translation, and this
variation unit concerns the position of this untranslatable particle. Because of
this, it might seem that there is no exegetical significance to the placement of
ἄν, but the position of the Greek particles is driven by prosodic concerns,
particularly in the placement of phrase and colon boundaries. Thus, particles
can serve as indirect evidence for “punctuation after the fact.” 92

89
Instead of ἐκ νόμου, the related P46 and B read the phrase ἐν νόμῳ (“by the Law”).
This sub-variant is discussed in Chapter 3.
90
WEISS 135 denigrates the external evidence in favor of ἦν ἄν as “nur nach  und 6
Min.,” but this judgment was made before the publication of the very good witnesses P46
and 1739.
91
It is unclear whether the omission of both ἂν ἦν by F G should count as support for
the omission of just ἄν in D; see WEISS 104: “Woher WH. Gal. 3, 21 das αν, das in D ne-
ben (wo nicht gar nach, vgl. ) ην aus Schreibeversehen ausfiel, einklammern, ist garnicht
abzusehen, da FG, in denen das αν ην fehlt, nicht einmal mit dafür zeugen.” ZAHN 178
n.43, however, takes the omission of ἄν to be original and more forceful; so also SCHLIER
121 n.2.
92
Helma Dik, Word Order in Ancient Greek: A Pragmatic Account of Word Order
Variation in Herodotus (ASCP 5; Amsterdam: J. C. Gieben, 1995), 35.
160 Chapter 4. Examining and Establishing the Text

According to the foundational research of Wackernagel and Fraenkel, 93 the

predominant position for the particle ἄν is second in its clause or colon. 94
Indeed, this is the most common position in the New Testament and other
first century literature such as Josephus and Epictetus. 95 It must also be kept
in mind that the notion of the “second position” is qualified in that ἄν is not
found after phonologically light words such as articles, prepositions, and
other proclitics. For example, in Rom 9:29 the particle ἄν is found in the
syntactically third position ὡς Σόδομα ἂν ἐγενήθημεν, since the word ὡς is
too phonologically light to host the particle. 96 Another important qualification
is that particularly emphatic constituents of a clause can be preposed before
the verb and placed in a colon of their own; when this preposing happens, the
particle ἄν is in second position of the colon containing the verb. For
example, in Gal 1:10 Χριστοῦ δοῦλος οὐκ ἂν ἤμην (“I would not be a slave
of Christ”), the predicate nominate noun phrase Χριστοῦ δοῦλος is fronted
into its own colon to confer emphasis, but the particle ἄν is placed in the
second position of the colon with the verb. With these principles in mind, let
us examine the various readings.
The Byzantine reading, ὄντως ἂν ἐκ νόμου ἦν ἡ δικαιοσύνη (“righteous-
ness would really be from the law”), is the most straightforward exegetically.
The particle ἄν is in the classic Wackernagel position and unambiguously
defines the apodosis to begin with the word ὄντως. Externally, on the other

93
Jacob Wackernagel, “Über ein Gesetz der indogermanischen Wortstellung” in Jacob
Wackernagel, Kleine Schriften (Göttingen: Vandenhoeck & Ruprecht, 1953), 1–104;
Eduard Fraenkel, “Kolon und Satz: Beobachtungen zur Gliederung des antiken Satz, I” in
Eduard Fraenkel, Kleine Beiträge zur klassischen Philologie (vol. 1; Raccolta di studi e
testi 95; Rome: Edizioni di storia e letteratura, 1964), 73–92; idem, “Kolon und Satz: Be-
obachtungen zur Gliederung des antiken Satz, II” in Fraenkel, Kleine Beiträge, 93–130.
For a treatment, informed by recent developments in pragmatics, see David Michael
Goldstein, “Wackernagel’s Law in Fifth-Century Greek” (Ph.D. diss., University of
California, Berkeley, 2010).
94
Because of this principle, arguments of this variant based on the position of ἄν with
respect to the verb are invalid. These include BURTON 194, who regards the preverbal
placement of ἄν as more Pauline, and SIEFFERT 218 n.*, who considers the placement of
ἄν after the verb anomalous (and dubiously the harder reading) because nowhere else in
Paul does this particle follows the verb (“bei Paulus nirgends sonst ἄν auf das Verbum
folgt”).
95
Based on my examination of ἄν in the TLG database. Of course, a large number of
these cases involve ἄν after the relative pronoun ὅς and similar words. But, even in an apo-
dosis, we have clear examples such as 1 Cor 2:8 (οὐκ ἂν τὸν κύριον τῆς δόξης ἐσταύρω-
σαν) and 11:31 (οὐκ ἂν ἐκρινόμεθα).
96
Other New Testament examples include John 4:10 καὶ ἔδωκεν ἄν σοι ὕδωρ ζῶν (be-
cause of the light καί) and John 15:19 ὁ κόσμος ἂν τὸ ἴδιον ἐφίλει (because of the light ὁ).
Multiple light words can coalesce together around one heavier word to host the particle, as
in John 8:19 καὶ τὸν πατέρα μου ἂν ᾔδειτε.
 4.2 Examination of Variation Units with Early, Split Support 161

hand, the support for this reading is late and appears to be an improvement in
clarity over the other attested positions for ἄν.
The reading of B and a subgroup of Alexandrians, which is adopted into
the Nestle-Aland text, places the particle after ἐκ νόμου, as ὄντως ἐκ νόμου
ἂν ἦν ἡ δικαιοσύνη, which means that the prepositional phrase goes into the
apodosis. More importantly, this placement indicates that the adverb ὄντως
belongs in a different colon, because it is too phonologically heavy to be
phrased with ἐκ νόμου. This placement confers a strong emphasis on the
word ὄντως, which can be paraphrased in English as follows: “For if a law
was given that could give life, it would really be true that righteousness is
from the law.” B’s closest relative P46, on the other hand, reverses the order
of the verb and the particle: ὄντως ἐκ νόμου ἦν ἂν ἡ δικαιοσύνη. In this
reading, the particle is placed after the verb, indicating that the phrase ὄντως
ἐκ νόμου belongs to its own colon and as a whole would be emphatic: “For if
a law was given that could give life, then it would really be from the law [as
opposed to somewhere else] that righteousness comes.”
Intrinsically, the reading ὄντως ἐκ νόμου ἂν ἦν ἡ δικαιοσύνη, with its
strong focus on ὄντως, fits the context well as an implied denial of the teach-
ing of Paul’s opponents that righteousness comes from the Torah. 97 Tran-
scriptionally, the ἂν ἦν reading is the most unusual and hence the mostly like-
ly to be changed. These internal considerations favor ἂν ἦν as the authorial
reading.

Gal 4:18 καλὸν δὲ ζηλοῦσθαι ἐν καλῷ πάντοτε καὶ μὴ μόνον ἐν τῷ παρεῖναί με πρὸς
ὑμᾶς.
Var. Western Eastern EDS.
ζηλοῦσθαι ×1175 A C 1241S 1739 SCC SBL NA
T
S BW WH T L
τὸ ζηλοῦσθαι D F G Chrys 1611 RP
(d est aemulari) Byz
ζηλοῦσθε (b ×vg aemulamini) B  33 T
M

? Marc P46 P Ψ

The weight of the external evidence favors the infinitive ζηλοῦσθαι, being
found in both the Western and Eastern branches, though the Western and
Byzantine texts add an article. The apparent support by the Latin reading
aemulamini for the reading ζηλοῦσθε is deceptive, because this Latin verb is
deponent while the proper Greek equivalent is active, not passive. 98 Thus, the

97
E.g. BETZ 174.
98
Neither BDAG “ζηλόω,” 427, nor LSJM “ζηλόω,” 755, admit a middle sense for this
verb. So also BURTON 247; LONGENECKER 194; Friedrich Zimmer, Exegetische Probleme
des Hebräer- und Galaterbriefs (Neutestamentliche Studien 1; Hilburghausen: Gadow,
1882), 193–194.
162 Chapter 4. Examining and Establishing the Text

Latin bonum autem aemulamini bears the sense of “but [you] strive after the
good,” a meaning that the Greek does not have. 99
These homonyms, at least from the second century on, 100 have different
senses. The infinitive ζηλοῦσθαι is the complement of καλόν (ἐστιν), with the
meaning: “but it is good to be sought after always in a good way.” By
contrast, the indicative ζηλοῦσθε is harder to construe, but the most
satisfactory analysis is to take the adjective καλόν as an accusative of respect
and render the clause as “but you are being properly sought after always in a
good way.” The indicative reading, however, does not fit the context because
of the tension with v.17 ζηλοῦσιν ὑμᾶς οὐ καλῶς (“they are seeking after
you, not properly”) and the redundancy of ἐν καλῷ in v.18. Transcriptionally,
the indicative ζηλοῦσθε is a common orthographic mistake among scribes. 101
Thus, the internal evidence also favors the infinitive ζηλοῦσθαι.

Gal 4:19b τέκνα μου, οὓς πάλιν ὠδίνω ἄχρις οὗ μορφωθῇ Χριστὸς ἐν ὑμῖν·
Var. Western Eastern EDS.
ἄχρις DFG C 33 ×1175 A C 1241S Ψ Chrys SCC ( RP ἄχρι) L
1611 Byz
μέχρις B * 1739 SBL NA S BW WH T
? Marc d b ×vg P46 P

This variation unit is not noted in the NA27 apparatus. The external evidence
in favor of ἄχρις is more genealogically coherent than that of the NA27
reading, μέχρις, despite its presence in the highly-regarded , B, and 1739. 102
These etymologically related prepositions have the same meaning: “until
when,” 103 which makes the internal evidence harder to evaluate than usual.
The conjunction ἄχρι(ς) οὗ is paralleled elsewhere in Paul (Rom 11:25; 1 Cor
11:26, 15:25; and Gal 3:19), while μέχρις οὗ would be its only occurrence in

99
ZAHN 220–221, who also points out that this meaning would also need an article, as
in τὸ κάλον.
100
Allen, Vox Graeca, 79.
101
BUSCEMI 413; EADIE 349; LIGHTFOOT 177–178; LONGENECKER 188 n.f; ROHDE
188. BUSCEMI 413 also thinks that ζηλοῦσθε is the easier reading because it would be a
direct exhortation.
102
WEISS 59 prefers μέχρις οὗ on the strength of his favorite B plus , noting that only
Lachmann adopted ἄχρις οὗ; so also SCHLIER 217. ZAHN 222 mildly favors ἄχρις but
without explanation.
103
See BDAG “ἄχρι,” 106; “μέχρι,” 664; Alfio Marcello Buscemi, L’uso delle preposi-
zioni nella lettera ai Galati (Analecta [Studium Biblicum Franciscanum] 17; Jerusalem:
Franciscan, 1987), 105, 107. Andrew L. Sihler, New Comparative Grammar of Greek and
Latin (New York: Oxford University Press, 1995), 441, § 409.6, notes that the initial sylla-
bles με- and ἄ- both derive from different vocalized forms (ablaut) of the initial element in
μετά, while the -χρις element is related to χείρ.
 4.2 Examination of Variation Units with Early, Split Support 163

Paul, 104 suggesting that ἄχρις οὗ is more Pauline than μέχρις οὗ – but the
mere handful of occurrences is too small to be definitive. It is also unlikely
that the ἄχρις οὗ reading is a scribal harmonization to the Gal 3:19 parallel,
ἄχρις οὗ ἔλθῃ τὸ σπέρμα, because the context is too far and too dissimilar to
trigger the harmonization. These factors suggest, weakly, that Gal 4:19 ought
to read ἄχρις οὗ.

Gal 4:25 24 ἅτινά ἐστιν ἀλληγορούμενα· αὗται γάρ εἰσιν δύο διαθήκαι, μία μὲν ἀπὸ
ὄρους Σινᾶ εἰς δουλείαν γεννῶσα, ἥτις ἐστιν Ἁγάρ. 25 τὸ γὰρ Σινᾶ ὄρος
ἐστὶν ἐν τῇ Ἀραβιᾳ· συστοιχεῖ δὲ τῇ νῦν Ἰερουσαλήμ, δουλεύει γὰρ μετὰ
τῶν τέκνων αὐτῆς.
Var. Western Eastern EDS.
omit v.25a SCC Schott cjj
T

γάρ D-Vorlage F G  C 1241S (1739 −ἐστίν) M M M T

SCC WH T L
b ×vg
γὰρ Ἁγὰρ (d −Σινᾶ) 33 P Ψ Chrys RP S T
T

1611 Byz
δὲ Ἁγὰρ D B ×1175 A SBL NA BW WH L
T M

δὲ P46 T
M

? Marc 105

The external evidence is all over the map, but the reading τὸ γὰρ Σινᾶ enjoys
plurality support among the early witnesses and branches. The evidence for
the NA27 reading τὸ δὲ Ἁγὰρ Σινᾶ is less weighty than it appears, because
the reading in D is secondary within its own tradition and therefore not a
witness to the earliest reading. H. J. Vogels has called attention to the fact
that, though D is usually arranged with a sense unit per line, the present text
departed from the arrangement by breaking 4:25a in an unusual location: Ητις
εστιν αγαρ το δε/Αγαρ σεινα ορος εστιν. 106 The words το δε at the end of the
line are not paralleled in the corresponding Latin column of d. Moreover, the
Latin on the following sense-line also reads enim, which corresponds to the
Greek γάρ. This evidence suggests that the το δε at the end of the line had
been added there in D’s exemplar, and the word γάρ had been deleted. Vogels
concludes that D’s exemplar originally agreed with F and G’s reading τὸ γὰρ

104
In fact, the only other occurrence of μέχρις οὗ in the New Testament is at Mark
13:30. In fact, Zimmer, Exegetische Probleme, 208 n.*, claims that, since μέχρις οὗ is un-
stable in Mark 13:30, it is likewise unstable in Gal 4:19.
105
Marcion’s text here is highly divergent and it has nothing corresponding to this
clause. See Schmid, Marcion, 317–318.
106
H. J. Vogels, “Der Codex Claromontanus der paulinischen Briefe,” in H. G. Wood,
ed., Amicitiae Corolla: A Volume of Essays Presented to James Rendel Harris, D. Litt., on
the Occasion of his Eightieth Birthday (London: University of London Press, 1933), 274–
299 at 294. So also Karl Th. Schäfer, “Der griechisch-lateinische Text des Galaterbriefes in
der Handschriftengruppe D E F G,” in Carl Feckes, ed., Scientia Sacra (Cologne: Bachem,
1935), 41–70, at 62–63.
164 Chapter 4. Examining and Establishing the Text

Σινᾶ. P46’s singular reading τὸ δὲ Σινᾶ could support either early reading, 107
but it seems best that P46 should be used to support the attestation of the
more substantive sub-variant, the omission of Ἁγὰρ, rather than the con-
junction. 108
There are two exegetical issues in this clause: the meaning of the particles
δέ and γάρ, and the inclusion or omission of the name Ἁγάρ. As for δέ, it
does not coordinate with the μέν in the preceding verse, μία μὲν ἀπὸ ὄρους
Σινᾶ, εἰς δουλείαν γεννῶσα, ἥτις ἐστὶν Ἁγάρ (“one [covenant] indeed is from
Mount Sinai, born into slavery, such is Hagar”), because the resulting μὲν . . .
δέ construction entails that the δέ clause refers to the other covenant, which
Gal 4:25a does not. 109 The conjunction δέ is not used adversatively, because
the information given in v. 25a does not contrast with that in v. 24; rather, v.
25a gives more information, so the conjunction δέ is best considered
connective (e.g., “now” or “and”). 110 As for γάρ, it is more specific about the
function of Gal 4:25a; it is causal, giving the reason why the covenant from
Mount Sinai is Hagar. 111
Without the name Ἁγάρ, the clause means “Sinai is a mountain in Arabia,”
because the noun ὄρος (“mountain”) functions as a predicate to Σινᾶ (“Sinai”)
in hyperbaton with its attributive preposition phrase ἐν τῇ Ἀραβίᾳ positioned
after the copula. 112 With the name Ἁγάρ, on the other hand, there are two
additional exegetical issues. First, the use of the neuter article τό before the
feminine Ἁγάρ is a quotative usage of the neuter article, 113 or perhaps a refer-

107
ROYSE 320 n.682, though Royse mistakenly thinks that P46’s τὸ δὲ Σεινᾶ is not
singular here due to misplaced reliance on Tischendorf (and Tregelles) for the claim that
33*vid also read τὸ δὲ Σινᾶ. Based on my inspection of the photograph available online at
the NT.VMR, however, no such reading is apparent for 33.
108
Pace METZGER 527.
109
In fact, it is only much later at v. 26 that Paul gets to expounding the other covenant:
ἡ δὲ ἄνω Ἰερουσαλὴμ ἐλευθέρα ἐστίν, ἥτις ἐστὶν μήτηρ ἡμῶν (“but the Jerusalem above is
free, such is our mother”), and even here it does not present a precisely parallel construc-
tion to the μέν clause. MUSSNER 320, argues that the structure of Paul’s argument is as if
he wrote: ἑτέρα δὲ εἰς ἐλευθερίαν γεννῶσα, ἥτις ἐστὶν Σαρά· συστοιχεῖ δὲ τῇ ἄνω Ἰερου-
σαλήμ, ἥτις ἐστὶν ἐλευθέρα καὶ μήτηρ ἡμῶν (“but the other is born into freedom, such is
Sarah; now she corresponds to the Jerusalem above, such is the free woman and our
mother”).
110
BETZ 244; LÉGASSE 356 n.1; LONGENECKER 198 n.d; MARTYN 438, and Herman
Ridderbos, The Epistle of Paul to the Churches of Galatia (NICNT; Grand Rapids, Mich.:
Eerdmans, 1953), 178 n.9.
111
Michael G. Steinhauser, “Gal 4,25a: Evidence of Targumic Tradition in Gal 4,21–
31?” Bib 70 (1989): 234–240 at 234.
112
LÉGASSE 356; LONGENECKER 198 n.e; and MARTYN 437.
113
BDF 140 at § 267(1). So also MARTYN 437 n.132; MATERA 1992:170; and Gerhard
Sellin, “Hagar und Sara. Religionsgeschichtliche Hintergründe der Schriftallegorese Gal
4,21–31” in Ulrich Mell and Ulrich B. Müller, eds., Das Urchristentum in seiner literari-
 4.2 Examination of Variation Units with Early, Split Support 165

ence to the name (i.e., τὸ ὄνομα) Hagar, rather than to the woman. 114 Second,
the inclusion of the name means that there are three nominative nouns in a
linking clause that normally takes two: Ἁγάρ (“Hagar”), Σινᾶ (“Sinai”), and
ὄρος (“mountain”). At least one of them must be an attributive noun (i.e.,
used adjectivally) or a noun in apposition. Even though the word order Σινᾶ
ὄρος for Mount Sinai is odd, 115 the least problematic of these options is to
take “Sinai” as attributive, so Σινᾶ ὄρος should mean “the Sinai mountain”
(or “Mount Sinai”), and the whole clause means, “this ‘Hagar’ is the Sinai
mountain in Arabia.” 116
Analysis of the transcriptional probabilities is complicated by the fact that
there is no simple, single path between the two earliest attested readings, τὸ
γὰρ Σινᾶ and τὸ δὲ Ἁγὰρ Σινᾶ. In particular any scenario involving an initial
accidental change must posit additional stages of scribal modification to
reach the other early reading. For example, if τὸ γὰρ Σινᾶ was the earliest
reading, then one would have to suppose that a scribe changed the γάρ to
Ἁγάρ and then supplied a conjunction such as δέ or γάρ to avoid the resulting
asyndeton. 117 Alternatively, if τὸ δὲ Ἁγὰρ Σινᾶ was the original reading, then
one would have to suppose that δέ first changed to γάρ and the resulting γὰρ
Ἁγάρ was simplified, perhaps by parablepsis. 118 Either multi-stage scenario
seems equally convoluted. The lack of Ἁγάρ seems to be the harder reading

schen Geschichte: Festschrift für Jürgen Becker zum 65. Geburtstag (BZNW 100; Berlin:
Walter de Gruyter, 1999), 59–84 at 74.
114
BETZ 244 n.65; and LÉGASSE , 354 n.5.
115
The word order of the LXX is consistently ὄρος Σινᾶ (Exod 19:11, 16, 18, 20, 23;
24:46; 31:18; 34:2, 4, 32; Lev 7:38; 25:1; 26:46; 27:34; Num 3:1; 28:6; 2 Esd 19:13; cf.
Acts 7:30, 38), 115 and, in fact, Paul had just used the usual order in the preceding verse (μία
μὲν ἀπὸ ὄρους Σινᾶ: “one from mount Sinai”). See G. I. Davies, “Hagar, El-Heğra and the
Location of Mount Sinai,” VT 22 (1972): 152–163 at 159 n.4; James D. G. Dunn, The
Epistle to the Galatians (BNTC; London: Black, 1993), 250; Susan M. Elliott, “Choose
Your Mother, Choose Your Master: Galatians 4:21–5:1 in the Shadow of the Anatolian
Mother of the Gods,” JBL 118 (1999): 661–683 at 668; and LÉGASSE 355.
116
BETZ 244; Bruce, Galatians, 219; Udo Borse, Der Brief an die Galater (RNT;
Regensburg: Pustet, 1984), 168–170; BURTON 258; LONGENECKER 198; MARTYN 432,
437; MATERA 167, 169; and Albrecht Oepke, Der Brief des Paulus an die Galater
(THKNT 9; 2d ed.; Berlin: Evangelische, 1957), 109.
117
LIGHTFOOT 193 (footnote omitted): “By the negligence or confusion of a scribe τὸ
Ἄγαρ might easily be substituted for τὸ γάρ, the word Ἄγαρ occurring in the immediate
context. As a next step a connecting particle must be supplied; and δὲ or γὰρ was inserted
according to the caprice or judgment of the transcriber, thus producing the second and third
readings.”
118
METZGER 527: “After γάρ had replaced δέ in some witnesses, the juxtaposition of
γὰρ Ἁγάρ led to the accidental omission sometimes of γάρ and sometimes of Ἁγάρ.” So
also MEYER 213.
166 Chapter 4. Examining and Establishing the Text

for a later scribe, and adding it would couple the parenthetical note of Gal
4:25a more closely to the immediate context. 119
As for the intrinsic probabilities, the attested readings hardly have any-
thing to commend themselves as fitting the context of Paul’s allegorical argu-
ment. The longer reading with Hagar is redundant at best and contradictory at
worst. In v.24, Paul had already identified Hagar with the Sinai covenant, but
the longer reading of v.25a redefines Hagar with Mount Sinai itself. More-
over, the new geographic information of Mount Sinai being in Arabia is
superfluous if Paul’s intention was to re-identify Hagar in his allegory. The
shorter reading, without Hagar, on the other hand, does convey the new in-
formation about the Arabian location without disturbing the allegory by any
redefinition of Hagar’s role in it. 120
Yet the shorter reading, locating Mount Sinai in Arabia, also has its
problems. One difficulty with the Hagar-less reading is the third clause of
v.25: δουλεύει γὰρ μετὰ τῶν τέκνων αὐτῆς (“for she is enslaved with her chil-
dren”). This clause needs to pick up Hagar, not Sinai, as its subject, and so
the back-reference to Hagar would have to skip over v.25a all the way to the
end of v.24: ἥτις ἐστὶν Ἁγάρ (“which is Hagar”). 121 Moreover, the short read-
ing is also unsuitable on intrinsic grounds, because it is merely a bald geo-
graphic statement that does hardly anything to advance Paul’s argument. 122
Another difficulty is that the way Paul thinks of Arabia in Gal 1 conflicts
with its use in Gal 4. 123 In Gal 1:15–16, Paul told the Galatians in defense of
his apostolic ministry that God was pleased to reveal his son in him so he

119
Pace MATERA 170 (quoting C. K. Barrett). BETZ 245 (apparently followed by
LONGENECKER 198 n.e) calls the inclusion of Hagar the harder reading under the assump-
tion that Paul must have been referring to some Arabic term; this argument falters,
however, on the ground that the canon of lectio difficilior looks at what is more difficult to
the scribe, which Betz himself concedes was not (“The name Hagar itself can easily be in-
terpreted as a later insertion, trying to help the argument by connecting more visibly Sinai
with Jerusalem.”).
120
See Allen Kerkeslager, “Jewish Pilgrimage and Jewish Identity in Hellenistic and
Early Roman Egypt” in Pilgrimage and Holy Space in Late Antique Egypt (ed. David
Frankfurter; Leiden: Brill, 1998), 99–225 at 184.
121
This consideration was decisive for MARTYN 438, who concluded that the presence
of Hagar was necessary.
122
Especially Charles Kingsley Barrett, “The Allegory of Abraham, Sarah, and Hagar
in the Argument of Galatians” in Rechtfertigung: Festschrift für Ernst Käsemann zum 70.
Geburtstag (ed. Johannes Friedrich et al.; Tübingen: J. C. B. Mohr [Paul Siebeck], 1976),
1–16 at 12.
123
Some exegetes and lexicographers have called attention to Paul’s only other mention
of Arabia in Gal 1:17, e.g., Udo Borse, “Ἀραβία,” Exegetical Dictionary of the New Testa-
ment (Horst Balz & Gerhard Schneider, eds.; vol. 1; Grand Rapids, Mich.: Eerdmans,
1990), 149; and N. T. Wright, “Paul, Arabia, and Elijah (Galatians 1:17),” JBL 115 (1996):
683–692 at 686.
 4.2 Examination of Variation Units with Early, Split Support 167

would proclaim Christ among the Gentiles. As a result, Paul did not go
straightaway to Jerusalem to confer with the apostles before him, but he went
to Arabia (v.17). So in Paul’s mind, Arabia is coded as Gentile territory. Yet
in the allegory of Sarah and Hagar, “Arabia” does not function as Gentile
territory at all, but as the location of Mount Sinai where the law was pro-
mulgated. 124
The note about Arabia is not merely logically superfluous but structurally
superfluous as well. In particular, there are corresponding mountain elements:
Sinai (v.24c) and the quotation of Isa 54:1 about Zion (v.27). There are also
corresponding motherhood elements: one gives birth into slavery (v.24d) and
the other is our mother (v.26c). The women correspond too: Hagar (v.24e)
and the free woman (v.26b). Finally, the the present Jerusalem (v.25b) and
the Jerusalem above (v.26a) correspond. In this tight structure, however, the
note about Arabia has no corresponding element.
These considerations raise the possibility that some or all of the v.25a
parenthesis is a marginal note that was interpolated into the text of Galatians,
as some textual critics have contended. 125 In particular, Heinrich Schott

124
Another problem is that in 1:17, Arabia is distinct from the present-day Jerusalem,
while in 4:25 it is said to correspond (via the location of Mount Sinai) to the present-day
Jerusalem.
125
Including myself, “‘For Sinai is a Mountain in Arabia’: A Note on the Text of
Galatians 4,25,” ZNW 105 (2014): 80–101. Earlier scholars include Heinrich Schott, Epi-
stolae Pauli ad Thessalonicenses et Galatas (Commentarii in Epistolas Novi Testamenti 1;
Leipzig: Barth, 1834), 533; Carl Holsten, Das Evangelium des Paulus (vol. 1; Berlin:
Reimer, 1880), 171–172; and S. A. Naber, “ϒΠΕΡ ΤΑ ΕΣΚΑΜΜΕΝΑ,” Mnemosyne n.s. 6
(1878): 85–104 at 102.
The first to propose a conjectural emendation was Richard Bentley, Epistola ad Joan-
nem Millium (Toronto: University of Toronto Press, 1962), photogr. repr. of Alexander
Dyce, ed., The Works of Richard Bentley, D.D. (vol. 2; London: Macpherson, 1836), 361–
365, found the mention of Hagar in the received text too difficult for the context because it
set up the anomalous, double identification of Mount Sinai with Hagar, so he proposed to
strike the sentence fragment Σινᾶ ὄρος ἐστὶν ἐν τῇ Ἀραβίᾳ (“Sinai is a mountain in
Arabia”) from the text with some further adjustment in the syntax of the resulting sentence,
to read as follows: Τῇ δὲ Ἄγαρ συστοιχεῖ ἡ νῦν Ἱερουσαλὴμ, δουλεύει γὰρ μετὰ τῶν τέκ-
νων αὐτῆς (“But to Hagar, corresponds the present Jerusalem, for she is enslaved with her
children”).
As another example, J. C. O’Neill, “‘For this Hagar is Mount Sinai in Arabia’ (Gala-
tians 4.25),” in Steve Moyise, ed., The Old Testament in the New Testament: Essays in
Honour of J.L. North (JSNTSS 189; Sheffield: Sheffield Academic Press, 2000), 210–219,
adopted P46’s singular reading of 4:25a without Hagar to avoid the identifcation with
Mount Sinai, but he then found himself exegetically compelled to add an explicit reference
to Hagar in the subsequent clauses of v.25 by adopting a singular reading from D* for the
second clause and conjecturally emending the γάρ following δουλεύει to Ἁγάρ in the third
clause of v.25. As a result, O’Neill’s text of v.25 read as follows: τὸ δὲ Σινᾶ ὄρος ἐστὶν ἐν
τῇ Ἀραβιᾳ. συστοιχοῦσα τῇ νῦν Ἱερουσαλήμ δουλεύει Ἁγὰρ μετὰ τῶν τέκνων αὐτῆς.
168 Chapter 4. Examining and Establishing the Text

conjectured in 1834 that the entire clause τὸ γὰρ [Ἄγαρ] Σινᾶ ὄρος ἐστὶν ἐν
τῇ Ἀραβίᾳ must have been a geographic gloss that crept into the text, because
it impeded and retarded Paul’s argument. 126 In favor of Schott’s proposal, the
resulting text without the clause does indeed read smoothly without detriment
to Paul’s allegorical argument. The great classical textual critic Paul Maas,
however, warns us that interpolations are “often very difficult to prove.” 127 In
particular, an interpolation should not be suspected merely for being super-
fluous, since “there is undoubtedly superfluous (or at least not demonstrably
indispensable) matter in every original.” 128 In the case of Gal 4:25, the paren-
thetical comment about Sinai being a mountain in Arabia is undoubtedly
superfluous: it does not contribute to the allegorical argument and in fact
complicates it.
There is also external evidence that v.25a once stood outside the main text.
Allen Kerkeslager has called attention to the overlooked textual variant
following Gal 4:25a. 129 Although Alexandrian and Byzantine witnesses read
συστοιχεῖ δὲ (“but it corresponds”) after the geographic note, the Western
witnesses read the feminine nominative participle ἡ συστοιχοῦσα instead. The
closest feminine nominative antecedent for this participle is not in v.25a at
all, but at the end of v.24 – the feminine Hagar. The longer reading of v.25a
with “Hagar” would not supply the proper antecedent, because that instance
of “Hagar” is actually neuter, taking the neuter article τό. On transcriptional
grounds, it is difficult to see why a scribe would modify the συστοιχεῖ δὲ in
v.25b to the participle ἡ συστοιχοῦσα, when the entirety of the clause v.25a
stands between the particle and its antecedent. Yet, if v.25a stood not in the
main text but in the margin,130 the connection between v.24 and the Western
variant in v.25b is very smooth indeed: μία μὲν ἀπὸ ὄρους Σινᾶ, εἰς δουλείαν
γεννῶσα, ἥτις ἐστὶν Ἄγαρ [. . .] ἡ συστοιχοῦσα τῇ νῦν Ἱερουσαλὴμ, . . . (“one
[covenant is] indeed from Mount Sinai, giving birth into slavery, which is

Unfortunately for O’Neill’s rather complicated and implausible proposal, the resulting
asyndeton between v.25a and v.25b is fairly harsh, and yet his text still does not account
for the presence of Arabia in v.25a.
126
Schott, Galatas, 533: “satis commode tamen abest, impediens quodammodo et
retardans argumentationem Paulinam.” Indeed, Schott, not Bentley, deserves the credit in
the apparatus of the Nestle-Aland 27th edition for the conjecture, though Bentley deserves
recognition for identifying the passage as difficult.
127
Paul Maas, Textual Criticism (trans. Barbara Flower; Oxford: Clarendon, 1958), 14.
128
Maas, Textual Criticism, 14–15.
129
Kerkeslager, “Jewish Pilgrimage,” 184–185.
130
Since all the Western witnesses have some version of v.25a, it is better to conclude
that it stood somewhere in their common ancestor, rather than they acquired the note
independently by contamination. A marginal note is the best place for this text to stand.
 4.2 Examination of Variation Units with Early, Split Support 169

Hagar [. . .] corresponding to the present Jerusalem”). 131 The attestation of the

geographic note of v.25a within all the witnesses of Galatians means that this
marginal note mostly likely had been present in the archetype of the textual
transmission of Galatians, which was perhaps the original edition of Paul’s
letter collection.
The possibility of a marginal note very early in the textual tradition of
Paul’s epistles is not an outlandish suggestion. Michael W. Holmes proposes
that some variant readings in P46 and other early witnesses are evidence of an
early commentary of Paul’s letters in the form of marginal notes in an early
second-century edition of the Pauline corpus. 132 That Gal 4:24a could be
another such marginal comment fits well with this proposed later context.
Therefore, the textual evidence as presented and argued above indicates
that the earliest form of Gal 4:25 is a marginal note that read τὸ γὰρ Σινᾶ
ὄρος ἐστὶν ἐν τῇ Ἀραβίᾳ (“for Sinai is a mountain in Arabia”). It does not be-
long in the main text of an edition purporting to present Paul’s authorial text
of Galatians.

Gal 5:7 Ἐτρέχετε καλῶς· τίς ὑμᾶς ἐνέκοψεν τῇ ἀληθείᾳ μὴ πείθεσθαι;

Var. Western Eastern EDS.
τῇ DFG P46 ² 33 ×1175 A 1241S 1739 Ψ Chrys T
SCC SBL NA RP
1611 Byz T
S L
− B * C M M
NA S BW WH T
? Marc d b ×vg P

There is genealogically coherent external support for the article τῇ before

ἀληθείᾳ, 133 and where the omission is attested, it is against the testimony of
the witness’s closest relative: B against P46, 134 * against 33, and C against
A. The fact that the lack of the article is attested by scattered unrelated
witnesses suggests that it has been coincidentally omitted multiple times.
The article on τῇ ἀληθείᾳ signals that it should be identifiable to the read-
ers. The previous mentions of “truth” are in 2:5 and 14, where it is called the

131
Kerkeslager, “Jewish Pilgrimage,” 184–185, considers the Western reading of the
participle to be original, but its smoothness suggests transcriptionally that it was a scribal
improvement. Kerkeslager’s intrinsic argument for a μέν . . . δέ falls short because, even
with his reading, Paul must have abandoned it.
132
Michael W. Holmes, “The Text of 46: Evidence of the Earliest ‘Commentary’ on
Romans?” in Tobias Nicklas, ed., New Testament Manuscripts: Their Text and Their
World (Leiden: Brill, 2006), 189–206.
133
Pace BURTON 282, SIEFFERT 305 n.***, and ZAHN 252, who adopt the anarthrous
reading on the strength of *AB, all before the discovery of P46.
134
According to ROYSE 268, 273, P46 tends to omit articles (36 times) twice as often as
when it adds them (15 times). This preference tends to strengthen P46’s testimony for the
presence of the article here.
170 Chapter 4. Examining and Establishing the Text

“truth of the gospel.” 135 Though the prior context is considerably removed
from the present context, the Gal 2 passage had been primed to some extent
in the previous sentence with its use of ἐτρέχετε καλῶς (“you were running
well”), which recalls 2:2 μή πως εἰς κενὸν τρέχω (“lest I somehow run in
vain”). The primary emphasis of the question falls on the interrogative, of
course, but there is a secondary emphasis on the final verb μὴ πείθεσθαι (“to
not obey”). 136 With this emphasis, Paul is expressing some surprise that the
Galatians have stopped obeying the truth.
Without the article, the previous discussion of the truth of the gospel in
Gal 2:5 and 14 allows for ἀληθείᾳ to have a definite force even without the
article. 137 If this is the case, omitting the article before an identifiable noun
makes the referent of the noun more prominent, especially when it is
preposed before the verb, 138 as is the case here. 139 With this understanding of
ἀληθείᾳ, the anarthrous reading would serve to emphasize the truth which the
Galatians are blocked from obeying. It is also possible to interpret the
anarthrous ἀληθείᾳ qualitatively: “who blocked you from obeying truth?” –
as if, perhaps, the Galatians were blocked from obeying more than merely the
truth of the gospel. 140
As for the internal evidence, the articulated reading τῇ ἀληθείᾳ appears to
be the harder reading. The article suggests that it is anaphoric but its closest
prior reference is back in chapter 2, despite some priming of the back-
reference with the common use of the metaphorical τρέχειν (“to run”) in both
2:5 and 5:7. 141 Moreover, the lack of the article would be an easier reading
for scribes because an anarthrous ἀληθείᾳ would tend to generalizes Paul’s

135
BETZ 265 n.111; BURTON 282; LONGENECKER 230; MATERA 183. There is an
intervening reference to Paul’s speaking the truth in 4:16 ὥστε ἐχθρὸς ὑμῶν γέγονα
ἀληθεύων ὑμῖν; (“Have I become your enemy by telling you the truth?”), but its content is
only implicit; it could also be the truth of the gospel.
136
Levinsohn, Discourse Features, 286. BUSCEMI 519 proposes that τῇ ἀληθείᾳ has
been raised by prolepsis from the infinitive μὴ πείθεσθαι to the main clause; this analysis is
possible because the use of the dative has been attested (in contemporary papyri) with the
verb of the main clause, ἐγκόπτω. Thus, under Buscemi’s analysis, the sentence would
mean, “Who blocked your way to the truth so that you not obey it?” As the dative is more
common with πείθω than ἐγκόπτω, it seems better, as with most commentators, to construe
τῇ ἀληθείᾳ with πείθεσθαι as: “Who blocked you from obeying the truth?”
137
LONGENECKER 230.
138
See Levinsohn, Discourse Features, 163–164: “once a concept like law or faith has
been activated, further anarthrous references to the concept may well have ‘definite’ force
if the reference occurs in a potentially focal position in the clause or sentence.”
139
Levinsohn, Discourse Features, 286.
140
BURTON 282.
141
LONGENECKER 221 n.f. suggests that the article was added by scribe to conform to
the use of the article there before ἀληθεία in Gal 2:5, 14.
 4.2 Examination of Variation Units with Early, Split Support 171

accusation against the Galatians, 142 either by making ἀληθείᾳ qualitative or

by giving it additional prominence. 143 These internal reasons confirm that the
article τῇ belongs in the critical text.

Gal 6:9 τὸ δὲ καλὸν ποιοῦντες μὴ ἐγκακῶμεν, καιρῷ γὰρ ἰδίῳ θερίσομεν μὴ

ἐκλυόμενοι.
Var. Western Eastern EDS.
θερίσομεν Marc D d b ×vg B A 1739 Chrys SCC SBL NA RP
1611 Byzpt S BW WH T L
θερίσωμεν F G  33 ×1175 C 1241S Ψ Byzpt
? P46 P

This variation unit is not noted in the NA27 apparatus. Both Western and
Eastern branches are split – as well as the basic witnesses B, and  33 –
between the future θερίσομεν and the subjunctive θερίσωμεν. Some of the
later variation may merely be orthographic as Greek had started losing the
length distinction between ο and ω in the second century. 144 Though the
future θερίσομεν is slightly favored over the subjunctive θερίσωμεν on
external grounds, 145 it is necessary to consider the internal evidence.
The future θερίσομεν nicely fits the context in that it provides support
(γάρ) for the preceding exhortation: 146 “But while we are doing good let us
not get discouraged, for we will reap (θερίσομεν) at our own time if we do
not give up.” The fit for hortatory subjunctive θερίσωμεν, on the other hand,
is more superficial as it continues Paul’s exhortations: “But while we are
doing good let us not get discouraged, for let us reap at our time if we do not
give up.” Transcriptionally, scribes would be likely to harmonize the tense of
the verb to the preceding hortatory subjunctive ἐγκακῶμεν. 147 Thus, the
internal evidence also favors the future θερίσομεν.

142
Pace Zahn 242, who asserts that the anarthrous reading is better because it fits the
following generalized statements better, but this is an acknowledgement that the anarthrous
reading is the easier reading.
143
Curiously, BURTON 282 claims that “some scribe, recognizing that the reference was
to the truth of the gospel, stumbled at the qualitativeness of the expression, but it is not
clear how that should be a stumbling block.
144
Allen, Vox Graeca, 94.
145
ZIMMER³ 305.
146
E.g., LONGENECKER 281–282.
147
Though the harmonization could conceivably go the other way to the ἔχομεν
following in v.10.
172 Chapter 4. Examining and Establishing the Text

Gal 6:10 Ἄρα οὖν ὡς καιρὸν ἔχομεν, ἐργαζώμεθα τὸ ἀγαθὸν πρὸς πάντας, μάλιστα δὲ
πρὸς τοὺς οἰκείους τῆς πίστεως.
Var. Western Eastern EDS.
ἔχομεν Marc D F G d b ×vg P46 ×1175 A C 1241S 1739 Ψ Chrys SCC SBL NA RP
1611 Byz S BW T L
ἔχωμεν B  33 WH
? P

The external evidence favors the indicative ἔχομεν, despite the fact that the
subjunctive ἔχωμεν is found in otherwise highly regarded witnesses B, , and
33. 148 This variation may merely be orthographic, however, for the vowels ο
and ω have had the same pronunciation throughout almost all of the
transmission history of the text. 149 In this case, the difference in spelling
reflects a difference in mood, between the indicative ἔχομεν (“as we have an
opportunity”) and the subjunctive ἔχωμεν (“in order that we have an
opportunity”). 150 Even though both readings make good sense in the
context, 151 transcriptionally the form ἔχωμεν appears to be a harmonization to
the following subjunctive ἐργαζώμεθα (“let us keep accomplishing”). 152
Thus, the external and transcriptional criteria point in the same direction, in
favor of the indicative ἔχομεν.

148
Pace BETZ 310 n.187, who evaluates the external evidence as “almost equally
divided between the two possibilities”; LONGENECKER 268 n.f (“equally strong
attestation”).
149
Allen, Vox Graeca, 94 (starting from the second century).
150
Cf. the differences in meaning of ὡς depending on the mood of the verb in BDAG
“ὡς,” 1103–1106, esp. 8b (“while, when, as long as” with present indicative) and 9a (“with
a view to, in order to” with subjunctive). There is an exegetical tendency to minimize the
semantic differences among the variant readings, however, leading some scholars to
propose implausible interpretations. For example, BURTON 345 interprets the subjunctive
in the clause as “as therefore we have opportunity” but this sense seems to require the
particle ἄν (BDAG, “ὡς” 8c, 1106), which is missing here. SCHLIER 205, based on a
suggestion now found in BDF 238 § 455(2), (3), interprets this ὡς with the subjunctive as a
variant of ἔως; thus, the clause would mean “as long as [solange als] we have an
opportunity.” So also BETZ 310 n.187; LÉGASSE 467 n.1; LONGENECKER 282.
151
MUSSNER 407 n.17, however, argues that Paul would not have used the subjunctive
here without ἄν; so also ZIMMER³ 306.
152
So SCHLIER 345 n.**; and WEISS 45. BURTON 345, however, with his own
understanding of ὡς, argues to the contrary: “The rarity of ὡς with the subjunctive without
ἄν probably led to the change to the easier indicative.”
 4.3 A Critical Text of Galatians 173

4.3 A Critical Text of Galatians

ΠΡΟΣ ΓΑΛΑΤΟΥΣ *

1:1 Παῦλος ἀπόστολος οὐκ ἀπ’ ἀνθρώπων οὐδὲ δι’ ἀνθρώπου ἀλλὰ διὰ
Ἰησοῦ Χριστοῦ καὶ θεοῦ πατρὸς τοῦ ἐγείραντος αὐτὸν ἐκ νεκρῶν, 2 καὶ οἱ
σὺν ἐμοὶ πάντες ἀδελφοὶ ταῖς ἐκκλησίαις τῆς Γαλατίας, 3 χάρις ὑμῖν καὶ
εἰρήνη ἀπὸ θεοῦ πατρὸς καὶ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ 4 τοῦ δόντος
ἑαυτὸν περὶ τῶν ἁμαρτιῶν ἡμῶν, ὅπως ἐξέληται ἡμᾶς ἐκ τοῦ αἰῶνος τοῦ
ἐνεστῶτος πονηροῦ κατὰ τὸ θέλημα τοῦ θεοῦ καὶ πατρὸς ἡμῶν, 5 ᾧ ἡ δόξα
εἰς τοὺς αἰῶνας τῶν αἰώνων, ἀμήν.
6 Θαυμάζω ὅτι οὕτως ταχέως μετατίθεσθε ἀπὸ τοῦ καλέσαντος ὑμᾶς ἐν
χάριτι Χριστοῦ εἰς ἕτερον εὐαγγέλιον, 7 ὃ οὐκ ἔστιν ἄλλο, εἰ μή τινές εἰσιν
οἱ ταράσσοντες ὑμᾶς καὶ θέλοντες μεταστρέψαι τὸ εὐαγγέλιον τοῦ Χριστοῦ.
8 ἀλλὰ καὶ ἐὰν ἡμεῖς ἢ ἄγγελος ἐξ οὐρανοῦ ὑμῖν εὐαγγελίζηται παρ’ ὃ
εὐηγγελισάμεθα ὑμῖν, ἀνάθεμα ἔστω. 9 ὡς προειρήκαμεν καὶ ἄρτι πάλιν
λέγω· εἴ τις ὑμᾶς εὐαγγελίζεται παρ’ ὃ παρελάβετε, ἀνάθεμα ἔστω.
10 Ἄρτι γὰρ ἀνθρώπους πείθω ἢ τὸν θεόν; ἢ ζητῶ ἀνθρώποις ἀρέσκειν; εἰ
ἔτι ἀνθρώποις ἤρεσκον, Χριστοῦ δοῦλος οὐκ ἂν ἤμην.

11 Γνωρίζω δὲ ὑμῖν, ἀδελφοί, τὸ εὐαγγέλιον τὸ εὐαγγελισθὲν ὑπ’ ἐμοῦ ὅτι

οὐκ ἔστιν κατὰ ἄνθρωπον· 12 οὐδὲ γὰρ ἐγὼ παρὰ ἀνθρώπου παρέλαβον
αὐτὸ οὐδὲ ἐδιδάχθην, ἀλλὰ δι’ ἀποκαλύψεως Ἰησοῦ Χριστοῦ.
13 Ἠκούσατε γὰρ τὴν ἐμὴν ἀναστροφήν ποτε ἐν τῷ Ἰουδαϊσμῷ, ὅτι καθ’
ὑπερβολὴν ἐδίωκον τὴν ἐκκλησίαν τοῦ θεοῦ καὶ ἐπόρθουν αὐτήν, 14 καὶ
προέκοπτον ἐν τῷ Ἰουδαϊσμῷ ὑπὲρ πολλοὺς συνηλικιώτας ἐν τῷ γένει μου,
περισσοτέρως ζηλωτὴς ὑπάρχων τῶν πατρικῶν μου παραδόσεων.
15 Ὅτε δὲ εὐδόκησεν ὁ θεὸς ὁ ἀφορίσας με ἐκ κοιλίας μητρός μου καὶ
καλέσας διὰ τῆς χάριτος αὐτοῦ 16 ἀποκαλύψαι τὸν υἱὸν αὐτοῦ ἐν ἐμοί, ἵνα

* MSS = Marc D F G d b ×vg; P46 B;  33 ×1175; A C P 1241S, 1739 Chrys hark 1611 Ψ
Byz cf. NA
1:3  p Rom 1:3, 1 Cor 1:3, 2 Cor 1.2, etc. ) ἡμῶν καὶ κυρίου b;  33; A P 1241S, Ψ NA ¦
καὶ κυρίου Chrys ¦ txt D F G d ×vg; P46 B; 1739 hark 1611 Byz ×1175
1:4  p 1 Cor 15:3) ὑπέρ B; 33 ×1175; P 1241S, Chrys NA ¦ txt D F G; P46; ; A, 1739 hark
1611 Ψ Byz
1:6  p 1:15) F* G b; P46vid ¦ [Χριστοῦ] ΝΑ ¦ txt B;  33 ×1175; A P, 1739 Ψ Chrys 1611
Byz (D d; 1241S, hark Ἰησοῦ Χριστοῦ), ×vg
1:8  εὐαγγ. ὑμῖν (D ὑμᾶς) d ×vg; 33; A P (1241S ἡμῖν), Chrys hark 1611 Byz ¦ εὐαγγ.
[ὑμιν] NA ¦ εὐαγγ. F G b; ; Ψ ¦ txt B; ×1175; 1739
1:11  γάρ D* F G d ×vg; B; ¹ 33 ×1175 NA ¦ − P, Chrys ¦ txt b; P46; *,²; A 1241S, 1739
Ψ hark 1611 Byz
1:12  οὔτε P46 B; hark 1611 Byz NA ¦ txt D* F G;  33 ×1175; A P 1241S, 1739 Ψ Chrys
174 Chapter 4. Examining and Establishing the Text

εὐαγγελίζωμαι αὐτὸν ἐν τοῖς ἔθνεσιν, εὐθέως οὐ προσανεθέμην σαρκὶ καὶ

αἵματι 17 οὐδὲ ἀπῆλθον εἰς Ἱεροσόλυμα πρὸς τοὺς πρὸ ἐμοῦ ἀποστόλους,
ἀλλὰ ἀπῆλθον εἰς Ἀραβίαν καὶ πάλιν ὑπέστρεψα εἰς Δαμασκόν.
18 Ἔπειτα μετὰ ἔτη τρία ἀνῆλθον εἰς Ἱεροσόλυμα ἱστορῆσαι Κηφᾶν καὶ
ἐπέμεινα πρὸς αὐτὸν ἡμέρας δεκαπέντε, 19 ἕτερον δὲ τῶν ἀποστόλων οὐκ
εἶδον εἰ μὴ Ἰάκωβον τὸν ἀδελφὸν τοῦ κυρίου. 20 ἃ δὲ γράφω ὑμῖν, ἰδοὺ
ἐνώπιον τοῦ θεοῦ ὅτι οὐ ψεύδομαι.
21 Ἔπειτα ἦλθον εἰς τὰ κλίματα τῆς Συρίας καὶ τῆς Κιλικίας· 22 ἤμην δὲ
ἀγνοούμενος τῷ προσώπῳ ταῖς ἐκκλησίαις τῆς Ἰουδαίας ταῖς ἐν Χριστῷ.
23 μόνον δὲ ἀκούοντες ἦσαν ὅτι ὁ διώκων ἡμᾶς ποτε νῦν εὐαγγελίζεται τὴν
πίστιν ἥν ποτε ἐπόρθει, 24 καὶ ἐδόξαζον ἐν ἐμοὶ τὸν θεόν.
2:1 Ἔπειτα διὰ δεκατεσσάρων ἐτῶν πάλιν ἀνέβην εἰς Ἱεροσόλυμα μετὰ
Βαρναβᾶ συμπαραλαβὼν καὶ Τίτον· 2 ἀνέβην δὲ κατὰ ἀποκάλυψιν· καὶ
ἀνεθέμην αὐτοῖς τὸ εὐαγγέλιον ὃ κηρύσσω ἐν τοῖς ἔθνεσιν, κατ’ ἰδίαν δὲ τοῖς
δοκοῦσιν, μή πως εἰς κενὸν τρέχω ἢ ἔδραμον. 3 ἀλλ’ οὐδὲ Τίτος ὁ σὺν ἐμοί,
Ἕλλην ὤν, ἠναγκάσθη περιτμηθῆναι· 4 διὰ δὲ τοὺς παρεισάκτους ψευδαδέλ-
φους, οἵτινες παρεισῆλθον κατασκοπῆσαι τὴν ἐλευθερίαν ἡμῶν ἣν ἔχομεν ἐν
Χριστῷ Ἰησοῦ, ἵνα ἡμᾶς καταδουλώσουσιν, 5 οἷς οὐδὲ πρὸς ὥραν εἴξαμεν τῇ
ὑποταγῇ, ἵνα ἡ ἀλήθεια τοῦ εὐαγγελίου διαμείνῃ πρὸς ὑμᾶς.
6 Ἀπὸ δὲ τῶν δοκούντων εἶναί τι, – ὁποῖοί ποτε ἦσαν οὐδέν μοι διαφέρει·
πρόσωπον ὁ θεὸς ἀνθρώπου οὐ λαμβάνει – ἐμοὶ γὰρ οἱ δοκοῦντες οὐδὲν
προσανέθεντο, 7 ἀλλὰ τοὐναντίον ἰδόντες ὅτι πεπίστευμαι τὸ εὐαγγέλιον τῆς
ἀκροβυστίας καθὼς Πέτρος τῆς περὶτομῆς, 8 ὁ γὰρ ἐνεργήσας Πέτρῳ εἰς
ἀποστολὴν τῆς περιτομῆς ἐνήργησεν καὶ ἐμοὶ εἰς τὰ ἔθνη, 9 καὶ γνόντες τὴν
χάριν τὴν δοθεῖσάν μοι, Ἰάκωβος καὶ Κηφᾶς καὶ Ἰωάννης, οἱ δοκοῦντες
στῦλοι εἶναι, δεξιὰς ἔδωκαν ἐμοὶ καὶ Βαρναβᾷ κοινωνίας, ἵνα ἡμεῖς εἰς τὰ
ἔθνη, αὐτοὶ δὲ εἰς τὴν περιτομήν· 10 μόνον τῶν πτωχῶν ἵνα μνημονεύωμεν,
ὃ καὶ ἐσπούδασα αὐτὸ τοῦτο ποιῆσαι.

1:15  F G b ×vg; P46 B; 1611 ¦ [ὁ θεός] NA ¦ txt D d;  33 ×1175; A P 1241S, 1739 Ψ

Chrys hark Byz
1:17  p 1:18, 2:1) ἀνῆλθον  33 ×1175; A C P 1241S, 1739 Ψ Chrys hark 1611 Byz NA ¦
ἦλθον P46; (d b ×vg veni) ¦ txt D F G; B
1:18  p 2:1)  33; A P 1241S, 1739 ¦ txt D F G d b ×vg; P46 B; ×1175; Ψ Chrys hark 1611
Byz
1:19  εἶδον οὐδένα D* F G (b) d ×vg ¦ οὐχ’ εἶδον P46 B ¦ txt  33 ×1175; A P 1241S, 1739
Ψ Chrys hark 1611 Byz
2:6  (D F G θεὸς ἀνθρ. πρ.); B; C, 1739 1611 Byz ¦ txt P46;  33 ×1175; A P 1241S (Ψ
Chrys πρ. ἀνθρ. ὁ θεός) ¦ [ὁ] NA
2:9  Πέτρος καὶ Ἰάκωβος Marc D F G d b ¦ Ἰάκ. καὶ Πέτρος P46 ¦ txt B;  33; ×vg, C
1241S, 1739 Ψ Chrys hark 1611 Byz
2:10  3 1 2 4 D F G d b ×vg; hark ¦ 3 4 1 2 Marc ¦ txt B;  33; A C 1241S, 1739 Chrys
1611 Byz
 4.3 A Critical Text of Galatians 175

11 Ὅτε δὲ ἦλθεν Κηφᾶς εἰς Ἀντιόχειαν, κατὰ πρόσωπον αὐτῷ ἀντέστην,

ὅτι κατεγνωσμένος ἦν. 12 πρὸ τοῦ γὰρ ἐλθεῖν τινας ἀπὸ Ἰακώβου μετὰ τῶν
ἐθνῶν συνήσθιεν· ὅτε δὲ ἦλθεν, ὑπέστελλεν καὶ ἀφώριζεν ἑαυτὸν φοβούμε-
νος τοὺς ἐκ περιτομῆς. 13 καὶ συνυπεκρίθησαν αὐτῷ καὶ οἱ λοιποὶ Ἰουδαῖοι,
ὥστε καὶ Βαρναβᾶς συναπήχθη αὐτῶν τῇ ὑποκρίσει. 14 ἀλλ’ ὅτε εἶδον ὅτι
οὐκ ὀρθοποδοῦσιν πρὸς τὴν ἀλήθειαν τοῦ εὐαγγελίου, εἶπον τῷ Κηφᾷ
ἔμπροσθεν πάντων·
εἰ σὺ Ἰουδαῖος ὑπάρχων ἐθνικῶς καὶ οὐχὶ Ἰουδαϊκῶς ζῇς, πῶς τὰ ἔθνη
ἀναγκάζεις ἰουδαΐζειν;
15 Ἡμεῖς φύσει Ἰουδαῖοι καὶ οὐκ ἐξ ἐθνῶν ἁμαρτωλοί· 16 εἰδότες δὲ ὅτι
οὐ δικαιοῦται ἄνθρωπος ἐξ ἔργων νόμου ἐὰν μὴ διὰ πίστεως Ἰησοῦ Χριστοῦ,
καὶ ἡμεῖς εἰς Χριστὸν Ἰησοῦν ἐπιστεύσαμεν, ἵνα δικαιωθῶμεν ἐκ πίστεως
Χριστοῦ καὶ οὐκ ἐξ ἔργων νόμου, ὅτι ἐξ ἔργων νόμου οὐ δικαιωθήσεται πᾶσα
σάρξ. 17 εἰ δὲ ζητοῦντες δικαιωθῆναι ἐν Χριστῷ εὑρέθημεν καὶ αὐτοὶ
ἁμαρτωλοί, ἆρα Χριστὸς ἁμαρτίας διάκονος; μὴ γένοιτο. 18 εἰ γὰρ ἃ κατέ-
λυσα ταῦτα πάλιν οἰκοδομῶ, παραβάτην ἐμαυτὸν συνιστάνω. 19 ἐγὼ γὰρ διὰ
νόμου νόμῳ ἀπέθανον, ἵνα θεῷ ζήσω. Χριστῷ συνεσταύρωμαι· 20 ζῶ δὲ
οὐκέτι ἐγώ, ζῇ δὲ ἐν ἐμοὶ Χριστός· ὃ δὲ νῦν ζῶ ἐν σαρκί, ἐν πίστει ζῶ τῇ τοῦ
θεοῦ καὶ Χριστοῦ τοῦ ἀγαπήσαντός με καὶ παραδόντος ἑαυτὸν ὑπὲρ ἐμοῦ.
21 Οὐκ ἀθετῶ τὴν χάριν τοῦ θεοῦ· εἰ γὰρ διὰ νόμου δικαιοσύνη, ἄρα
Χριστὸς δωρεὰν ἀπέθανεν.

3:1 Ὦ ἀνόητοι Γαλάται, τίς ὑμᾶς ἐβάσκανεν, οἷς κατ’ ὀφθαλμοὺς Ἰησοῦς
Χριστὸς προεγράφη ἐσταυρωμένος; 2 τοῦτο μόνον θέλω μαθεῖν ἀφ’ ὑμῶν· ἐξ
ἔργων νόμου τὸ πνεῦμα ἐλάβετε ἢ ἐξ ἀκοῆς πίστεως; 3 οὕτως ἀνόητοί ἐστε,
ἐναρξάμενοι πνεύματι νῦν σαρκὶ ἐπιτελεῖσθε; 4 τοσαῦτα ἐπάθετε εἰκῇ; εἴ γε
καὶ εἰκῇ. 5 ὁ οὖν ἐπιχορηγῶν ὑμῖν τὸ πνεῦμα καὶ ἐνεργῶν δυνάμεις ἐν ὑμῖν,
ἐξ ἔργῶν νόμου ἢ ἐξ ἀκουῆς πίστεως;
6 Καθὼς Ἀβραὰμ ἐπίστευσεν τῷ θεῷ, καὶ ἐλογίσθη αὐτῷ εἰς δικαιοσύνη·
7 γινώσκετε ἄρα ὅτι οἱ ἐκ πίστεως, οὗτοι υἱοί εἰσιν Ἀβραάμ. 8 προϊδοῦσα

2:12  p ἦλθεν) τινα P46 d

2:12  p τινας) ἦλθον b ×vg; ×1175; A C 1241S, 1739 Ψ Chrys hark 1611 Byz NA ¦ txt D*
F G d; P46 B;  33
2:13  ×vg; P46 B; 1739 ¦ [καί] NA
2:16  P46; 33 ×1175; A P, 1739 Ψ hark 1611 Byz ¦ [δέ] NA ¦ txt D* F G d b ×vg; B; ; C
1241S, Chrys
2:16  P46 B; 33 ×1175, 1241S, 1739 hark 1611 ¦ Χριστόν Chrys ¦ txt D F G d b ×vg; ; A
C P, Ψ Byz
2:20  τοῦ υἱοῦ τοῦ θεοῦ ×vg;  33 ×1175; A C 1241S; 1739 Ψ Chrys hark 1611 Byz NA ¦
τοῦ θεοῦ 330 ¦ τοῦ Χριστοῦ καὶ θεοῦ b ¦ txt D* F G d; P46 B
3:3  ex itac.) ἐπιτελεῖσθαι D* F G;  33 ×1175; 1241S ¦ txt d b ×vg; P46 B; A C, 1739 Ψ
Chrys hark 1611 Byz
3:7  D F G d b ×vg; 33 ×1175; A C, 1739 hark 1611 Byz ¦ txt P46 B; ; P 1241S, Ψ Chrys
176 Chapter 4. Examining and Establishing the Text

δὲ ἡ γραφὴ ὅτι ἐκ πίστεως δικαιοῖ τὰ ἔθνη ὁ θεός, προευηγγελίσατο τῷ

Ἀβραάμ ὅτι ἐνευλογηθήσονται ἐν σοὶ πάντα τὰ ἔθνη· 9 ὥστε οἱ ἐκ πίστεως
εὐλογοῦνται σὺν τῷ πιστῷ Ἀβραάμ.
10 Ὅσοι γὰρ ἐξ ἔργων νόμου εἰσιν, ὑπὸ κατάραν εἰσίν· γέγραπται γὰρ ὅτι
ἐπικατάρατος πᾶς ὃς οὐκ ἐμμένει πᾶσιν τοῖς γεγραμμένοις ἐν τῷ βιβλίῳ τοῦ
νόμου τοῦ ποιῆσαι αὐτά. 11 ὅτι δὲ ἐν νόμῳ οὐδεὶς δικαιοῦται παρὰ τῷ θεῷ
δῆλον, ὅτι ὁ δίκαιος ἐκ πίστεως ζήσεται· 12 ὁ δὲ νόμος οὐκ ἔστιν ἐκ πίστεως,
ἀλλ’ ὁ ποιήσας αὐτὰ ζήσεται ἐν αὐτοῖς. 13 Χριστὸς ἡμᾶς ἐξηγόρασεν ἐκ τῆς
κατάρας τοῦ νόμου γενόμενος ὑπὲρ ἡμῶν κατάρα, ὅτι γέγραπται·
ἐπικατάρατος πᾶς ὁ κρεμάμενος ἐπὶ ξύλου, 14 ἵνα εἰς τὰ ἔθνη ἡ εὐλογία τοῦ
Ἀβραὰμ γένηται ἐν Χριστῷ Ἰησοῦ, ἵνα τὴν ἐπαγγελίαν τοῦ πνεύματος
λάβωμεν διὰ τῆς πίστεως.
15 Ἀδελφοί, κατὰ ἄνθρωπον λέγω· ὅμως ἀνθρώπου κεκυρωμένην διαθή-
κην οὐδεὶς ἀθετεῖ ἢ ἐπιδιατάσσεται. 16 τῷ δὲ Ἀβραὰμ ἐρρέθησαν αἱ
ἐπαγγελίαι καὶ τῷ σπέρματι αὐτοῦ, οὐ λέγει· καὶ τοῖς σπέρμασιν, ὡς ἐπὶ
πολλῶν ἀλλ’ ὡς ἐφ’ ἑνός· καὶ τῷ σπέρματί σου, ὅς ἐστιν Χριστός. 17 τοῦτο δὲ
λέγω· διαθήκην προκεκυρωμένην ὑπὸ τοῦ θεοῦ ὁ μετὰ τετρακόσια καὶ
τριάκοντα ἔτη γεγονὼς νόμος οὐκ ἀκυροῖ εἰς τὸ καταργῆσαι τὴν ἐπαγγελίαν.
18 εἰ γὰρ ἐκ νόμου ἡ κληρονομία, οὐκέτι ἐξ ἐπαγγελίας· τῷ δὲ Ἀβραὰμ δι’
ἐπαγγελίας κεχάρισται ὁ θεός.
19 Τί οὖν ὁ νόμος; τῶν παραβάσεων χάριν προσετέθη, ἄχρις οὗ ἔλθῃ τὸ
σπέρμα ᾧ ἐπήγγελται, διαταγεὶς δι’ ἀγγέλων ἐν χειρὶ μεσίτου. 20 ὁ δὲ μεσίτης
ἑνὸς οὐκ ἔστιν, ὁ δὲ θεὸς εἷς ἐστιν. 21 ὁ οὖν νόμος κατὰ τῶν ἐπαγγελιῶν τοῦ
θεοῦ; μὴ γένοιτο. εἰ γὰρ ἐδόθη νόμος ὁ δυνάμενος ζῳοποιῆσαι, ὄντως ἐκ
νόμου ἂν ἦν ἡ δικαιοσύνη· 22 ἀλλὰ συνέκλεισεν ἡ γραφὴ τὰ πάντα ὑπὸ
ἁμαρτίαν, ἵνα ἡ ἐπαγγελία ἐκ πίστεως Ἰησοῦ Χριστοῦ δοθῇ τοῖς πιστεύουσιν.
23 Πρὸ τοῦ δὲ ἐλθεῖν τὴν πίστιν ὑπὸ νόμον ἐφρουρούμεθα συγκλειόμενοι
εἰς τὴν μέλλουσαν πίστιν ἀποκαλυφθῆναι, 24 ὥστε ὁ νόμος παιδαγωγὸς
ἡμῶν γέγονεν εἰς Χριστόν, ἵνα ἐκ πίστεως δικαιωθῶμεν· 25 ἐλθούσης δὲ τῆς
πίστεως οὐκέτι ὑπὸ παιδαγωγόν ἐσμεν.
26 Πάντες γὰρ υἱοὶ θεοῦ ἐστε διὰ τῆς πίστεως ἐν Χριστῷ Ἰησοῦ· 27 ὅσοι
γὰρ εἰς Χριστὸν ἐβαπτίσθητε, Χριστὸν ἐνεδύσασθε. 28 οὐκ ἔνι Ἰουδαῖος οὐδὲ
Ἕλλην, οὐκ ἔνι δοῦλος οὐδὲ ἐλεύθερος, οὐκ ἔνι ἄρσεν καὶ θῆλυ· πάντες γὰρ
ὑμεῖς εἷς ἐστε ἐν Χριστῷ Ἰησοῦ. 29 εἰ δὲ ὑμεῖς Χριστοῦ, ἄρα τοῦ Ἀβραὰμ
σπέρμα ἐστέ, κατ’ ἐπαγγελίαν κληρονόμοι.
4:1 Λέγω δέ, ἐφ’ ὅσον χρόνον ὁ κληρονόμος νήπιός ἐστιν, οὐδὲν διαφέρει
δούλου κύριος πάντων ὤν, 2 ἀλλὰ ὑπὸ ἐπιτρόπους ἐστὶν καὶ οἰκονόμους ἄχρι

3:14  p) εὐλογίαν Marc D* F G d b; P46

3:21  d; P46 B ¦ [τοῦ θεοῦ] NA
3:21  ἐκ νόμου ἦν ἂν (ἐν νόμῳ P46);  33 ×1175; 1739 Ψc ¦ ἂν ἐκ νόμου ἦν Chrys 1611
Byz ¦ ἐκ νόμου ἦν D* ¦ ἐκ νόμου F G d ¦ txt (ἐν νόμῳ B); A C 1241S
3:24  ἐγένετο P46 B
 4.3 A Critical Text of Galatians 177

τῆς προθεσμίας τοῦ πατρός. 3 οὕτως καὶ ἡμεῖς, ὅτε ἦμεν νήπιοι, ὑπὸ τὰ
στοιχεῖα τοῦ κόσμου ἤμεθα δεδουλωμένοι· 4 ὅτε δὲ ἦλθεν τὸ πλήρωμα τοῦ
χρόνου, ἐξαπέστειλεν ὁ θεὸς τὸν υἱὸν αὐτοῦ, γενόμενον ἐκ γυναικός, γενό-
μενον ὑπὸ νόμον, 5 ἵνα τοὺς ὑπὸ νόμον ἐξαγοράσῃ, ἵνα τὴν υἱοθεσίαν
ἀπολάβωμεν.
6 Ὅτι δὲ ἐστε υἱοί , ἐξαπέστειλεν ὁ θεὸς τὸ πνεῦμα τοῦ υἱοῦ αὐτοῦ εἰς
τὰς καρδίας ἡμῶν κράζον· αββα ὁ πατήρ. 7 ὥστε οὐκέτι εἶ δοῦλος ἀλλὰ υἱός·
εἰ δὲ υἱός, καὶ κληρονόμος διὰ θεοῦ.
8 Ἀλλὰ τότε μὲν οὐκ εἰδότες θεὸν ἐδουλεύσατε τοῖς φύσει μὴ οὖσιν
θεοῖς· 9 νῦν δὲ γνόντες θεόν, μᾶλλον δὲ γνωσθέντες ὑπὸ θεοῦ, πῶς ἐπιστρέ-
φετε πάλιν ἐπὶ τὰ ἀσθενῆ καὶ πτωχὰ στοιχεῖα οἷς πάλιν ἄνωθεν δουλεύειν
θέλετε; 10 ἡμέρας παρατηρεῖσθε καὶ μῆνας καὶ καιροὺς καὶ ἐνιαυτούς,
11 φοβοῦμαι ὑμᾶς μή πως εἰκῇ κεκοπίακα εἰς ὑμᾶς.
12 Γίνεσθε ὡς εγώ, ὅτι κἀγὼ ὡς ὑμεῖς, ἀδελφοί, δέομαι ὑμῶν. οὐδέν με
ἠδικήσατε· 13 οἴδατε δὲ ὅτι δι’ ἀσθένειαν τῆς σαρκὸς εὐηγγελισάμην ὑμῖν τὸ
πρότερον, 14 καὶ τὸν πειρασμὸν ὑμῶν ἐν τῇ σαρκί μου οὐκ ἐξουθενήσατε
οὐδὲ ἐξεπτύσατε, ἀλλὰ ὡς ἄγγελον θεοῦ ἐδέξασθέ με, ὡς Χριστὸν Ἰησοῦν.
15 ποῦ οὖν ὁ μακαρισμὸς ὑμῶν; μαρτυρῶ γὰρ ὑμῖν ὅτι εἰ δυνατὸν τοὺς
ὀφθαλμοὺς ὑμῶν ἐξορύξαντες ἐδώκατέ μοι. 16 ὥστε ἐχθρὸς ὑμῶν γέγονα
ἀληθεύων ὑμῖν; 17 ζηλοῦσιν ὑμᾶς οὐ καλῶς, ἀλλὰ ἐκκλεῖσαι ὑμᾶς θέλουσιν,
ἵνα αὐτοὺς ζηλοῦτε· 18 καλὸν δὲ ζηλοῦσθαι ἐν καλῷ πάντοτε καὶ μὴ μόνον
ἐν τῷ παρεῖναί με πρὸς ὑμᾶς. 19 τέκνα μου, οὓς πάλιν ὠδίνω ἄχρις οὗ
μορφωθῇ Χριστὸς ἐν ὑμῖν· 20 ἤθελον δὲ παρεῖναι πρὸς ὑμᾶς ἄρτι καὶ
ἀλλάξαι τὴν φωνήν μου, ὅτι ἀποροῦμαι ἐν ὑμῖν.
21 Λέγετέ μοι, οἱ ὑπὸ νόμον θέλοντες εἶναι, τὸν νόμον οὐκ ἀκούετε;
22 γέγραπται γὰρ ὅτι Ἀβραὰμ δύο υἱοὺς ἔσχεν, ἕνα ἐκ τῆς παιδίσκης καὶ ἕνα
ἐκ τῆς ἐλευθέρας. 23 ἀλλ’ ὁ μὲν ἐκ τῆς παιδίσκης κατὰ σάρκα γεγέννηται, ὁ
δὲ ἐκ τῆς ἐλευθέρας δι’ ἐπαγγελίας. 24 ἅτινά ἐστιν ἀλληγορούμενα· αὗται
γάρ εἰσιν δύο διαθήκαι, μία μὲν ἀπὸ ὄρους Σινᾶ εἰς δουλείαν γεννῶσα, ἥτις
ἐστὶν Ἁγάρ. 25  συστοιχεῖ δὲ τῇ νῦν Ἰερουσαλήμ, δουλεύει γὰρ μετὰ τῶν
τέκνων αὐτῆς. 26 ἡ δὲ ἄνω Ἰερουσαλὴμ ἐλευθέρα ἐστίν, ἥτις ἐστὶν μήτηρ

4:3  p ἦμεν) ἦμεν B; A C 1241S, 1739 Ψ Chrys 1611 Byz ¦ txt D* F G; P46;  33 ×1175
4:6  θεοῦ Marc D F G d b
4:8  2–6 1 (Marc 2 τῇ 3 5 6 δουλεύτε) D (F G 4 3) d b ×vg
4:10  3 4 1 2 D F G d b ¦ txt Marc ×vg; P46 B;  33 ×1175; A C 1241S, 1739 Ψ Chrys
hark 1611 Byz
4:18  ex itac.) ζηλοῦσθε (b ×vg aemulamini); B;  33
4:19  τεκνία C 33 ×1175; A C 1241S, Ψ Chrys 1611 Byz ¦ txt D* F G d; B; *, 1739
4:19  μέχρις B; *; 1739 NA
4:23  P46 B; Marc ×vg
4:25  τὸ γὰρ Σινᾶ ὄρος ἐστὶν ἐν τῇ Ἀραβίᾳ D-Vorlage F G b ×vg; ; C 1241S, (1739
−ἐστίν) ¦ τὸ δὲ Σινᾶ etc. P46 ¦ τὸ δὲ Ἁγὰρ Σινᾶ etc. D; B; ×1175; A NA ¦ τὸ γὰρ Ἁγὰρ
Σινᾶ etc. (d −Σινᾶ); 33; P, Ψ Chrys (hark ὄρος Σινᾶ) Byz | txt Schott cj ex Bentley
178 Chapter 4. Examining and Establishing the Text

ἡμῶν· 27 γέγραπται γάρ· εὐφράνθητι, στεῖρα ἡ οὐ τίκτουσα, ῥῆξον καὶ βόησον,

ἡ οὐκ ὠδίνουσα· ὅτι πολλὰ τὰ τέκνα τῆς ἐρήμου μᾶλλον ἢ τῆς ἐχούσης τὸν
ἄνδρα.
28 Ὑμεῖς δέ, ἀδελφοί, κατὰ Ἰσαὰκ ἐπαγγελίας τέκνα ἐστέ. 29 ἀλλ’
ὥσπερ τότε ὁ κατὰ σάρκα γεννηθεὶς ἐδίωκεν τὸν κατὰ πνεῦμα, οὕτως καὶ
νῦν. 30 ἀλλὰ τί λέγει ἡ γραφή; ἔκβαλε τὴν παιδίσκην καὶ τὸν υἱὸν αὐτῆς· οὐ
γὰρ μὴ κληρονομήσῃ ὁ υἱὸς τῆς παιδίσκης μετὰ τοῦ υἱοῦ τῆς ἐλευθέρας.
31 διό, ἀδελφοί, οὐκ ἐσμὲν παιδίσκης τέκνα ἀλλὰ τῆς ἐλευθέρας. 5:1 τῇ
ἐλευθερίᾳ ἡμᾶς Χριστὸς ἠλευθέρωσεν· στήκετε οὖν καὶ μὴ πάλιν ζυγῷ
δουλείας ἐνέχεσθε.

2 Ἴδε ἐγὼ Παῦλος λέγω ὑμῖν ὅτι ἐὰν περιτέμνησθε, Χριστὸς ὑμᾶς οὐδὲν
ὠφελήσει. 3 μαρτύρομαι δὲ πάλιν παντὶ ἀνθρώπῳ περιτεμνομένῳ ὅτι ὀφειλέ-
της ἐστὶν ὅλον τὸν νόμον ποιῆσαι. 4 κατηργήθητε ἀπὸ Χριστοῦ, οἵτινες ἐν
νόμῳ δικαιοῦσθε, τῆς χάριτος ἐξεπέσατε. 5 ἡμεῖς γὰρ πνεύματι ἐκ πίστεως
ἐλπίδα δικαιοσύνης ἀπεκδεχόμεθα. 6 ἐν γὰρ Χριστῷ Ἰησοῦ οὔτε περιτομή τι
ἰσχύει οὔτε ἀκροβυστία ἀλλὰ πίστις δι’ ἀγάπης ἐνεργουμένη.
7 Ἐτρέχετε καλῶς· τίς ὑμᾶς ἐνέκοψεν τῇ ἀληθείᾳ μὴ πείθεσθαι; 8 ἡ
πεισμονὴ οὐκ ἐκ τοῦ καλοῦντος ὑμᾶς. 9 μικρὰ ζύμη ὅλον τὸ φύραμα ζυμοῖ.
10 ἐγὼ πέποιθα εἰς ὑμᾶς ἐν κυρίῳ ὅτι οὐδὲν ἄλλο φρονήσετε· ὁ δὲ ταράσσων
ὑμᾶς βαστάσει τὸ κρίμα, ὅστις ἐὰν ᾖ.
11 Ἐγὼ δέ, ἀδελφοί, εἰ περιτομὴν ἔτι κηρύσσω, τί ἔτι διώκομαι; ἄρα κατή-
ργηται τὸ σκάνδαλον τοῦ σταυροῦ.
12 Ὄφελον καὶ ἀποκόψονται οἱ ἀναστατοῦντες ὑμᾶς.

13 Ὑμεῖς γὰρ ἐπ’ ἐλευθερίᾳ ἐκλήθητε, ἀδελφοί· μόνον μὴ τὴν ἐλευθερίαν εἰς
ἀφορμὴν τῇ σαρκί, ἀλλὰ διὰ τῆς ἀγάπης δουλεύετε ἀλλήλοις. 14 ὁ γὰρ πᾶς
νόμος  ἐν ἑνὶ λόγῳ πεπλήρωται, ἐν τῷ· ἀγαπήσεις τὸν πλησίον σου ὡς
σεαυτόν. 15 εἰ δὲ ἀλλήλους δάκνετε καὶ κατεσθίετε, βλέπετε μὴ ὑπ’ ἀλλήλων
ἀναλωθῆτε.

4:28  …  p v.31) ἡμεῖς . . . ἐσμέν ×vg; ; A C P 1241S; Ψ Chrys hark 1611 Byz ¦ txt D* F
G d b; P46 B; 33 ×1175; 1739
4:30  p Gen 21:20) κληρονομήσει D; (P46 −μή) B;  33 ×1175; 1241S NA ¦ txt (F G −μή);
A C, 1739 Ψ Chrys 1611 Byz
5:1  ᾗ Marc F G (d b ×vg qua) ¦ ἐπ’ Hort cj
5:1  Marc; ×1175; C P 1241S, 1739 Chrys hark 1611 Byz ¦ txt D* F G d b ×vg; B;  33; A
5:1  D F G d b ¦ txt Marc ×vg; B;  33 ×1175; A C P 1241S, 1739 Ψ Chrys hark 1611 Byz
5:7  Β; *; C ¦ [τῇ] NA
5:14  ἐν ὑμῖν Marc D* F G d b
5:14  πληροῦται D F G d b ×vg; Ψ Chrys 1611 Byz ¦ txt Marc; P46 B;  33 ×1175; A C
1241S, 1739
5:14  Marc D* F G d b ×vg
 4.3 A Critical Text of Galatians 179

16 Λέγω δέ, πνεύματι περιπατεῖτε καὶ ἐπιθυμίαν σαρκὸς οὐ μὴ τελέσητε.

17 ἡ γὰρ σὰρξ ἐπιθυμεῖ κατὰ τοῦ πνεύματος, τὸ δὲ πνεῦμα κατὰ τῆς σαρκός,
ταῦτα γὰρ ἀλλήλοις ἀντίκειται, ἵνα μὴ ἃ ἐὰν θέλητε ταῦτα ποιῆτε. 18 εἰ δὲ
πνεύματι ἄγεσθε, οὐκ ἐστὲ ὑπὸ νόμον. 19 φανερὰ δέ ἐστιν τὰ ἔργα τῆς
σαρκός, ἅτινά ἐστιν  πορνεία, ἀκαθαρσία, ἀσέλγεια, 20 εἰδωλολατρία, φαρ-
μακεία, ἔχθραι, ἔρις, ζῆλος, θυμοί, ἐριθεῖαι, διχοστασίαι, αἱρέσεις, 21 φθό-
νοι,  μέθαι, κῶμοι καὶ τὰ ὅμοια τούτοις, ἃ προλέγω ὑμῖν, καθὼς προεῖπον
ὅτι οἱ τὰ τοιαῦτα πράσσοντες βασιλείαν θεοῦ οὐ κληρονομήσουσιν. 22 ὁ δὲ
καρπὸς τοῦ πνεύματός ἐστιν ἀγάπη χαρὰ εἰρήνη, μακροθυμία χρηστότης
ἀγαθωσύνη, πίστις 23 πραΰτης ἐγκράτεια· κατὰ τῶν τοιούτων οὐκ ἔστιν
νόμος. 24 οἱ δὲ τοῦ Χριστοῦ  τὴν σάρκα ἐσταύρωσαν σὺν τοῖς παθήμασιν
καὶ ταῖς ἐπιθυμίαις.
25 Εἰ ζῶμεν πνεύματι, πνεύματι καὶ στοιχῶμεν. 26 μὴ γινώμεθα κενό-
δοξοι, ἀλλήλους προκαλούμενοι, ἀλλήλοις φθονοῦντες.
6:1 Ἀδελφοί, ἐὰν καὶ προλημφθῇ ἄνθρωπος ἔν τινι παραπτώματι, ὑμεῖς οἱ
πνευματικοὶ καταρτίζετε τὸν τοιοῦτον ἐν πνεύματι πραΰτητος, σκοπῶν σεαυ-
τὸν μὴ καὶ σὺ πειρασθῇς.
2 Ἀλλήλων τὰ βάρη βαστάζετε καὶ οὕτως ἀναπληρώσετε τὸν νόμον τοῦ
Χριστοῦ. 3 εἰ γὰρ δοκεῖ τις εἶναί τι μηδὲν ὤν, φρεναπατᾷ ἑαυτόν. 4 τὸ δὲ
ἔργον ἑαυτοῦ δοκιμαζέτω ἕκαστος, καὶ τότε εἰς ἑαυτὸν μόνον τὸ καύχημα
ἕξει καὶ οὐκ εἰς τὸν ἕτερον· 5 ἕκαστος γὰρ τὸ ἴδιον φορτίον βαστάσει.
6 Κοινωνείτω δὲ ὁ κατηχούμενος τὸν λόγον τῷ κατηχοῦντι ἐν πᾶσιν
ἀγαθοῖς.
7 Μὴ πλανᾶσθε, θεὸς οὐ μυκτηρίζεται. ὃ γὰρ ἐὰν σπείρῃ ἄνθρωπος, τοῦτο
καὶ θερίσει· 8 ὅτι ὁ σπείρων εἰς τὴν σάρκα ἑαυτοῦ ἐκ τῆς σαρκὸς θερίσει
φθοράν, ὁ δὲ σπείρων εἰς τὸ πνεῦμα ἐκ τοῦ πνεύματος θερίσει ζωὴν αἰώνιον.

5:17  p) δέ ×1175; A C P 1241S, 1739 Ψ Chrys hark 1611 Byz ¦ txt D* F G d b ×vg; P46
B;  33
5:17  ἃ ἄν CC, Ψ 1611 Byz ¦ ὃ ἄν D* F G ¦ ἄν B* ¦ ἅ C* 1739 Chrys ¦ txt BC;  ×1175; A
1241S
5:19  μοιχεία D (μοιχεἶαι F G d b); Ψ ( Chrys) hark 1611 Byz ¦ txt Marc ×vg; B;  33
×1175; A C P 1241S, 1739
5:20  p) ζῆλοι Marc (F G ζήλους) b ×vg;  ×1175; C 1241S, Ψ Chrys hark 1611 Byz ¦ txt
D* d; B; 33; P, 1739
5:21  p Rom 1:29?) φόνοι D F G d b ×vg; ×1175; A C 1241S, 1739 Ψ Chrys hark 1611
Byz ¦ txt (Marc); P46 B;  33; P
5:21  προείρηκα D F G ¦ txt Marc; P46 B; (* εἶπον) 33 ×1175; A C 1241S, (1739
προείπομεν) Ψ Chrys hark 1611 Byz
5:26  p) ἀλλήλους P46 B; G; P 1241S, Byzpt
6:2  p βαστάζετε) ἀναπληρώσατε D;  33 ×1175; A C 1241S, 1739 Ψ Chrys hark 1611
Byz ¦ txt Marc F G d b ×vg; (P46 ἀπο-) B
6:4  P46 B
180 Chapter 4. Examining and Establishing the Text

9 τὸ δὲ καλὸν ποιοῦντες μὴ ἐγκακῶμεν, καιρῷ γὰρ ἰδίῳ θερίσομεν μὴ ἐκλυ-

όμενοι.
10 Ἄρα οὖν ὡς καιρὸν ἔχομεν, ἐργαζώμεθα τὸ ἀγαθὸν πρὸς πάντας,
μάλιστα δὲ πρὸς τοὺς οἰκείους τῆς πίστεως.

11 Ἴδετε πηλίκοις ὑμῖν γράμμασιν ἔγραψα τῇ ἐμῇ χειρί.

12 Ὅσοι θέλουσιν εὐπροσωπῆσαι ἐν σαρκί, οὗτοι ἀναγκάζουσιν ὑμᾶς
περιτέμνεσθαι, μόνον ἵνα τῷ σταυρῷ τοῦ Χριστοῦ  μὴ διώκωνται. 13 οὐδὲ
γὰρ οἱ περιτεμνόμενοι αὐτοὶ νόμον φυλάσσουσιν ἀλλὰ θέλουσιν ὑμᾶς περι-
τέμνεσθαι, ἵνα ἐν τῇ ὑμετέρᾳ σαρκὶ καυχήσωνται.
14 Ἐμοὶ δὲ μὴ γένοιτο καυχᾶσθαι εἰ μὴ ἐν τῷ σταυρῷ τοῦ κυρίου ἡμῶν
Ἰησοῦ Χριστοῦ, δι’ οὗ ἐμοὶ κόσμος ἐσταύρωται κἀγὼ κόσμῳ. 15 οὔτε γὰρ
περιτομή τί ἐστιν οὔτε ἀκροβυστία ἀλλὰ καινὴ κτίσις. 16 καὶ ὅσοι τῷ κανόνι
τούτῳ στοιχήσουσιν, εἰρήνη ἐπ’ αὐτοὺς καὶ ἔλεος καὶ ἐπὶ τὸν Ἰσραὴλ τοῦ
θεοῦ.
17 Τοῦ λοιποῦ κόπους μοι μηδεὶς παρεχέτω· ἐγὼ γὰρ τὰ στίγματα
τοῦ Ἰησοῦ ἐν τῷ σώματί μου βαστάζω.

18 Ἡ χάρις τοῦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ μετὰ τοῦ πνεύματος ὑμῶν,
ἀδελφοί· ἀμήν.

6:9  p ἐγκακῶμεν) θερίσωμεν F G;  33 ×1175; C 1241S, Ψ Byzpt ¦ txt Marc D d b ×vg;

B; A, 1739 Chrys hark 1611 Byzpt
6:10  p ἐργαζώμεθα) ἔχωμεν B;  33
6:11  ἡλίκοις P46 B; 33
6:12  Ἰησοῦ P46 B; ×1175
6:13  περιτετμημένοι F G d b; P46 B; ×1175; Byz ¦ txt D ×vg;  33; A C 1241S, 1739
Chrys hark 1611
6:17  Χριστοῦ Marc; ×1175; P, Ψ ¦ κυρίου Ἰησοῦ 1739 hark 1611 Byz ¦ κυρίου Ἰησοῦ
Χριστοῦ d b;  ¦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ D* F G; Chrys ¦ txt ×vg; P46 B; 33; A
C 1241S
 Chapter 5

A History of Textual Variation in Galatians

5.1 Introduction

Just as the evidence for reconstructing the critical text of Galatians can be
divided into external and internal evidence, the main topics of inquiry for the
study of the textual history of Galatians can be divided into external and
internal questions. The “external” questions of textual history relate to the
origin of the external evidence – how the manuscripts and their groupings
came to be and how they are interrelated among each other. The “internal”
questions of textual history pertain to the various causes of origin for the
variant readings of the text and what these variant readings can tell us about
the readers and copyists of the text and their historical situation. Throughout
much of the course of the textual criticism of the New Testament, the external
questions of textual history have dominated the discussion because they have
been felt to be essential for establishing the text. Within the past generation of
scholarship, however, the internal questions of textual history have risen to
the fore, especially in the light of the “orthodox corruption” approaches of
Bart D. Ehrman and others working in this vein. Both aspects of the history
of the text are important, so this chapter concerns the internal questions of the
textual history, while the following concluding chapter wraps up this study of
the text of Galatians by looking at the external questions of the textual
history.
With regard to the internal questions of textual history, the variants genera-
ted in the course of the transmission of the text are the result of a long and
extended process of copying different forms of the text from one or more ex-
emplars. As the differing forms of the text are copied, changes are introduced
into the textual traditions, and these textual changes can be studied to deter-
mine what they have to say about the scribes and readers of the texts at dif-
ferent times and places. Even before the introduction of stemmatics to the
study of the New Testament text, internal evidence has been used to establish
the text, identify which readings are secondary, assess the nature and causes
of the textual variants, and determine how these readings function theologi-
cally. With a stemmatic history of the text, moreover, the stemma can be used
to refine these findings further by investigating whether there is a pattern to
this textural variant generally and to these theological effects in particular.
182 Chapter 5. A History of Textual Variation in Galatians

For example, if theologically potent variations are clustered together, then it

may help to localize a theological controversy in the history of the text and its
readers.
Several kinds of theological effects have been fruitfully explored in the
textual variation of the New Testament. One of them is Christological, where
textual variants affect what the text says about the nature of Jesus Christ. In a
seminal study, Bart D. Ehrman has classified the function of Christological
variants as anti-adoptionistic (tending to emphasize that Jesus was more than
mere man), anti-separationist (tending to emphasize that Jesus and the Christ
are the same), anti-docetic (tending to emphasize the humanity of Jesus), and
anti-Patripassionist (tending to emphasize the subordination of the Son to the
Father). 1
Another theological effect of textual variation has been labeled as “anti-
Judaic,” where variant readings affect what the text says about Jew and
Judaism. In Eldon Jay Epp’s study of Anti-Judaic tendencies in the Western
text of Acts, he classified anti-Judaic variants based on (a) how they relate
Jews to Jesus, (b) Jews to Gentiles, (c) Jews to the apostles. 2 Epp saw any
textual variant as anti-Judaic if its effect made the portrayal of Jews in Acts
more negative or made Jesus, the Gentiles, and apostles more positive in
comparison to Acts’ portrayal of Jews.
Given the size of the textual tradition, this chapter concentrates on two of
the most important branches of the text: the Western text and the portion of
the Eastern branch that led to the Byzantine text, paying particular attention
to possible theological effects in the textual variation.

5.2 The Western Branch and Prototype

The Western branch is one of the oldest branches in the textual transmission
of Galatians. As detailed in the stemma depicted by FIG. 20, the Western
branch may be defined as the set of witnesses more closely related to the
Greek diglots D*, F, and G, than to its closest relative P46–B. This set of
witnesses include the Greek diglots by definition, but it also includes the Old
Latin versions b and d, as well as mixed representatives DC, the Vulgate, and
the Latin version of Codex Augiensis (F). According to this definition, the
text of Marcion is also a member of the Western branch, because it is more
closely related to the common ancestor of D*, F, and G (labeled δ in FIG. 20)

1
Bart D. Ehrman, The Orthodox Corruption of Scripture: The Effect of Early Christo-
logical Controversies on the Text of the New Testament (New York: Oxford University
Press, 1993).
2
Eldon Jay Epp, The Theological Tendency of Codex Bezae Cantabrigiensis in Acts
(SNTSMS 3; Cambridge: Cambridge University Press, 1966).
 5.2 The Western Branch and Prototype 183

than to P46 or B. Unfortunately, our knowledge of the text of Marcion is

fragmentary, known only through quotations by Tertullian, Epiphanius, and
other authors, 3 so the text of the common ancestor of Marcion and D* is dif-
ficult to reconstruct with much precision. Where the testimony on Marcion’s
text is silent, it is impossible to determine whether a reading that agrees with
D*, F, and G but disagrees with P46, B, and  was introduced before or after
the common ancestor of Marcion and D* was written. Accordingly, this study
of the textual variation in the Western branch focuses on the readings of the
common ancestor of D*, F, G, and the Latin versions that differ from
branches most closely related to it, including the P46–B cluster. The text of
this common ancestor is referred to as the Western prototype and it is
represented by the siglum δ in the stemma depicted in FIG. 20, as follows:

ω
•
ε
•
β
P46-B
Marc α ℵ 33
Western Eastern
Branch δ Branch

• D*
• κ Byzantine
• Text
d •
G Dc
vg
b
F
f

FIG. 20: Stemma of Galatians with Detail on the Western Branch

In accordance with the stemma, the text of the Western prototype δ is deter-
mined by the agreement of D* and the joint agreement of F and G, on the one
hand, and of Latin versions d, b, and the Vulgate, on the other hand. It should
be kept in mind that this determination may both understate and overstate the
amount of textual change for this particular hypothetical ancestor. In addition,

3
See generally, Schmid, Marcion.
184 Chapter 5. A History of Textual Variation in Galatians

it is unclear how many generations of copying separate the creation of the

Western prototype δ from their common ancestor with Marcion or from its
common ancestor with P46 and B. Thus, the changes should not all be
attributed to a single scribe, the one who penned the most recent common
ancestor of the Greek-Latin diglots, for some of those changes could have
been made along the way to the Western prototype δ. Nevertheless, the
following seventy textual variants have been identified as present within the
Western prototype.

Gal 1:4 ὅπως ἐξέληται ἡμᾶς ἐκ τοῦ αἰῶνος τοῦ ἐνεστῶτος αἰῶνος πονηροῦ
τοῦ ἐνεστῶτος αἰῶνος πονηροῦ D F G d b ×vg; Ψ Chrys (hark) 1611 Byz
τοῦ αἰῶνος τοῦ ἐνεστῶτος πονηροῦ P46 B,  33, A P 1241S, 1739 pesh
The change in the position of the noun αἰῶνος is found in all the Western
witnesses, and it is also attested in the common ancestor of the Byzantine text
and the Athos codex Ψ. 4 This transposition changes the phrase from the
syntactically ambiguous τοῦ αἰῶνος τοῦ ἐνεστῶτος πονηροῦ (either “the age
of the present evil one” or, as it is usually taken, “the present age of evil” 5) to
the unambiguous τοῦ ἐνεστῶτος αἰῶνος πονηροῦ “the present age of evil.” It
is possible that this transposition came from a corrected leap, in which the
scribe leapt from one τοῦ to the next and, after writing ἐνεστῶτος, added back
the omitted word αἰῶνος. Even so, the theological effect is that it forecloses a
possible interpretation that the present age is under the rule of an evil
demiurge, such as the one found in Marcionite, Gnostic, and other dualist
systems. 6

4
Whether the transposition of the adjective in the Western and Byzantine branches is a
coincidence is addressed later in the chapter.
5
BURTON 13 states that the position of πονηροῦ in the noun phrase τοῦ αἰῶνος τοῦ
ἐνεστῶτος πονηροῦ “gives it special emphasis,” but the basis for this claim about the
predicate position seems dubious, for it would be more emphatic to prepose the adjective
before the noun, as if it had read τοῦ πονηροῦ αἰῶνος τοῦ ἐνεστῶτος πονηροῦ. See Stépha-
nie Bakker, “Adjective Ordering in Herodotus: A Pragmatic Explanation” in Rutger Allan
and Michel Buijs, eds., The Language of Literature: Linguistic Approaches to Classical
Texts (ASCP 13; Leiden: Brill, 2007), 188–210 at 195.
6
See John Riches, Galatians Through the Centuries (BBC; Malden, Mass.: Blackwell,
2008), 77–78, quoting Jerome, Comm. Gal. (PL 26.338) on Gal 1:4, as saying that “the
heretics usually take this as an opportunity to assert that there are two creators, one of light
and the world to come, another of darkness and the present age.”
 5.2 The Western Branch and Prototype 185

Gal 1:8 ἀλλὰ καὶ ἐὰν ἡμεῖς ἢ ἄγγελος ἐξ οὐρανοῦ ὑμῖν εὐαγγελίζηται ὑμῖν παρ’ ὃ
εὐηγγελισάμεθα ὑμῖν, ἀνάθεμα ἔστω.
εὐαγγελίζηται ὑμῖν d ×vg; 33, A P (1241S ἡμῖν), Chrys pesh hark 1611 Byz
εὐαγγελίζηται ὑμᾶς D*
εὐαγγελίζηται FG b; , Ψ
ὑμῖν εὐαγγελίζηται B; ×1175; 1739
The reading of the Western prototype is difficult to reconstruct because of the
instability of the text at this location, and indeed throughout the entire textual
history. The accusative ὑμᾶς by D* along with the dative vobis in the Old
Latin d are indirect testimony for the presence of the dative ὑμῖν in the
Western prototype. 7 As explained in Chapter 4, the movement of the pronoun
to after the verb serves to decontextualize the anathema and thereby make it
more pertinent to a broader audience. The omission of the pronoun altogether
in F, G, and b further helps to remove the historical contingency of the
anathema.

Gal 1:11 Γνωρίζω δὲ γὰρ ὑμῖν, ἀδελφοί, τὸ εὐαγγέλιον τὸ εὐαγγελισθὲν ὑπ’ ἐμοῦ
γάρ D* F G d ×vg; B ¹ 33
− P, Chrys
δέ b; P46, *, A 1241S, 1739 Ψ pesh hark 1611 Byz
The change in the connective particle from δέ to γάρ appears to give a reason
for Paul’s statement that he is Christ’s slave in the preceding v.10. 8

Gal 1:18 ἔπειτα μετὰ ἔτη τρία ἀνῆλθον εἰς Ἱεροσόλυμα ἱστορῆσαι Κηφᾶν Πέτρον
Πέτρον D F G d b ×vg; P, Ψ Chrys hark 1611 Byz
Κηφᾶς P46 B,  33, A 1241S, 1739 ×pesh
This substitution replaces the Aramaic name Cephas with the Greek Peter. 9
This change is also found in other manuscripts related to the Byzantine text.

7
There are two other possibilities. One is that the Western prototype read the accusative
ὑμᾶς, which was changed to the dative in the Old Latin d. Since the Vulgate is a mixture of
Western and Alexandrian forms of the text, its attestation of the dative vobis is problematic
because it could have had its origin from the Alexandrian ὑμῖν. Another is that the pronoun
was omitted in the prototype, but D and d added one back into the text, albeit in different
cases.
8
See discussion in Chapter 3.
9
BUSCEMI 126; METZGER 522.
186 Chapter 5. A History of Textual Variation in Galatians

Gal 1:19 ἕτερον δὲ τῶν ἀποστόλων οὐκ εἶδον οὐδένα εἰ μὴ Ἰάκωβον τὸν ἀδελφὸν
τοῦ κυρίου.
εἶδον οὐδένα D* F G (b neminem vidi) d ×vg
οὐκ εἶδον (P46 B οὐχ)  33 ×1175, A P 1241S,
1739 Ψ Chrys pesh hark 1611 Byz
The use of the negative pronoun οὐδένα (“no one”) serves to strengthen
Paul’s denial about not seeing any other apostle. 10

Gal 1:24 καὶ ἐν ἐμοὶ ἐδόξαζον ἐν ἐμοὶ τὸν θεόν.

ἐν ἐμοὶ ἐδόξαζον D F G b d ×vg
ἐδόξαζον ἐν ἐμοί P46 B,  33, A C P 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
Attested only in the Western branch, this transposition of ἐν ἐμοί to a pre-
verbal position serves to emphasize the reference to Paul and thereby lend
him more prominence. 11

Gal 2:1 ἔπειτα διὰ δεκατεσσάρων ἐτῶν πάλιν ἀνέβην πάλιν εἰς Ἱεροσόλυμα μετὰ
Βαρναβᾶ
ἀνέβην πάλιν DFGdb
πάλιν ἀνέβην ×vg; P46 B,  33, A C P 1241S, 1739 Ψ (pesh) hark 1611 Byz
This transposition could be the result a corrected leap, from the nu of ἐτῶν to
the nu of πάλιν, but the movement of the adverb πάλιν to a post-verbal
position has the effect of deemphasizing it. 12

Gal 2:6 θεὸς πρόσωπον ὁ θεὸς ἀνθρώπου οὐ λαμβάνει

θεὸς πρόσωπον D F G d b ×vg; (pesh)
πρόσωπον θεός B, C, 1739 1611 Byz
πρόσωπον ὁ θεός P46,  33, A P 1241S, Ψ Chrys
As discussed in Chapter 4, both the omission of the article and the movement
of God to the beginning of the sentence serve to emphasize God and thereby
make a contrast between God’s lack of partiality and human favoritism: “It is
God who does not accept a person’s appearance.” These changes serve to
make the statement more independent of the context.

10
Cf. BETZ 77 n.201, who considers this an “attempt[] to clarify the text.”
11
Stephen H. Levinsohn, Discourse Features of New Testament Greek: A Coursebook
on the Information Structure of New Testament Greek (2d ed.; Dallas: SIL International,
2000), 39. See also BDF 248, § 472: “Any emphasis on an element in the sentence causes
that element in the sentence to be moved forward.”
12
BUSCEMI 149 suggests that the transposition was stylistic, meant to break up the
accumulation of adverbs and temporal phrases.
 5.2 The Western Branch and Prototype 187

Gal 2:8 ὁ γὰρ ἐνεργήσας Πέτρῳ εἰς ἀποστολὴν τῆς περιτομῆς ἐνήργησεν καὶ ἐμοὶ
κἀμοὶ εἰς τὰ ἔθνη,
κἀμοί D* F G; P46, 33, A C P 1241S, al 13
καὶ ἐμοί B, , 1739 Ψ Chrys pesh hark 1611 Byz
The use of the crasis form κἀμοί appears to be a common, independent
change, for it is also found in other branches of the textual tradition.

Gal 2:9 καὶ γνόντες τὴν χάριν τὴν δοθεῖσάν μοι, Πέτρος καὶ Ἰάκωβος καὶ Κηφᾶς
καὶ Ἰωάννης, οἱ δοκοῦντες στῦλοι εἶναι,
Πέτρος καὶ Ἰάκωβος Marc D F G d b
Ἰάκωβος καὶ Πέτρος P46
Ἰάκωβος καὶ Κηφᾶς×vg; B,  33, C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
This change replaces the name Cephas with Peter and moves him to the be-
ginning of the list, demoting James to second place. 14 For the Western text of
Acts, Eldon Jay Epp has noticed a tendency to increase the prominence of
Peter, who like Paul, takes the gospel to the gentiles. 15 Although Peter does
not have a mission to the uncircumcised in Galatians (see especially vv.8 and
10), this transposition does lend more prominence to Peter. To the extent that
James represents Jewish Christianity, this word order variation at the expense
of James could be characterized as having an anti-Judaic effect, but, then
again, there were people who viewed Peter as the champion of Jewish
Christianity (for example, in the Pseudo-Clementines), so little can be made
of the word order in terms of an anti-Judaic effect. 16

Gal 2:10 μόνον ἵνα τῶν πτωχῶν ἵνα μνημονεύωμεν,

ἵνα τῶν πτωχῶν μνημονεύωμεν D F G d b ×vg; pesh hark
τῶν πτωχῶν ἵνα μνημονεύωμεν B,  33, A C 1241S, 1739 Chrys 1611 Byz
ἵνα μνημονεύωμεν τῶν πτωχῶν Marc 17
This transposition changes the prolepsis of τῶν πτωχῶν ἵνα to a more typical
word order. 18 The apparent Syriac and Latin support for the Western word

13
Also 075 69 88 104 223 330 489 927 1175 1319 1448 1573 1799 1837 2400 2892.
14
METZGER 523, who also suggests that it has the additional effect of “bringing
together the familiar pair of names, James and John.”
15
Epp, Theological Tendency, 154. See also Howard Eshbaugh, “Theological Variants
in the Western Text of the Pauline Corpus” (Ph.D. diss., Case Western Reserve University,
1975), 145–148.
16
I am grateful to Joel Marcus for this point.
17
Schmid, Marcion, 316, prints his reconstruction for Gal 2:10 as μόνον τῶν πτωχῶν
ἵνα μνηνομνεύωμεν, with italics indicating an unsecure reading. In fact, the evidence from
Tertullian, Marc. 5.3.6 is tantum ut meminissent egenorum, which means that, if any order
is indicated, it would be μόνον ἵνα μηηνομνεύωμεν τῶν πτωχῶν.
188 Chapter 5. A History of Textual Variation in Galatians

order is due to the fact that these languages do not have native structures
corresponding the Greek syntactic construction.

Gal 2:11 ὅτε δὲ ἦλθεν Κηφᾶς Πέτρος εἰς Ἀντιόχειαν,

Πέτρος D F G d; Chrys, hark, 1611, Byz
Κηφᾶς b ×vg; B,  33, A C P 1241S, 1739 Ψ pesh
As with Gal 1:18 and 2:9, the Aramaic name Cephas is replaced with the
Greek Peter. 19

Gal 2:14 εἶπον τῷ Κηφᾷ Πέτρῳ ἔμπροσθεν πάντων·

Πέτρῳ D F G b d; Chrys, hark 1611 Byz
Κηφᾳ ×vg; P46 B,  33, A C P 1241S, 1739 Ψ pesh
As with Gal 1:18, 2:9, and 2:11, this substitution replaces the Aramaic name
Cephas with the Greek Peter. 20

Gal 3:1 . . ., οἷς κατ’ ὀφθαλμοὺς Ἰησοῦς Χριστὸς προεγράφη ἐν ὑμῖν ἐσταυρωμένος;
ἐν ὑμῖν D F G d b; Chrys hark 1611 Byz
− ×vg; B,  33, A C 1241S, 1739 Ψ
The verb προεγράφη can have either a temporal sense (“portrayed pre-
viously”) or a locative sense (“portrayed publicly”). 21 The addition of the pre-
positional phrase ἐν ὑμῖν (“among you”) makes sure that the clause carries a
locative meaning, namely, that Jesus was visually and publicly portrayed
among the Galatians.” As such, this addition can be seen as clarifying an am-
biguity in the text. 22 This addition’s presence in the precursor to the
Byzantine text is discussed later in this chapter.

18
According to David Michael Goldstein, “Wackernagel’s Law in Fifth-Century Greek”
(Ph.D. diss., University of California, Berkeley, 2010), 76–79 only strong topics are al-
lowed to be fronted before complementizers like ἵνα. Thus, the difference may be prag-
matic: μόνον τῶν πτωχῶν ἵνα μνημονεύωμεν can be paraphrased as “as for the poor, only
that we remember them,” while μόνον ἵνα τῶν πτωχῶν μνημονεύωμεν would be “only that
it is the poor that we should remember.” The latter implies a contrast of the poor with some
other group of people.
19
METZGER 523.
20
METZGER 524.
21
BDAG “προγράφω” 867; Alfio Marcello Buscemi, L’uso delle preposizioni nella
lettera ai Galati (Analecta [Studium Biblicum Franciscanum] 17; Jerusalem: Franciscan
Printing Press, 1987), 89–90. See also Stephen C. Carlson, “Clement of Alexandria on the
‘Order’ of the Gospels,” NTS 47 (2001): 118–125.
22
By contrast, BETZ 131 n.39 argues that the addition of ἐν ὑμῖν enables the verb
προεγράφη to be understood in a temporal sense.
 5.2 The Western Branch and Prototype 189

Gal 3:2 τοῦτο μόνον θέλω μαθεῖν θέλω ἀφ’ ὑμῶν·

μόνον μαθεῖν θέλω ἀφ’ ὑμῶν D F G d
μόνον θέλω ἀφ’ ὑμῶν μαθεῖν ×vg
μόνον θέλω μαθεῖν ἀφ’ ὑμῶν b; B,  33, A C 1241S, 1739 Ψ Chrys hark 1611 Byz
The word order change affects which verb μόνον modifies. The Western
reading applies μόνον to μαθεῖν (“only to learn”), while the previous reading
applies μόνον to θέλω (“I only want”). The transposition allows Paul to want
other things, such as being with the Galatians (4:20 ἤθελον δὲ παρεῖναι πρὸς
ὑμᾶς), and thus this change possibly avoids making Paul sound disingenuous.

Gal 3:7 γινώσκετε ἄρα ὅτι οἱ ἐκ πίστεως, οὗτοι υἱοί εἰσιν υἱοὶ Ἀβραάμ.
εἰσιν υἱοί D F G d b ×vg; 33, A C, 1739 pesh hark 1611 Byz
υἱοί εἰσιν P46 B, , P 1241S, Ψ Chrys
As discussed in Chapter 3, this common transposition could be either
accidental (a corrected leap) or done to give the clause a more typical word
order.

Gal 3:10 ἐπικατάρατος πᾶς ὃς οὐκ ἐμμένει ἐν πᾶσιν τοῖς γεγραμμένοις ἐν τῷ βιβλίῳ
ἐν D F G d b ×vg; A, Chrys Byz
− P46 B,  33, C 1241S, 1739 Ψ hark pesh
The addition of the preposition ἐν appears to be a minor harmonization to the
Septuagintal text of Deut 27:26 ἐπικατάρατος πᾶς ἄνθρωπος, ὃς οὐκ ἐμμένει
ἐν πᾶσιν τοῖς λόγοις τοῦ νόμου τούτου. 23 It also disambiguates the dative
πᾶσιν to be locative (as a “dative of rule”) rather than instrumental.

Gal 3:11a ὅτι δὲ ἐν νόμῳ οὐδεὶς δικαιοῦται παρὰ τῷ θεῷ δῆλον,

τῷ D* F G; 1241S, 1611
− P46 B,  33, A C, 1739 Byz
God is a definite, uniquely identifiable entity, so the presence or absence of
the article does not change what it refers to, but it does give a clue as to its
prominence of θεός in the sentence and, indirectly, as to the structure of the
sentence. With the article, the reference to God lacks prominence and it is lo-
cated in the usual place after the verb in the ὅτι clause for a constituent famil-
iar to the hearer. Indeed, the main emphasis of the clause is on οὐδείς instead:
“That, in the Law, no one is being made right before God is clear.” Without
the article, in contrast, the reference to God becomes more prominent, 24 rais-

23
BETZ 144 n.60; BUSCEMI 257.
24
Levinsohn, Discourse Features, 162: “If an anarthrous substantive has a unique ref-
erent and is activated, then its referent is prominent.” As a “Very Important Participant” in
Paul’s letters, God is always activated.
190 Chapter 5. A History of Textual Variation in Galatians

ing the possibility that it construes with δῆλον rather than with δικαιοῦται. 25
Indeed, the clause with δικαιοῦται as the main verb already locates the
emphasis on οὐδείς: “It is clear with God that, in the Law, no one is being
made right.” A motive for this change may be to characterize the following
quotation from Hab 2:4 as explicitly coming from God, which has an anti-
Marcionite theological effect. 26 Another possibility is that the omission of the
article is a Latinism since Latin lacks both definite and indefinite articles. 27

Gal 3:11b ὅτι . . . δῆλον, γέγραπται γὰρ ὅτι ὁ δίκαιος ἐκ πίστεως ζήσεται·
γέγραπται γάρ D* F G (d scriptum est); pesh
− ×vg; P46 B,  33, A C 1241S, 1739 Chrys hark 1611 Byz
Galatians 3:11 has a difficult syntactic ambiguity in how to understand each
ὄτι. 28 Is the first ὅτι a complement for δῆλον and the second ὅτι causal (“That
. . . is clear, because . . .”)? Or, is the first causal and the second ὅτι a
complement (“Because . . ., it is clear that . . .”)? The Western addition
resolves the ambiguity in favor of the former possibility, by marking the
clause following δῆλον as explicitly causal with the particle γάρ.

Gal 3:14 ἵνα εἰς τὰ ἔθνη ἡ εὐλογία τοῦ Ἀβραὰμ γένηται ἐν Χριστῷ Ἰησοῦ, ἵνα τὴν
ἐπαγγελίαν εὐλογίαν τοῦ πνεύματος λάβωμεν διὰ τῆς πίστεως.
εὐλογίαν Marc D* F G d b; P46
ἐπαγγελίαν ×vg; B,  33, A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
As discussed in Chapter 3, the unexpected appearance of “promise” probably
induced scribes to assimilate it to εὐλογίαν in the immediately preceding
clause.

Gal 3:16 τῷ δὲ Ἀβραὰμ ἐρρέθησαν αἱ ἐπαγγελίαι καὶ τῷ σπέρματι αὐτοῦ,

− D* F G d b ×vg; C, Chrys
δέ B,  33, A 1241S, 1739 Ψ pesh hark 1611 Byz
The asyndeton of the Western branch creates a closer connection to the
preceding verse. The omission of the particle in C and Chrysostom appears to
be coincidental.

25
F and G, however, omit δῆλον, so this interpretive possibility is not available.
26
Unfortunately, Marcion’s text of Galatians is not extant for this verse.
27
See Bonifatius Fischer, “Limitations of Latin in Representing Greek,” in Bruce M.
Metzger, The Early Versions of the New Testament: Their Origin, Transmission, and
Limitations (Oxford: Clarendon Press, 1977), 362–374 at 366.
28
BURTON 166.
 5.2 The Western Branch and Prototype 191

Gal 3:17 διαθήκην προκεκυρωμένην ὑπὸ τοῦ θεοῦ εἰς Χριστὸν ὁ . . . νόμος οὐκ
ἀκυροῖ
εἰς Χριστὸν D F G d; Chrys pesh hark 1611 Byz
− b ×vg; P46 B,  33, A C 1241S, 1739 Ψ
There are a number of different options for construing εἰς Χριστόν. Eshbaugh
makes two proposals, both of which have theological implications. 29 One
possibility is that the preposition εἰς is temporal, that is, with the meaning
“until”: “a covenant previously (or publicly) ratified by God until Christ.” In
this case, the theological effect is anti-Judaic in that the Law was only to last
until Christ. Another possibility, actually favored by Eshbaugh, is that the
preposition εἰς is agentive, on the grounds that in later Koine Greek εἰς and ἐν
have converged and that ἐν can designate an agent. If this possibility is right,
then Christ would be the agent through whom God had ratified the covenant,
and its theological point would be Christ’s pre-existence and his “acting in
history prior to the incarnation.” 30 The difficulty with this proposal, however,
is that εἰς is a poor way to signal the behavior of an agent; the preposition διά
would be much more appropriate.
A better option is to construe εἰς as indicating a goal. 31 Such an
understanding is attested for the unprefixed verb in 2 Cor 2:8 παρακαλῶ ὑμᾶς
κυρῶσαι εἰς αὐτὸν ἀγάπην (“I urge you to ratify/affirm love for him”). In this
case, the addition makes it clear that the ratified covenant was for Christ. It
had Christ as the goal. This change seems to have been inspired by the
exegesis of the previous verse that Christ was Abraham’s seed to whom the
promises were spoken, 32 thereby making Paul’s point more explicitly than
Paul did. This has the consequence of having the God of Abraham operate in
history with a view to Christ, thereby making the connection between the Old
Testament God and Christ closer and, as a result, more pointedly anti-
Marcionite. 33
There is another theological effect of this textual addition. For example, in
a portion of his commentary disputing Jewish views of the Law,
Ambrosiaster construes his text of Gal 3:17, testamentum confirmatum a deo
in Christo (“a testament affirmed by God in Christ”), as teaching that “the
blessing of the promise is confirmed only in Christ” (in solo Christo firmatam
benedictionem promissionis docet). In other words, Ambrosiaster understands

29
Eshbaugh, “Theological Variation,” 70–73, esp. 72.
30
Eshbaugh, “Theological Variation,” 73.
31
See BDAG “εἰς 4cβ,” 290.
32
METZGER 525 is somewhat vaguer about the effect: “to introduce into the argument a
reference to Χριστός of the preceding verse.” So also BURTON 183.
33
Ehrman, Orthodox Corruption, 241. Much of this passage was objectionable to
Marcion and omitted by him, including Gal 3:15–17, according to Tertullian, Marc. 5.4.8
(Schmid, Marcion, 316, 318 n.25).
192 Chapter 5. A History of Textual Variation in Galatians

his text to affirm explicitly the supersessionist position that Christ exclusively
fulfills the promise God spoke to Abraham. 34 Thus, the addition is not only
anti-Marcionite in effect but also anti-Judaic.
This insertion also shows up in the precursor to the Byzantine text.

Gal 3:19 Τί οὖν ὁ νόμος; τῶν παραβάσεων χάριν προσετέθη ἔτεθη, ἄχρις οὗ ἔλθῃ τὸ
σπέρμα ᾧ ἐπήγγελται, διαταγεὶς δι’ ἀγγέλων ἐν χειρὶ μεσίτου.
τῶν παραβάσεων χάριν ἔτεθη (×vg Quid igitur lex? propter transgressiones
posita est.) 35
τῶν παραδόσεων χάριν ἔτεθη D*
τῶν πράξεων χάριν ἔτεθη (d quid igitur lex? factorum gratia posita est)
τῶν πράξεων ἔτεθη F G (b quid igitur lex factorum? posita est)
τῶν πράξεων P46
τῶν παραβάσεων χάριν προσέτεθη B,  33, A C 1241S, 1739 Ψ pesh hark 1611 Byz
προσέτεθη τῶν παραβάσεων χάριν Chrys 36

The prototypical Western reading for this verse is difficult to reconstruct. It is

tempting to view the agreement between P46 and F–G–b for πράξεων
(“deeds”) as evidence that the noun is a reading inherited by the Western
branch, but this would make it hard to account for the reading in D* of παρα-
δόσεων (“traditions”), which appears to be derived visually from παρα-
βάσεων (“transgressions”) as attested elsewhere (B  33 etc.; cf. Vulgate).
Thus, it seems best to conclude that the earliest Western reading is that of the
putative source for D*, while the other attested readings in Western witnesses
are later modifications. 37
Nevertheless, the change in the verb from προσέτεθη to ἔτεθη is strongly
attested in the Western branch. The loss of the prefix changes the meaning of
the verb from “added” to “put into place” or “established.” Ernest de Witt
Burton argues that this change was made to avoid an “apparent contradiction”
with vv.15–17. 38 While Burton’s proposed motive seems plausible, this
change also has two theological implications. In particular, this change raises

34
See also the discussion in Martin Meiser, Galater (NTP 9; Göttingen: Vandenhoeck
& Ruprecht, 2007), 149–150.
35
This is also the reading of Clement of Alexandria, Strom. 1.26.167, who then con-
nects 3:19 with vv.23–24.
36
Instead of τί οὖν ὁ νόμος; “Why, then, the law?”, Chrysostom’s text reads τίνος οὖν
ἕνεκεν τὸν νόμον ἔδωκε; “On account of what, then, did he give the law?”.
37
It could not have been very late, however, for “deeds” is attested in Irenaeus, Adv.
haer. 3.7.2: Quid ergo lex factorum? posita est, usquequo veniat semen cui promissum est,
disposita per angelos in manu Mediatoris.
38
BURTON 188; so also BUSCEMI 316. Specifically, the contradiction occasioned by
προσέτεθη would appear to be with v.15 that no one amends (ἐπιδιατάσσεται) a previously
ratified covenant, and with v.17 that the Law cannot void God’s previously ratified cove-
nant so as to invalidate the promise (εἰς τὸ καταργῆσαι τὴν ἐπαφφελίαν). Cf. de Boer, Gal-
atians, 228–229, who points out that προστίθημι is a near synonym of ἐπιδιατάσσομαι and
suggests that the distinction in this context is that it refers to an unauthorized addition.
 5.2 The Western Branch and Prototype 193

the prestige of the Law before Christ by stating that the Law was established
(not merely added). 39 The change heightens a supersessionist reading of the
passage, in that the text now implies that, after Christ, the Law is no longer
considered to be established. 40

Gal 3:21 εἰ γὰρ ἐδόθη νόμος ὁ δυνάμενος ζῳοποιῆσαι, ὄντως ἐκ νόμου ἂν ἦν ἡ

δικαιοσύνη·
− D* F G
ἄν P46 B,  33, A C 1241S, 1739 Chrys 1611 Byz
The position of ἄν is somewhat unusual, 41 but the omission may simply have
been a mechanical error by the similarity of the letters between ΑΝ and ΗΝ.

Gal 3:24 ὥστε ὁ νόμος παιδαγωγὸς ἡμῶν γέγονεν εἰς Χριστόν Ἰησοῦν, ἵνα ἐκ
πίστεως δικαιωθῆμεν.
Ἰησοῦν D* F G d b
− ×vg; P46 B,  33, A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
The addition of Jesus could be a harmonization to Jesus Christ in v.22 or
Christ Jesus in v.26. Generally, the addition of “Jesus” to “Christ” has a
potentially anti-Separationist function in the context of the crucifixion. 42
Such a theological effect may indeed be present in this context if ἐκ πίστεως
is given a Christocentric reference, namely, to Christ’s faithfulness even to
the point of death at the cross (cf. Gal 2:20). 43

39
Eshbaugh, “Theological Variants,” 105, arguing instead that the change from προσἔ-
τεθη to ἔτεθη was meant to avoid the implications that the Law was an afterthought.
40
E.g., Irenaeus, Adv. haer. 3.7.2, who argues that Gal 3:19 features a hyperbaton due
to Paul’s rapid writing in the Spirit and that its logical order is really as follows: “What is
the law of deeds? Ordained by angels in the hand of the mediator, it was established until
the promised seed comes” (Quid ergo lex factorum? disposita per angelos in manu Media-
toris posita est usquedum veniat semen cui promissum est). Irenaeus’s focus on the
chronological order of events further supports a supersessionist interpretation.
41
See discussion in Chapter 4 for more details.
42
See generally Ehrman, Orthodox Corruption, 143–155.
43
See generally Douglas A. Campbell, The Deliverance of God: An Apocalyptic
Rereading of Justification in Paul (Grand Rapids, Mich.: Eerdmans, 2009), 867–875.
194 Chapter 5. A History of Textual Variation in Galatians

Gal 3:29a εἰ δὲ ὑμεῖς Χριστοῦ εἷς ἐστε ἐν Χριστῷ Ἰησοῦ,

εἷς ἐστε ἐν Χριστῷ Ἰησοῦ D*
ἕν ἐστε ἐν Χριστῷ Ἰησοῦ F G d
omit εἰ δὲ ὑμεῖς Χριστοῦ b
Χριστοῦ ×vg; B,  33, A C 1241S, 1739 Chrys pesh hark 1611 Byz
This lengthy substitution is clearly an assimilation to the preceding v.28
πάντες γὰρ ὑμεῖς εἷς ἐστε ἐν Χριστῷ Ἰησοῦ, 44 explaining what it means to
belong to Christ.

Gal 3:29b ἄρα οὖν τοῦ Ἀβραὰμ σπέρμα ἐστέ,

ἄρα οὖν D* F G
ἄρα B,  33, A C 1241S, 1739 Ψ Chrys 1611 Byz
The earliest manuscripts lacked accents, so αρα could be either ἄρα, a marker
of inference, or ἆρα, and interrogative marker. 45 The addition of the
inferential particle οὖν removes this ambiguity.

Gal 4:6 Ὅτι δὲ ἐστε υἱοὶ θεοῦ, ἐξαπέστειλεν ὁ θεὸς τὸ πνεῦμα τοῦ υἱοῦ αὐτοῦ εἰς
τὰς καρδίας ἡμῶν κράζον· αββα ὁ πατήρ.
υἱοὶ θεοῦ Marc D F G d b
υἱοί ×vg; P46 B,  33, A C P 1241S, 1739 Ψ pesh hark 1611 Byz
As discussed in Chapter 3, the addition of θεοῦ appears to be a local
harmonization to 3:26 πάντες γὰρ υἱοὶ θεοῦ ἐστε.

Gal 4:8 Ἀλλὰ τότε μὲν οὐκ εἰδότες θεὸν ἐδουλεύσατε τοῖς φύσει μὴ οὖσιν
θεοῖς ἐδουλεύσατε·
τοῖς . . . θεοῖς ἐδουλεύσατε D F G d b ×vg
τοῖς . . . θεοῖς δουλεύετε Marc 46
ἐδουλεύσατε τοῖς . . . θεοῖς P46 B,  33, A C 1241S,
1739 Ψ Chrys pesh hark 1611 Byz
As discussed in Chapter 3, the word order is more highly marked with the
verb at the end of the clause, and it serves to emphasize the fact that they
were enslaved to false gods. The effect of this change seems to heighten the
anti-pagan force of the verse.

44
BUSCEMI 346.
45
BDAG 127. For those later manuscripts that do write accents, ἄρα is read by 1611
and most Byz, while ἆρα is read by 33 and 1241S, as well as 88 104 460 618 910 1315
1424 1837 1874.
46
Tertullian, Marc. 5.4.5: si ergo his, qui ‘in natura sunt’ dei, servitis. Schmid, Mar-
cion, 116, suggests that the omission of μή is due to a corruption of τοῖς μὴ φύσει into τοῖς
τῇ φύσει.
 5.2 The Western Branch and Prototype 195

Gal 4:9 νῦν νυνὶ δὲ γνόντες θεόν ... πῶς ἐπιστρέφετε πάλιν ἐπὶ τὰ ἀσθενῆ καὶ πτωχὰ
στοιχεῖα ...;
νυνί D* F G; Ψ
νῦν P46 B,  33, A C 1241S, 1739 Chrys pesh hark 1611 Byz
The adverb νυνί is a strengthened form of νῦν, and at other locations the
manuscripts often vary between them, in either direction. 47 It seems plausible
that the strengthened νυνί (“now”) was adopted to heighten the contrast with
the τότε (“then”) of the previous sentence.

Gal 4:10 ἡμέρας παρατηρεῖσθε καὶ μῆνας καὶ καιροὺς καὶ ἐνιαυτοὺς καὶ καιρούς,
καὶ ἐνιαυτοὺς καὶ καιρούς DFGdb
καὶ καιροὺς καὶ ἐνιαυτούς Marc 48 ×vg; P46 B,  33, A C 1241S,
1739 Ψ Chrys pesh hark 1611 Byz
The attestation by Marcion is evidence that this transposition is a later
development in the Western branch. The time references in the earliest form
of the text, “days . . . months . . . seasons . . . years,” suggest a connection to
the Jewish cultic calendar, 49 so the transposition of seasons and years might
indicate that this connection had been lost in the mind of the scribe of the
common ancestor of the Western branch.

Gal 4:13 οἴδατε δὲ ὅτι δι’ ἀσθένειαν τῆς σαρκὸς εὐηγγελισάμην ὑμῖν τὸ πρότερον,
− D* F G d b
δέ ×vg; P46 B,  33, A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz 50
Like the omission in Gal 3:16, the asyndeton created by the omission of δέ
creates a closer connection to the preceding verse.

47
When used with δέ both adverbs appear within Paul’s letters with some frequency:
νυνὶ δέ in Rom 3:21; 7:6 (νῦν F G), 17; 15:23, 25 (νῦν F G); 1 Cor 12:18 (νῦν A B D* F G
1611); 13:13; 15:20 (νῦν F G); 2 Cor 8:11, 22 (νῦν Ψ); Phlm 9 (νῦν A), 11; cf. Eph 2:13;
Col 1:22 (νῦν D* F G); 3:8; and νῦν δέ in Rom 11:30 (νυνί B); 1 Cor 5:11 (νυνί * C D*);
7:14 (νυνί D* F G); 12:20 (νυνί F G P 1241S 1611 1739); 14:6 (νυνί Byz); Gal 4:9 (νυνί
D* F G Ψ); Phil 3:18; cf. Eph 5:8 (νυνί F G); Col 1:26.
48
Schmid, Marcion, 317 and 347, considers this reading secure based on Tertullian,
Marc. 5.4.5: ipse declaret: dies observatis et menses et tempora et annos, but it should be
noted that when Tertuallian is less careful he quotes these temporal terms in varying orders
(see Schmid, Marcion, 102).
49
So BURTON 234; LONGENECKER 182.
50
Some Byzantine manuscripts omit the particle: 1242 1448 1734 1735.
196 Chapter 5. A History of Textual Variation in Galatians

Gal 4:16 ὥστε ἐγὼ ἐχθρὸς ὑμῶν γέγονα ἀληθεύων ὑμῖν;

ἐγώ D* F G d b
− ×vg; P46 B,  33, A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
The presence or absence of the personal pronoun ἐγώ affects the placement of
emphasis and hence the implied contrast in the sentence. Without the pro-
noun, the emphasis falls on the fronted ἐχθρὸς ὑμῶν – “So that I became an
enemy of yours by telling you the truth?” – and it contrasts with the
Galatians’ prior reception of Paul “like God’s angel” (v.13). With the pro-
noun, the emphasis falls on Paul, heightening the contrast between him and
his (presumably Judaizing) opponents – “So that I’m the one who became
your enemy by telling you the truth?” Despite the apparent plausibility that
the shorter reading is the result of a leap from one epsilon to another, the ad-
dition of the pronoun seems more suitable to a later historical context in
which the parting of the ways between Christianity and Judaism has carried
on to a further extent when it is more salient to pit Paul against the Jews.

Gal 4:17 ἀλλὰ ἐκκλεῖσαι ὑμᾶς θέλουσιν, ἵνα αὐτοὺς ζηλοῦτε· ζηλοῦτε δὲ τὰ κρείττω
χαρίσματα
ζηλοῦτε δὲ τὰ κρείττω 51 χαρίσματα D* F G d b
− ×vg; B,  33, A C 1241S,
1739 Chrys pesh hark 1611 Byz
This additional phrase, which means “but desire the better gifts,” is a remini-
scence of the Western reading of 1 Cor 12:31, ζηλοῦτε δὲ τα χαρίσματα τα
κρείσσονα. 52 The difference in the word order and use of the Attic form
κρείττω demonstrates that this addition comes from a common origin rather
than from independent incorporations of 1 Cor 12:31 into this context.

51
This is the standard spelling of the diphthong. D* F G, however, use itacistic
spellings with ι instead of ει. The itacistic spelling κρίσσονα for κρείσσονα also occurs for
D* F G in 1 Cor 12:31.
52
BUSCEMI 413. Instead of κρείσσονα, P46vid B  33 read μείζονα.
 5.2 The Western Branch and Prototype 197

Gal 4:18 καλὸν δὲ τὸ ζηλοῦσθαι ἐν καλῷ πάντοτε καὶ μὴ οὐ μόνον ἐν τῷ παρεῖναί με

πρὸς ὑμᾶς.
− D* F G d b
δέ ×vg; B, , A C 1241S, 1739 Chrys pesh hark 1611 Byz
γάρ 33

τό D F G; Chrys 1611 Byz

− B,  33, A C 1241S, 1739

οὐ DFG
μή B,  33, A C 1241S, 1739 Chrys 1611 Byz
These three little changes to this verse are collected together. Though un-
related in their functions, their patterned agreement against non-Western wit-
nesses nonetheless demonstrates a common origin. As with Gal 3:16 and
4:13, the deletion of δέ creates an asyndeton. The addition of the article τό is
anaphoric, 53 referring back to the statement ζηλοῦσιν ὑμᾶς οὐ καλῶς in v.17;
it also makes it clear that the following word is an infinitive rather than the
homophonous indicative ζηλοῦσθε, which is the actual reading of B, , and
33. 54 Finally, the change from μὴ μόνον to οὐ μόνον makes this negative con-
form to more common usage. 55

Gal 4:21 Λέγετέ μοι, οἱ ὑπὸ νόμον θέλοντες εἶναι, τὸν νόμον οὐκ ἀκούετε
ἀναγινώσκετε;
ἀναγινώσκετε 56 D F G d b ×vg
ἀκούετε P46 B,  33, A C 1241S, 1739 Chrys pesh hark 1611 Byz
The change from “do you not hear (ἀκούετε) the Law?” to “do you not read
(ἀναγινώσκετε) the Law?” is striking. Origen, Contra Celsum 2.3.8, quotes
the first part of the sentence as οἱ τὸν νόμον ἀναγινώσκοντες (“those who
read the law”) relating to Jews, 57 which suggests that a reference to people
reading the law connotes Jews in early Christianity. Similarly, this change
from hearing the law to reading the law would also make it clear that Paul is
addressing Jews or those engaging in Jewish practices. As a result, this
change can be seen as part of a number of variants in the Western prototype
that are arguably anti-Judaic in effect. 58

53
BDF 205–206 § 399(1).
54
For more details, see the discussion in Chapter 3.
55
See A. T. Robertson, A Grammar of the Greek New Testament in the Light of Histori-
cal Research (4th ed.; New York: Hodder & Stoughton, 1923), 1162.
56
Also in 104 and 1175.
57
Also quoted in the same form in Contra Celsum 4.44.24.
58
Steven Di Mattei, “Paul’s Allegory of the Two Covenants (Gal 4.21–31) in Light of
First-Century Hellenistic Rhetoric and Jewish Hermeneutics,” NTS 52 (2006): 102–122 at
102–103 n.2, suspects that the Western reading was influenced by Origen’s usage, but it
198 Chapter 5. A History of Textual Variation in Galatians

Gal 4:23 ὁ δὲ ἐκ τῆς ἐλευθέρας δι’ διὰ τῆς ἐπαγγελίας.

διὰ τῆς D F G; B; 1739 1611 Byz
δι’ P46,  33, A C 1241S, 1739 Ψ Chrys
This is the first occurrence of “promise” since 3:29. As the pattern of
manuscript attestation indicates, adding an article to refer back to the promise
spoken to Abraham 59 appears to be a common, independent change.

Gal 4:25 τὸ γὰρ Σινᾶ ὄρος ἐστὶν ἐν τῇ Ἀραβίᾳ· συστοιχεῖ δὲ συστοιχοῦσα τῇ νῦν
Ἰερουσαλήμ,
συστοιχοῦσα D* F G (d quae consonant)
σύστοιχον (×vg qui [scil. mons] conjunctus est)
συστοιχεῖ δέ (b conjugitur autem); P46 B,  33, A C P 1241S,
1739 Chrys pesh hark 1611 Byz
This textual variant is significant because it indicates that the first part of the
clause about Sinai being a mountain in Arabia did not yet stand in the main
text. 60 Although non-Western witnesses read συστοιχεῖ δὲ (“but it corres-
ponds”) after the geographic note, the Western Greek witnesses read the
feminine nominative participle ἡ συστοιχοῦσα instead. The closest feminine
nominative antecedent for this participle is not in v.25a at all, but the femi-
nine Hagar at the end of v.24: ἥτις ἐστὶν Ἁγάρ. This antecedent is too far
away from the participle, unless v.25a stood not in the main text but in the
margin. 61 At some point after this change, the marginal gloss was inserted
into the main text but without patching up the participle. The masculine par-
ticiple of the Vulgate represents an attempt to make it relate to a closer
antecedent, i.e., the masculine mons (“mountain”).

Gal 4:27 εὐφράνθητι, στεῖρα ἡ οὐ μὴ τίκτουσα,

μή DFG
οὐ P46 B,  33, A C 1241S, 1739 Chrys 1611 Byz
Except for Hebraisms from the Septuagint (this is a quotation from Isa 54:1),
the regular Hellenistic negative with an articular participle is μή, not οὐ. 62

seems better to view them as independent – and different – attempts to make explicit the
Jewish nature of Paul’s opponents.
59
BUSCEMI 449.
60
Allen Kerkeslager, “Jewish Pilgrimage and Jewish Identity in Hellenistic and Early
Roman Egypt” in Pilgrimage and Holy Space in Late Antique Egypt (ed. David Frank-
furter; Leiden: Brill, 1998), 99–225 at 184–185. BUSCEMI 450, on the contrary, merely
concedes that v.25a is to be conceived of as a parenthesis and suggests that the participial
form was to disambiguate the subject.
61
For more detail, see the discussion in Chapter 3.
62
BDF 222–223 § 430(3).
 5.2 The Western Branch and Prototype 199

Thus, the change in the Western branch from οὐ to μή is in conformation to

the standard idiom.

Gal 4:30 οὐ γὰρ μὴ κληρονομήσῃ ὁ υἱὸς τῆς παιδίσκης μετὰ τοῦ υἱοῦ τῆς ἐλευθέρας
μου Ἰσαάκ.
τοῦ υἱοῦ μου Ἰσαάκ D* F G d b
τοῦ υἱοῦ τῆς ἐλευθέρας×vg; P46 B,  33, A C P 1241S,
1739 Ψ Chrys pesh hark 1611 Byz
This substitution is a harmonization to the Septuagintal text of Gen 21:10. 63

Gal 5:1a τῇ ᾗ ἐλευθερίᾳ ἡμᾶς Χριστὸς ἠλευθέρωσεν·

ᾗ ἐλευθερίᾳ Marc F G (d b ×vg qua)
τῇ ἐλευθερίᾳ D; B,  33, A C P 1241S, 1739 Ψ
τῇ ἐλευθερίᾳ ᾗ Chrys pesh hark 1611 Byz
The agreement between FG-db and Marcion indicates either that the Western
prototype featured the substitution of the relative pronoun ᾗ for the article τῇ
and it was reversed by D, or that it was independently made by both Marcion
(as quoted by Tertullian 64) and the common ancestor of F, G, and the Old
Latin witnesses b and d. The asyndeton at Gal 5:1 is ambiguous as to whether
there is a close connection between 5:1 and 4:31 or whether 5:1 begins a new
section. 65 The substitution of the relative pronoun ᾗ for the article τῇ resolves
this ambiguity in favor of the former option by making the connection syntac-
tically close with a relative clause. By contrast, the ambiguity of the asyn-
deton is resolved in the other direction within the Eastern branch as discussed
later in this chapter.

63
BUSCEMI 451 also suggests a harmonization to v.28.
64
Tertullian, Marc 5.4.9 attests the relative pronoun, qua libertate Christus nos manu-
misit, but Schmid, Marcion, 74, thinks that this evidence is also compatible with the pre-
sence of the article in Marcion’s Greek text, following T. Baarda, “Gal. 5.1a: ᾗ ἐλευθερίᾳ
. . . Over de ‘Westerse Tekst’ en de Tekst van Marcion” in C. J. den Heyer, ed., Christo-
logische perspectieven: Exegetische en hermeneutische studies (Kampen: Kok, 1992),
173–193. Specifically, Baarda argues that Tertullian’s use of manumisit indicates that he
had made an independent translation of his own Greek text and that qua libertate can also
translate τῇ ἐλευθερία ᾗ. Given the ambiguity of Tertullian’s Latin, however, it seems best
to relate it to the form of its closest relatives in the Western branch rather than to the form
of its distant relatives in the Byzantine and Harklean texts.
65
See generally Levinsohn, Discourse Features, 118–119; and Stephanie L. Black, Sen-
tence Conjunctions in the Gospel of Matthew: καί, δέ, τότε, γάρ, οὖν and Asyndeton in Nar-
rative Discourse (JSNTSS 216; STG 9; London: Sheffield Academic Press, 2002), 179–
217, esp. 216–217.
200 Chapter 5. A History of Textual Variation in Galatians

Gal 5:1b στήκετε οὖν καὶ μὴ πάλιν δουλείας ζυγῷ δουλείας ἐνέχεσθε.
δουλείας ζυγῷ DFGdb
ζυγῷ δουλείας Marc 66 ×vg; B,  33, A C P 1241S, 1739 Ψ Chrys pesh hark
1611 Byz
This transposition gives more prominence to slavery (δουλείας) than to yoke
(ζυγῷ). By emphasizing that the yoke (of the Law, cf. Acts 15:10) is slavery,
this reading appears to heighten a perceived anti-Judaic element in the text.

Gal 5:3 μαρτύρομαι δὲ πάλιν παντὶ ἀνθρώπῳ περιτεμνομένῳ

− D F G d b; 1739 Chrys
πάλιν ×vg; P46 B,  33, A C, Ψ pesh hark 1611 Byz
This omission appears to be due to an accidental leap from ΠΑ of πάλιν to the
ΠΑ of παντί. It has the side-effect of not implying that there was a previous
time when Paul warned men undergoing circumcision.

Gal 5:10 ὁ δὲ ταράσσων ὑμᾶς βαστάσει τὸ κρίμα, ὅστις ἐὰν ἂν ᾖ.

ἄν D F G; Ψ Chrys 1611 Byz
ἐάν P46 B,  33, A C 1241S, 1739
The change from ἐάν to ἄν is in the direction of using a more classical form
for this particle after an indefinite pronoun. 67 The same change in this particle
occurs in v.17.

Gal 5:11 Ἐγὼ δέ, ἀδελφοί, εἰ περιτομὴν ἔτι κηρύσσω, τί ἔτι διώκομαι;
− D* F G d b; 1739
ἔτι ×vg; P46 B,  33, A C 1241S, Ψ Chrys pesh hark 1611 Byz
The omission of ἔτι (“still”) avoids the implication that Paul had once
preached circumcision after his call. 68 Because of its apparent apologetic
motive in preserving Paul’s consistency, the same omission in 1739 ought to
be considered coincidental rather than genetic.

66
Schmid, Marcion, 319, prints the words ζυγῷ δουλείας in italics, indicating un-
certainty, but Tertullian, Marc 5.4.9, attests this order: et merito nondecebat manumissos
rursus iugo servitutis, id est, adstringi.
67
See BDAG “ἐάν,” 267–268; BDF 57 § 107; LSJ “ἐάν,” 465.
68
Douglas A. Campbell, “Galatians 5.11: Evidence of an Early Law-observant Mission
by Paul?” NTS 57 (2011): 325–347.
 5.2 The Western Branch and Prototype 201

Gal 5:12 Ὄφελον καὶ ἀποκόψονται ἀποκόψωνται οἱ ἀναστατοῦντες ὑμᾶς.

ἀποκόψωνται D F G (d b ×vg abscidantur); P46
ἀποκόψονται B,  33, A C P 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
The particle ὄφελον is construed with the indicative, 69 so the change from a
future indicative ἀποκόψονται to an aorist subjunctive ἀποκόψωνται is
unusual, if not ungrammatical. 70 Nevertheless, these forms were pronounced
identically by the second century, 71 so this difference may merely be ortho-
graphic as far the Western scribes were concerned. Because the Latin ver-
sions use a subjunctive form, abscidantur, it is tempting to see the ἀποκό-
ψωνται spelling in the Greek-Latin diglots as being influenced by the Latin.

Gal 5:13 μόνον μὴ τὴν ἐλευθερίαν εἰς ἀφορμὴν τῇ σαρκί, ἀλλὰ διὰ τῆς ἀγάπης τῇ
ἀγάπῃ τοῦ πνεύματος δουλεύετε ἀλλήλοις.
τῇ ἀγάπῃ τοῦ πνεύματος D F G (d b per caritatem spiritus)
διὰ τῆς ἀγάπης P46 B,  33, A C 1241S,
1739 Ψ Chrys pesh hark 1611 Byz
This Western form of the text, “by the love of the spirit,” is striking. The only
other occurrence of “the love of the Spirit” in Paul’s letters is Rom 15:30 Πα-
ρακαλῶ δὲ ὑμᾶς, ἀδελφοί, διὰ τοῦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ καὶ διὰ τῆς
ἀγάπης τοῦ πνεύματος, where the Romans were encouraged to pray on Paul’s
behalf as he goes to Jerusalem. In this case, the immediately preceding men-
tion of the flesh (τῇ σαρκί) and the subsequent contrast of the spirit and the
flesh in vv.16–26 appears to have induced a scribe to introduce an explicit re-
ference to the spirit. 72

69
BDF 181 § 359(1). BDF 194 § 384 cites this very variant but does not comment on
the mood.
70
The aorist subjunctive in P46, however, is legitimate because it changes ὄφελον to
ἄρα. This suggests that P46’s agreement here is unrelated.
71
W. Sidney Allen, Vox Graeca: The Pronunciation of Classical Greek (3d ed.; Cam-
bridge: Cambridge University Press, 1987), 94. Geoffrey Horrocks, Greek: A History of
the Language and its Speakers (New York: Longman, 1997), 75–76, dates the change even
earlier.
72
BUSCEMI 527.
202 Chapter 5. A History of Textual Variation in Galatians

Gal 5:14 ὁ γὰρ πᾶς νόμος ἐν ὑμῖν ἐν ἑνὶ λόγῳ πεπλήρωται πληροῦται, ἐν τῷ·
ἀγαπήσεις τὸν πλησίον σου ὡς σεαυτόν.
ἐν ὑμῖν (Marc −ἐν ἑνὶ λόγῳ) D* F G d b
− ×vg; P46 B,  33, A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz

πληροῦται D F G d b ×vg; Ψ Chrys 1611 Byz

πεπλήρωται Marc; P46 B,  33, A C 1241S, 1739

− Marc D* F G d b ×vg
ἐν τῷ P46 B,  33, A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
These three variants are collected together because they occur within a short
span of text. Two of these are supported by Marcion and one is not. 73 As for
the first reading ἐν ὑμῖν, the effect of this change, in light of its context (e.g.
v.13 ὑμεῖς γὰρ ἐπ’ ἐλευθερίᾳ ἐκλήθητε), is to make the statement more anti-
Judaic, in that the law is specified as being fulfilled among the gentile
Galatians. 74 The third reading, also attested in Marcion, can also be seen as
anti-Judaic, because the lack of introductory signal for the quotation means
that the positive citation of Lev 19:18 in the Torah is not explicitly marked. 75
The failure to mark the Levitical love injunction as a quotation opens a her-
meneutical space to reinterpret the command, not necessarily as command
from the Torah, but from another source such as Jesus (cf. Matt 5:43; Luke
10:27, etc.).
As for the second reading, the perfect tense πεπλήρωται is usually taken to
be gnomic, but with Paul’s strong focus on Christ it should be construed
Christologically, that, the whole law has been fulfilled by Jesus’s love for his
neighbor (as by giving himself on the cross cf. Gal 1:4, 2:20). 76 The substitu-
tion of the present tense πληροῦται (perhaps prompted by the contemporary
sense of the added phrase ἐν ὑμῖν) has the effect of reinforcing the gnomic
interpretation.

73
Schmid, Marcion, 130–131 and 182, citing Tertullian, Marc. 5.4.12: tota enim, inquit,
lex ‘in vobis’ adimpleta est: dileges proximum tuum tamquam te, and Epiphanius, Pan.
42.11.8, 42.12.3 ὁ γὰρ πᾶς νόμος ὑμῖν πεπλήρωται· ἀγαπήσεις τὸν πλησίον ὡς σεαυτόν, re-
spectively. Schmid, Marcion, 318, prints πεπλήρωται in italic to indicate that it is an un-
certain reading, but both Tertullian and Epiphanius have perfect forms: adimpleta est and
πεπλήρωται, respectively (see Schmid, Marcion, 182).
74
Eshbaugh, “Theological Variants,” 187. Pace Schmid, Marcion, 130–131, who dis-
agrees with Harnack’s view that this change is Marcionite and anti-Jewish on the grounds
that the inference of “not the Jews” drawn from ἐν ῦμιν is much too subtle (“viel zu
subtil”).
75
Pace Schmid, Marcion, 130–131, who cites the Rom 13:9 parallel to argue that there
is no unified tendency behind the omission.
76
MARTYN, 486–491.
 5.2 The Western Branch and Prototype 203

According to the stemma, the fact that two of these anti-Judaic variants are
also attested in Marcion suggests that they were already present in the com-
mon ancestor of Marcion and the Western branch.

Gal 5:15 εἰ δὲ ἀλλήλους δάκνετε καὶ κατεσθίετε ἀλλήλους, βλέπετε μὴ ὑπ’ ἀλλήλων
ἀναλωθῆτε ὑπ’ ἀλλήλων.
δάκν. κ. κατ. ἀλλήλους DFGdb
ἀλλήλους δάκν. κ. κατ. ×vg; P46 B,  33, A C 1241S,
1739 Ψ Chrys pesh hark 1611 Byz

ἀναλωθῆτε ὑπ’ ἀλλήλων D F G d; 1241S

ὑπ’ ἀλλήλων ἀναλωθῆτε b ×vg; P46 B,  33, A C,
1739 Ψ Chrys pesh hark 1611 Byz
These two transpositions appear to be related, for they both move the recipro-
cal pronoun from a marked position before the verb to a more typical position
afterwards. These changes move the emphasis from the Galatians (referred to
by the reciprocal pronouns) to the actions of the verb. As a result, the verse is
less contingent on the original occasion of the letter and more applicable for a
general audience.

Gal 5:17 ἵνα μὴ ἃ ἐὰν ὃ ἂν θέλητε ταῦτα ποιῆτε.

ὃ ἄν D* F G
ἃ ἄν (d b ×vg quaecumque); CC, Ψ 1611 Byz
ἃ ἐάν ΒC, , A 1241S
ἄν B*
ἅ C*, 1739 Chrys
There are two changes here, both attested clearly in the Greek Western wit-
nesses. First, they change the relative pronoun from the plural to the singular,
despite its lack of agreement with the plural ταῦτα. 77 Second, they change the
Hellenistic particle ἐάν after a relative pronoun to the classical particle ἄν.
This same change also occurs at Gal 6:7, and the substitution of the Hellen-
istic particle ἐάν in this construction is also found at 5:10.

77
Even though Western witnesses are even divided between the singular ὅ (D* F G)
and the plural ἅ (d b ×vg), the witnesses in favor of the plural are versional and thus unlike
to convey grammatical irregularities in the Greek. For a fuller discussion of this variant,
see Chapter 3.
204 Chapter 5. A History of Textual Variation in Galatians

Gal 5:19 φανερὰ δέ ἐστιν τὰ ἔργα τῆς σαρκός, ἅτινά ἐστιν μοιχεία πορνεία
ἀκαθαρσία ἀσέλγεια
μοιχεία πορνεία D; Ψ hark 1611 Byz
μοιχεῖαι πορνεῖαι FGdb
πορνεία μοιχεία Chrys pc 78
πορνεία Marc 79 ×vg; B,  33, A C P 1241S, 1739 pesh
The addition of “adultery” (μοιχεία) before “fornication” (πορνεία) reinforces
the list of the deeds of the flesh with a serious sexual transgression, 80 perhaps
influenced by their collocation in other vice lists (e.g. Matt 15:19; Mark
7:21–22). 81 The plural form μοιχεῖαι in F G d b is a later development as-
sociated with making other entries on the list plural.

Gal 5:21a φθόνοι φόνοι μέθαι κῶμοι καὶ τὰ ὅμοια τούτοις,

φόνοι D F G d b (×vg homicidia); A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
− Marc 82; P46 B,  33, P pc 83
This addition is an unusual Western-Eastern branch agreement against the
central, basic members of the stemma. Although it is possible that the word
φόνοι (“murders”) could have been accidentally omitted after its rhyme
φθόνοι (“envies”) among these basic members, 84 it is more likely that such a
serious offense was added to the list of fleshy deeds in order to strengthen the
list, just as with the addition of “adultery” at the beginning of the list (v.19). 85

Gal 5:21b ἃ προλέγω ὑμῖν, καθὼς προεῖπον προείρηκα ὅτι . . .

προείρηκα DFG
προεῖπον Marc 86; P46 B, (* εἶπον) 33, A C 1241S,
(1739 προείπομεν) Ψ Chrys pesh hark 1611 Byz
The aorist προεἶπον can have either a temporal sense, “I said beforehand,” or
a locative sense, “I said before [the public], i.e., I declare.” 87 Although the
preceding present tense form προλέγω must have the locative, declaratory
sense of the prefix, the force of the aorist προεἶπον is ambiguous. The change

78
That is, 1270 1319 1573.
79
Schmid, Marcion, 179, citing Epiphanius, Pan. 42.11.8 and 42.12.3.
80
BUSCEMI 527.
81
BETZ 283 n.106; LONGENECKER 248 n.a.
82
Schmid, Marcion, 192, however, argues that this reading in Marcion is uncertain
because of variants in the quotations by Epiphanius.
83
Including 323 876 945.
84
Indeed, this probably accounts for the omission in some Byzantine manuscripts.
85
See the fuller discussion in Chapter 3.
86
Schmid, Marcion, 163, citing Epiphanius, Pan. 42.11.8 and 42.12.3.
87
LSJ “προεἶπον,” 1476, although BDAG “προεἶπον,” 867–868 only gives temporal or
prophetic meanings.
 5.2 The Western Branch and Prototype 205

to the perfect tense προείρηκα (cf. 2 Cor 13:2 προείρηκα καὶ προλέγω)
buttresses the declaratory sense by implying that what Paul had said is still in
effect.

Gal 5:23 πραΰτης ἐγκράτεια ἁγνεία· κατὰ τῶν τοιούτων οὐκ ἔστιν νόμος.
ἁγνεία D* F G b d
− ×vg; P46 B,  33, A C P 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
Similar to the addition of the vices of adultery and murder to the deeds of the
flesh in vv.21 and 23, respectively, this addition appends “chastity” (ἁγνεία)
to the list of the fruits of the spirit. 88

Gal 5:25 Εἰ πνεύματι ζῶμεν πνεύματι, πνεύματι καὶ στοιχῶμεν.

πνεύματι ζῶμεν DFGd
ζῶμεν πνεύματι b ×vg; P46 B,  33, A C P 1241S,
1739 Ψ Chrys pesh hark 1611 Byz
The position of the words πνεύματι (“by the Spirit”) and ζῶμεν (“we live”)
affect the relative prominence and hence the implied contrasts the two con-
stituents have. If πνεύματι is first, then the contrast is with the flesh, as it had
been expounded in the contrast of vv.19–21 and vv.22–23: “If it is by the
Spirit [rather than by the flesh] we’re living, . . .” If ζῶμεν is first, then the
contrast is between mere living (ζῶμεν) and walking in line (στοιχῶμεν). 89 As
both Paul’s argument and the καί before the verb στοιχῶμεν are ascensive,
then placing the verb first in the apodosis is more fitting to the context. 90
Thus, this transposition of the Western appears to be a superficial harmo-
nization to its earlier context or to the same order of the apodosis. It also
avoids having the word πνεύματι written twice in a row.

88
BETZ 288 n.162 notes that it “concludes the list with a religious rather than philo-
sophical concept.”
89
The notion of living is also anticipated by the statement of crucifying the flesh in
v.24.
90
The fronting of πνεύματι in the apodosis provides an emphatic point of departure: “it
is with the [same] Spirit that we walk in line.”
206 Chapter 5. A History of Textual Variation in Galatians

Gal 6:1 σκοπῶν σεαυτὸν μὴ καὶ σὺ αὐτὸς πειρασθῇς.

αὐτὸς πειρασθῇς D*
αὐτὸς πειρασθῇ F G (d ipse temptetur)
σὺ πειρασθῇς b ×vg; P46 B,  33, A C 1241S,
1739 Ψ Chrys pesh hark 1611 Byz
The change of the personal pronoun σύ to an intensive pronoun αὐτός appears
to be a harmonization with the preceding σεαυτόν. 91 The apparent mismatch
in person in D*’s text was further modified in favor of the third person in the
later Western stream of text.

Gal 6:3 εἰ γὰρ δοκεῖ τις εἶναί τι μηδὲν ὤν, ἑαυτὸν φρεναπατᾷ ἑαυτόν.
ἑαυτὸν φρεναπατᾷ D F G d b ×vg; Ψ pesh hark 1611 Byz
φρεναπατᾷ ἑαυτόν P46 B,  33, A C 1241S, 1739 Chrys
The fronting of the reflexive pronoun ἑαυτόν emphasizes the self-deception
of the someone who falsely considers himself important. This emphasis
makes the statement more self-contained and appropriate as a proof text.

Gal 6:7 ὃ γὰρ ἐὰν ἂν σπείρῃ ἄνθρωπος, τοῦτο ταῦτα καὶ θερίσει·
ὃ ἄν D* F G; B
ἃ ἄν (d b ×vg quae)
ὃ ἐάν  33, A C P 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
ἃ ἐάν P46

ταῦτα D* F G d b ×vg; P46

τοῦτο B,  33, A C P 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
The first variant is just like 5:10 and 5:17, where the Hellenistic particle ἐάν
after a relative pronoun has been substituted with a more classical ἄν. The
second variant changes the singular resumptive pronoun τοῦτο to the plural
ταῦτα, creating a lack of number concord with the pendent nominative ante-
cedent. 92 A similar lack of agreement is apparent at 5:17.

91
A bare intensive αὐτοί is used with second-person plural verbs in Acts 18:15, Rom
15:14, and 1 Thess 3:3.
92
BUSCEMI 568.
 5.2 The Western Branch and Prototype 207

Gal 6:8 ὅτι ὁ σπείρων εἰς τὴν σάρκα ἑαυτοῦ αὐτοῦ, ἐκ τῆς σαρκὸς αὐτοῦ θερίσει
φθοράν
σάρκα αὐτοῦ D* F G; 1241S, Ψ pc 93
σάρκα ἑαυτοῦ B,  33, A C, 1739 Chrys 1611 Byz

σαρκὸς αὐτοῦ D* F G d
σαρκὸς − b ×vg; B,  33, A C P 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
These changes produce a more parallel construction.

Gal 6:11 Ἴδετε πηλίκοις ὑμῖν γράμμασιν ὑμῖν ἔγραψα τῇ ἐμῇ χειρί.
γράμμασιν ὑμῖν DFGd
ἔγραψα ὑμῖν (b ×vg scripsi vobis)
ὑμῖν γράμμασιν P46 B,  33, A C 1231S, 1739 Ψ Chrys 1611 Byz
The personal pronoun ὑμῖν appears to have been transposed to keep the
phrase πηλίκοις γράμμασιν together or, perhaps, to confer extra emphasis on
πηλίκοις (cf. a similar word order in Matt 25:20 Κύριε, πέντε τάλαντά μοι
παρέδωκας, with a strong emphasis on the number of talents).

Gal 6:15 ἐν γὰρ Χριστῷ Ἰησοῦ οὔτε γὰρ περιτομή τί ἐστιν οὔτε ἀκροβυστία ἀλλὰ
καινὴ κτίσις.
ἐν γὰρ Χριστῷ Ἰησοῦ οὔτε D F G d b ×vg; , A C 1241S, hark Byz
οὔτε γάρ P46 B, 33, Ψ Chrys pesh 1611
The insertion of the phrase ἐν Χριστῷ Ἰησοῦ, with the adjustment of the posi-
tion of the particle γάρ, is an apparent harmonization to Gal 5:6 ἐν γὰρ
Χριστῷ Ἰησοῦ οὔτε περιτομή τι ἰσχύει οὔτε ἀκροβυστία. 94

Gal 6:16a καὶ ὅσοι τῷ κανόνι τούτῳ στοιχήσουσιν στοίχουσιν,

στοίχουσιν D F G; A C, 1739 Chrys
στοιχήσωσιν P46
στοιχήσουσιν B,  33, 1241S, Ψ 1611 Byz
The change in tense from the future στοιχήσουσιν to the present στοίχουσιν
appears to lend Paul’s final benediction more contemporary relevance. 95

93
Also 056 075 618 1424 1505–2495 1648 1780 1837 2147 2464 2815.
94
BETZ 319 n.77; BUSCEMI 598.
95
BUSCEMI 598.
208 Chapter 5. A History of Textual Variation in Galatians

Gal 6:16b εἰρήνη ἐπ’ αὐτοὺς καὶ ἔλεος καὶ ἐπὶ τὸν Ἰσραὴλ τοῦ θεοῦ κυρίου.
κυρίου DFGd
θεοῦ b ×vg; P46 B,  33, A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz
Taken out of context, the change from “Israel of God” to “Israel of the Lord”
would not be anti-Judaic because Israel’s God was routinely called the
“Lord.” In this particular context, with the addition of “Lord” to Jesus Christ
in the following verse, however, this substitution does have an anti-Judaic
theological effect of subjugating Israel to the Lord Jesus Christ.

Gal 6:17 ἐγὼ γὰρ τὰ στίγματα τοῦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ ἐν τῷ σώματί μου
βαστάζω.
κυρίου ἡμῶν Ἰησοῦ Χριστοῦ D* F G; Chrys
κυρίου Ἰησοῦ Χριστοῦ d b; 
κυρίου Ἰησοῦ 1739 hark 1611 Byz
κυρίου ἡμῶν Ἰησοῦ pesh
Χριστοῦ Marc, 96 P, Ψ
Ἰησοῦ ×vg; P46 B, 33, A C 1241S
The bare use of the name Jesus here was particularly susceptible to expansion
among the manuscripts, and the fullest forms are found in the Western
branch. A theological effect of these additions is anti-Separationist because it
makes it clear that Jesus as Christ the Lord has the marks (στίγματα) of the
cross. 97

Seven of these Western textual changes are also found in Marcion (2:9; 3:14;
4:6, 4:8; 5:14a, 14c; and 6:17), but Marcion’s text lacks some later changes of
the Western branch (4:10, 5:19, 5:21a, 5:21b, 6:17). These facts suggest that
Marcion’s base text came from an early point in the stream of the textual
tradition that survived in the Greek-Latin diglots and some Old Latin versions
of Galatians.
Many of the textual changes are theological in effect. The epistle of Gala-
tians covers both Christological and Judaic topics, but the textual variation is
not evenly distributed. The largest group of potentially theologically signi-
ficant changes (not all equally plausible, however) are those which are anti-
Judaic in effect (3:17, 3:19, 4:21, 5:1b, 5:14a, 5:14c, 6:16b, less likely also
2:9). The largest group of these changes concern the role of the Torah (ὁ νο-
μός) in the text of Galatians, and these changes denigrate it. For example, at
3:17 the addition of the words εἰς Χριστὸν (“for Christ”) implies that only
Christ, not the law, was the fulfillment of God’s promise to Abraham. At

96
Schmid, Marcion, 64–65, cautions that Tertullian’s quoting practice may not abso-
lutely reflect Marcion’s use of Christological titles.
97
Ehrman, Orthodox Corruption, 162.
 5.2 The Western Branch and Prototype 209

3:19, the change of the verb to ἔτεθη (“was established”) implies that the Law
was no longer established after Christ. The word order change at 5:1 gives
more emphasis to the slavery aspect of the Law. Even when the Law is
quoted favorable at Gal 5:14, one textual change makes the Levitical love
commandment fulfilled among Gentiles and another change, the loss of the
explicit quotation signal ἐν τῷ, divorces its connection with the Torah.
Less common are those variants that anti-Judaic pertaining to Jews as a
people, but there are two. At Gal 4:21, the change from those who hear the
Law to those who read the Law has been viewed, certainly in the third
century, as applying to Jews, who read from the Torah. This change has the
effect of applying the negative part of the allegory in vv.21–31 to Jews as a
people. Another change occurs at 6:16 where the phrase “Israel of God” is
changed to “Israel of the Lord,” which is Jesus Christ in context. This seems
to have effect of making Paul’s blessing subjugate the nation of Israel to the
Lord Jesus Christ.
Somewhat surprisingly, considering the Christocentric focus of the epistle,
there are relatively few Christologically charged textual variants. To be sure,
some appear to be anti-Separationist (3:24, and 6:17) and some are anti-
Demiurgic (1:4, 3:11a, and 3:17); one is anti-pagan in effect (4:8). Yet these
are fairly insignificant compared to the intensity of the anti-Judaic or at least
anti-Torah tendencies of the textual variation in the Western branch.
There are other substantive changes to the text. Several changes strengthen
the portrayal of Paul in the letter (1:19, 1:24, 3:2, 4:16, 5:11). Peter, too, sees
his importance enhanced within the Western text (2:9), and the replacement
of the name Cephas by Peter (1:18, 2:9, 2:11, 2:14) may be related to this.
Some are additions to the vice and virtue lists (5:19, 5:21a, 5:23). Some de-
contextualize a verse to broaden its applicability, perhaps for homiletical
purposes (1:8, 2:6, 5:15, 6:3). Others simply enhance the contrast already
inherent in the text (4:9, 6:11).
Many of the changes foster a smoother, clearer text. Several resolve ap-
parent ambiguities in the text (1:4, 3:1, 3:10, 3:11b, 4:18b, 4:23, 5:1, 5:21b).
Some changes avoid unusual or non-classical constructions (2:10, 3:7, 4:18c,
4:27, 5:10, 5:14b, 5:15a, 5:15b, 5:17, 6:7, 6:8). A few attempt to clarify
obscure portions of the text (1:11, 4:10, 6:16b). Harmonization is also a
frequent textual change, mostly to the immediate context (3:14, 3:29a, 3:29b,
4:6, 5:13, 5:25, 6:1,), but also to the LXX (3:10, 4:30), to another Pauline
epistle (4:17 to 1 Cor 12:31), or to another page of Galatians (6:15 to 5:6).
Other changes are hard to classify. Four involve avoidance of the particle
δέ (3:16, 4:13, 4:18a, 4:25). Two could be mechanical errors (3:21, 5:3). One
orthographic change may have been due to Latin influence (5:12), and an-
other involves crasis (2:8). The effect of the word order change at 2:1 is pre-
sently unclear.
210 Chapter 5. A History of Textual Variation in Galatians

5.3 The Eastern Branch and the Road to the Byzantine Text

The Eastern branch is by far the largest branch of the textual tradition of
Galatians. It not only includes the vast majority of the medieval manuscripts
known as the Byzantine text-type, it also includes witnesses traditionally
identified as “Alexandrian” plus some related to the Syriac versions. This
section traces through the textual development that led to the Byzantine text
and assesses what editorial or scribal activities were dominant at each stage
of the process.

ω
•
Western • Eastern ε
Branch β Branch
Marc P46-B α ℵ 33
δ
γ A-C-P-1241S
vg
ψ 1739

Byzantine
σ Ψ
Text
κ Chrys-(pesh)-hark-
K-L-226- 1611-1505-2495
547-1854 etc.

FIG. 21: Stemma of Galatians with Detail on the Rise of the Byzantine Text

As can be seen in the detail of the stemma depicted in FIG. 21, the road to the
Byzantine Text begins within the Eastern branch and goes through a long de-
velopment consisting of a number of different stages labeled α, γ, ψ, σ, and κ.
Anticipating the results of this investigation, the stemma also depicts a low-
level of contamination from the Western branch δ to stage σ in the form of a
dotted line. This link was not found by the cladistic software because it fell
below the reticulation cutoff but it is nonetheless inferred from a close textual
analysis of the variation.
 5.3 The Eastern Branch and the Road to the Byzantine Text 211

5.3.1 Stage α
Stage α is an early stage of the Eastern branch before it diverged into two
streams, an “Alexandrian” stream (comprising A C P 1241S and various other
mixed manuscripts) and a “Syrian” stream (comprising 1739, Ψ, Syriac and
Byzantine texts). 98 The leading witnesses for the Alexandrian stream are A
and C, while the leading witnesses for the Syrian stream are 1739 and Ψ. Ev-
idence for the readings introduced at stage α lies mainly in the agreement of
A and C, on the one hand, and of 1739 and Ψ on the other hand. All of these
variation units here at stage α had been discussed in detail back in Chapter 3.
Thus, the discussion here is limited to assessing the scribal tendencies of this
stage in the development of the text.

Gal 1:8 ἀλλὰ καὶ ἐὰν ἡμεῖς ἢ ἄγγελος ἐξ οὐρανοῦ ὑμῖν εὐαγγελίζηται ὑμῖν παρ’ ὃ
εὐηγγελισάμεθα ὑμῖν,
εὐαγγ. ὑμῖν A P (1241S ἡμῖν), 1739 Ψ Chrys pesh hark 1611 Byz, ×vg;
33, (D ἡμᾶς) d
εὐαγγ. Ψ; , F G b
ὑμῖν εὐαγγ. 1739; B ×1175
As discussed in greater detail in Chapter 3, this transposition makes the
statement less coupled to the context and more appropriate for a standalone
proof text.

Gal 1:17 οὐδὲ ἀπῆλθον ἀνῆλθον εἰς Ἱεροσόλυμα πρὸς τοὺς πρὸ ἐμοῦ ἀποστόλους,
ἀνῆλθον A C P 1241S, 1739 Ψ Chrys hard 1611 Byz;  33
ἀπῆλθον pesh; B, D F G
ἦλθον (×vg b d veni); P46
This substitution is an assimilation to the more usual verb in v.18 with
Jerusalem as the destination. 99

Gal 1:18 Ἔπειτα μετὰ τρία ἔτη τρία ἀνῆλθον εἰς Ἱεροσόλυμα ἱστορῆσαι Κηφᾶν
μετὰ τρία ἔτη A P 1241S, 1739 pesh;  33
μετὰ ἔτη τρία Ψ Chrys hark 1611 Byz, ×vg; P46 B, D F G d b
This transposition appears to be an assimilation to the numeral-noun order of
2:1 διὰ δεκατεσσάρων ἐτῶν. 100

98
Specifically, the mixed Alexandrian manuscripts comprise 88 (39% Alexandrian,
61% Byzantine), 104 (56% Alexandrian, 44% Byzantine), 1837 (49% Alexandrian, 51%
Byzantine), 2464 (48% Alexandrian, 52% Byzantine).
99
See discussion in Chapter 3.
100
See discussion in Chapter 3.
212 Chapter 5. A History of Textual Variation in Galatians

Gal 2:12 πρὸ τοῦ γὰρ ἐλθεῖν τινας ἀπὸ Ἰακώβου μετὰ τῶν ἐθνῶν συνήσθιεν· ὅτε δὲ
ἦλθεν ἦλθον, ὑπέστελλεν καὶ ἀφώριζεν ἑαυτὸν
ἦλθον A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz, ×vg
ἦλθεν  33, P46 B, D* F G d
As discussed in Chapter 3, this substitution, from the singular ἦλθεν to the
plural ἦλθον, is a harmonization to the preceding ἐλθεῖν τινας.

Gal 2:20 ἐν πίστει ζῶ τῇ τοῦ υἱοῦ τοῦ θεοῦ καὶ Χριστοῦ τοῦ ἀγαπήσαντός με
τοῦ υἱοῦ τοῦ θεοῦ A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz, ×vg;  33
τοῦ θεοῦ καὶ Χριστοῦ P46 B, D* F G d (b)
This modification may have been done to avoid a possible Patripassionist in-
terpretation. For further discussion, see Chapter 3.

Gal 4:3 ὅτε ἦμεν νήπιοι, ὑπὸ τὰ στοιχεῖα τοῦ κόσμου ἤμεθα ἦμεν δεδουλωμένοι·
ἦμεν A C 1241S, 1739 Ψ Chrys 1611 Byz; B
ἤμεθα  33, P46 B, D* F G
This is a clear assimilation to the more classical active voice form ἦμεν ear-
lier in the verse.

Gal 4:19 τέκνα τεκνία μου, οὓς πάλιν ὠδίνω ἄχρις οὗ μορφωθῇ Χριστὸς ἐν ὑμῖν·
τεκνία A C 1241S, Ψ Chrys 1611 Byz, ×vg; 33 b
τέκνα 1739; , B, D* F G d
As discussed in Chapter 3, this substitution of τέκνα for the diminutive
τεκνία, perhaps inspired by Paul’s birthing imagery (ὠδίνω “I am groaning in
labor”), has the effect of aggrandizing Paul.

Gal 4:28 ὑμεῖς ἡμεῖς δέ, ἀδελφοί, κατὰ Ἰσαὰκ ἐπαγγελίας τέκνα ἐστέ ἐσμέν.
ἡμεῖς . . . ἐσμέν A C P 1241S, Ψ Chrys pesh hark 1611 Byz, ×vg; 
ὑμεῖς . . . ἐστέ 1739; 33, P46 B, D* F G d b
This substitution appears to be a harmonization to v. 26 or v. 31 (οὐκ
ἐσμέν), 101 and it has the effect of explicitly including Paul among the children
of the promise according to Isaac.

101
METZGER 528.
 5.3 The Eastern Branch and the Road to the Byzantine Text 213

Gal 5:17 ταῦτα γὰρ δὲ ἀλλήλοις ἀντίκειται, ἵνα μὴ ἃ ἐὰν θέλητε ταῦτα ποιῆτε.
δέ A C P 1241S, 1739 Ψ Chrys (pesh καί) hark 1611 Byz
γάρ ×vg;  33, P46 B, D* F G d b
As discussed in Chapter 3, changing the connective particle avoids having a
consequence (the ἴνα-clause) in what appears to be a confirmatory statement
(signaled by γάρ).

Gal 5:21 φθόνοι φόνοι μέθαι κῶμοι καὶ τὰ ὅμοια τούτοις,

φόνοι A C 1241S, 1739 Ψ Chrys pesh hark 1611 Byz, (×vg homicidia); D F G d b
− P pc;  33, P46 B
The addition of murders (φόνοι) strengthens the vice list. For further
discussion, see Chapter 3.

Gal 5:24 οἱ δὲ τοῦ Χριστοῦ Ἰησοῦ τὴν σάρκα ἐσταύρωσαν

Χριστοῦ Ἰησοῦ A C 1241S, 1739 Ψ; (* κυρίου χ. ι.) 33, B
Χριστοῦ P, Chrys pesh hark 1611 Byz, ×vg; P46, D F G d b
The addition of “Jesus” to “Christ” belongs to the earliest stages of the East-
ern branch because it is found in the leading members of both Eastern sub-
branches (A–C and 1739–Ψ) as well in their closest relatives (, 33 and B).
Both expansion and contraction of references to Christ are found in the
tradition; indeed, there are Eastern branch members that omit the “Jesus”
(e.g., P, Byz).

Gal 6:2 Ἀλλήλων τὰ βάρη βαστάζετε καὶ οὕτως ἀναπληρώσετε ἀναπληρώσατε τὸν
νόμον τοῦ Χριστοῦ.
ἀναπληρώσατε A C 1241S, 1739 Ψ Chrys hark 1611 Byz;  33, D
ἀναπληρώσετε P46 B ×pesh, F G b ×vg
This change appears to be a scribal harmonization to the imperative mood of
the preceding verb βαστάζετε.

Twelve textual variants have been identified and analyzed for this stage of the
transmission of the text. Compared to other places in the tradition, this num-
ber of textual variants is fairly low; for example, it is about half of the textual
distance between B and its common ancestor with P46. Nevertheless, the
scribes who transmitted the text up to this stage are not perfect. They tend to
be influenced by textual harmonization or to make improvements in the
theological substance of the text.
Textual harmonization appears to be a common source of textual variation
at this stage. In particular, at 1:17 the non-standard verb ἀπῆλθον is harmo-
nized to more common ἀνῆλθον in v.18. At 1:18 the numeral-noun word
214 Chapter 5. A History of Textual Variation in Galatians

order is assimilated to that of 2:1. At 2:12, the singular ἦλθεν was conformed
to the plural under the influence of ἐλθεῖν τινας earlier in the verse. At 4:3,
the non-classical middle imperfect ἤμεθα was assimilated to the preceding ac-
tive form. At 5:17 the change in the connective particle could be a local har-
monization, though it could also have been done to improve the flow of the
discourse superficially. At Gal 6:2 the mood of the verb appears to be a har-
monization to the mood of the preceding verb.
Some apparent harmonizations also perform other functions. For example,
the change in the personal pronoun and corresponding verb from the second-
person plural to the first-person plural in 4:28 could well be a harmonization
based on v.31, but it could also have been motivated to make Paul’s inclusion
among the children of the promise explicit. The addition of “Jesus” to
“Christ” at 5:24 could have been influenced by similar collocations through
the epistle or as part of a scribal tendency to toy with these titles.
Nevertheless, the non-harmonistic changes seem theologically or homileti-
cally motivated. For example, the change in the position of the pronoun at
Gal 1:8 makes the statement more suitable for a standalone proof text. At
2:20, the change from “God and Christ” to “Son of God” appears to be theo-
logically motivated (particularly anti-Patripassionist). Another change, to the
diminutive τεκνία in 4:19, has the effect of increasing Paul’s stature. The ad-
dition of murders (φόνοι) to the vice list of Gal 5:19–21 appears theologically
motivated as it strengthens the list of the deeds of the flesh.

5.3.2 Stage γ
Stage γ relates to state of the text for the common ancestor of 1739 and the
Byzantine text. According to the stemma, other descendants of this common
ancestor include the Athos Codex Ψ, John Chrysostom’s text in his com-
mentary on Galatians, and the Syriac (Peshitta and Harklean) versions.

Gal 2:4 κατασκοπῆσαι τὴν ἐλευθερίαν . . . ἵνα ἡμᾶς καταδουλώσουσιν

καταδουλώσωσιν,
καταδουλώσωσιν 1739; 33, P46, F G b d ×vg
καταδουλώσωνται Ψ 1611 Byz
καταδουλώσουσιν A C 1241S, , P46 B, D
The use of the future indicative after ἵνα is non-classical, 102 so it not
surprising to see a variety of unrelated witnesses all modify the form of the
verb to the more classical aorist subjunctive. The change in voice from the
active to the middle occurs at a later stage on the road to the Byzantine text.

102
BDF 186–187 § 369(2).
 5.3 The Eastern Branch and the Road to the Byzantine Text 215

Gal 2:16 εἰδότες δὲ ὅτι οὐ δικαιοῦται ἄνθρωπος ἐξ ἔργων νόμου

δέ Chrys; C 1241S, , B, D* F G b d ×vg
− 1739 Ψ pesh hark 1611 Byz; A P, 33, P46
As discussed in greater detail in Chapter 3, the motivation for the omission of
the connective particle δέ is difficult to determine. It has the effect of creating
a closer connection of this participial phrase to the preceding statement. It is
also possible that the omission occurred earlier at stage α and later reversed,
but it was felt that transcriptionally it is easier to have multiple omissions of
little particles.

Gal 3:13 ὅτι γέγραπται γάρ· ἐπικατάρατος πᾶς ὁ κρεμάμενος ἐπὶ ξύλου,
γέγραπται γάρ 1739 Ψ Chrys pesh hark 1611; 
ὅτι γέγραπται A C 1241S, 33, P46 B, D* F G
This changes does not affect the sense, for both ὅτι and γάρ are causal
conjunctions. Even so, it does appear to be a harmonization to the γέγραπται
γάρ formula, most recently found in v.10.

Gal 4:14 καὶ τὸν πειρασμὸν ὑμῶν τὸν ἐν τῇ σαρκί μου οὐκ ἐξουθενήσατε
τόν 1739 Ψ Chrys hark 1611 Byz; C P 1241S
− A,  33, P46 B, D* F G
The agreement of 1739–Ψ–Byz with C–P–1241S against A might mean that
the addition of the article occurred at an earlier stage and reversed in A, or it
could mean that it was added twice independently. Without the article, the as-
sociation of the prepositional phrase ἐν τὴν σαρκί μου is syntactically ambig-
uous. On the one hand, it could be construed with the preceding τὸν πειρα-
σμὸν ὑμῶν with the meaning “and you did not disdain your trial in my flesh.”
On the other hand, it could be construed with the following verb οὐκ ἐξουθε-
νήσατε with the meaning “as for your trial, you did not disdain it with my
flesh.” Both constructions of the phrase are difficult, but only the former is
sensical. 103 The addition of the article resolves this ambiguity in favor of the
former possibility.

103
LONGENECKER 191 suggests that it means something like, “and though my illness
was a temptation for you [to reject me], you did not despise or disdain me,” by taking ἐν τῇ
σαρκί μου as a reference to δι’ ἀσθένειαν τῆς σαρκὸς in the previous verse.
216 Chapter 5. A History of Textual Variation in Galatians

Gal 5:1 τῇ ἐλευθερίᾳ ἡμᾶς Χριστὸς ἡμᾶς ἠλευθέρωσεν· στήκετε οὖν καὶ μὴ πάλιν
ζυγῷ δουλείας ἐνέχεσθε.
Χριστὸς ἡμᾶς 1739 Chrys hark 1611 Byz; C P 1241S, Marc
ἡμᾶς Χριστός A,  33, B, D* F G d b ×vg
Both modern-day exegetes and late antique scribes have been puzzled about
how to understand this clause, and this transposition here is but the beginning
of a cascade of changes to make this difficult verse more comprehensible. In
particular, the critical text of Gal 5:1 has three features that have hindered the
interpretation of this verse: its asyndeton, its dative phrase τῇ ἐλευθερίᾳ, and
its unusual word order. First, the asyndeton of the sentence is ambiguous as
its connection to the discourse; this sentence could be either closely connec-
ted to the preceding verse, 104 or it could be the start of a new section of the
letter. 105 Second, the dative phrase τῇ ἐλευθερίᾳ is difficult to construe, and
scholars are conflicted as to whether the dative here signify means, 106 man-
ner, 107 advantage, 108 destination, 109 or goal. 110 Third, the verb-final word
order is unusual and highly marked, so it is necessary to account for the
position of the various pre-verbal constituents in the sentence. 111 All of these
possibilities are interrelated and a decision on one issue tends to affect the
decision on another.
The word order of Gal 5:1a nicely fits the pragmatic “Topic – Focus –
Verb – Remainder” template identified for ancient Greek. 112 In this case, τῇ

104
Those who connect Gal 5:1 with 4:31 include: de Boer, Galatians, 290; Pierre Bon-
nard, L’Épître de Saint Paul aux Galates (2d ed.; Commentaire du Nouveau Testament 9;
Neuchâtel: Delachaux et Niestlé, 1972), 101; BRUCE 226; LÉGASSE 342–343; MARTYN
468;
105
Those who consider Gal 5:1 to be the start of a new section include: BETZ 255;
BORSE 178; BUSCEMI 493–494; Dunn, Galatians, 261; LONGENECKER 223–224; MATERA
180; MUSSNER 342; ROHDE 212; SCHLIER 163;
106
That is, “by or with freedom.” So BRUCE 226; BURTON 271, “by (bestowing) the
freedom (spoken above).”
107
This is, “freely,” used intensively. So Bonnard, Galates, 101–102, “pour la liberté.”
108
That is, “for (the benefit of) freedom.” So BDR 152 § 188 n.1; ROHDE 212, “zur
Freiheit.”
109
That is, “into freedom.” So MARTYN 447, “into the realm of freedom.”
110
That is, “for freedom, in order to be free.” So BETZ 255–256; BORSE 178, “zur
Freiheit”; BUSCEMI 503, “per la libertà”; de Boer, Galatians, 309 n.443; Dunn, Galatians,
262; LÉGASSE 367, “en vue de l’acquisition de la liberté”; LONGENECKER 224; MATERA
180; MUSSNER 342–343, “als Ziel die Freiheit des Menschen”; SCHLIER 163, “zur
Freiheit.”
111
The default word order in Koine Greek is verb first, implying that preverbal
elements are significant. See BDF 248 § 472.
112
Helma Dik, Word Order in Ancient Greek: A Pragmatic Account of Word Order
Variation in Herodotus (ASCP 5; Amsterdam: J. C. Gieben, 1995), 12; Stephen H. Lev-
insohn, “Self-Instruction Materials on Non-Narrative Discourse” (SIL International, 2008),
 5.3 The Eastern Branch and the Road to the Byzantine Text 217

ἐλευθερίᾳ corresponds to the topic – what the sentence is about – and the
article here indicates that it is anaphoric, pointing back to the freedom of the
previous verse, namely, being a descendant of the free woman. In addition,
Χριστός corresponds to the focus – the most informative part of the sentence
– and the idea is that Christ, rather than someone or something else, is the one
who liberates. The placement of the unemphatic pronoun ἡμᾶς, on the other
hand, as a supportive constituent is determined by prosodic concerns, and it
gravitates to the so-called Wackernagel “second position,” after the initial,
stressed word of its intonation unit or colon. 113 Thus, the placement of the
pronoun ἡμᾶς suggests that the entire clause is a single intonation unit, and
that the purpose of the pre-verbal placement of the topic τῇ ἐλευθερίᾳ is to
switch the center of attention from the subject of 4:31 (“us”) to the state of
freedom. 114 Neither the asyndeton nor the topicalized dative phrase here
indicates how strong of a break in the discourse lies between 4:31 and 5:1.
The word order of this transposition is less ambiguous. By placing ἡμᾶς
after Χριστός, it marks Χριστός as the first constituent of its intonation unit,
and it puts the dative phrase τῇ ἐλευθερίᾳ into its own intonation unit. This
change in the prosody of the sentence, signaled by the position of the unem-
phatic ἡμᾶς, converts the dative phrase into a “strong topic,” 115 and it has the
effect of announcing a greater discontinuity in the subject of the discourse. 116
As a result, before the transposition of the pronoun ἡμᾶς, the word order of
the sentence was ambiguous on the surface as to where the break in the
discourse occurs, either right before or right after 5:1. After the transposition,
the word order more clearly indicates that Gal 5:1 begins a new section of the
letter. Thus, the transposition has the effect of reducing the ambiguity of this
difficult verse, especially in understanding how this verse relates structurally
to the rest of the letter.

43, online: http://www.sil.org/~levinsohns/NonNarr.pdf, accessed June 5, 2010; Steven E.

Runge, A Discourse Grammar of the Greek New Testament: A Practical Introduction for
Teaching & Exegesis (Lexham Bible Reference Series; Bellingham, Wash.: Logos
Research Systems, 2010), 226–239, esp. 228.
113
Jacob Wackernagel, “Über ein Gesetz der indogermanischen Wortstellung” in Jacob
Wackernagel, Kleine Schriften (Göttingen: Vandenhoeck & Ruprecht, 1953), 1–104;
Eduard Fraenkel, “Kolon und Satz: Beobachtungen zur Gliederung des antiken Satz, I” in
Eduard Fraenkel, Kleine Beiträge zur klassischen Philologie (vol. 1; Raccolta di studi e
testi 95; Rome: Edizioni di storia e letteratura, 1964), 73–92; idem, “Kolon und Satz: Be-
obachtungen zur Gliederung des antiken Satz, II” in Fraenkel, Kleine Beiträge, 93–130.
For a treatment informed by recent developments in functional linguistics, see David
Michael Goldstein, “Wackernagel’s Law in Fifth-Century Greek” (Ph.D. diss., University
of California, Berkeley, 2010).
114
Levinsohn, “Self-Instruction Materials,” 49–50.
115
See Goldstein, “Wackernagel’s Law,” 81–82 and 121–122.
116
See Goldstein, “Wackernagel’s Law,” 142.
218 Chapter 5. A History of Textual Variation in Galatians

Gal 5:14 ἀγαπήσεις τὸν πλησίον σου ὡς σεαυτόν ἑαυτόν.

ἑαυτόν 1739 Chrys 1611 Byz; 1241S, P46
σεαυτόν A C,  33, B, D* F G
The loss of the initial aspirate and doubled consonants in later Greek pronun-
ciation means that these two variants were pronounced the same by the fourth
century. 117 Thus, this change could merely be orthographic, a simple haplo-
graphy of the sigma, 118 or a reflection of the practice – already begun during
Roman times but more common in later Greek – of using forms of ἑαυτοῦ for
the first and second person reflexive pronouns. 119

Gal 6:9 τὸ δὲ καλὸν ποιοῦντες μὴ ἐγκακῶμεν ἐκκακῶμεν, καιρῷ γὰρ ἰδίῳ θερίσομεν
μὴ ἐκλυόμενοι.
ἐκκακῶμεν 1739 Ψ Chrys 1611 Byz; C 1241S, (F G ἐκκακήσωμεν)
ἐγκακῶμεν A P,  33, B, D* (×vg d b deficiamus)
The semantic difference between ἐγκακέω (“lose enthusiasm, be discour-
aged”) and ἐκκακέω (“lose heart”) is about as subtle as their difference in
spelling. 120 The variant reading μὴ ἐκκακῶμεν appears to be occasioned by a
phonetic or graphic harmonization to μὴ ἐκλυόμενοι (“get tired, give out”) in
the following clause. 121 The presence of this reading in C and 1241S is con-
sistent with the idea that it was made at an earlier stage α but reversed for A
and P, but the transcriptional unlikelihood of the reversal suggests that this
stage is one of multiple occasions for this error.

Gal 6:14 εἰ μὴ ἐν τῷ σταυρῷ τοῦ . . . Χριστοῦ, δι’ οὗ ἐμοὶ κόσμος ἐσταύρωται κἀγὼ
τῷ κόσμῳ.
τῷ 1739 Ψ Chrys 1611 Byz; 1241S
− A C,  33, P46 B, D* F G
The omission of the article for uniquely identifiable referents, such as the
world, serves to lend them additional prominence in the sentence. 122 Without
the article on κόσμῳ, there are two constituents in the final verbless 123 clause

117
Specifically, both ὡς σεαυτόν and ὡς ἑαυτόν would be pronounced as [oseaf'ton].
See Allen, Vox Graeca, 53; Horrocks, Greek, 113.
118
BRUCE 239; BUSCEMI 527.
119
Robertson, Grammar, 688–689. Similarly, BRUCE 239; BUSCEMI 527. See also BDF
35 § 64(1).
120
Compare BDAG “ἐγκακέω,” 272–273, with BDAG “ἐκκακέω,” 303. BURTON 344
considers the two forms to be synonymous.
121
Pace BUSCEMI 568, who attributes the change to conformation with later post-New
Testament usage.
122
Levinsohn, Discourse Features, 162–165. So also BURTON 355: “The use of κόσμος
and κόσμῳ without the article gives to both words a qualitative emphasis.”
123
Presumably, there is an ellipsis of the verb ἐσταύρωμαι.
 5.3 The Eastern Branch and the Road to the Byzantine Text 219

that could be emphatic: the world or Paul himself, due to the explicit personal
pronoun κἀγώ. If the emphasis falls upon κόσμῳ in the clause, then Paul’s re-
lationship to the world would be highlighted twice, first that the world has
been crucified to him, and then it was to the world that Paul has been cruci-
fied. With the article on κόσμῳ, the information structure of the clause is less
ambiguous – the emphasis falls on Paul via the term κἀγώ. Thus, this change
has the effect of lending more prominence to Paul and reducing the ambiguity
of the clause.

Gal 6:17 ἐγὼ γὰρ τὰ στίγματα τοῦ κυρίου Ἰησοῦ ἐν τῷ σώματί μου βαστάζω.
κυρίου Ἰησοῦ 1739 hark 1611 Byz
κυρίου ἡμῶν Ἰησοῦ Χριστοῦ D* F G; Chrys
κυρίου Ἰησοῦ Χριστοῦ ; d b
κυρίου ἡμῶν Ἰησοῦ pesh
Χριστοῦ Ψ; P, Marc
Ἰησοῦ A C 1241S, ×vg, 33, P46 B
Although the addition of κυρίου is attested in several other witnesses in other
branches of the textual transmission, this change appears to be independent of
them, perhaps suggested by the use of κύριος (albeit in longer formulae) in
vv.14 and 18. 124 This “edifying expansion” 125 does not have quite the anti-
Separationist theological effect of the Western branch, which adds “Christ” to
the text.
Some of the changes made by stage γ relate more to the form of the Greek
than its meaning. For example, the change at 2:4 in the mood of the verb after
ἵνα is Atticizing, but the change at 5:14 appears to be in the direction of a
more vernacular Greek. Hardly affecting the meaning are the harmonizations
at Gal 3:12, 6:9, and 6:17.
In at least three cases, the introduced textual variants have the effect of
reducing the ambiguity of the text. At Gal 4:14, the placement of the article
disambiguates how a prepositional phrase is to be construed; the word order
change at Gal 5:1 affects how the asyndeton is to be understood; and the
addition of the article at Gal 6:14 unambiguously places the prominence on
Paul. More difficult to assess, however, is the omission of the particle δέ in
2:16. This could have been done to have a closer connection to the previous
verse, or it could just be accidental.
Among these ten textual changes, there is little evidence of intentional
interference with the substance of the text. The relative paucity of textual
changes and the lack of substantive modifications of the text’s meaning even
among these changes suggest that the stage that has produced the common

124
BUSCEMI 599.
125
METZGER 530.
220 Chapter 5. A History of Textual Variation in Galatians

ancestor of 1739 and the Byzantine text was carefully copied, with only
minimal and usually clarifying editorial intervention into the text.

5.3.3 Stage ψ
Stage ψ relates to the text of the common ancestor of the Athos Codex Ψ and
the Byzantine text. Also included here are the text of John Chrysostom and
the Syriac versions.

Gal 1:4 ὅπως ἐξέληται ἡμᾶς ἐκ τοῦ αἰῶνος τοῦ ἐνεστῶτος αἰῶνος πονηροῦ
τοῦ ἐνεστῶτος αἰῶνος πονηροῦ Ψ Chrys (hark) 1611 Byz; D F G d b ×vg
τοῦ αἰῶνος τοῦ ἐνεστῶτος πονηροῦ 1739, A P 1241S,  33, P46 B ×pesh
As discussed earlier with respect to the Western prototype, this transposition
changes the phrase from the syntactically ambiguous τοῦ αἰῶνος τοῦ ἐνεστῶ-
τος πονηροῦ (either “the age of the present evil one” or, more usually, “the
present age of evil”) to the unambiguous τοῦ ἐνεστῶτος αἰῶνος πονηροῦ “the
present age of evil,” with an anti-Gnostic or Marcionite theological effect of
foreclosing a possible interpretation that the present age is under the rule of
an evil demiurge. The agreement here with the Western branch of the text
could be coincidental or it could reflect an awareness and adoption of the
Western reading.

Gal 1:18a Ἔπειτα μετὰ τρία ἔτη τρία ἀνῆλθον εἰς Ἱεροσόλυμα
μετὰ ἔτη τρία Ψ Chrys hark 1611 Byz, ×vg; P46 B, D F G d b
μετὰ τρία ἔτη A P 1241S, 1739 pesh;  33
This change reverses the transposition at an earlier stage of the transmission
of the text, in assimilation to the numeral-noun order of 2:1 διὰ δεκα-
τεσσάρων ἐτῶν. Although it is possible that this change is itself an indepen-
dent harmonization to the noun-number order of ἡμέρας δεκαπέντε later in
v.18, the fact that this change too agrees with Western witnesses (cf. 1:4) may
suggest contamination as a factor for this stage of the transmission.

Gal 1:18b ἱστορῆσαι Κηφᾶν Πέτρον καὶ ἐπέμεινα πρὸς αὐτὸν ἡμέρας δεκαπέντε,
Πέτρον Ψ Chrys hark 1611 Byz; P,D F G d b ×vg
Κηφᾶς 1739 ×pesh, A 1241S,  33, P46 B
This substitution replaces the Aramaic name Cephas with the more accessible
Greek name Peter, 126 though the occurrences of the name Cephas in Gal 2:11
and 14 are not substituted until the next stage. As with the changes at vv.4
and 18a, this change is also attested among Western witnesses.

126
METZGER 522.
 5.3 The Eastern Branch and the Road to the Byzantine Text 221

Gal 2:4 κατασκοπῆσαι τὴν ἐλευθερίαν . . . ἵνα ἡμᾶς καταδουλώσωσιν

καταδουλώσωνται,
καταδουλώσωνται Ψ 1611 Byz
καταδουλώσωσιν 1739; 33, P46, F G b d ×vg
καταδουλώσουσιν A C 1241S, , P46 B, D
This variant concerns the change in voice from active to middle, the change
in mood having occurred at a previous stage of the transmission. The use of
the middle voice here suggests that the false brothers who intend to enslave
Paul’s party are doing so for their own benefit, 127 that is, so that Paul’s party
would be enslaved to them. This change has the effect of heightening the self-
interest of the false brothers and to the extent that they would be identified
with Jews, this change is anti-Jewish in character. 128

Gal 2:18 εἰ γὰρ ἃ κατέλυσα ταῦτα πάλιν οἰκοδομῶ, παραβάτην ἐμαυτὸν συνιστάνω
συνίστημι.
συνίστημι Ψ Chrys 1611 Byz; 1241S
συνιστάνω 1739, A C,  33, P46 B, D* F G
This change replaces the Hellenistic form συνιστάνω with the classical
συνίστημι. 129

Gal 3:1 Ὦ ἀνόητοι Γαλάται, τίς ὑμᾶς ἐβάσκανεν τῇ ἀληθείᾳ μὴ πείθεσθαι, οἷς κατ’
ὀφθαλμοὺς Ἰησοῦς Χριστὸς προεγράφη ἐσταυρωμένος;
τῇ ἀληθείᾳ μὴ πείσθεσθαι Ψ hark 1611 Byz; C 1241S
− Chrys ×pesh; 1739, A,  33, B, D* F G b d ×vg
The infinitive phrase, “from obeying the truth,” was obtained from a similar
question in Gal 5:7 τίς ὑμᾶς ἐνέκοψεν τῇ ἀληθείᾳ με πείθεσθαι (“who
blocked you from obeying the truth”) and added to the question of 3:1, “who
bewitched you?” 130 This reading was known as a variant to Jerome, 131 which
suggests that stage ψ goes back at least to the fourth century.

127
LSJ “καταδουλόω,” 890; also LONGENECKER 52. BUSCEMI 150 understands this
change as polemical.
128
Pace LONGENECKER 52, who merely sees this as a change to the more common
voice for this verb.
129
BURTON 131; BUSCEMI 218; LONGENECKER 91.
130
BETZ 131 n.35; BUSCEMI 229; LONGENECKER 99.
131
BUSCEMI 229 citing Jerome, In Gal. 1.3 (PL 26.373 = CCL 77A.68). See also Martin
Meiser, Galater, 122 n.185.
222 Chapter 5. A History of Textual Variation in Galatians

Gal 3:29 εἰ δὲ ὑμεῖς Χριστοῦ, ἄρα τοῦ Ἀβραὰμ σπέρμα ἐστέ, καὶ κατ’ ἐπαγγελίαν
κληρονόμοι.
καί Ψ Chrys pesh hark 1611 Byz; 1241S, F G
− 1739, A C,  33, B, D* d b ×vg
The addition of καί clarifies the function of the appositional phrase κατ’
ἐπαγγελίαν κληρονόμοι as being additive, namely, that those (Galatians) who
are of Christ are also heirs according to the promise in addition to being the
seed of Abraham. Though this change is also paralleled in two Western wit-
nesses, F and G, it appears to be independently motivated.

Gal 4:6 Ὅτι δὲ ἐστε υἱοί, ἐξαπέστειλεν ὁ θεὸς τὸ πνεῦμα τοῦ υἱοῦ αὐτοῦ εἰς τὰς
καρδίας ἡμῶν ὑμῶν κράζον· αββα ὁ πατήρ.
ὑμῶν Ψ pesh hark 1611 Byz; 33
ἡμῶν 1739, A C P 1241S, , P46 B, Marc D* F G d b ×vg
The modification from ἡμῶν to ὑμῶν is a harmonization that reduce the
harshness of the shift from the second-person plural in the causal clause
(“because you are sons”) to the first-person plural in the main clause (“God
sent the spirit of his sons into our hearts crying ‘Abba, Father’”), 132 despite
the implicit exclusion of Paul getting the spirit of God’s son.

Gal 4:7 εἰ δὲ υἱός, καὶ κληρονόμος διὰ θεοῦ διὰ Χριστοῦ.

θεοῦ διὰ Χριστοῦ Chrys pesh hark 1611 1505 2495 Byz; P 1241S, D d
κληρονόμος μὲν θεοῦ, συγκληρονόμος δὲ Χριστοῦ Ψ
διὰ Ἰησοῦ Χριστοῦ 1739C
διὰ θεοῦ 1739*vid, A C,  33, P46 B, F G b ×vg
The full reading for Ψ, ostensibly derived from the same phrase in Rom
8:17, 133 includes references to both God and Christ, and thus it seems best to
assume the existence of a reference to Christ in its exemplar. The motivation
for the addition of Christ into this sentence is the apparent difficulty with
having God be the agent (διά), rather than the source of the inheritance. 134
This textual change may also have an anti-Judaic effect in denigrating the
Law. By specifying that the inheritance is through Christ (rather than though
God), it implicitly sets up a contrast with another means opposed to Christ,
namely, the Law – a contrast exploited by Theodoret of Cyrus as follows:

132
BETZ 210 n.86; BUSCEMI 365; METZGER 526; LONGENECKER 161.
133
BUSCEMI 366; METZGER 527.
134
BETZ 212 n.99; METZGER 528. BUSCEMI 366 notes that Paul has no problem with
naming God as an agent, citing Gal 1:9 and 1 Cor 1:9.
 5.3 The Eastern Branch and the Road to the Byzantine Text 223

“For not every son is an heir too, but you are both a son and an heir, and you
turned out to be both through Christ, not through the Law.” 135

Gal 4:8 Ἀλλὰ τότε μὲν οὐκ εἰδότες θεὸν ἐδουλεύσατε τοῖς μὴ φύσει μὴ οὖσιν θεοῖς·
τοῖς μὴ φύσει οὖσιν Ψ Chrys 1611 Byz; F G
τοῖς τῇ φύσει οὖσιν Marc 136
τοῖς μὴ οὖσιν K 1319, d b
τοῖς φύσει μὴ οὖσιν 1739, A C 1241S,  33, P46 B, D
The placement of the negative particle μή can be a clue to where the em-
phasis of the sentence lies, particularly when the negative particle precedes a
pre-verbal constituent. 137 More specifically, the transposition of the negative
μή to the beginning of the participial clause has the effect of emphasizing
φύσει (“by nature”). 138 This change also has the effect of reducing the
ambiguity of the pre-verbal negative particle.

Gal 4:14 καὶ τὸν πειρασμὸν ὑμῶν μου τὸν ἐν τῇ σαρκί μου οὐκ ἐξουθενήσατε
τὸν πειρασμόν μου Ψ Chrys pesh hark 1611 Byz; C, P46
τὸν πειρασμὸν ὑμῶν 1739, A,  33, B, D* F G d b ×vg
The small change in the pronoun – from “your” to “my” – has a large effect.
Instead of not disdaining their own trial with Paul’s flesh (presumably sick-
ness, cf. v. 13), which scribes may have found to be obscure, 139 the Galatians
did not disdain Paul’s trial in his own flesh. This change has the effect of
highlighting Paul’s suffering for the sake of those he evangelizes. Given the
appeal of this variant reading, it is unnecessary to suppose that its presence
here is a re-emergence of an older reading (as found in C and P46) but the
coincidental recreation of it.

135
Meiser, Galater, citing Theodoret, In Gal. (PG 82, 488A): «Ὥστε οὐκ ἔτι εἶ δοῦλος,
ἀλλὰ υἱός· εἰ δὲ υἱὸς, καὶ κληρονόμος Θεοῦ διὰ Χριστοῦ.» Οὐ γὰρ πᾶς υἱὸς καὶ κληρο-
νόμος. Σὺ δὲ καὶ υἱὸς καὶ κληρονόμος· ἔτυχες δὲ τούτων διὰ τοῦ Χριστοῦ, οὐ διὰ τοῦ
νόμου.
136
Schmid, Marcion, 116, proposes that the Tertullian’s quotation in Marc. 5.4.5 of qui
in natura sunt reflects a corruption of τοῖς μὴ φύσει οὖσιν into τοῖς τῇ φύσει οὖσιν.
137
See generally Levinsohn, Discourse Features, 48–53.
138
Levinsohn, Discourse Features, 49: “One way of specifically bringing a nonverbal
constituent into focus while negating it is to place it immediately after the negative particle
and before the verb.”
139
BUSCEMI 413; METZGER 527; MUSSNER 307 n.71. Both BETZ 224–225 n.56 and
LONGENECKER 191 claim that τὸν πειρασμὸν ὑμῶν ἐν τῇ σαρκί μου is “grammatically
awkward,” but its grammar and syntax seem straightforward enough (indeed, BUSCEMI 424
analyzes these phrases as proleptic)–it is the meaning of “your trial in my flesh” that is
unclear.
224 Chapter 5. A History of Textual Variation in Galatians

Gal 4:31 διό ἄρα, ἀδελφοί, οὐκ ἐσμὲν παιδίσκης τέκνα ἀλλὰ τῆς ἐλευθέρας.
ἄρα Chrys 1611 Byz; A C P 1241S
− Ψ
διό 1739,  33, B, Marc 140 D*
The change from διό to ἄρα is reconstruct to happen here instead of later at
stage σ in order to account better for the omission of the particle altogether in
Ψ. Specifically, the omission in Ψ can easily be considered to be an ac-
cidental omission of ἄρα due to a leap from the first alpha of ἄρα to the
initial alpha of ἀδελφοί. An omission of διό on the other hand lacks such a
convenient explanation.
Although διό and ἄρα are both inferential particles, they have somewhat
different connotations and they relate to the structure of the discourse in
somewhat different ways. According to Stephen H. Levinsohn’s analysis of
inferential particles, the particle διό constrains the interpretation of its clause
as inferential and continuative (like the conjunction καί), while the particle
ἄρα constrains the interpretation of its clause as inferential and as a conse-
quence of what was stated earlier. 141 In other words, διό is a more appropriate
particle within a section, 142 while ἄρα is more appropriate for concluding the
section. 143 With the textual change at 5:1 in the previous stage of the textual
transmission to support the interpretation that it begins a new section, the
sentence in 4:31 has now become the concluding sentence of its section 4:21–
31, making the διό a mismatch for its new place in the discourse. 144 The
substitution of ἄρα for διό provides a more suitable inferential particle for
this new structure of Galatians, where 5:1 begins, rather than ends, a major
section of the letter.

140
Schmid, Marcion, 130, is uncertain whether Tertullian, Marc. 5.4.8, propter quod,
reflects διό or ἄρα (οὖν) but favors the former.
141
Stephen H. Levinsohn, “‘Therefore’ or ‘Wherefore’: What’s the Difference?” (paper
presented at the annual meeting of the Society for Biblical Literature; San Francisco, No-
vember 2011), 3–6; online: http://www.sil.org/~levinsohns/InferentialsPaper.pdf, accessed
February 6, 2012.
142
To be sure, BURTON 269 argues that διό may be used for summations, but his
examples are not compelling. For example, διό is found at the end of a section at 2 Cor
12:10, but it seems that the actual conclusion begins at v.9 with Ἥδιστα οὖν μᾶλλον καυ-
χήσομαι. His other example, 1 Thess 5:11 does not summarize its section but provides a
parenetic inference at the end of the section.
143
LONGENECKER 218.
144
The distinctive, inferential particle οὖν in 5:1b is a very appropriate choice to end a
discourse section.
 5.3 The Eastern Branch and the Road to the Byzantine Text 225

Gal 5:10 ὁ δὲ ταράσσων ὑμᾶς βαστάσει τὸ κρίμα, ὅστις ἐὰν ἂν ᾖ.

ἄν Ψ Chrys 1611 Byz; D F G
ἐάν 1739, A C 1241S,  33, P46 B
The change from ἐάν to ἄν is in the direction of using a more classical form
for this particle after an indefinite pronoun. 145 Although the same change oc-
curs in the Western text, it is unnecessary to appeal to contamination for this
probably coincidental change.

Gal 5:14 ὁ γὰρ πᾶς νόμος ἐν ἑνὶ λόγῳ πεπλήρωται πληροῦται, ἐν τῷ· ἀγαπήσεις τὸν
πλησίον σου ὡς ἑαυτόν.
πληροῦται Ψ Chrys 1611 Byz; 1241S, D* F G d b ×vg
πεπλήρωται 1739, A C,  33, P46 B
The change from the perfect tense πεπλήρωται (“has been fulfilled”) to the
present πληροῦται (“is being fulfilled”) is also paralleled in the Western text.
The perfect tense is ambiguous as to whether it is gnomic or referring to a
specific event in the act that has a continuing effect. Though Paul may have
intended the latter possibility in Christ’s own fulfillment of the law, those
who understood it as gnomic changed the tense of the verb to the more appro-
priate present tense.

Gal 5:17a ταῦτα δὲ ἀλλήλοις ἀντίκειται ἀλλήλοις,

ἀντίκειται ἀλλήλοις Ψ Chrys pesh hark 1611 Byz; , P46
ἀλλήλοις ἀντίτκειται 1739, A C P 1241S, 33, B, D* F G d b ×vg
The movement of the reciprocal pronoun ἀλλήλοις to after the verb has the
effect of de-emphasizing it. With the new word order, the verb ἀντίκειται ap-
pears to introduce a new, distinct point, namely, that the flesh and the spirit
are actually in opposition to each other, instead of merely being against
(κατά) each other. Such a strengthening of the point appears to be sufficiently
well-motived transcriptionally that it is best to see similar transpositions in 
and P46 independent.

145
See BDAG “ἐάν,” 267–268; BDF 57 § 107; LSJ “ἐάν,” 465.
226 Chapter 5. A History of Textual Variation in Galatians

Gal 5:17b ἵνα μὴ ἃ ἐὰν ἂν θέλητε ταῦτα ποιῆτε.

ἃ ἄν Ψ 1611 Byz; CC, (d b ×vg quaecumque)
ἅ Chrys; 1739, C*
ἃ ἐάν ΒC, , A 1241S
ἄν B*
ὃ ἄν D* F G
Although there are many other readings for this part of the verse, the
Hellenistic particle ἐάν after a relative pronoun was substituted with a more
classical ἄν at this stage of the transmission of the text.

Gal 5:19 φανερὰ δέ ἐστιν τὰ ἔργα τῆς σαρκός, ἅτινά ἐστιν μοιχεία πορνεία
ἀκαθαρσία ἀσέλγεια
μοιχεία πορνεία Chrys hark 1611 Byz; D*
μοιχεῖαι πορνεῖαι FGdb
πορνεία μοιχεία Ψ
πορνεία pesh; 1739, A C P 1241S,  33, B, Marc
The addition of adultery (μοιχεία) to a list of the deeds of the flesh serves to
strengthen the representation of sexual vices. 146 The fact that this addition is
also paralleled in Western witnesses is suggestive of a Western influence on
the text at this stage of the transmission.

Gal 5:20ab εἰδωλολατρία φαρμακεία ἔχθραι, ἔρις ἔρεις ζῆλος ζῆλοι θυμοί, ἐριθεῖαι
διχοστασίαι αἱρέσεις
ἔρεις Ψ Chrys Byzpt; C, P, 1241S, , B, D*
ἔρις hark 1611 Byzpt; 1739, A, 33, F G

ζῆλοι Ψ Chrys hark 1611 Byz; C 1241S, , Marc b ×vg

ζήλους FG
ζῆλος 1739, P, 33, B, D* d
The change from the singular ζῆλος (“jealousy”) to the plural ζῆλοι (“jeal-
ousies”) appears to have been triggered by the following plural θυμοί (“an-
gers”). 147 This also appears to be the case for the change from the singular
ἔρις (“discord”) to the plural ἔρεις (“discords”), though the itacism of the ει
diphthong in the plural form, which became to be pronounced identically to
the singular, may only indicate that this is an orthographic variant.

146
See the discussion above on this variant in the Western branch.
147
BUSCEMI 528; BURTON 304.
 5.3 The Eastern Branch and the Road to the Byzantine Text 227

Gal 6:3 εἰ γὰρ δοκεῖ τις εἶναί τι μηδὲν ὤν, ἑαυτὸν φρεναπατᾷ ἑαυτόν.
ἑαυτὸν φρεναπατᾷ Ψ pesh hark 1611 Byz; D F G d b ×vg
φρεναπατᾷ ἑαυτόν 1739 Chrys, A C 1241S,  33, P46 B
This is another reading that agrees with Western witnesses, in which the
fronting of the reflexive pronoun ἑαυτόν emphasizes the self-deception of the
one who falsely considers himself important.

Gal 6:12 Ὅσοι θέλουσιν εὐπροσωπῆσαι ἐν σαρκί, οὗτοι ἀναγκάζουσιν ὑμᾶς περι-
τέμνεσθαι, μόνον ἵνα μὴ τῷ σταυρῷ τοῦ Χριστοῦ μὴ διώκωνται.
μὴ τῷ σταυρῷ Ψ Chrys hark 1611 Byz; F G
μὴ διώκωνται pesh; 1739, A C 1241S,  33, P46 B, D* d b ×vg
The placement of the negative particle μή can be a clue to where the em-
phasis of the sentence lies. 148 More specifically, when a negative particle
precedes a pre-verbal constituent, that constituent is emphasized. 149 Thus, the
variant reading with μή at the beginning of the clause places the emphasis on
the cross of Christ. When the negative particle precedes the verb, on the con-
trary, it is ambiguous whether the verb or the pre-verbal constituent bears the
emphasis. 150 In this case where the variant reading with μή is before the final
verb διώκωνται, the word order does not specify whether the emphasis falls
on the verb or on the cross of Christ. The transposition at this stage in the
transmission of the text resolves the ambiguity of the inherited reading in
favor of Christ’s cross.

Gal 6:15 οὔτε γὰρ περιτομή τί ἐστιν ἰσχύει οὔτε ἀκροβυστία ἀλλὰ καινὴ κτίσις.
ἰσχύει Ψ Chrys hark 1611 Byz; b ×vg
ἐστιν 1739, A C 1241S,  33, P46 B, D* F G d
The substitution of the verb is an apparent harmonization to Gal 5:6 ἐν γὰρ
Χριστῷ Ἰησοῦ οὔτε περιτομή τι ἰσχύει οὔτε ἀκροβυστία. 151 The change of the
verb from ἐστιν (“is”) to ἰσχύει (“is in force”) strengthens the statement.
Interestingly, Gal 5:6 has been the source of another popular harmonization,
namely, the addition of the phrase ἐν Χριστῷ Ἰησοῦ. This addition, in fact,
will occur at the next stage of the transmission of the text.
The twenty-one textual changes identified for stage ψ affect both the form
and the substance of the text. In terms of form, some of the changes reflect a
more classical idiom at 2:19 (συνίστημι not συνιστάνω) and both 5:10 and
5:17b (ἄν not ἐάν). 152 Reduction of ambiguity is still a source of textual

148
See generally Levinsohn, Discourse Features, 48–53.
149
Levinsohn, Discourse Features, 49.
150
Levinsohn, Discourse Features, 48.
151
BETZ 319 n.77; BUSCEMI 598.
152
Gal 6:7 still reads ἐάν, however.
228 Chapter 5. A History of Textual Variation in Galatians

variation at 1:4 (position of adjective), 3:29 (apposition), 4:8 (position of

negative); 6:12 (position of negative). Some textual changes appear to clean
up after the effects of earlier changes to text: 4:31 (different inferential
particle), 5:17a (word order). Harmonization continues to play a large role, in
about a quarter of the variations: 1:18 (word order, to 2:1), 3:1 (to 5:7), 4:6
(grammatical person), 5:20ab (ἔρεις ζῆλοι), and 6:15 (ἰσχύει from 5:6).
In terms of substance, the introduced variants are more common, but their
theological effects are diverse. The textual variants at 2:4 and 4:7 are anti-
Jewish in effect. Paul’s stature as the suffering apostle is elevated by the
change at 4:14. The change at 1:4 could be anti-dualist (i.e. anti-Marcionite
or anti-Gnostic), while the addition of adultery to 5:14 makes the vice list
more homiletically useful.
A number of the textual changes at this stage are paralleled in Western
witnesses: 1:4 (word order), 1:18a (word order), 1:18b (Peter), 5:14
(πληροῦται); 6:3 (word order). These six changes are fall well beneath the
reticulation cutoff of 25 for generating the initial stemma, and none of these
are so striking as to demand a genetic relationship. Thus, it seems better to
consider them as coincidental improvements to the text, though sporadic,
low-grade influence by Western readings cannot be ruled out on this state of
the evidence. In particular, it must be noted that the support for the existence
of stage ψ in the bootstrap analysis is fairly low at 74%, on a cutoff of 70%. It
is sufficiently close to the cutoff that it could well disappear in a larger study.
In other words, there is a low but non-negligible possibility that this stage is a
merely mirage of the analysis and that Ψ is in fact a mixed Byzantine-
Alexandrian text to be placed lower on the stemma than here. If it is true that
stage ψ is an artifact of the analysis, the changes here can then be attributed
to the next stage σ, where, as will be seen in the following subsection, the
supposition of a low level of contamination from a Western source already
reasonably supported by the evidence. In such a case, then, it would become
economical to chalk these six changes up to the same incident of Western
contamination in stage σ.

5.3.4 Stage σ
Stage σ relates to the state of the text for the common ancestor of the
Byzantine text and the Harklean group (hark 1611 1505 2495). Also included
here are the text of John Chrysostom and the Peshitta, which, according to the
stemma, is mixed with older readings.
 5.3 The Eastern Branch and the Road to the Byzantine Text 229

Gal 1:10 εἰ γὰρ ἔτι ἀνθρώποις ἤρεσκον, Χριστοῦ δοῦλος οὐκ ἂν ἤμην.
γάρ Chrys pesh hark 1611 1505 2495 Byz
− Ψ 1739, A P 1241S,  33, P46 B, D* F G d b ×vg
The addition of the connective particle γάρ provides a more explicit indica-
tion of the connection between this statement than the asyndeton of the older
form of the text. 153 More particularly, it indicates that Paul’s being a slave of
Christ answers the rhetorical question in v.10b as to whether he is a people-
pleaser.

Gal 1:12 οὐδὲ γὰρ ἐγὼ παρὰ ἀνθρώπου παρέλαβον αὐτὸ οὐδὲ οὔτε ἐδιδάχθην,
οὔτε pesh hark 1611 1505 2495 Byz; P46 B
οὐδέ Chrys; Ψ 1739, A P 1241S,  33, D* F G
There is a difference in nuance between the two readings, and the substitution
of οὔτε for οὐδέ tends to indicate that receiving a tradition and being taught
are synonyms for the same event. 154 The difference is so subtle semantically
and so close phonetically that the presence of the same variant in P46–B
appears to be a coincidence.

Gal 2:11 Ὅτε δὲ ἦλθεν Κηφᾶς Πέτρος εἰς Ἀντιόχειαν, κατὰ πρόσωπον αὐτῷ
ἀντέστην
Πέτρος Chrys hark 1611 1505 2495 Byz; D F G d
Κηφᾶς pesh; Ψ 1739, A C P 1241S,  33, B, b ×vg
Here, the Aramaic name Cephas is replaced with the more accessible Greek
name Peter, a change also attested among Western witnesses.

Gal 2:14a εἶπον τῷ Κηφᾷ Πέτρῳ ἔμπροσθεν πάντων·

Πέτρῳ Chrys hark 1611 1505 2495 Byz; D F G d
Κηφᾷ pesh; Ψ 1739, A C P 1241S,  33, B, b ×vg
As with the variant at 2:11, the Greek name Peter again replaces the Aramaic
name Cephas, a change also attested among Western witnesses.

153
BETZ 56 n.116.
154
After a negative particle, οὐδέ adds another negative thought while οὔτε subdivides
the earlier negative thought. See the discussion in Chapter 3.
230 Chapter 5. A History of Textual Variation in Galatians

Gal 2:14b εἰ σὺ Ἰουδαῖος ὑπάρχων ἐθνικῶς ζῇς καὶ οὐχὶ Ἰουδαϊκῶς ζῇς,
ἐθνικῶς ζῇς Chrys pesh hark 1611 1505 2495 Byz; D
Ἰουδαϊκῶς ζῇς Ψ 1739, A C P 1241S,  33, B, F G
The transposition of the verb ζῇς before the phrase καὶ οὐχὶ Ἰουδαϊκῶς tends
to de-emphasize the phrase because it is no longer pre-verbal. 155 This has the
effect of lending relatively more prominence to ἐθνικῶς, perhaps to
underscore that Peter too had been living like a gentile.

Gal 2:14c πῶς τί τὰ ἔθνη ἀναγκάζεις ἰουδαΐζειν;

τί Chrys hark 1611 1505 2495 Byz
πῶς Ψ 1739, A C P 1241S,  33, P46 B, D* F G d b ×vg
The interrogative πῶς (“how”) has connotations of surprise and disapproval
(e.g. Gal 4:9; 1 Cor 15:12). 156 The interrogative τί (“why”), on the other hand,
presupposes the truth of the assertion, namely, that Cephas had been
pressuring the gentiles at Antioch to be Torah-observant. This substitution
slightly strengthens the Judaizing valence of the passage and is therefore
slightly more anti-Jewish in effect.

Gal 3:1 . . ., οἷς κατ’ ὀφθαλμοὺς Ἰησοῦς Χριστὸς προεγράφη ἐν ὑμῖν ἐσταυρωμένος;
ἐν ὑμῖν Chrys hark 1611 1505 2495 Byz; D F G d b
− ×pesh; Ψ 1739, A C 1241S,  33, B, ×vg
Also paralleled by Western witnesses, this addition makes the image more
vivid: “to whom Jesus Christ was visually portrayed as crucified among
you.” 157 From a transcriptional viewpoint, multiple omissions are easier to
understand as coincidental than multiple additions of the same exact text, this
addition has the appearance of being genetically significant.

Gal 3:17 διαθήκην προκεκυρωμένην ὑπὸ τοῦ θεοῦ εἰς Χριστὸν ὁ . . . νόμος οὐκ
ἀκυροῖ
εἰς Χριστὸν Chrys pesh hark 1611 1505 2495 Byz; D F G d
− ×pesh; 1739 Ψ, A C 1241S,  33, P46 B, b ×vg
This addition is also found among Western witnesses and it clarifies that the
previously ratified covenant was for Christ. As discussed earlier, this change
is anti-Jewish in effect by enhancing a supersessionist understanding of the
text. Similarly, the agreement in addition appears genetically significant.

155
Stephen H. Levinsohn, “Los rasgos discursivos comparativos aplicados a la traduc-
ción de Gálatas,” 35; online: http://www.recursosteologicos.org/Documents/Galatas_tra-
duccion_discursivo.pdf; accessed June 1, 2010.
156
BDAG “πῶς 1aβ,γ” 901. Also BURTON 114.
157
See the discussion above for the same variant in the Western branch.
 5.3 The Eastern Branch and the Road to the Byzantine Text 231

Gal 3:21 εἰ γὰρ ἐδόθη νόμος ὁ δυνάμενος ζῳοποιῆσαι, ὄντως ἂν ἐκ νόμου ἂν ἦν ἡ

δικαιοσύνη·
ὄντως ἄν Chrys 1611 1505 2495 Byz
ἐκ νόμου ἄν A C 1241S, B
ἦν ἄν 1739,  33, P46
− Ψ, D* F G
The transposition of the clausal particle ἄν to the second position of the
apodosis makes the word order more standard and clarifies that ὄντως indeed
does not belong to the protasis. 158

Gal 4:15a ποῦ τίς οὖν ὁ μακαρισμὸς ὑμῶν;

τίς Chrys hark 1611 1505 2495 Byz; D d b
ποῦ ×pesh; Ψ 1739, A C 1241S,  33, P46 B, F G ×vg
Paralleled in some Western witnesses (D d b), this is another substitution of
an interrogative (see 2:14b), this time from “where” to “what” or “who.” If
“who” is meant (that is, “So who is your blessing?”), then the question would
presume that Paul is the answer. Thus, this change has the effect of enhancing
Paul’s stature. 159

Gal 4:15b μαρτυρῶ γὰρ ὑμῖν ὅτι εἰ δυνατὸν τοὺς ὀφθαλμοὺς ὑμῶν ἐξορύξαντες ἂν
ἐδώκατέ μοι.
ἐξορύξαντες ἄν Chrys 1611 1505 2495 Byz
ἐδώκατε ἄν μοι 1241S
− Ψ 1739, A C,  33, P46 B, D F G
The lack of ἄν in a counterfactual statement is non-classical, 160 and this
addition adds the particle in the second position of the apodosis. 161

Gal 4:18 καλὸν δὲ τὸ ζηλοῦσθαι ἐν καλῷ πάντοτε καὶ μὴ μόνον ἐν τῷ παρεῖναί με

πρὸς ὑμᾶς.
τό Chrys 1611 1505 2495 Byz; D F G
− 1739, A C 1241S,  33, B
Also attested among Western witnesses, the addition of the article τό makes it
clear that the following word is an infinitive rather than the homophonous

158
See the discussion of this variant in Chapter 3.
159
BURTON 243, however, posits an “unintentional clerical corruption, ΠΟ being con-
verted into ΤΙΣ, and Υ omitted to make sense,” but this sequence of events seems rather
far-fetched.
160
BDF 182 § 360(1). Also BURTON 244.
161
The placement of ἄν in 1241S implies that ἐξορύξαντες belongs to its own intonation
unit as a temporal frame rather than an attendant circumstances participle.
232 Chapter 5. A History of Textual Variation in Galatians

indicative ζηλοῦσθε, which is the actual reading of B, , and 33. 162 As for the
function of the article, it is anaphoric, 163 referring back to the statement
ζηλοῦσιν ὑμᾶς οὐ καλῶς in v.17.

Gal 5:1 τῇ ἐλευθερίᾳ ᾗ Χριστὸς ἡμᾶς ἠλευθέρωσεν στήκετε οὖν καὶ μὴ πάλιν ζυγῷ
δουλείας ἐνέχεσθε.
τῇ ἐλευθερίᾳ ᾗ Chrys pesh hark 1611 1505 2495 Byz
τῇ ἐλευθερίᾳ Ψ 1739, A C P 1241S,  33, B, D*
ᾗ ἐλευθερίᾳ Marc 164 F G (d b ×vg qua)

− Chrys hark 1611 1505 2495 Byz; P, D d b ×vg

οὖν ×pesh; Ψ 1739, A C 1241S,  33, B, F G
With the addition of the relative pronoun ᾗ, the problematic dative τῇ
ἐλευθερίᾳ can now be construed as a locative with στήκετε as “on the free-
dom with which Christ freed us, stand and do not be subject again to a yoke
of slavery.” It is possible that this reading is a conflation of the Eastern and
Western branch readings with just the article and relative pronoun, re-
spectively. In this new syntactic configuration, the inferential particle οὖν is
out of place and deleted.

Gal 5:4 κατηργήθητε ἀπὸ τοῦ Χριστοῦ,

τοῦ Chrys 1611 1505 2495 Byz; 1739, A 1241S, 33
− Ψ, C, , P46 B, D* F G
The lack of coherence in the inclusion of the article before Χριστοῦ, suggests
that its addition (or omission) occurred multiple times in the transmission of
the text. Indeed, with the recent reference to Christ in v.1, the use of an article
here in v.4 to point back to v.1 may seem natural, particularly if 5:1 was con-
strued as the beginning of a new section of the letter rather than the end of an
old section. In other words, the addition of an article here is a clue to how
scribes understood the structure of the epistle around 5:1 – those that under-
stood 5:1 to start a new section would be motivated to mark this understand-
ing by placing an anaphoric article on Χριστοῦ in v.4.

162
For more details, see the discussion in Chapter 3.
163
BDF 205–206 § 399(1).
164
For Marcion as a witness to this reading, see the discussion above for the Western
branch reading.
 5.3 The Eastern Branch and the Road to the Byzantine Text 233

Gal 5:24 οἱ δὲ τοῦ Χριστοῦ Ἰησοῦ τὴν σάρκα ἐσταύρωσαν

Χριστοῦ Chrys pesh hark 1611 Byz; P, P46, D F G d b ×vg
Χριστοῦ Ἰησοῦ 1739 Ψ, A C 1241S, (* κυρίου χ. ι.) 33, B
The deletion of Ἰησοῦ after Χριστοῦ reverses a change earlier in the trans-
mission of the text. Although both expansion and contraction of references to
Christ are found in the tradition, it is striking that this deletion is also
paralleled among Western witnesses, suggesting some low-level correction of
the text by reference to a Western text.
The fifteen textual variants introduced at stage σ show more significant
intervention into the text, albeit over fewer variation units than the previous
stage. Many of these improve the clarity of the text, such as the use of the
name “Peter” (2:11 and 2:14a), the standard placement of the particle ἄν
(3:21 and 4:15b), the addition of γάρ at 1:10 to reduce the ambiguity of the
asyndeton, and the addition of an article with a homophonous infinitive
(4:18). Similarly, the function of the obscure dative of 5:1 is clarified by add-
ing a relative pronoun that makes the dative phrase into a complement of the
verb στήκετε. In four places, the text is strengthened to become more forceful
or vivid (1:12, 2:14b, 2:14c, 3:1). Two of these changes, at 5:1 and 5:4,
belong to a cascade of changes over the infamously obscure 5:1.
Two theological effects among the variant readings are detectable at this
stage. First, there is a set of anti-Judaic textual changes (2:14b, 2:14c, and
3:17). Second, there is an attempt to aggrandize Paul’s stature (4:15a).
Half of the changes are paralleled in Western witnesses, and at least three
are significant, those at 3:1 (addition of ἐν ὑμῖν), 3:17 (addition of εἰς
Χριστόν), and 5:24 (omission of Ἰησοῦ, to result in a shorter, more primitive
text). These changes, though not numerous enough to affect the stemma,
seem unlikely candidates for coincidence and therefore seem to represent a
very low level of influence from a Western text. If such a Western influence
occurred, then 2:11, 2:14a, 4:15a, and 4:18 may be added to the list.

5.3.5 Stage κ: The Byzantine Text and Prototype

This is the last stage of the transmission of the text until the large diversi-
fication into twenty or so apparently independent lineages of the Byzantine
text. To keep the citation of evidence manageable, the witness list for the
Byzantine text is augmented by [K L 225 547 1854], of which 1854 is the
closest to the prototype of the Byzantine text. The ninth century uncials K
and L are the oldest witnesses in the Byzantine Branch, though they are not
the best representatives of the Byzantine prototype.
234 Chapter 5. A History of Textual Variation in Galatians

Gal 2:6 πρόσωπον ὁ θεὸς ἀνθρώπου οὐ λαμβάνει

πρόσωπον θεὸς ἀνθρώπου Byz [K L 226 547 1854]; 1611 1505 2495, 1739
πρόσωπον ὁ θεὸς ἀνθρώπου hark 165; A P 1241S,  33, P46
πρόσωπον ἀνθρώπου ὁ θεὸς Ψ Chrys
ὁ θεὸς γὰρ πρόσωπον ἀνθρώπου pesh 166
θεὸς πρόσωπον ἀνθρώπου (b ×vg deus personam hominis)
θεὸς ἀνθρώπου πρόσωπον D F G (d deus hominis personam)
It is unclear whether the omission of the article occurs here or a stage or two
earlier, since both Ψ and Chrysostom include the article, but it seems best to
take its omission in the Byzantine prototype and the Greek Harklean witness-
es (1611 1505 2495) as coincidental. As discussed in more detail in Chapter
3, the omission of the article before θεός gives more prominence to God and
consequently makes the statement sound more proverbial.

Gal 2:14 εἰ σὺ Ἰουδαῖος ὑπάρχων ἐθνικῶς ζῇς καὶ οὐχὶ οὐκ Ἰουδαϊκῶς,
οὐκ Byz [L 226 547 1854]; Chrys, 1739, F G (×vg non)
οὐχί K; Ψ 1505 1611 2495, A C 1241S,  33, B, D*
omit καὶ οὐκ Ἰουδαϊκῶς P46, d b
In the previous stage, the transposition of the verb ζῇς had the effect of de-
emphasizing the phrase καὶ οὐχὶ Ἰουδαϊκῶς. Here, the substitution of the less
intensive but more common negative οὐκ for οὐχί completes this trans-
formation. 167

Gal 2:16ab ὅτι διότι ἐξ ἔργων νόμου οὐ δικαιωθήσεται ἐξ ἔργων νόμου πᾶσα σάρξ.
διότι Byz [K L 226 547 1854]; Ψ, C P, ×vg
ὅτι pesh hark 1611 1505 2495 1739, A 1241S,  33, P46 B, D* F G b
ὃς (d qui)

οὐ δικαιωθήσεται ἐξ ἔργων νόμου Byz [K L 226 547 1854]

οὐ δικαιωθήσεται Ψ
ἐκ ἔργων νόμου οὐ δικαιωθήσεται pesh hark 1611 1505 2495 1739,
A C P 1241S,  33, P46 B, D* F G b d
There are two changes here. The first is the substitution of an inferential sub-
ordinator διότι for the causal subordinator ὅτι, perhaps under the influence of
Rom 3:20 διότι ἐξ ἔργων νόμου οὐ δικαιωθήσεται. This change has the effect
of making the appeal to Ps 143:2 a conclusion of this part of Paul’s argu-

165
The emphatic form for “God” in Syriac may not be diagnostic of the Greek (definite)
article. See P. J. Williams, Early Syriac Translation Technique and the Textual Criticism
of the Greek Gospels (TS3 2; Piscataway, N.J.: Gorgias, 2004), 141–142.
166
Also the emphatic form for “God.”
167
LONGENECKER 63 n.f suggests that the change stemmed from a haplography of the
successive iotas.
 5.3 The Eastern Branch and the Road to the Byzantine Text 235

mentation, rather than a premise. The other change is the transposition of the
phrase ἐξ ἔργων νόμου after the verb, which is a partial harmonization to the
Greek of Ps 143 [LXX 142]: 2, ὅτι οὐ δικαιωθήσεται ἐνώπιόν σου πᾶς ζῶν.

Gal 3:10a γέγραπται γὰρ ὅτι ἐπικατάρατος πᾶς . . .

− Byz [K L 226 547 1854]; 1241S, b ×vg
ὅτι 1611 1505 2495 Ψ 1739, A C,  33, P46 B, D* F G d
The omission of ὅτι is a harmonization to the standard γέγραπται γάρ formula
(without ὅτι), as found, for example, at 3:13 in the Eastern branch. 168

Gal 3:10b ἐπικατάρατος πᾶς ὃς οὐκ ἐμμένει ἐν πᾶσιν τοῖς γεγραμμένοις ἐν τῷ βιβλίῳ
ἐν πᾶσιν Byz [K L 225 547 1854]; Chrys, A, D F G d b ×vg
πᾶσιν hark pesh Ψ 1739, C 1241S,  33, P46 B
− 1611* 1505 2495
Also found in Western witnesses, the addition of the preposition ἐν appears to
be a minor harmonization to the Septuagintal text of Deut 27:26 ἐπικατάρα-
τος πᾶς ἄνθρωπος, ὃς οὐκ ἐμμένει ἐν πᾶσιν τοῖς λόγοις τοῦ νόμου τούτου. 169

Gal 3:12 ἀλλ’ ὁ ποιήσας αὐτὰ ἄνθρωπος ζήσεται ἐν αὐτοῖς.

ἄνθρωπος Byz [K L 226 547 1854]; 1611 1505 2495
− hark pesh Ψ 1739, A C 1241S,  33, P46 B, Marc D* F G d b ×vg
The addition of the noun ἄνθρωπος seems to be a harmonization to the quo-
tation in Rom 10:5, 170 or a partial harmonization to the Septuagintal text of
Lev 18:5 ἃ ποήσας ἄνθρωπος ζήσεται ἐν αὐτοῖς. 171

Gal 3:16 τῷ δὲ Ἀβραὰμ ἐρρέθησαν ἐρρήθησαν αἱ ἐπαγγελίαι καὶ τῷ σπέρματι αὐτοῦ,

ἐρρήθησαν Byzpt [K L 226 547 1854]; 1739
ἐρρέθησαν Byzpt 172; 1611 1505 2495 Chrys Ψ, A C 1241S,  33, B, D* F G
This textual change replaces the Koine and late Greek form ἐρρέθησαν with
the Attic form ἐρρήθησαν, 173 though many Byzantine manuscripts restore the
latter form.

168
BETZ 144 n.59 points out that the γέγραπται γὰρ ὅτι quotation formula is also found
at Gal 4:22.
169
BETZ 144 n.60; BUSCEMI 257.
170
BUSCEMI 257.
171
LONGENECKER 108 n.e; cf. BETZ 147 n.92.
172
Including 056 1 ×88 ×104 131 204 223 330 460 517 876 945 1022 ×1175 1243 1245
1270 1836 ×1837 1874 1960 2125 2138 2400 ×2464 2815.
173
ZIMMER² 337–338.
236 Chapter 5. A History of Textual Variation in Galatians

Gal 3:17 ὁ μετὰ ἔτη τετρακόσια καὶ τριάκοντα ἔτη γεγονὼς νόμος οὐκ ἀκυροῖ
μετὰ ἔτη 430 Byz [K L 223 547 1854]; 1505 1611 2495 Ψ
μετὰ 430 ἔτη pesh hark Chrys 1739, A C 1241S,  33, P46 B, D* F G b d ×vg
The default order for noun phrases is to have the noun before the modifier.
This transposition moves ἔτη into the default position before the number, and
it has the effect of removing the emphasis put upon the large number. The
presence of the same reading in close relatives of the Byzantine branch (that
is, the Greek Harkleans and Ψ) may be due to accidental coincidence or low-
level contamination.

Gal 4:15 τίς οὖν ἦν ὁ μακαρισμὸς ὑμῶν;

τίς οὖν ἦν Byz [K 226 547 1854]; D (d quae ergo erat)
τίς οὖν ἐγένετο (b quae ergo fuit)
τίς οὖν L 174; Chrys hark 1611 1505 2495
ποῦ οὖν Ψ 1739, A C 1241S,  33, P46 B ×pesh
ποῦ οὖν ᾖ FG
ποῦ ἐστιν οὖν (×vg ubi est ergo)
Various witnesses add forms of the verb “to be” to clarify this verbless ques-
tion in various ways, 175 and the choice for the Byzantine prototype was the
imperfect ἦν. The variation between ποῦ and τίς occurred at an earlier stage
in the transmission.

Gal 4:23 ὁ δὲ ἐκ τῆς ἐλευθέρας δι’ διὰ τῆς ἐπαγγελίας.

διὰ τῆς Byz [K L 226 547 1854]; 1611 1505 2495, 1739, B, D F G
δι’ Chrys Ψ 1739, A C 1241S,  33, P46
This is the first occurrence of “promise” since 3:29. Adding an article to refer
back to the promise of chapter 3 appears to be a common, independent
change.

174
Other Byzantine supporting this reading include: 075 205 330 440 796 876 1022
1319–1573 1780 1827 2138 2344 2400.
175
BURTON 243; BUSCEMI 413; LONGENECKER 188 n.d.
 5.3 The Eastern Branch and the Road to the Byzantine Text 237

Gal 4:25 συστοιχεῖ δὲ τῇ νῦν Ἰερουσαλήμ, δουλεύει γὰρ δὲ μετὰ τῶν τέκνων αὐτῆς.
δέ Byzpt [226 1854]; 176 1611 1505 2495
νῦν Byzpt [K L 547]; 177 Ψ
καί ×pesh, b ×vg
γάρ hark 1739, A C P 1241S,  33, P46 B, D* F G d
The disruption in Paul’s allegorical argument due to the note about Sinai
being a mountain earlier in the verse means that the appropriate function of
the connective particle is no longer plain. This lack of clarity has led to
various attempts to improve the connection, apparently by harmonization to
other connectives particles (δέ, νῦν) in the context.

Gal 4:26 ἡ δὲ ἄνω Ἰερουσαλὴμ ἐλευθέρα ἐστίν, ἥτις ἐστὶν μήτηρ πάντων ἡμῶν·
πάντων Byz [K L 226 547 1854]; A, b, ²
− Chrys pesh 1611 1505 2495 Ψ 1739, C 1241S, * 33, P46 B, D F G d ×vg
The addition of πάντων “gives the text a broader, pastoral application.” 178 Of
particular interest here is that the phrase ἥτις ἐστὶν μήτηρ πάντων ἡμῶν also
occurs in Polycarp, Phil. 3.3, in a context referring to Paul’s letters and the
faith given to the Philippians, “which is the mother of us all.” Zuntz holds
that Polycarp’s proof text had been this interpolated form of Gal 4:26, which
managed to survive in the Byzantine text, 179 but it not clear that Gal 4:26 is
the text that Polycarp had in mind. 180 After all, Rom 4:16 has a similar notion
applied to Sarah’s husband Abraham, καὶ τῷ ἐκ πίστεως Ἀβραάμ, ὅς ἐστιν
πατὴρ πάντων ἡμῶν (“and to one out of the faith of Abraham, who is the
father of us all”), and Polycarp’s context is about faith, not Jerusalem. There-
fore, it seems best to take both the variant reading in Gal 4:26 and in Polycarp
as being influenced by Rom 4:16. 181

176
Also 69 223 876 910 1022 1243 1270 1315 1352 1424 1448 1646 1734 1735 1738
1780 1799 1827 1854 1874 1891 1960 1982 2138 2147 2401 2412 2423 2815 2892.
177
Also K L 1 131 205 209 323 440 460 489 517 614 618 796 927 945 999 1242 1245
1836 2125.
178
Günther Zuntz, The Text of the Epistles: A Disquisition upon the Corpus Paulinum
(The Schweich Lectures of the British Academy 1946; London: Oxford University Press,
1953), 223. So also BUSCEMI 450; METZGER 528.
179
Zuntz, Text, 223. It should be noted that the Latin translation of Irenaeus, Adv. haer.
5.35.2, has the addition: quae est mater omnium nostrum.
180
E.g., Michael W. Holmes, “Polycarp’s Letter to the Philippians and the Writings that
later formed the New Testament,” in Andrew F. Gregory and Christopher M. Tuckett, eds.,
The Reception of the New Testament in the Apostolic Fathers (Oxford: Oxford University
Press, 2005), 187–227 at 209–210.
181
BURTON 263 alternatively suggests that Gal 4:26 was influenced by Polycarp, but
given the relative popularity of Galatians and Polycarp, this seems unlikely.
238 Chapter 5. A History of Textual Variation in Galatians

Gal 5:1 τῇ ἐλευθερίᾳ οὖν ᾗ Χριστὸς ἡμᾶς ἠλευθέρωσεν στήκετε καὶ μὴ πάλιν ζυγῷ
δουλείας ἐνέχεσθε.
οὖν Byz [K L 226 547 1854]
− Chrys ×pesh hark 1611 1505 2495 Ψ 1739,
A C P 1241S,  33, B, D F G b ×vg
This addition restores the οὖν that had been deleted after στήκετε in the pre-
vious stage and positions the connective particle toward the beginning of the
sentence, finally getting rid of the asyndeton. Interestingly, the inferential
particle οὖν now connects the verse to the previous section, effectively un-
doing the earlier attempts to relate it to the following material.

Gal 6:13 οὐδὲ γὰρ οἱ περιτεμνόμενοι περιτετμημένοι αὐτοὶ νόμον φυλάσσουσιν

περιτετμημένοι Byzpt [L 226 1854]; 182 2495 Ψ, P46 B, F G d b
περιτεμνόμενοι Byzpt [K 547]; 183 1739 Chrys pesh hark 1611 1505,
A C 1241S,  33, D ×vg
Even Byzantine manuscripts vary between the perfect and present participles,
though the preponderance of the evidence favors the perfect participial read-
ing of περιτετμημένοι for the Byzantine prototype. As noted above in con-
nection with the same change in P46 B, the change in tense of the participle
has an anti-Judaic effect in making the referent of the participle more clearly
refer to Judaizers, i.e. those who had been circumcised. While it is possible
that this change can be a re-emerged of an early reading as attested to be P46
and B, it is unnecessary to do so since the change appears to be well-
motivated transcriptionally, especially for such an unstable variation unit.

Gal 6:15 ἐν γὰρ Χριστῷ Ἰησοῦ οὔτε γὰρ περιτομή τί ἰσχύει οὔτε ἀκροβυστία ἀλλὰ
καινὴ κτίσις.
ἐν γὰρ Χριστῷ Ἰησοῦ οὔτε Byz [K L 226 547 1854];
2495 hark, A C P 1241S, , D F G d b ×vg
οὔτε γάρ 1611 1505 pesh Chrys Ψ 1739, 33, P46 B
Already helped by the replacement of ἐστιν with ἰσχύει in the previous stage,
the harmonization to Gal 5:6 (ἐν γὰρ Χριστῷ Ἰησοῦ οὔτε περιτομή τι ἰσχύει
οὔτε ἀκροβυστία) continues with the insertion of the phrase ἐν Χριστῷ Ἰησοῦ
and the adjustment of the position of the particle γάρ. 184

182
Also 056 1 ×6 69 205 209 223 330 440 489 517 614 618 796 927 999 1022 ×1175
1270 1319 1352 1573 1646 1734 1735 1738 1780 1799 1827 1891 1960 2138 2400 2401
2412 2423 2815 2892.
183
Also 075 131 323 460 876 910 945 1242 1243 1315 1424 1448 1836 1874 1982
2125 2147.
184
BETZ 319 n.77; BUSCEMI 598.
 5.4 Conclusions 239

Of the sixteen changes for the final stage of the transmission of the text to the
Byzantine prototype, the most frequent textual change is harmonization. In
fact, harmonization is a factor in over half of the introduced variant readings,
including 2:16a (to Rom 3:20), 2:16b (to Ps 143:2), 3:10a (to Gal 3:13),
3:10b (to Deut 27:26), 3:12 (to Rom 10:5 or Lev 18:5), 4:23 (to Gal 3:29?),
4:25 (to Gal 4:26), 4:26 (to Rom 4:10), and 6:15 (to Gal 5:6). The large
amount of harmonization demonstrates the scribes’ deep familiarity with Bib-
lical texts.
Aside from harmonization, many of the changes are cosmetic to make a
more smoothly reading text, such as those at 2:14 (more common word), 3:16
(Atticism), 3:17 (default noun-modifier order), and 4:15 (fill in ellipsis).
Some of the changes to smooth the text also make it more suitable for pas-
toral application, including those at 2:6 and 4:26. A possible anti-Jewish
theological effect may be discernable in 6:13 in the change of the tense of the
participle.
As opposed to the previous stage, there is little evidence of a Western
influence on the text. Though both the changes in 3:10a and 6:13 are
paralleled in Western witnesses, they are also found elsewhere and could be
independently motivated.

5.4 Conclusions

According to this analysis of over a hundred and twenty textual changes with-
in the two main branches of the transmission of the text of Galatians, some
conclusions can be made. First, scribal behavior was not uniform. Each
branch of the text has identifiably distinct patterns of scribal behavior. For
example, the Western branch has the largest concentration of theologically
significant variants, the majority of which were anti-Jewish in effect. This
suggests that the Western branch of the text developed within a milieu char-
acterized by a polemic engagement with multiple opponents, especially Jews
but also demiurgic and other non-orthodox Christians. As for the Eastern
branch, a different set of concerns rose to the fore. There, the strongest influ-
ence appears to be the need to clarify the text and reduce its ambiguity. At
one step on the road to the Byzantine text, a number of variants also paral-
leled in Western witnesses were introduced into the text, though not at a level
high enough to trigger a reticulating link in the initial construction of the
stemma. The final step for the Byzantine text has an unusually large number
of harmonizations to rather distant parallels, suggesting that its scribes had a
great familiarity with the Christian scriptures.
Despite all these differences, there are some recognizable commonalities
in Christian scribal behavior across the centuries and the regions. For exam-
ple, harmonization to the local context is common in every branch, indicating
240 Chapter 5. A History of Textual Variation in Galatians

that scribes were not copying the text as if they were reading it for the first
time but a text with which they were familiar. As another example, there was
a tendency to clarify the text, whether to remove its ambiguities or to
strengthen its message. This behavior suggests that its scribes desired a useful
text. Yet the differences in the scribal behavior demonstrates that the scribes
were not only copying a familiar and useful text, but that they were also
copying a living text that reflected the scribes’ environment.
 Chapter 6

Contrasts and Conclusions

6.1 An External History of the Text of Galatians

This is the first study of the text of Galatians to propose a detailed stemma
codicum of a significant number of the leading manuscripts and versions of
Galatians. In all, 92 witnesses were collated and analyzed at 1624 variation
units. Given the immense nature of this analytical task, the initial shape of the
stemma was constructed with the help of a computer program that imple-
ments “cladistic” algorithms devised by computational biologists to classify
organisms based on common descent. It should be noted that the use of cla-
distics for textual criticism is not completely new. Deep philosophical sim-
ilarities between cladistics and stemmatics have long been recognized, and
some manuscript traditions had already been analyzed using cladistics prior
to this study. Nevertheless, this study advances the state of the art in comput-
er-assisted stemma construction by proposing and implementing methods to
account for significant levels of mixture within the history of the text. These
methods too have their theoretical antecedents in the work of textual criti-
cism, but their implementation in a computer program is novel. Accordingly,
this chapter describes the external history of the text of Galatians based on
the stemma constructed in Chapter 2, oriented in Chapter 3, and examined in
Chapter 4.
The textual transmission of Galatian divides into two large textual streams.
One of them leads to the “Western” branch. Its members include D, F, G, and
the Old Latin witnesses b and d. The text of this branch is the most divergent
in the transmission of Galatians, with more harmonizations and theologically
significant variation, especially those with an anti-Judaic effect, as described
in Chapter 5. Nonetheless, there are numerous, clear Leitfehler that indicate
that this branch has a common origin subsequent to the archetype. The Vul-
gate is also a member of this branch, but it is mixed; about four-fifths of its
text is Western. An early relative of this group is Marcion, corroborating the
views of other scholars that Marcion’s text is an early witness or precursor to
the Western branch. Marcion flourished in the mid-second century, so the
common ancestor of Marcion and the Western branch must be at least that
old. This indicates that the stream that led to the Western text diverged from
its closest relatives no later than to the first half of the second century.
242 Chapter 6. Contrasts and Conclusions

An ancient group is the cluster of P46 and B. Its textual transmission is

generally strict, and its most common errors are omissions of little words and
harmonizations to the local context. P46 has the distinction of being the
oldest direct witness to the text of Galatians, in the early part of the third
century. Accordingly, the common archetype behind P46 and B must be
earlier, somewhere in the second century at the latest. The closest relative to
P46–B is the Western branch, which implies that the common ancestor dates
back to the beginnings of the second century.
The largest textual stream in Galatians is the one, for want of a better term,
that gave rise to the “Eastern” branch. Though its leading witnesses are not as
ancient as those in the Western branch (with Marcion) and the P46–B cluster,
its text is independent of those parts of the transmission of the text and must
therefore ultimately originate from a manuscript at least as old as the
common ancestor of the Western branch and P46–B. The earliest surviving
precursors to the Eastern branch are considerably later than this ancestral
manuscript, however. Codex Sinaiticus () is from the mid-fourth century
and the Queen of the Cursives 33 is from the ninth century. Despite the
lateness of these witnesses, their textual quality is very good. Except for four
arguably orthographic textual variants, the agreements between  and 33
constitute the earliest stratum of the Eastern branch.
After the stage of the textual tradition supported by  and 33, the Eastern
branch splits into two sub-branches. One sub-branch comprises the fifth-
century manuscripts A and C as well as P and 1241s of the Middle Ages.
Most of the other members of this sub-branch have been mixed with readings
from the Byzantine text that dominated the transmission of the text within the
Middle Ages. The other sub-branch is a large collection of witnesses that
includes family 1739, the Athos Codex Ψ, a “Syrian” group (comprising the
text of John Chrysostom, the Peshitta, and Harklean witnesses), and the
Byzantine text proper. As for the common ancestor of the Byzantine text, it
has been found to be the result of a long process of transmission within the
Eastern branch involving more than fifty newly introduced textual variants. 1
There is fairly little evidence that the Byzantine text is the result of a recen-
sion aside from a small influx of Western readings into the common ancestor
of the Syrian group and the Byzantine text. The apparent use of this sub-
stream by Chrysostom and the Peshitta indicate that this text was prevalent in
the fourth century. Nevertheless, its leading subgroup, family 1739, has affin-

1
In this regard, this finding coheres with that of Klaus Wachtel, Der byzantinsiche Text
der katholischen Briefe: Eine Untersuchung zur Entstehung der Koine des Neuen Testa-
ments (ANTT 24; Berlin: de Gruyter, 1995), for the Catholic Epistles.
 6.1 An External History of the Text of Galatians 243

ities with Origen’s text of Paul, 2 so the source of this sub-branch probably
goes back to the early third century.
It should be kept in mind that, although 1739 is at the head of a stream that
eventually gave rise to the Byzantine text, it is textually much closer to the
stream’s ancestors than to its descendants. Stage γ, the immediate ancestor of
1739, differs from other early manuscripts in ten variation units, while it
differs from the Byzantine prototype in another fifty-five. Raw statistics in-
volving overall similarity certainly show 1739 as more similar to early manu-
scripts than to Byzantine one, 3 but as explained earlier in section 1.4.3, over-
all similarity can be misleading as to genealogical relationships. The stemma
here shows that 1739 is genealogically more related to Byzantine prototype
because they have an ancestor in common, despite the fact that the Byzantine
text is the product of so much more scribal activity that 1739 appears to have
more in common with early witnesses as a result.
It must be pointed out that nowhere in this description has been mentioned
an “Alexandrian” text. The reason for this omission is that the Alexandrian
text – as usually conceived with P46, B, , A, C, 33, and 1739 – is not a
stemmatically coherent text. These witnesses are merely those with a high
quality text but there is no special genealogical relationship among them.
They do not a share a common ancestor below the archetype; in fact, their
most recent common ancestor is the archetype. Every other surviving witness
to the text – whether Western, Byzantine, or something else – is also a de-
scendant of their common ancestor, and hence also a member of their
common genealogy.
Although this study is the first to propose a precise history of the text in
the form of a stemma, it is important to keep in mind that the text of the
Pauline corpus has been investigated before by some of the most prominent
textual critics within the field of New Testament textual criticism, namely,
Günther Zuntz and F. J. A. Hort. Consequently, the history of the text
proposed by this study is also compared and contrasted with the work of these
important pioneers.

2
See J. Neville Birdsall, “The Text and Scholia of the Codex von der Goltz and its Al-
lies, and Their Bearing upon the Texts of the Works of Origen, Especially the Commentary
on Roman,” in J. Neville Birdsall, ed., Collected Papers in Greek and Georgian Textual
Criticism (Piscataway, N.J.: Gorgias, 2006), 81–86. Birdsall also found close affinities be-
tween 1739 and 424** and 6 in his study. Unfortunately, the corrections to 424 were too
few to include in this study, but the affiliation of 6 with 1739 is confirmed.
3
For example, Parker, Introduction, 263, lists the “closest associates” of 1739 based on
percentage agreement data taken from Text und Textwert in descending order as 0243,
1506, C, , B, A, and P46. Byzantine witnesses are far down the list. Percentage agreement
data is overall similarity data.
244 Chapter 6. Contrasts and Conclusions

6.2 Zuntz and Hort

[T]he textual criticism of the New Testament cannot be carried out by statistical methods.
. . . None but commensurable entities can be reduced to figures, and no two variants are
strictly commensurable. Readings of all shades between good and bad; slips of the pen and
intentional alterations; attestation by anything between one and a thousand witnesses: what
is their common denominator? Variant readings can fruitfully be compared and grouped on
more than one principle, but they cannot reasonably be added up or reduced to percentages
like the factor of an arithmetical sum. What is the sum total of, say, an egg plus a grape
plus a unicorn? 4

So states Günther Zuntz in his second Schweich lecture of the British acade-
my in 1946. This statement neatly articulates an objection to the application
of computational techniques to the craft of textual criticism. 5 After all, the
cladistic method adopted for this study considers all collated textual variants
regardless of their nature and tries to minimize the number of accidental co-
incidences among them under the maximum parsimony principle. The result
of this procedure is a stemma that shows the proposed genealogical relation-
ships between the surviving manuscripts. By contrast, Zuntz’s own approach
to the text of Paul is qualitative, not quantitative. He is skeptical of being able
to the recover the genealogical history of the text because of the tradition was
“beset with contamination.” 6 Rather, Zuntz deploys the metaphor the trans-
mission of the text as a stream, whose readings mix together, and different
manuscripts represent different samples from the stream. 7 Nevertheless, his
study remains the most detailed examination of the text of Paul over the past
sixty years. 8 Thus, it would be illuminating to compare and contrast this
study with that of Zuntz both in terms of methods and results.
Zuntz’s point of entry into his study of the text of the Pauline corpus is an
examination of the oldest manuscript of the collection, the then-recently
discovered and published Chester Beatty papyrus P46. 9 This manuscript was
unavailable to Westcott and Hort and it cut by half the gap in time – from 250
years to about 125 years – between the archetype of the letter collection and

4
ZUNTZ 58.
5
See also ZUNTZ 12: “It follows that textual criticism, in our field, still can, and must,
use the traditional methods (if adapted to its subject); and that it cannot be carried out
mechanically. At every stage the critic has to use his brains. Were it different, we could put
the critical slide-rule into the hands of any fool and leave it to him to settle the problems of
the New Testament text.” The modern rhetorical equivalent of “slide-rule” is the computer.
6
ZUNTZ 9.
7
ZUNTZ 264.
8
Michael W. Holmes, “The Text of Epistles Sixty Years After: An Assessment of
Günther Zuntz’s Contribution to Text-Critical Methodology and History,” in J. W.
Childers and D. C. Parker, eds., Transmission and Reception: New Testament Text-Critical
and Exegetical Studies (TS3 4; Piscataway, N.J.: Gorgias, 2006), 89–113.
9
ZUNTZ 17–18.
 6.2 Zuntz and Hort 245

the earliest substantial manuscript. 10 Accordingly, the first step that Zuntz
performs is an examination of the quality of P46 and he finds that P46’s text
has been “beset with a great number of scribal slips.” 11 Nevertheless, Zuntz
also holds that P46 sometimes preserves readings of “outstanding quality.” 12
Zuntz bases this conclusion on readings in 1 Cor 2:4, Heb 12:1, and 1 Cor
14:39, 13 as well as various instances where he concludes that P46 correctly
has the shorter reading in 1 Corinthians and Hebrews. 14 As a result, Zuntz
concludes, “P⁴⁶, indeed, proves to be a uniquely important manuscript. While
it is true that it abounds with surface errors, its basic text is of supreme
quality.” 15
In many respects, Zuntz’s study of the scribal habits of P46 has been su-
perseded by the detailed investigation of P46’s singular readings by James R.
Royse, 16 but the main point of Zuntz’s (and Royse’s) assessment remains –
P46 is a poor copy of an excellent exemplar. This conclusion is corroborated
by the present study of the text of Galatians. According to the stemma gener-
ated in Chapter 2, P46 diverges from its common ancestor with its closest re-
lative B in almost 100 variation units. By contrast, B itself diverges from that
same ancestor in 26 places, which is about a quarter of the number of chang-
es, despite the fact that B was copied more than a hundred years later. Never-
theless, P46’s place in the stemma is very high, near the archetype, which
means that the text it inherited is of high quality. As a result, this study cor-
roborates Zuntz’s assessment of P46, both in terms of P46’s scribal character
as well as the nature of its source.
Zuntz’s next step is to compare the text of P46 with that of leading wit-
nesses to the text of Paul. The first such witness is Codex Vaticanus (B).
Here, Zuntz looks at nine instances, mainly in 1 Corinthians and Hebrews,
where P46 and B agree on “definitely spurious” readings, 17 and concludes
from them that these “special agreements in error (Leitfehler) demonstrate the
close interrelationship between P⁴⁶ and B; . . . . these two manuscripts belong
to one and the same ancient and narrow branch of the tradition.” 18 The stem-
ma proposed in this study supports Zuntz’s conclusion here. According to the

10
ZUNTZ 17.
11
ZUNTZ 23.
12
ZUNTZ 23.
13
ZUNTZ 23–31.
14
ZUNTZ 31–34, paying particular attention to 1 Cor 8:2–3; Heb 2:8; 3:6; 11:4, and 39.
Zuntz also examines instances of corruption in P46’s exemplar (34–39).
15
ZUNTZ 56. By “basic text,” Zuntz means the text of its exemplar.
16
James R. Royce, Scribal Habits in Early Greek New Testament Papyri (NTTSD 36;
Leiden: Brill, 2008), 199–358.
17
ZUNTZ 61–62. These instances are found at 1 Cor 1:8, 18; 3:12; 7:5; 13:5; Heb 5:1;
7:2; 8:10; and Col 1:12.
18
ZUNTZ 62.
246 Chapter 6. Contrasts and Conclusions

cladistic analysis of Chapter 2, P46 is more closely related to B than to any

other witness in this study. Indeed, P46 and B are the only witnesses to share
certain striking scribal errors in Galatians (Gal 1:19 οὐχ for οὐκ; 3:21 ἐν νό-
μῳ for ἐκ νόμου; 3:24 ἐγένετο for γέγονεν; and 6:4 omit ἕκαστος), as well as
some sub-singular readings (Gal 3:21 omit τοῦ θεοῦ with d; 6:11 ἡλίκοις for
πηλίκοις with 33). As a result, this study of Galatians confirms the conclusion
of a “close interrelationship” between P46 and B from Zuntz’s study else-
where in Paul.
Zuntz also compared the tenth-century minuscule 1739 with P46 and B. 19
Zuntz first acknowledges that one finds “1739 in frequent agreement with the
bulk of the ‘Alexandrians’ as against all the other evidence,” but he then as-
serts that “wherever the less distinguished members of the ‘Alexandrian’
group oppose its leaders P⁴⁶ B , the Athos manuscript [scil. 1739] hardly
ever joins the former.” 20 With this in mind, Zuntz goes on to examine 1739’s
possible relationship with P46 and B specifically. Between 1739 and P46,
Zuntz noted four special agreements in error that were singular or sub-sin-
gular, 21 and four such special agreements in error between 1739 and B. 22
Zuntz also pointed out two more shared special agreements in error among
1739, P46, and B. 23 Based on this evidence, Zuntz concludes that 1739 is a
descendant of an old manuscript that “in turn joins itself to a branch of the
tradition which centres on P⁴⁶ B; indeed, it is, or was, even nearer to the pa-
pyrus than to codex Vaticanus.” 24 In other words, 1739 is an important wit-
ness for Zuntz of Paul’s text close to the common ancestor of P46 and B and
close to the archetype of the tradition.
Regrettably, Zuntz’s estimation of the relationship between 1739 and P46–
B has not been borne out by this study of the text in Galatians. According to
the stemma constructed in Chapter 2, 1739 is not closely related to P46–B;
rather, it is closely related in a genealogical sense, on the one hand, to the
common ancestor of Ψ and the Byzantine text and, on the other hand, to the
secondary Alexandrians A, C, P, and 1241S. This conclusion stands in stark
contrast with Zuntz’s assertion that 1739 “hardly ever joins” the secondary
Alexandrians. In Galatians, however, there are some striking agreements of

19
ZUNTZ 78–84.
20
ZUNTZ 78.
21
ZUNTZ 79. Specifically, 1 Cor 15:31; Heb 5:1; 13:5, and 6. Zuntz also argued that
P46 and 1739 shared two special agreements in genuine readings at Heb 1:3 and 11:39, but
agreements in genuine readings are not evidence of a genetic relationship at any level
lower than the archetype.
22
ZUNTZ 79. Specifically, Rom 5:11; 1 Cor 8:8; 13:6; and Gal 4:6.
23
ZUNTZ 80. Specifically, at 1 Cor 6:14 and Col 3:15. Zuntz also noted four agreements
in what he considered genuine readings but these are not evidence of a special stemmatic
relationship.
24
ZUNTZ 81.
 6.2 Zuntz and Hort 247

1739 with the secondary Alexandrians against P46, B, and , including Gal
2:12 ἦλθον for ἦλθεν, 4:3 ἦμεν for ἤμεθα; 5:17 δέ for γάρ; and 5:21 add φό-
νοι. 25 Moreover, there are additional agreements between 1739 and the se-
condary Alexandrians with  against P46 and B, including Gal 1:17 ἀνῆλθον
for ἀπῆλθον, 1:18 μετὰ τρία ἔτη for μετὰ ἔτη τρία, 2:20 τοῦ υἱοῦ τοῦ θεοῦ for
τοῦ θεοῦ καὶ Χριστοῦ, and 6:2 ἀναπληρώσατε for ἀναπληρώσετε.
Two possibilities come to mind for the differing conclusions about 1739
between Zuntz’s study and this study. One possibility is that Zuntz does not
consider the agreements between 1739 and the secondary Alexandrians to
have outweighed the agreements between 1739 and P46 and B. Unfortunate-
ly, Zuntz does not specify what the few agreements were between 1739 and
the secondary Alexandrians that he found, so it is hard to tell whether this
difference in our conclusions derives from the number of agreements or the
weight that Zuntz places upon them. Another possibility is that these two
studies focus on different parts of the Pauline corpus, which may have had
different textual histories. Zuntz concentrates on 1 Corinthians and Hebrews,
while this study focused on Galatians. Since many of the striking agreements
occurred in Hebrews, it is possible that the history of the text for Hebrews
may differ from that of Galatians, in that 1739 may have had an exemplar for
Hebrews that was unrelated to its exemplar for Galatians. A follow-up study
on the text of Hebrews could clarify these questions by showing whether the
history of the text for these epistles differs from that of Galatians.
The next focus of Zuntz’s investigation is the Western text. 26 He defines
the term “Western” geographically as referring to those witnesses that are
found in the Western parts of the Roman Empire. Thus, for Zuntz, Western
witnesses comprise D, F, G, Tertullian, the Old Latin d, and non-Vulgate
quotations in the Latin Fathers. Recognizing its mixed character, Zuntz also
counts the Vulgate as a Western witness except when it supports Alexandrian
readings. Based on this definition of “Western,” Zuntz then sorts their read-
ings into three sets: W, W+, and Wω. The first set W consists of those read-
ings that are exclusively found in Western witnesses. The second set W+ in-
cludes those readings that occur in both Western and non-Western witnesses.
The third set Wω applies to Western-Byzantine readings. 27 With this arrange-
ment of Western readings, Zuntz proceeds to examine them separately.
As for the exclusively Western readings, Zuntz is hardly impressed as to
their originality. He notes that “these variants which . . . fail to find non-
Western support, are all of them evident errors.” 28 Zuntz also attributes many

25
These are all discussed in more detail in Chapter 3.
26
ZUNTZ 84–142.
27
ZUNTZ 85. Note that Zuntz’s use of the siglum ω for the Byzantine text differs from
this study’s use of it as standing for the archetype.
28
ZUNTZ 89.
248 Chapter 6. Contrasts and Conclusions

of them to apparent Latinisms. His characterization of the exclusively West-

ern readings is generally borne out by this study, but not his reasons for them.
According to the investigation in Chapter 5, only a couple of the Western
readings appear to be Latinisms, but many of them functioned to improve the
clarity of the text or to heighten the perceived anti-Judaic nature of some
parts of Paul’s argument.
Zuntz is considerably more impressed with Western readings that are
attested elsewhere, especially in P46 and B. Zuntz concludes that “Western
witnesses joined by P46 or B or 1739 are more often right than wrong; they
are hardly ever wrong when joined by the whole ‘proto-Alexandrian’
group.” 29 If anything, this study is even more optimistic about the value of
Western readings when they are supported by both P46 and B, including the
following: Gal 1:3 κυρίου ἡμῶν (also with 33), 1:17 ἀπῆλθον, 1:18 μετὰ ἔτη
τρία, 2:20 θεοῦ καὶ Χριστοῦ, 4:3 ἤμεθα, 4:19 τέκνα (also with ), and 6:2
ἀναπληρώσετε.
As for Western-Byzantine readings, Zuntz holds, “Byzantine readings must
be ancient. They go back to the time before the Chester Beatty papyrus [i.e.
P46] was written; the time before the emergence of separate Eastern and
Western traditions; in short, they reach back into the second century.” 30 In
light of this study of the text of Galatians, Zuntz’s conclusion is surprising.
Although it is true that the Eastern and Western branches of the text devel-
oped largely independently of one another, there are still some cases of cross-
contamination. The Vulgate, for example, is a mixture of a Western text and
an Eastern text, which Zuntz acknowledges. 31 Furthermore, according to this
study, the Byzantine text is the result of a long process of transmission within
the Eastern branch, whose development can be traced through a number of
different intermediate stages. If these Western readings were actually present
all along in the Eastern stream as Zuntz supposed, it is curious that it is only
at one stage, between Ψ and the Syrian group, that certain striking Western
readings appear in the textual stream. Although this amount of Western
readings was insufficient to trigger an explicit link for contamination in the
stemma, these Western readings seem striking enough to call a conclusion of
coincidental scribal activity into question. Rather, the evidence studied here
in Galatians suggests that there may have been some low-level of influence
from a Western source at a particular stage in the history of the Byzantine
text. 32

29
ZUNTZ 158.
30
ZUNTZ 150–151.
31
ZUNTZ 85.
32
ZUNTZ 151 ridicules such a notion, however: “Could a Byzantine patriarch in the
eighth or ninth century be supposed to have sent envoys to some Greek monastery in Sicily
or south Italy in order thence to procure some obsolete manuscripts and from them to
 6.2 Zuntz and Hort 249

Zuntz’s history of the Pauline text differs from that of his predecessor Hort
in two key respects. First, Zuntz disagrees with Hort that the Byzantine text is
the result of a recension that conflated readings from an Eastern (“Neutral”)
text and a Western Text. 33 Rather, Zuntz sees apparent Western readings in
the Byzantine text as the independent survival of ancient readings. 34 The re-
sults of this study suggest that the truth may lie somewhere between them.
Although there is insufficient evidence that the Byzantine text was created by
a thorough-going recension or conflation of Eastern and Western texts, it
does appear that several Western readings entered the pre-Byzantine trans-
mission of the text at a particular stage.
Second, Zuntz and Hort disagree over the value of the Western text. Spe-
cifically, Hort had a low opinion of the Western text throughout the New Tes-
tament, holding that it was affected by a scribal freedom to paraphrase the
text. 35 Even when the Western text agrees with B in Paul, Hort holds that this
is the result of contamination from a Western manuscript. 36 In other words,
B’s agreements with Western readings, according to Hort, do not enhance
their value, but instead they lessen B’s worth as a witness to the authorial text
of Paul. 37 Zuntz disagrees with this assessment, arguing that Western agree-
ments with B, especially where supported by P46, are ancient, pre-Western
readings. 38 The results of this study, especially in Chapter 3, support Zuntz
over Hort in their dispute. At every place but one in the text of Galatians

intrude a number of Western readings into that sacred text which his authority made
prevail among the Orthodox”? Based on this study’s stemma, however, the stage at which
the Western readings entered the pre-Byzantine is ancestral to the text of Chrysostom and
the Peshitta and, therefore, must pre-date the end of the fourth century. It is within this
period that Jerome was conflating a Western text with an early Eastern text.
33
Zuntz’s disagreement with Hort more explicit in his article “The Text of the Epistles”
in Günther Zuntz, Opuscula Selecta: Classica, Hellenistica, Christiana (Manchester:
Manchester University Press, 1972), 252–268 at 253.
34
ZUNTZ 150–151.
35
HORT 122–124, 170–171. The striking exception to Hort’s general disapproval of the
Western text is the inaptly called “Western non-interpolations” (175–177).
36
HORT 150–151, 166, 171, 224, 240–241. Hort’s leading example for Western
contamination in B is Col 1:12, where B reads καλέσαντι καὶ ἰκανώσαντι against D* F G
καλέσαντι and P46  A C 1739 Byz (HORT 240). Hort did not know of P46, of course, but
Zuntz points out that P46’s failure to confirm B’s reading shows that it can help assess B’s
quality (ZUNTZ 41). In other words, P46’s agreement with B on an apparent Western
reading means that the reading is pre-Western. (Hort’s other example is 2 Thess 3:4, at
which P46 is unfortunately lacking.)
37
In one case, Hort held that even the Western agreement with both  and B at Gal 2:12
(ἦλθεν for ἦλθον) was “unquestionably wrong” (224). Chapter 3 of this study, however,
presents a case for the ἦλθεν reading in Gal 2:12 based on internal evidence, against Hort’s
conclusion.
38
E.g., ZUNTZ 158.
250 Chapter 6. Contrasts and Conclusions

where P46 and B agree with Western witnesses against Eastern witnesses
such as , A, C, 33, and 1739, this study concludes that the internal evidence
favors the P46–B-Western variant reading.
Despite the difference in attitude toward computational and genealogical
approaches to textual criticism between Zuntz and this study, the results are
remarkably congruent. Both this study and that of Zuntz divide the textual
tradition of Galatians into two main branches, a Western branch and an East-
ern Branch. Both studies see the Byzantine text as the result of a process of
transmission within the Eastern branch. Both studies also have a high regard
for P46 when it is supported by B, and for the readings of the Western branch
when it is supported by P46, B, and other early witnesses. Despite these
broad areas of agreement, there are two main differences between Zuntz and
this study. First, Zuntz rates 1739 more highly than the present study. Second,
Zuntz considers Western readings in the Byzantine to be ancient survivals,
rather than a case of low-level contamination.

6.3 Contrasts with the Critical Texts of the Galatians

This study proposes a critical text that differs from the critical text of the
Nestle-Aland edition in a number of important ways. There are two main
reasons for this difference. First, the critical text of this study is edited based
on an explicit history of the text, precisely depicted in the form of a stemma
codicum. Second, this study relies on recent research on linguistic pragmatics
of the Greek article, word order, and conjunctive particles to come to a
theoretically informed decision about certain difficult variants.
There are twelve textual differences between this study and the Nestle-
Aland edition. The most textually significant one is the emendation at 4:25,
relegating to the apparatus the statement that Sinai is a mountain in Arabia.
This statement has long been seen as a marginal gloss that crept into the text,
but the analysis in this study that the note does not belong in the main text of
a critical edition is the fullest to date. Specifically, the note is both argumen-
tatively and structurally superfluous and its conception about Arabia is in
tension with Arabia’s role in Gal 1:17. Even if one were to retain the note in
the text and thereby avoid the use of conjecture, this study still argues for a
form of the statement different from that of the Nestle-Aland text, namely, it
would be the shorter version without the explicit reference to Hagar.
The most historically significant difference between this study’s critical
text and the text of the Nestle-Aland edition is the change of a single letter at
Gal 2:12. Rather than stating “when they came” (ἦλθον), referring to some
people from James, the best attested reading states, “when he came” (ἦλθεν),
referring to Cephas. Yet this tiny difference in the text results in a markedly
different understanding of the Antioch incident. With the reading of the
 6.3 Contrasts with the Critical Texts of the Galatians 251

Nestle-Aland text, on the one hand, Cephas came to Antioch, ate with the
local gentiles, but then was intimidated into changing his mind. With this
study’s critical text, on the other hand, Cephas came to Antioch with no
intention of eating with the gentiles, and this is what Paul objected to.
Perhaps the most theologically significant difference between the critical
text of this study and that of the Nestle-Aland is the variation unit in Gal
2:20, where Paul no longer lives by the faith of the son of God, but by the
faith of God and Christ. This reading has a significant bearing on the pistis
Christou debate, which concerns whether the genitive in the various πίστις
Χριστοῦ constructions is objective (faith in Christ) or subjective (faithfulness
of Christ). What makes the “God and Christ” important is that its formulation
is more amenable to a subjective genitive interpretation than the Nestle-Aland
reading (cf., Rom 3:3 τὴν πίστιν τοῦ θεοῦ “the faithfulness of God”). This
strengthens the exegetical evidentiary base for the subjective genitive.
Some of the textual differences involve more subtle theological issues. For
example, the word order variant at 1:3 gives the opening greeting a somewhat
stronger Christological cast to the greeting than does the Nestle-Aland text.
The word order variant at 1:8 presupposes that the non-Pauline preaching is a
distortion of the gospel of Christ (v.6). The different connective particle in
1:11 has a structural implication, namely, that v.11, not v.10, begins a major
section of the letter. The difference between “went away” (ἀπῆλθον) and
“went up” (ἀνῆλθον) in 1:17 slightly strengthens Paul’s claim of
independence from Jerusalem. The addition of Jesus to 5:24 has an anti-
Separationist effect.
To be sure, other textual variants have barely any discernable difference.
The difference in preposition at 1:4 between περί and ὑπέρ is virtually
untranslatable into English. The difference between οὐδέ (“and not”) and
οὔτε (“nor”) in 1:12 might suggest that Paul received his traditions and
teachings on difficult occasions, or it might suggest an afterthought on Paul’s
part. The difference between ἄχρις and μέχρις in 4:19 is so small that this
textual variation was not even included in the Nestle-Aland apparatus. Also
virtually non-existent is the distinction in meaning between the aorist
subjunctive κληρονομήσῃ and the future κληρονομήσει in 4:30.
Another set of changes proposed in this study is simply due to a difference
in editorial philosophy. This study adopts the philosophy of the recent SBL
edition of the New Testament that it is an editor’s duty to decide what the text
says based on the evidence and the editor’s expertise, and thus brackets
should be used sparingly, if at all. There are nine places in the Nestle-Aland
text that contain bracketed readings. Of these nine bracketed readings, the
longer reading is chosen eight times (1:6, 1:8, 1:15, 2:6, 2:13, 2:16, 3:21, and
252 Chapter 6. Contrasts and Conclusions

5:7), and the shorter reading is chosen (5:24) once. 39 For one of the bracketed
readings, this study proposes an original and more plausible exegesis of the
verse with the longer reading. According to exegetical arguments made for
Gal 1:6, the most satisfying interpretation of the construction ἀπὸ τοῦ
καλέσαντος ὑμῶν ἐν χάριτι Χριστοῦ is “from the one who called you with
grace to belong to Christ.” Understanding Χριστοῦ as a second complement
of καλέσαντος avoids the problems of the more common “by the grace of
Christ” interpretation.
The text of the Nestle-Aland edition is very good, but there is room for
improvement. To be sure, most of the textual differences hardly change our
understanding of the letter except in very subtle ways, but at least three of the
readings are historically and theologically significant. The results of this
study on reconstructing a critical text suggest that there may be room for im-
provement in other letters of Paul as well, if not in the rest of the New
Testament. Part of the reason is that our understanding of the Greek language
continues to improve, allowing us to appreciate finer nuances in the different
forms of the text to a degree never before possible.

6.4 Suggestions for Further Study

This study has shown the promising value of computer-assisted stemma con-
struction in the textual criticism of Galatians. The use of a computer, pro-
grammed to implement cladistic algorithms and taking contamination into
account, has helped greatly in the construction of a stemma representing the
textual history of the transmission of Galatians. Moreover, the generated
stemma enables the textual critic to focus on a particular group of readings
for assessing their value as a witness to Paul’s authorial text. Obviously, this
method can be applied to other portions of the text of Paul and indeed the
New Testament. It would be interesting to determine if the other books of the
Pauline corpus exhibit the same or similar histories of the text. Moreover, the
application of stemmatics outside of Paul 40 can help to resolve specific text-
critical controversies in those texts, as well as to get a greater understanding
of the history of textual variation within the transmission of the text.
Even without the application of the computer program, this study suggests
that the textual criticism of the Pauline corpus ought to value the agreements

39
By way of comparison, the SBL edition opts for the shorter reading in 1:15, 2:6, and
5:24.
40
Indeed, an earlier version of this software used in this study was applied to the ques-
tion of the Caesarean text in Mark. Stephen C. Carlson, “The Origin(s) of the ‘Caesarean’
Text” (paper presented at the annual meeting of the Society of Biblical Literature, San
Antonio, 20 Nov. 2004).
 6.4 Suggestions for Further Study 253

between P46, B, and the Western text more highly than the current practice as
reflected in the Nestle-Aland edition. This study has corroborated Zuntz’s
finding that such agreements demonstrate that they are ancient readings going
back to the second century. Therefore, they should be given more weight than
suggested by Hort and his successors.
This study has also shown that the recent advances in linguistics, especial-
ly in the field of pragmatics, enables the textual critic to make theoretically
informed judgments about the effect of word order changes, the presence or
absence of the article, and the use of discourse conjunctions. These theoreti-
cal advances in the exegesis of the text and its variants can be applied else-
where in New Testament and even in classical textual criticism.
 Appendix A

Errata for Swanson

Verse Page Group Line Relevant Reading MS Correction

1:1 1 3 8 αὐτὸν c Change to ¹
1:2 2 3 1 ὑπὲρ c Change to ²
1:3 2 1 10 Πχάρεις F Replace w/ χάρεις
1:3 2 1 10 Πχάρεις G Replace w/ χάρεις
1:4 2 3 4 ἐξέληται 33 Delete 33 & add line
for 33 w/ ἐξελεῖται
1:4 2 3 10 ἁμαρτελῶν Fc Replace w/ ἁμαρτειῶν
1:4 3 1 12 ἐνεστῶτος αἰῶνος c Change to ²
1:6 3 4 12 Χριστοῦ Fc Change to F³
1:7 4 2 1 καὶ θέλοντες c Change to ¹
1:8 4 3 5 εὐαγγελίζηται Add Gc to Dc
1:8 4 3 11 εὐαγγελίζαηται G Change to G*
1:8 5 1 11 ὑμᾶς D*.c2 Change to D*
1:8 5 1 12 ὑμῖν Dc.1 Change to D²
1:8 5 1 21 ὑμῖν c Change to ²
1:9 5 2 1 προειρήκαμεν c Change to ²
1:10 6 2 12 αἴτει Add Gc to Fc
1:10 6 2 13 αἰτεί α G Change to G*
1:11 7 1 1 τὸ εὐαγγέλιον (3×) Β First two are deleted
1:11 7 1 2 γὰρ c Change to ²
1:11 7 1 7 γὰρ ὑμεῖν D*.C2 Change to D* (no Dc2)
1:11 7 1 12 δὲ * Change to *,²
1:11 7 1 12 δὲ ὑμῖν Dc1 Change to Dc
1:11 7 2 1 ἔστι Β Change to Β* per NA27
1:11 7 2 4 ἔστιν Add Βc to  per NA27
1:12 8 1 1 οὔτε Dc1 Change to D²
1:12 8 1 13 οὐδὲ D*.c2 Change to D*
1:14 9 2 1 περισσοτέρως Add 46c to B
1:14 9 2 4 περισσότερον ὡς 46 Change to 46* & omit
ὡς
1:15 10 1 13 ἀφωρίσας Dc Change to D²
1:16 10 3 1 εὐαγγελίζωμαι Add D² to B
1:16 10 3 10 εὐαγγελίζομαι Dc Delete Dc
1:17 10 4 3 ···λθον 46 Replace w/ ·λθον
1:18 11 4 13 Πέτρον c Change to ²
1:19 12 1 12 ἶδον Add D¹ to 
1:19 12 1 12 ἶδον Add 1424c to 
1:19 12 1 18 εἶδον Dc Change to D¹
1:19 12 1 18 εἶδον 1424 Change to 1424*
256 Appendix A

Verse Page Group Line Relevant Reading MS Correction

1:20 12 2 6 ὑμῖν F* Delete F* & add line
for F* w/ υμην
1:21 12 3 6 κλίματα F Delete F
1:21 12 3 7 κλίμματα Add F to Gc
1:21 13 1 1 τῆς Κιλικίας c Change to ¹
1:22 13 2 14 ταῖς Dc Change to D²
2:2 15 2 1 ἔθνεσιν Β Change to Β*
2:2 15 2 12 ἔθνεσι Add Βc to 075
2:4 16 3 1 καταδουλώσουσιν Β Change to Β*
2:4 16 3 7 καταδουλώσωσιν Add Βc to 33
2:5 17 1 1 ἀλήθεια Add Dc to B
2:5 17 1 2 ········ε···ῃ C Remove line
2:5 17 1 20 ἀλήθια Dc Delete Dc
2:5 17 1 23 διαμένῃ Add Cc to A
2:6 17 2 1 εἶναί Add Fc to B
2:6 17 2 1 ποτε c Change to ²
2:6 17 2 2 εἶνέ F Change to F*
2:6 18 1 1 πρόσωπον θ̅ς̅ Add D¹ to B
ἀνθρώπου
2:6 18 1 7 θ̅ς̅ ἀνθρώπου D Change to D*,²
πρόσωπον
2:7 18 3 12 εἰδότες Add C to 46
2:7 18 3 13 εἰδόντες C Delete C
2:8 18 4 1 ὁ γὰρ ἐνεργήσας c Change to ¹
2:8 19 1 1 ἐνήργησε Β Change to Βc per NA27
2:8 19 1 7 ἐνήργησεν Add Β* to 910 per
NA27
2:8 19 1 14 Πέτρῳ – περιτομῆς c Change to ¹
2:8 19 1 15 ἐνήργησεν καὶ ἐμοὶ Dc Change to D²
2:8 19 1 26 καὶ μοὶ Fc Replace w/ καί μοι
2:8 19 2 5 ἐθνην F Change to Fc & add line
for F* w/ εἐθνην
2:9 20 1 14 μὲν c Change to ¹
2:10 20 2 8 μνημονεύωμεν Add Gc to F
2:10 20 2 9 μνημονεύμωμεν G Change to G*
2:11 21 1 1 ἦλθεν Β Change to Β*
2:11 21 1 5 ἦλθε Add Βc to 104
2:11 21 1 9 Ἀντιόχειαν Dc Change to D²
2:12 22 2 1 ὑπέστελλεν Β Change to Β*
2:12 22 2 20 ὑπέστελλε Add Βc to 075
2:13 22 3 11 καὶ c Change to ²
2:13 23 1 2 συναπήχθη  Change to * & add
line for ² w/
συνυπήχθη.
2:14 23 2 1 ὀρθοποδοῦσιν Β Change to Β*
2:14 23 2 1 εἶδον ὅτι οὐκ Dc Change to D² & add
line for D¹ w/ ἶδον ὅτι
οὐχ
 Errata for Swanson 257

Verse Page Group Line Relevant Reading Correction

MS
2:14 23 2 7 ἀλήθειαν G Delete G & add line for
G w/ ἀλήθιαν
2:14 23 2 11 ὀρθοποδοῦσι Add Βc to 
2:14 23 3 17 ἔμπροσθεν / ὑπάχων Dc Change to D²
2:14 24 1 1 οὐχὶ c Delete c & add line for
² w/ οὐχεὶ.
2:16 25 1 2 διακαιοῦται G Change to G*
2:16 25 1 3 δικαιοῦται Add Gc to C
2:16 25 1 5 εἰδότες ὅτι Dc Change to D²
2:16 26 1 1 δικαιωθήσεται Β Remove line, add B to
next.
2:16 26 1 10 διότι Dc Change to D²
2:18 27 1 18 συνίστημι Dc Change to D²
2:20 27 3 12 ζῶ c Change to ¹
2:20 28 1 12 καὶ 1424 Change to 1424c & add
line for 1424*vid w/out
καὶ
3:1 29 2 11 κατ’ Dc Change to D²
3:2 30 1 4 μαθεῖν θέλω Dc Change to D²
3:2 30 1 14 θέλω μαθεῖν Add D¹ to 
3:7 32 1 1 Γεινώσκετε * Delete *
3:7 31 1 1 Γεινώσκετε Β Change to Β*
3:7 32 1 3 Γινώσκετε Add * to Ψ
3:7 32 1 3 Γινώσκετε Add Βc to Ψ
c
3:7 32 1 16 Γεινώσκετε  Delete c
3:7 32 1 23 Γινώσκετε Add ² to K
3:8 33 1 1 Ἐνευλογηθήσονται F* Delete F*
3:8 33 1 10 Εὐλογηθήσονται Fc Change to F
3:10 33 3 1 εἰσίν Β Change to Β*
3:10 33 3 8 εἰσί Add Βc to Ψ
3:10 33 3 9 ἐμμένη 1243* Replace w/ ἐμμένῃ
3:10 33 3 10 ἐμμένη 1243c Replace w/ ἐμμένῃ
3:10 33 3 10 ἐμμένη 1424* Replace w/ ἐμμένῃ
3:10 33 3 11 ············ 1424* Replace w/ ὁ
κρεμάμενος
3:10 33 3 12 ἐμμένη 2464 Replace w/ ἐμμένῃ
3:10 33 3 17 ἐμμένι ἐν C Replace w/ ἐμμένει
3:10 33 4 19 ἐν c Change to ²
3:10 33 4 19 ὅτι Add 1874c to ²
3:10 33 4 20 γὰρ Ἐπικαράτος 1874 Change to 1874*
3:11 34 1 10 νόμῳ 33 Replace w/ νόμου
3:11 34 2 1 δῆλον, ὅτι Add D² to B
3:11 34 2 17 παρὰ τῷ θ̅ω̅ Dc Change D¹ and omit τῷ
3:12 34 3 1 ἀλλ’ Dc Change to D²
3:12 35 1 1 αὐτα ζήσεται Add D¹ to B
3:12 35 1 14 αὐτα αν̅ο̅ς ζήσεται Dc Change to D²
3:13 35 2 1 ἡμῶν F Delete F & add line for
F w/ ὑμῶν
258 Appendix A

Verse Page Group Line Relevant Reading MS Correction

3:13 35 2 1 τοῦ Add Gc
3:13 35 2 4 τοῦτ’ G Change to G* and
replace w/ τοῦ τ
3:13 35 2 9 γέγραπται γὰρ Dc Change to D²
3:14 35 4 7 γενήσεται C Delete C
3:14 35 4 7 γενήσεται 33 Delete 33
3:14 35 4 15 γένηται Add C to D
3:14 35 4 15 γένηται Add 33 to D
3:14 36 1 10 εὐλογίαν D*.c2 Change to D*
3:14 36 1 12 ἐπαγγελίαν Dc.1 Change to D²
3:16 37 1 1 τῷ δὲ Ἀβραὰμ C Delete C
3:16 37 1 1 τῷ δὲ Ἀβραὰμ Add D¹ to B*
3:16 37 1 2 τῷ Ἀβραὰμ Add C to D*
3:16 37 1 12 ἐρρήθησαν Dc Change to D²
3:16 37 2 13 σπέρμασιν, ὡς Dc Change to D²
3:16 37 3 1 ἔστιν Β Change to Β*
3:16 37 3 19 ἔστι Add Βc to C
3:17 38 2 8 λ̅ G Replace w/ α
3:18 38 3 1 ἡ 1424c Delete 1424c
3:18 38 3 3 ἡ 1424* Omit ἡ
3:18 38 3 5 νόμου κληρονομία Add 1424c to 1234
3:19 39 3 19 ἀγγέλου Add C to 945
3:19 39 3 28 ἀγγέλων C Change to Cc
3:19f 40 1 1 μεσείτου/μεσείτης Β Change to Β*
3:20 40 1 3 εἷς ἐστιν F Change to Fc & add line
for F* εετιν
3:19f 40 1 13 μεσίτου/μεσίτης Add Βc to A
3:20 40 1 13 οὐκ Dc Change to D² & add
line for D¹ w/ οὐχ
3:20 40 2 2 οὖ νόμος F* Replace w/ οὖν ομος
3:21 40 3 1 δυνάμενος C Change to Cc & add
line for C*vid w/
δυναμος
3:22 41 1 16 συνέκλεισεν Dc Change to D²
3:22 41 2 1 ἐκ πίστεως ι̅υ̅ Add 1424c to B
3:22 41 2 3 ἐκ πίστεως 1424 Change to 1424*
3:22 41 3 1 χ̅υ̅ 1424* Change to 1424
3:22 41 3 9 χ̅ν̅ 1424c Remove line
3:23 42 2 9 συνκλιόμενοι D* Add line for D¹ w/
συνκεκλίσμενοι
3:23 42 2 17 συγκεκλείσμενοι Dc Change to D²
3:24 43 1 8 εἰς χ̅ν̅ ι̅ν̅ D Change to D*
3:24 43 1 20 εἰς χ̅ν̅ Add D² to 
3:26 43 3 1 θ̅υ̅ c Change to ¹
3:28 45 1 1 καὶ 1874 Delete 1874 & add line
for 1874 w/ ἢ
3:28 45 1 16 ἐστε ἐν χ̅υ̅ * Add line for ¹ w/out
ἐν
3:28 45 1 17 εἷς ἐστε ἐν χ̅ω̅ c Change to ²
 Errata for Swanson 259

Verse Page Group Line Relevant Reading MS Correction

3:29 45 2 1 ὑμεῖς χ̅υ̅, αρα τοῦ Dc.1 Delete Dc.1
3:29 45 2 3 ὑμεῖς A Delete A & add line for
A w/ ·μεῖς
3:29 45 2 3 ὑμεῖς χ̅υ̅, αρα τοῦ Add D² to 
3:29 45 2 4 αρα οὖν τοῦ Dc.2 Remove line
3:29 45 2 5 ὑμεῖς εἷς χ̅ω̅ ι̅υ̅ αρα D* Add line for D¹ w/out
οὖν τοῦ οὖν
3:29 45 3 9 κατὰ G Change to G* & add
line for Gc w/ κατ’
4:1 46 2 19 πάντων ὤν, C Change to Cc & add
line for C* w/out ὤν
4:2 46 2 9 ἐπιτρόπους Add Gc to F
4:2 46 2 10 ἐπιπτρόποις G Change to G*
4:2 46 3 1 ἔστιν Β Change to Β*
4:2 46 3 23 ἔστι Add Βc to C
4:4 47 2 1 ἦλθεν Β Change to Β*
4:4 47 2 10 ἦλθε Add Βc to 056
4:5 48 2 1 ἀπολάβωμεν 1424 Change to 1424c
4:6 48 2 1 ἐξαπέστειλεν Β Change to Β* & add
line for Βc w/
ἐξαπέστειλε
4:5 48 2 5 λάβωμεν Add 1424* to 1242
4:6 48 3 8 υἱοῖς F Change to F*
4:6 48 3 9 υἱός Add Fc to G
4:6 48 3 9 κράξομεν G Change to G* & add
line for Gc w/ κράζομεν
4:6 48 3 20 ὑμῶν Dc Change to D²
4:7 49 1 1 ἀλλὰ D Change to D*
4:7 49 1 13 ἀλλ’ Add D² to 69
4:7 49 2 13 θ̅υ̅ διὰ χ̅υ̅ c Change to ²
4:8 50 1 11 ἰδότες D Change to D* & add
line for D² w/ εἰδότες
4:8 50 1 13 φύσειν Add line for D¹ w/
φύσει
4:8–9 50 2 5 νυνεὶ Add line for D¹ w/ νῦν
4:8–9 50 2 5 οὖσι / νυνεὶ Add line for D² w/
οὖσιν / νῦν
4:8–9 50 2 20 οὖσιν / νυνεὶ Dc.2 Remove line
4:12 52 2 1 Γείνεσθε Β Change to Β*
4:12 52 2 3 δέομαι F Change to Fc & add line
for F* w/ δθομαι
4:12 52 2 16 Γίνεσθε Add Βc to C
4:13 52 3 1 ἀσθένειαν Dc Change to D²
4:13 52 3 4 ἀσθένιαν Add D¹ to 1175
4:14 53 2 2 οὐκ Add ¹ to A
4:14 53 2 10 τὸν c Change to ²
4:14 53 3 1 ἐδέξασθέ Add C to B
4:14 53 3 4 ἐξεδέξασθέ C Remove line
4:15 54 2 4 ἀνεδώκατέ c Delete c
260 Appendix A

Verse Page Group Line Relevant Reading MS Correction

4:15 54 2 8 ἀν ἐδώκατέ / ἐγὼ Dc Remove line
4:15 54 2 15 ἀν ἐδώκατέ Add ² to K
4:15 54 2 15 ἀν ἐδώκατέ Add D² to K
4:17 55 1 1 ἀλλὰ ἐκκλεῖσαι Dc Change to D²
4:17 55 2 8 Dc Change to D²
4:19 56 1 1 ὠδείνω Β Change to Β*
4:19 56 1 3 ὠδίνω Add Βc to *
4:19 56 1 23 τεκνία / ἄχρις c Change to ²
4:22 57 2 9 ἔσχεν F Change to Fc & add line
for F* w/ εεχεν
4:24 58 3 1 εἰσιν / Σεινᾶ Β Change to Β*
4:24 58 3 3 εἰσιν δύο c Change to ²
4:24 58 3 14 εἰσι / Σινᾶ Add Βc to C
4:25 59 1 1 Σεινᾶ Β Change to Β*
4:25 59 1 1 ὄρος ἐστὶν D Change to D*,² & add
line for D¹ w/ ἐστιν
ὄρος
4:25 59 1 3 Σινᾶ Add Βc to A
4:25 59 2 1 συνστοιχεῖ Β Change to Β*
4:25 59 2 29 συστοιχεῖ Add Βc to 69
4:25 60 1 8 δὲ Dc Change to D²
4:26 60 2 1 ἥτις ἐστὶν Add ¹ to B
4:26 60 2 1 ἐστὶν Β Change to Β*
4:26 60 2 5 ἐστὶ Add Βc to Cc
4:26 60 2 11 πάντων c Change to ²
4:26 60 2 14 πάντων Cc Replace w/ πάντον
4:27 60 4 1 ὠδείνουσα Β Change to Β*
4:27 60 4 8 ὠδίνουσα Add Βc to 
4:27 61 1 1 ἢ τῆς Add Gc to B
4:27 61 1 6 ἢι G Change to G* &
replace w/ ης
4:28 61 2 3 ὑμῖς Add line for D¹ w/
ὑμεῖς
4:28 61 2 15 Ἰσαὰκ c Change to ²
4:28 61 2 15 ἡμεῖς Dc Change to D²
4:30 62 3 1 τοῦ υἱοῦ c Change to ¹
4:31 63 1 1 δίο Add D¹ to B
4:31 63 1 18 ἄρα Dc Change to D²
5:1 63 2 13 χ̅ς̅ ἡμᾶς c Change to ²
5:1 64 1 19 ἐνέχεσθε Dc Change to D¹ & add
line for D² w/ ἀνέχεσθε
5:2 65 1 15 Παῦλος c Change to ¹
5:2 65 1 15 Παῦλος 1424 Change to 1424c & add
line for 1424* w/
Παλος
5:2 65 2 1 ὠφελήσει C Delete C & add line for
C w/ ὠφειλήσει
5:3 65 2 1 πάλιν Dc Change to D²
 Errata for Swanson 261

Verse Page Group Line Relevant Reading MS Correction

5:3 65 3 1 ὀφειλέτης Add Dc to B
5:3 65 3 5 ἀνθρώπῳ / ὅτι Add ² to F
5:3 65 3 6 ὀφιλέτης D Change to D*
5:3 65 3 14 ἀνθρώπων * Replace w/ ἀνθρώπῳ
5:3 65 3 15 ἀνθρώπων / ὅτι c Remove line
5:4 66 1 10 τοῦ Dc Change to D²
5:4 66 2 1 ἐξεπέσατε Add D¹ to B
5:4 66 2 10 δικαιοῦσθε G Replace w/ δικαιοῦσθαι
5:4 66 2 17 ἐξεπέσετε Dc Change to D²
5:5 66 3 13 ἀπεκδεχόμεθα c Change to ²
5:6 67 1 9 ἐν γὰρ Add Gc to F
5:6 67 1 10 ἐν G Change to G*
5:7 67 3 3 τῇ c Change to ²
5:7 68 1 1 πείθεσθαι Β Change to Βc
5:7 68 1 7 πείθεσθαι Dc Change to D²
5:7 68 1 14 πείθεσθε Add Β* to 614
5:7 68 1 18 τοῦ F Change to Fc & add line
for F* w/ τοι
5:7 68 1 21 πίθεσθαι D* Add line for D¹ w/
πείθεσθε
5:9 68 2 1 μεικρά Β Change to Β*
5:9 68 2 5 δόλοι D* Replace w/ δολοῖ
5:9 68 2 16 μικρά Add Βc to 
5:9 68 2 16 ζυμοῖ Dc Change to D²
5:10 68 2 13 ἐγὼ δὲ πέποιθα C Change to C*
5:10 68 2 16 ἐγὼ πέποιθα Add Cc to 
5:10 68 3 10 φρονήσεται· ὁ δὲ G Change to G*
5:10 68 3 11 χ̅ρ̅υ̅ F Add line for Gc &
replace w/ χ̅υ̅
5:11 69 2 1 ἔτι κηρύσσω Add Cc to B
5:11 69 2 1 ἔτι κηρύσσω Dc Change to D²
5:11 69 2 3 κατήργηται F Change to Fc & add line
for F*vid w/ κατιργηται
5:11 69 2 9 εἴ τι C Change to C*
5:12 70 1 15 ὤφελον Dc Change to D²
5:12 70 2 1 οἱ G Change to Gc & add
line for G* w/ οὐ
5:13 70 3 1 Ὑμεῖς Dc Change to D²
5:14 71 2 6 νόμος Dc Change to D²
5:14 71 3 15 Ἀγαπήσεις Add D¹ to F
5:14 71 3 20 ἐν τῷ· Ἀγαπήσεις Dc Change to D²
5:15 72 2 4 ὑπὸ Dc Change to D²
5:15 72 2 13 ὑπὸ c Change to ²
5:15 72 2 13 ὑπὸ Add D¹ to c
5:16 72 3 6 τελέσετε Dc Change to D²
5:16 72 3 7 τελέσηται F Change to Fc & add line
for F* w/ γελεσηται
5:17 72 4 11 ἐπιθυμῆ L Replace w/ ἐπιθυμῇ
262 Appendix A

Verse Page Group Line Relevant Reading MS Correction

5:17 73 1 4 τὸ δὲ τὸ πν̅α̅ 46* Change to 46c
5:17 73 1 5 τὸ δὲ πν̅α̅ 46c Change to 46*
5:17 73 1 12 δὲ Dc Change to D²
5:17 73 1 16 δὲ c Change to ²
5:17 73 2 11 ἃ θέλητε C Change to C*
5:17 73 2 27 ἃ ἂν θέλητε Add Cc to Dc
5:18 73 2 4 π̅ν̅ι̅ 1874c Delete 1874c & add line
for 1874c w/ π̅ν̅τι
5:18 73 3 4 οὐκέτι ἐστὲ Add ² to 1175
5:18 73 3 9 οὐκ τις c Remove line
5:19 73 3 1 ἐστι Β Change to Bc & add
line for B* w/ ἐστὶν̣ <
NA27
5:19–20 74 1 1 ἐστιν / ἀκαθαρσεία / Β Change to Β*
εἰδωλολατρεία
5:19–20 74 1 5 ἐστι / ἀκαθαρσία / Add Βc to 1241s
εἰδωλολατρία
5:19 74 1 19 μοιχία c Change to ²
5:19 74 1 20 ἀσέλγια Add line for D¹ w/
ἀσέλγεια
5:19–20 74 1 21 μοιχεία, πορνεία / Dc Change to D²
εἰδωλολατρεία
5:20 74 2 1 ἐρειθεῖαι Β Change to Β* & add
line for Bc w/ ἐριθεῖαι
5:20 74 2 1 ἐρειθεῖαι u Replace w/ ἐριθεῖαι
5:20 74 2 24 φαρμακία / ἐριθῖαι Add D¹ to [w]
5:20 74 2 30 φαρμακεία / ἐριθεῖαι Dc Change to D²
5:21 75 1 21 μέθαι Dc Change to D¹
5:21 75 2 12 καὶ c Change to ²
5:21 75 2 12 προεῖπον Dc Change to D²
5:23 76 2 8 πραότης 1735 Delete 1735 & add line
for 1735 w/ πραότις
5:23 76 3 1 ἔστιν Β Change to Β*
5:23 76 3 1 ἔστιν C Delete C
5:23 76 3 15 ἔστι Add Βc to L
5:23 76 3 15 ἔστι Add C to L
5:24 77 1 1 παθήμασιν Β Change to Β*
5:24 77 1 4 τοῦ χ̅υ̅ c Change to ¹
5:24 77 1 4 παθήμασι Add Βc to c
5:25 77 3 21 στοιχοῦμεν Dc Change to D²
5:26 77 3 1 γεινώμεθα Β Change to Β*
5:26 77 3 3 γεινώμεθα G Change to Gc & add
line for G* w/
γενώμεθα
5:26 77 3 28 γινώμεθα Add Βc to 
5:26 77 3 28 γινώμεθα Add D¹ to 
6:1 78 2 1 προλημφθῇ Β Change to Β*
6:1 78 2 12 προληφθῇ Add Βc to L
 Errata for Swanson 263

Verse Page Group Line Relevant Reading MS Correction

6:1 78 4 8 πειρασθεῖς Add 1874 to L
6:1 78 4 9 σὺ πειρασθῃς Dc Change to D²
6:1 78 4 10 ἔκαστος F Replace w/ ἕκαστος
6:1 78 4 10 πιρασθῇ F Change to Fc & add line
for F* w/ πιραθη
6:1 78 4 11 πιρασθῇς D* Add line for D¹ w/
πειρασθῃς
6:2 79 1 11 ἀναπληρώσατε c Change to ²
6:3 79 3 8 αυτοῦ F* Remove line
6:3 79 3 9 ἑαυτοῦ Fc Change to F
6:7 80 3 5 μυκτηρίζεται Dc Change to D¹
6:7 81 1 1 ἂν σπείρῃ Add D¹ to B
6:7 81 1 9 ἐὰν σπείρῃ Dc Change to D²
6:8 81 2 1 ἑαυτοῦ Dc Change to D²
6:8 81 2 1 ἑαυτοῦ 1424 Change to 1424c
6:8 81 2 1 θερίσει 1735 Change to 1735c
6:8 81 2 5 θερήσει Add 1735*vid to 88
6:8 81 2 8 θερίσει c Change to ²
6:8 81 2 9 σπίρων D* Add line for D¹ w/
σπείρων
6:8 81 2 14 αυτοῦ Add 1424* to Ψ
6:8 81 3 8 εἰς D* Change to ἐκ
6:11 83 2 8 πιλείκοις G Replace w/ πηλείκοις
6:13 84 4 3 μόνον ... 1874 Change to 1874*
6:14 85 2 1 καυχᾶσθαι Dc Change to D²
6:14 85 2 12 καυχήσασθαι D*.c2 Change to D*
6:14 85 3 1 κἀγὼ κόσμῳ C Change to C*
6:14 85 3 11 κἀγὼ τῷ κόσμῳ Add Cc to Dc
6:14 85 3 11 κἀγὼ τῷ κόσμῳ Dc Change to D²
6:15 86 1 19 ἰσχυει c Change to ²
6:15 86 1 19 ἰσχυει Dc Change to D²
6:16 86 2 1 στοιχήσουσιν Add Cc to B
6:16 86 2 6 στοιχ·············· C Remove line
6:16 86 2 22 στοίχουσιν Add C* to D
6:16 86 3 4 ἐπὶ Add line for D¹ w/ καὶ
ἐπὶ
6:16 86 3 8 τοῦ θ̅υ̅ Dc Change to D²
6:17 87 1 1 λοιποῦ F Delete F & add line for
F w/ λοιποὶ
6:17 87 2 6 χ̅υ̅ Dc.1 Delete Dc.1
6:17 87 2 18 τοῦ κ̅υ̅ ι̅υ̅ Dc.2 Change to Dc
 Appendix B

Critical Text of the Byzantine Prototype

ΠΡΟΣ ΓΑΛΑΤΟΥΣ *

1 Παῦλος ἀπόστολος οὐκ ἀπ’ ἀνθρώπων οὐδὲ δι’ ἀνθρώπου ἀλλὰ διὰ Ἰησοῦ
Χριστοῦ καὶ θεοῦ πατρὸς τοῦ ἐγείραντος αὐτὸν ἐκ νεκρῶν, 2 καὶ οἱ σὺν ἐμοὶ
πάντες ἀδελφοὶ ταῖς ἐκκλησίαις τῆς Γαλατίας, 3 χάρις ὑμῖν καὶ εἰρήνη ἀπὸ
θεοῦ πατρὸς καὶ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ 4 τοῦ δόντος ἑαυτὸν περὶ
τῶν ἁμαρτιῶν ἡμῶν, ὅπως ἐξέληται ἡμᾶς ἐκ τοῦ ἐνεστῶτος αἰῶνος πονηροῦ
κατὰ τὸ θέλημα τοῦ θεοῦ καὶ πατρὸς ἡμῶν, 5 ᾧ ἡ δόξα εἰς τοὺς αἰῶνας τῶν
αἰώνων, ἀμήν.
6 Θαυμάζω ὅτι οὕτως ταχέως μετατίθεσθε ἀπὸ τοῦ καλέσαντος ὑμᾶς ἐν
χάριτι Χριστοῦ εἰς ἕτερον εὐαγγέλιον, 7 ὃ οὐκ ἔστιν ἄλλο, εἰ μή τινές εἰσιν οἱ
ταράσσοντες ὑμᾶς καὶ θέλοντες μεταστρέψαι τὸ εὐαγγέλιον τοῦ Χριστοῦ.
8 ἀλλὰ καὶ ἐὰν ἡμεῖς ἢ ἄγγελος ἐξ οὐρανοῦ εὐαγγελίζηται ὑμῖν παρ’ ὃ
εὐηγγελισάμεθα ὑμῖν, ἀνάθεμα ἔστω. 9 ὡς προειρήκαμεν καὶ ἄρτι πάλιν
λέγω· εἴ τις ὑμᾶς εὐαγγελίζεται παρ’ ὃ παρελάβετε, ἀνάθεμα ἔστω.
10 Ἄρτι γὰρ ἀνθρώπους πείθω ἢ τὸν θεόν; ἢ ζητῶ ἀνθρώποις ἀρέσκειν; εἰ
γὰρ ἔτι ἀνθρώποις ἤρεσκον, Χριστοῦ δοῦλος οὐκ ἂν ἤμην.

11 Γνωρίζω δὲ ὑμῖν, ἀδελφοί, τὸ εὐαγγέλιον τὸ εὐαγγελισθὲν ὑπ’ ἐμοῦ ὅτι

οὐκ ἔστιν κατὰ ἄνθρωπον· 12 οὐδὲ γὰρ ἐγὼ παρὰ ἀνθρώπου παρέλαβον αὐτὸ
οὔτε ἐδιδάχθην, ἀλλὰ δι’ ἀποκαλύψεως Ἰησοῦ Χριστοῦ.
13 Ἠκούσατε γὰρ τὴν ἐμὴν ἀναστροφήν ποτε ἐν τῷ Ἰουδαϊσμῷ, ὅτι καθ’
ὑπερβολὴν ἐδίωκον τὴν ἐκκλησίαν τοῦ θεοῦ καὶ ἐπόρθουν αὐτήν, 14 καὶ
προέκοπτον ἐν τῷ Ἰουδαϊσμῷ ὑπὲρ πολλοὺς συνηλικιώτας ἐν τῷ γένει μου,
περισσοτέρως ζηλωτὴς ὑπάρχων τῶν πατρικῶν μου παραδόσεων.
15 Ὅτε δὲ εὐδόκησεν ὁ θεὸς ὁ ἀφορίσας με ἐκ κοιλίας μητρός μου καὶ
καλέσας  διὰ τῆς χάριτος αὐτοῦ 16 ἀποκαλύψαι τὸν υἱὸν αὐτοῦ ἐν ἐμοί,

* MSS = DC L 131 226 547 876 910 927 1270 1315 1352 1448 1734 1735 1854 2125
2147 2344 2401 2423 … Chrys–hark–1611 Ψ
1:3  1 2 547 910 1448 2147 2423 … Chrys comm ¦ 3 1 2 131 876 2125 … Chrys lem Ψ
1:4  ὑπέρ 547 876 927 1315 1352 1448 1735 … Chrys
1:6  οὕτω 547 1352 2423 HF
1:8  εὐαγγελίζεται 131 910 927 1270 1315 1352 1734 2125 2147 … Chrys RP-MARG ¦
txt DC L 226 547 (εὐη- 876) 1448 1735 1854 2401 2423 … 1611 Ψ RP-TEXT
1:11  εὐηγγελισθὲν 131 1735
 Critical Text of the Byzantine Prototype 265

ἵνα εὐαγγελίζωμαι αὐτὸν ἐν τοῖς ἔθνεσιν, εὐθέως οὐ προσανεθέμην σαρκὶ

καὶ αἵματι 17 οὐδὲ ἀνῆλθον εἰς Ἱεροσόλυμα πρὸς τοὺς πρὸ ἐμοῦ ἀποστό-
λους, ἀλλὰ ἀπῆλθον εἰς Ἀραβίαν καὶ πάλιν ὑπέστρεψα εἰς Δαμασκόν.
18 Ἔπειτα μετὰ ἔτη τρία ἀνῆλθον εἰς Ἱεροσόλυμα ἱστορῆσαι Πέτρον καὶ
ἐπέμεινα πρὸς αὐτὸν ἡμέρας δεκαπέντε, 19 ἕτερον δὲ τῶν ἀποστόλων οὐκ
εἶδον εἰ μὴ Ἰάκωβον τὸν ἀδελφὸν τοῦ κυρίου. 20 ἃ δὲ γράφω ὑμῖν, ἰδοὺ
ἐνώπιον τοῦ θεοῦ ὅτι οὐ ψεύδομαι.
21 Ἔπειτα ἦλθον εἰς τὰ κλίματα τῆς Συρίας καὶ τῆς Κιλικίας· 22 ἤμην δὲ
ἀγνοούμενος τῷ προσώπῳ ταῖς ἐκκλησίαις τῆς Ἰουδαίας ταῖς ἐν Χριστῷ.
23 μόνον δὲ ἀκούοντες ἦσαν ὅτι ὁ διώκων ἡμᾶς ποτε νῦν εὐαγ-
γελίζεται τὴν πίστιν ἥν ποτε ἐ πόρθει, 24 καὶ ἐδόξαζον ἐν ἐμοὶ τὸν θεόν.
2:1 Ἔπειτα διὰ δεκατεσσάρων ἐτῶν πάλιν ἀνέβην εἰς Ἱεροσόλυμα μετὰ
Βαρναβᾶ συμπαραλαβὼν καὶ Τίτον· 2 ἀνέβην δὲ κατὰ ἀποκάλυψιν· καὶ
ἀνεθέμην αὐτοῖς τὸ εὐαγγέλιον ὃ κηρύσσω ἐν τοῖς ἔθνεσιν, κατ’ ἰδίαν δὲ τοῖς
δοκοῦσιν, μή πως εἰς κενὸν τρέχω ἢ ἔδραμον. 3 ἀλλ’ οὐδὲ Τίτος ὁ σὺν ἐμοί,
Ἕλλην ὤν, ἠναγκάσθη περιτμηθῆναι· 4 διὰ δὲ τοὺς παρεισάκτους ψευδαδέλ-
φους, οἵτινες παρεισῆλθον κατασκοπῆσαι τὴν ἐλευθερίαν ἡμῶν ἣν ἔχομεν ἐν
Χριστῷ Ἰησοῦ, ἵνα ἡμᾶς καταδουλώσωνται, 5 οἷς οὐδὲ πρὸς ὥραν εἴξαμεν τῇ
ὑποταγῇ, ἵνα ἡ ἀλήθεια τοῦ εὐαγγελίου διαμείνῃ πρὸς ὑμᾶς.
6 Ἀπὸ δὲ τῶν δοκούντων εἶναί τι, – ὁποῖοί ποτε ἦσαν οὐδέν μοι διαφέρει·
πρόσωπον  θεὸς ἀνθρώπου οὐ λαμβάνει – ἐμοὶ γὰρ οἱ δοκοῦντες οὐδὲν
προσανέθεντο, 7 ἀλλὰ τοὐναντίον ἰδόντες ὅτι πεπίστευμαι τὸ εὐαγγέλιον τῆς
ἀκροβυστίας καθὼς Πέτρος τῆς περὶτομῆς, 8 ὁ γὰρ ἐνεργήσας Πέτρῳ εἰς
ἀποστολὴν τῆς περιτομῆς ἐνήργησεν καὶ ἐμοὶ εἰς τὰ ἔθνη, 9 καὶ γνόντες
τὴν χάριν τὴν δοθεῖσάν μοι, Ἰάκωβος καὶ Κηφᾶς καὶ Ἰωάννης, οἱ δοκοῦντες
στῦλοι εἶναι, δεξιὰς ἔδωκαν ἐμοὶ καὶ Βαρναβᾷ κοινωνίας, ἵνα ἡμεῖς  εἰς τὰ
ἔθνη, αὐτοὶ δὲ εἰς τὴν περιτομήν· 10 μόνον τῶν πτωχῶν ἵνα μνημονεύωμεν,
ὃ καὶ ἐσπούδασα αὐτὸ τοῦτο ποιῆσαι.

1:15  με 1734 1854

1:16  εὐαγγελίζομαι L 131 927 1315 2147
1:17  ἀπῆλθον DC 2344
1:21  κλήματα L 910 1315 1735 2147 2401
1:23  1315 2401
1:23  πιστήν 2125 2147
1:23  ἐπόρθῃ L 1315 1735 2125
2:4  ἔχωμεν 1270 1352 1735 2147
2:5  διαμείνει 927 1315
2:6  ὁ 1270 1734 … Chrys–hark Ψ
2:8  Πέτρον L 2401
2:8  ἐνέργησεν 910 1448
2:8  κἀμοί 927 1448
2:9  μὲν DC 226 547 1315 1734 2344 RP-TEXT ¦ txt L 131 876 910 927 1270 1352 1448
1735 1854 2125 2147 2401 2423 RP-MARG
2:10  μνημονεύομεν DC 1315 1735 2401
266 Appendix B

11 Ὅτε δὲ ἦλθεν Πέτρος εἰς Ἀντιόχειαν, κατὰ πρόσωπον αὐτῷ ἀντέστην,

ὅτι κατεγνωσμένος ἦν. 12 πρὸ τοῦ γὰρ ἐλθεῖν τινας ἀπὸ Ἰακώβου μετὰ τῶν
ἐθνῶν συνήσθιεν· ὅτε δὲ ἦλθον, ὑπέστελλεν καὶ ἀφώριζεν ἑαυτὸν φοβού-
μενος τοὺς ἐκ περιτομῆς. 13 καὶ συνυπεκρίθησαν αὐτῷ καὶ οἱ λοιποὶ
Ἰουδαῖοι, ὥστε καὶ Βαρναβᾶς συναπήχθη αὐτῶν τῇ ὑποκρίσει. 14 ἀλλ’ ὅτε
εἶδον ὅτι οὐκ ὀρθοποδοῦσιν πρὸς τὴν ἀλήθειαν τοῦ εὐαγγελίου, εἶπον τῷ
Πέτρῳ ἔμπροσθεν πάντων·
εἰ σὺ Ἰουδαῖος ὑπάρχων ἐθνικῶς ζῇς καὶ οὐκ Ἰουδαϊκῶς, τί τὰ
ἔθνη ἀναγκάζεις ἰουδαΐζειν;
15 Ἡμεῖς φύσει Ἰουδαῖοι καὶ οὐκ ἐξ ἐθνῶν ἁμαρτωλοί· 16 εἰδότες  ὅτι οὐ
δικαιοῦται ἄνθρωπος ἐξ ἔργων νόμου ἐὰν μὴ διὰ πίστεως Ἰησοῦ Χριστοῦ, καὶ
ἡμεῖς εἰς Χριστὸν Ἰησοῦν ἐπιστεύσαμεν, ἵνα δικαιωθῶμεν ἐκ πίστεως
Χριστοῦ καὶ οὐκ ἐξ ἔργων νόμου, διότι οὐ δικαιωθήσεται ἐξ ἔργων νόμου
πᾶσα σάρξ. 17 εἰ δὲ ζητοῦντες δικαιωθῆναι ἐν Χριστῷ εὑρέθημεν καὶ αὐτοὶ
ἁμαρτωλοί, ἆρα Χριστὸς ἁμαρτίας διάκονος; μὴ γένοιτο. 18 εἰ γὰρ ἃ
κατέλυσα ταῦτα πάλιν οἰκοδομῶ, παραβάτην ἐμαυτὸν συνίστημι. 19 ἐγὼ γὰρ
διὰ νόμου νόμῳ ἀπέθανον, ἵνα θεῷ ζήσω. Χριστῷ συνεσταύρωμαι· 20 ζῶ δὲ
οὐκέτι ἐγώ, ζῇ δὲ ἐν ἐμοὶ Χριστός· ὃ δὲ νῦν ζῶ ἐν σαρκί, ἐν πίστει ζῶ τῇ τοῦ
υἱοῦ τοῦ θεοῦ τοῦ ἀγαπήσαντός με καὶ παραδόντος ἑαυτὸν ὑπὲρ ἐμοῦ.
21 Οὐκ ἀθετῶ τὴν χάριν τοῦ θεοῦ· εἰ γὰρ διὰ νόμου δικαιοσύνη, ἄρα
Χριστὸς δωρεὰν ἀπέθανεν.

3:1 Ὦ ἀνόητοι Γαλάται, τίς ὑμᾶς ἐβάσκανεν τῇ ἀληθείᾳ μὴ πείσθεσθαι, οἷς

κατ’ ὀφθαλμοὺς Ἰησοῦς Χριστὸς προεγράφη ἐν ὑμῖν ἐσταυρωμένος; 3 τοῦτο
μόνον θέλω μαθεῖν ἀφ’ ὑμῶν· ἐξ ἔργων νόμου τὸ πνεῦμα ἐλάβετε ἢ ἐξ ἀκοῆς
πίστεως; 3 οὕτως ἀνόητοί ἐστε, ἐναρξάμενοι πνεύματι νῦν σαρκὶ ἐπιτελεῖσθε;
4 τοσαῦτα ἐπάθετε εἰκῇ; εἴ γε καὶ εἰκῇ. 5 ὁ οὖν ἐπιχορηγῶν ὑμῖν τὸ πνεῦμα
καὶ ἐνεργῶν δυνάμεις ἐν ὑμῖν, ἐξ ἔργῶν νόμου ἢ ἐξ ἀκουῆς πίστεως;
6 Καθὼς Ἀβραὰμ ἐπίστευσεν τῷ θεῷ, καὶ ἐλογίσθη αὐτῷ εἰς δικαιοσύνη·
7 γινώσκετε ἄρα ὅτι οἱ ἐκ πίστεως, οὗτοί εἰσιν υἱοὶ Ἀβραάμ. 8 προϊδοῦσα δὲ
ἡ γραφὴ ὅτι ἐκ πίστεως δικαιοῖ τὰ ἔθνη ὁ θεός, προευηγγελίσατο τῷ Ἀβραάμ

2:12  2344 2423

2:12  ἀφόριζεν L 226 1315 2147 ¦ ὑφώριζεν 927
2:13  2 3 1 DC 1270 1315
2:14  οὐχί 131 … 1611 Ψ
2:14  πῶς DC 1270
2:14  ἀναγνάζῃς L 1315
2:16  δέ L 1734 2147
2:16  547 2401
2:16  ὄτι … hark–1611
2:16  3 4 5 1 2 ... hark–1611 ¦ 1 2 Ψ
3:1  ἐβάσκηνε 547 927 1352
 Critical Text of the Byzantine Prototype 267

ὅτι ἐνευλογηθήσονται ἐν σοὶ πάντα τὰ ἔθνη· 9 ὥστε οἱ ἐκ πίστεως εὐλογοῦνται

σὺν τῷ πιστῷ Ἀβραάμ.
10 Ὅσοι γὰρ ἐξ ἔργων νόμου εἰσιν, ὑπὸ κατάραν εἰσίν· γέγραπται γὰρ 
ἐπικατάρατος πᾶς ὃς οὐκ ἐμμένει ἐν πᾶσιν τοῖς γεγραμμένοις ἐν τῷ βιβλίῳ τοῦ
νόμου τοῦ ποιῆσαι αὐτά. 11 ὅτι δὲ ἐν νόμῳ οὐδεὶς δικαιοῦται παρὰ τῷ θεῷ
δῆλον, ὅτι ὁ δίκαιος ἐκ πίστεως ζήσεται· 12 ὁ δὲ νόμος οὐκ ἔστιν ἐκ πίστεως,
ἀλλ’ ὁ ποιήσας αὐτὰ ἄνθρωπος ζήσεται ἐν αὐτοῖς. 13 Χριστὸς ἡμᾶς ἐξηγό-
ρασεν ἐκ τῆς κατάρας τοῦ νόμου γενόμενος ὑπὲρ ἡμῶν κατάρα, γέγραπται
γάρ· ἐπικατάρατος πᾶς ὁ κρεμάμενος ἐπὶ ξύλου, 14 ἵνα εἰς τὰ ἔθνη ἡ εὐλογία
τοῦ Ἀβραὰμ γένηται ἐν Χριστῷ Ἰησοῦ, ἵνα τὴν ἐπαγγελίαν τοῦ πνεύματος
λάβωμεν διὰ τῆς πίστεως.
15 Ἀδελφοί, κατὰ ἄνθρωπον λέγω· ὅμως ἀνθρώπου κεκυρωμένην διαθή-
κην οὐδεὶς ἀθετεῖ ἢ ἐπιδιατάσσεται. 16 τῷ δὲ Ἀβραὰμ ἐρρήθησαν αἱ ἐπαγ-
γελίαι καὶ τῷ σπέρματι αὐτοῦ, οὐ λέγει· καὶ τοῖς σπέρμασιν, ὡς ἐπὶ πολλῶν
ἀλλ’ ὡς ἐφ’ ἑνός· καὶ τῷ σπέρματί σου, ὅς ἐστιν Χριστός. 17 τοῦτο δὲ λέγω·
διαθήκην προκεκυρωμένην ὑπὸ τοῦ θεοῦ εἰς Χριστὸν ὁ μετὰ ἔτη τετρα-
κόσια καὶ τριάκοντα γεγονὼς νόμος οὐκ ἀκυροῖ εἰς τὸ καταργῆσαι τὴν
ἐπαγγελίαν. 18 εἰ γὰρ ἐκ νόμου ἡ κληρονομία, οὐκέτι ἐξ ἐπαγγελίας· τῷ δὲ
Ἀβραὰμ δι’ ἐπαγγελίας κεχάρισται ὁ θεός.
19 Τί οὖν ὁ νόμος; τῶν παραβάσεων χάριν προσετέθη, ἄχρις οὗ ἔλθῃ τὸ
σπέρμα ᾧ ἐπήγγελται, διαταγεὶς δι’ ἀγγέλων ἐν χειρὶ μεσίτου. 20 ὁ δὲ
μεσίτης ἑνὸς οὐκ ἔστιν, ὁ δὲ θεὸς εἷς ἐστιν. 21 ὁ οὖν νόμος κατὰ τῶν
ἐπαγγελιῶν τοῦ θεοῦ; μὴ γένοιτο. εἰ γὰρ ἐδόθη νόμος ὁ δυνάμενος ζῳοποιῆ-
σαι, ὄντως ἂν ἐκ νόμου ἦν ἡ δικαιοσύνη· 22 ἀλλὰ συνέκλεισεν ἡ γραφὴ τὰ
πάντα ὑπὸ ἁμαρτίαν, ἵνα ἡ ἐπαγγελία ἐκ πίστεως Ἰησοῦ Χριστοῦ δοθῇ τοῖς
πιστεύουσιν.
23 Πρὸ τοῦ δὲ ἐλθεῖν τὴν πίστιν ὑπὸ νόμον ἐφρουρούμεθα συγκεκλεισμέ-
νοι εἰς τὴν μέλλουσαν πίστιν ἀποκαλυφθῆναι, 24 ὥστε ὁ νόμος παιδαγωγὸς

3:10  ὄτι DC … 1611 Ψ

3:10  2344 ... hark–1611 Ψ
3:14  γενήσεται L 226
3:16  ἐρρέθησαν 131 876 1270 2125 HF … Chrys–1611 Ψ
3:16  πέρμασιν 910 1735
3:17  κεκυρωμένην 876 1448 ¦ προκεκρυμμένην 1735
3:17  1270 1448
3:17  4 1 2 3 DC 1270 … Chrys–hark Ψ
3:17  τριακόσια 876 1315 1352
3:17  γέγονος 226 1734
3:19  τῷ 1315 1735
3:19  ὃ L 927 1270 1315 1734 2147
3:19  δέ 910 ¦ − 927 1315 1734 2147
3:23  2 1 547 876 ¦ 1 γάρ 927 ¦ 1 1315
268 Appendix B

ἡμῶν γέγονεν εἰς Χριστόν, ἵνα ἐκ πίστεως δικαιωθῶμεν· 25 ἐλθούσης δὲ τῆς

πίστεως οὐκέτι ὑπὸ παιδαγωγόν ἐσμεν.
26 Πάντες γὰρ υἱοὶ θεοῦ ἐστε διὰ τῆς πίστεως ἐν Χριστῷ Ἰησοῦ· 27 ὅσοι
γὰρ εἰς Χριστὸν ἐβαπτίσθητε, Χριστὸν ἐνεδύσασθε. 28 οὐκ ἔνι Ἰουδαῖος οὐδὲ
Ἕλλην, οὐκ ἔνι δοῦλος οὐδὲ ἐλεύθερος, οὐκ ἔνι ἄρσεν καὶ θῆλυ· πάντες γὰρ
ὑμεῖς εἷς ἐστε ἐν Χριστῷ Ἰησοῦ. 29 εἰ δὲ ὑμεῖς Χριστοῦ, ἄρα τοῦ Ἀβραὰμ
σπέρμα ἐστέ, καὶ κατ’ ἐπαγγελίαν κληρονόμοι.
4:1 Λέγω δέ, ἐφ’ ὅσον χρόνον ὁ κληρονόμος νήπιός ἐστιν, οὐδὲν διαφέρει
δούλου κύριος πάντων ὤν, 2 ἀλλὰ ὑπὸ ἐπιτρόπους ἐστὶν καὶ οἰκονόμους ἄχρι
τῆς προθεσμίας τοῦ πατρός. 3 οὕτως καὶ ἡμεῖς, ὅτε ἦμεν νήπιοι, ὑπὸ τὰ
στοιχεῖα τοῦ κόσμου ἦμεν δεδουλωμένοι· 4 ὅτε δὲ ἦλθεν τὸ πλήρωμα τοῦ
χρόνου, ἐξαπέστειλεν ὁ θεὸς τὸν υἱὸν αὐτοῦ, γενόμενον ἐκ γυναικός,
γενόμενον ὑπὸ νόμον, 5 ἵνα τοὺς ὑπὸ νόμον ἐξαγοράσῃ, ἵνα τὴν υἱοθεσίαν
ἀπολάβωμεν. 6 Ὅτι δὲ ἐστε υἱοί, ἐξαπέστειλεν ὁ θεὸς τὸ πνεῦμα τοῦ υἱοῦ
αὐτοῦ εἰς τὰς καρδίας ὑμῶν κράζον· αββα ὁ πατήρ. 7 ὥστε οὐκέτι εἶ δοῦλος
ἀλλὰ υἱός· εἰ δὲ υἱός, καὶ κληρονόμος θεοῦ διὰ  Χριστοῦ.
8 Ἀλλὰ τότε μὲν οὐκ εἰδότες θεὸν ἐδουλεύσατε τοῖς μὴ φύσει οὖσιν
θεοῖς· 9 νῦν δὲ γνόντες  θεόν, μᾶλλον δὲ γνωσθέντες ὑπὸ θεοῦ, πῶς ἐπι-
στρέφετε πάλιν ἐπὶ τὰ ἀσθενῆ καὶ πτωχὰ στοιχεῖα οἷς πάλιν ἄνωθεν
δουλεύειν θέλετε; 10 ἡμέρας παρατηρεῖσθε καὶ μῆνας καὶ καιροὺς καὶ
ἐνιαυτούς, 11 φοβοῦμαι ὑμᾶς μή πως εἰκῇ κεκοπίακα εἰς ὑμᾶς.
12 Γίνεσθε ὡς εγώ, ὅτι κἀγὼ ὡς ὑμεῖς, ἀδελφοί, δέομαι ὑμῶν. οὐδέν με
ἠδικήσατε· 13 οἴδατε δὲ ὅτι δι’ ἀσθένειαν τῆς σαρκὸς εὐηγγελισάμην ὑμῖν
τὸ πρότερον, 14 καὶ τὸν πειρασμόν μου τὸν ἐν τῇ σαρκί μου οὐκ ἐξου-
θενήσατε οὐδὲ ἐξεπτύσατε, ἀλλὰ ὡς ἄγγελον θεοῦ ἐδέξασθέ με, ὡς Χριστὸν
Ἰησοῦν. 15 τίς οὖν ἦν ὁ μακαρισμὸς ὑμῶν; μαρτυρῶ γὰρ ὑμῖν ὅτι εἰ δυνατὸν
τοὺς ὀφθαλμοὺς ὑμῶν ἐξορύξαντες ἂν ἐδώκατέ μοι. 16 ὥστε ἐχθρὸς ὑμῶν
γέγονα ἀληθεύων ὑμῖν; 17 ζηλοῦσιν ὑμᾶς οὐ καλῶς, ἀλλὰ ἐκκλεῖσαι ὑμᾶς

3:29  DC 910
3:29  κατεπαγγελίαν 910 2125
4:3  οὕτω 547 1352 2401 2423 HF
4:4  γεννώμενον (-ό- 2147) 226 910 2401
4:5  ἐξαγοράσηται 131 2147
4:6  ἡμῶν 1315 2344
4:6  κράζων 876 1315 1352 1734 1735 ¦ κράζοντας 2344
4:7  Ἰησοῦ 927 2147
4:8  2 1 DC 927
4:9  γνώντες 226 1735
4:9  τόν 1448 1854
4:9  ἀσθενεῖ 1352 2147
4:13  1448 1734 1735
4:15  L 876 2344 … Chrys–hark–1611 Ψ
 Critical Text of the Byzantine Prototype 269

θέλουσιν, ἵνα αὐτοὺς ζηλοῦτε· 18 καλὸν δὲ τὸ ζηλοῦσθαι ἐν καλῷ πάντοτε

καὶ μὴ μόνον ἐν τῷ παρεῖναί με πρὸς ὑμᾶς. 19 τεκνία μου, οὓς πάλιν ὠδίνω
ἄχρις οὗ μορφωθῇ Χριστὸς ἐν ὑμῖν· 20 ἤθελον δὲ παρεῖναι πρὸς ὑμᾶς ἄρτι
καὶ ἀλλάξαι τὴν φωνήν μου, ὅτι ἀποροῦμαι ἐν ὑμῖν.
21 Λέγετέ μοι, οἱ ὑπὸ νόμον θέλοντες εἶναι, τὸν νόμον οὐκ ἀκούετε;
22 γέγραπται γὰρ ὅτι Ἀβραὰμ δύο υἱοὺς ἔσχεν, ἕνα ἐκ τῆς παιδίσκης καὶ ἕνα
ἐκ τῆς ἐλευθέρας. 23 ἀλλ’ ὁ μὲν ἐκ τῆς παιδίσκης κατὰ σάρκα γεγέννηται, ὁ
δὲ ἐκ τῆς ἐλευθέρας διὰ τῆς ἐπαγγελίας. 24 ἅτινά ἐστιν ἀλληγορούμενα·
αὗται γάρ εἰσιν  δύο διαθήκαι, μία μὲν ἀπὸ ὄρους Σινᾶ εἰς δουλείαν γεν-
νῶσα, ἥτις ἐστὶν Ἁγάρ. 25 τὸ γὰρ Ἁγὰρ Σινᾶ ὄρος ἐστὶν ἐν τῇ Ἀραβίᾳ·
συστοιχεῖ δὲ τῇ νῦν Ἰερουσαλήμ, δουλεύει δὲ μετὰ τῶν τέκνων αὐτῆς. 26 ἡ
δὲ ἄνω Ἰερουσαλὴμ ἐλευθέρα ἐστίν, ἥτις ἐστὶν μήτηρ πάντων ἡμῶν·
27 γέγραπται γάρ· εὐφράνθητι, στεῖρα ἡ οὐ τίκτουσα, ῥῆξον καὶ βόησον, ἡ οὐκ
ὠδίνουσα· ὅτι πολλὰ τὰ τέκνα τῆς ἐρήμου μᾶλλον ἢ τῆς ἐχούσης τὸν ἄνδρα.
28 Ἡμεῖς δέ, ἀδελφοί, κατὰ Ἰσαὰκ ἐπαγγελίας τέκνα ἐσμέν. 29 ἀλλ’
ὥσπερ τότε ὁ κατὰ σάρκα γεννηθεὶς ἐδίωκεν τὸν κατὰ πνεῦμα, οὕτως καὶ
νῦν. 30 ἀλλὰ τί λέγει ἡ γραφή; ἔκβαλε τὴν παιδίσκην καὶ τὸν υἱὸν αὐτῆς· οὐ
γὰρ μὴ κληρονομήσῃ ὁ υἱὸς τῆς παιδίσκης μετὰ τοῦ υἱοῦ τῆς ἐλευθέρας.
³¹ ἄρα, ἀδελφοί, οὐκ ἐσμὲν παιδίσκης τέκνα ἀλλὰ τῆς ἐλευθέρας.
5:1 Τῇ ἐλευθερίᾳ οὖν ᾗ Χριστὸς ἡμᾶς ἠλευθέρωσεν στήκετε καὶ μὴ πάλιν
ζυγῷ δουλείας ἐνέχεσθε. 2 ἴδε ἐγὼ Παῦλος λέγω ὑμῖν ὅτι ἐὰν περιτέμνησθε,
Χριστὸς ὑμᾶς οὐδὲν ὠφελήσει. 3 μαρτύρομαι δὲ πάλιν παντὶ ἀνθρώπῳ
περιτεμνομένῳ ὅτι ὀφειλέτης ἐστὶν ὅλον τὸν νόμον ποιῆσαι. 4 κατηργήθητε
ἀπὸ τοῦ Χριστοῦ, οἵτινες ἐν νόμῳ δικαιοῦσθε, τῆς χάριτος ἐξεπέσατε.
5 ἡμεῖς γὰρ πνεύματι ἐκ πίστεως ἐλπίδα δικαιοσύνης ἀπεκδεχόμεθα. 6 ἐν γὰρ
Χριστῷ Ἰησοῦ οὔτε περιτομή τι ἰσχύει οὔτε ἀκροβυστία ἀλλὰ πίστις δι’
ἀγάπης ἐνεργουμένη.

4:18  131 2147

4:20  1734 2147
4:23  δι’ 1315 … Chrys–1611 Ψ
4:24  αἱ 1352 2147
4:25  1352 2344
4:25  τήν L 131 1315
4:25  δέ … 1611 ¦ νῦν L 131 547 927 2125 … Ψ
4:26  DC ... Chrys–1611 Ψ
4:28  ὑμεῖς 1270 2147
4:28  2147 2401
4:28  ἐστέ 1270
4:29  οὕτω 2344 2423 HF
4:30  κληρονομήσει DC 927 1734 1735 2423
5:1  DC … Chrys–hark–1611 Ψ
5:1  ἀνέχεσθε DC 2147
5:3  μαρτύρωμαι 1315 2344
5:4  ἐξεπέσετε DC 131 910 927 1352 1448 2147 HF RP-MARG
270 Appendix B

7 Ἐτρέχετε καλῶς· τίς ὑμᾶς ἐνέκοψεν τῇ ἀληθείᾳ μὴ πείθεσθαι; 8 ἡ πει-

σμονὴ οὐκ ἐκ τοῦ καλοῦντος ὑμᾶς. 9 μικρὰ ζύμη ὅλον τὸ φύραμα ζυμοῖ.
10 ἐγὼ  πέποιθα εἰς ὑμᾶς ἐν κυρίῳ ὅτι οὐδὲν ἄλλο φρονήσετε· ὁ δὲ
ταράσσων ὑμᾶς βαστάσει τὸ κρίμα, ὅστις ἂν ᾖ. 11 Ἐγὼ δέ, ἀδελφοί, εἰ
περιτομὴν ἔτι κηρύσσω, τί ἔτι διώκομαι; ἄρα κατήργηται τὸ σκάνδαλον τοῦ
σταυροῦ. 12 Ὄφελον καὶ ἀποκόψονται οἱ ἀναστατοῦντες ὑμᾶς.

13 Ὑμεῖς γὰρ ἐπ’ ἐλευθερίᾳ ἐκλήθητε, ἀδελφοί· μόνον μὴ τὴν ἐλευθερίαν εἰς
ἀφορμὴν τῇ σαρκί, ἀλλὰ διὰ τῆς ἀγάπης δουλεύετε ἀλλήλοις. 14 ὁ γὰρ
πᾶς νόμος ἐν ἑνὶ λόγῳ πληροῦται, ἐν τῷ· ἀγαπήσεις τὸν πλησίον σου
ὡς ἑαυτόν. 15 εἰ δὲ ἀλλήλους δάκνετε καὶ κατεσθίετε, βλέπετε μὴ ὑπ’
ἀλλήλων ἀναλωθῆτε.
16 Λέγω δέ, πνεύματι περιπατεῖτε καὶ ἐπιθυμίαν σαρκὸς οὐ μὴ τελέσητε.
17 ἡ γὰρ σὰρξ ἐπιθυμεῖ κατὰ τοῦ πνεύματος, τὸ δὲ πνεῦμα κατὰ τῆς σαρκός,
ταῦτα δὲ ἀντίκειται ἀλλήλοις, ἵνα μὴ ἃ ἂν θέλητε ταῦτα ποιῆτε. 18 εἰ
δὲ πνεύματι ἄγεσθε, οὐκ ἐστὲ ὑπὸ νόμον. 19 φανερὰ δέ ἐστιν τὰ ἔργα τῆς
σαρκός, ἅτινά ἐστιν πορνεία, μοιχεία, ἀκαθαρσία, ἀσέλγεια, 20 εἰδωλο-
λατρία, φαρμακεία, ἔχθραι, ἔρεις, ζῆλοι, θυμοί, ἐριθεῖαι, διχοστασίαι,
αἱρέσεις, 21 φθόνοι, φόνοι, μέθαι, κῶμοι καὶ τὰ ὅμοια τούτοις, ἃ προλέγω
ὑμῖν, καθὼς καὶ προεῖπον ὅτι οἱ τὰ τοιαῦτα πράσσοντες βασιλείαν θεοῦ οὐ
κληρονομήσουσιν. 22 ὁ δὲ καρπὸς τοῦ πνεύματός ἐστιν ἀγάπη χαρὰ εἰρήνη,
μακροθυμία χρηστότης ἀγαθωσύνη, πίστις 23 πραΰτης ἐγκράτεια· κατὰ τῶν
τοιούτων οὐκ ἔστιν νόμος. 24 οἱ δὲ τοῦ Χριστοῦ τὴν σάρκα ἐσταύρωσαν σὺν
τοῖς παθήμασιν καὶ ταῖς ἐπιθυμίαις.
25 Εἰ ζῶμεν πνεύματι, πνεύματι καὶ στοιχῶμεν. 26 μὴ γινώμεθα κενό-
δοξοι, ἀλλήλους προκαλούμενοι, ἀλλήλοις φθονοῦντες.

5:10  δέ 1734 2344

5:10  φρονήσητε 1735 2401
5:14  λόγος L 1270
5:14  ἀγαπήσῃς 1315 2147
5:14  σεαυτοῦ DC 1315 2147
5:17  ἐπιθύμῃ L 2147
5:17  DC 2147
5:17  ἀντίκεινται L 1315 1734 2147
5:17  θέλετε 1735 2401
5:17  ποιεῖτε 226 1270
5:20  ἔχθρα 1315 1735
5:20  ἔρις 131 226 876 927 1315 1352 1448 2147 … hark–1611
5:21  131 1448
5:25  στοιχοῦμεν DC L 1735 2147
5:26  γινόμεθα L 2147 ¦ γενόμεθα 131 1315 2344 2401 ¦ γενώμεθα 910 927 1270
5:26  προσκαλούμενοι L 131 927 1315 1734 2147 2401 ¦ − 876
5:26  ἀλλήλους 131 547 910 927 1315 1734 2125 2147 ¦ − 876
 Critical Text of the Byzantine Prototype 271

6:1 Ἀδελφοί, ἐὰν καὶ προληφθῇ ἄνθρωπος ἔν τινι παραπτώματι, ὑμεῖς οἱ

πνευματικοὶ καταρτίζετε τὸν τοιοῦτον ἐν πνεύματι πραΰτητος, σκοπῶν σεαυ-
τὸν μὴ καὶ σὺ πειρασθῇς.
2 Ἀλλήλων τὰ βάρη βαστάζετε καὶ οὕτως ἀναπληρώσατε τὸν νόμον τοῦ
Χριστοῦ. 3 εἰ γὰρ δοκεῖ τις εἶναί τι μηδὲν ὤν, ἑαυτὸν φρεναπατᾷ. 4 τὸ δὲ
ἔργον ἑαυτοῦ δοκιμαζέτω ἕκαστος, καὶ τότε εἰς ἑαυτὸν μόνον τὸ καύχημα
ἕξει καὶ οὐκ εἰς τὸν ἕτερον· 5 ἕκαστος γὰρ τὸ ἴδιον φορτίον βαστάσει.
6 Κοινωνείτω δὲ ὁ κατηχούμενος τὸν λόγον τῷ κατηχοῦντι ἐν πᾶσιν ἀγα-
θοῖς.
7 Μὴ πλανᾶσθε, θεὸς οὐ μυκτηρίζεται. ὃ γὰρ ἐὰν σπείρῃ ἄνθρωπος, τοῦτο
καὶ θερίσει· 8 ὅτι ὁ σπείρων εἰς τὴν σάρκα ἑαυτοῦ ἐκ τῆς σαρκὸς θερίσει
φθοράν, ὁ δὲ σπείρων εἰς τὸ πνεῦμα ἐκ τοῦ πνεύματος θερίσει ζωὴν αἰώνιον.
9 τὸ δὲ καλὸν ποιοῦντες μὴ ἐκκακῶμεν, καιρῷ γὰρ ἰδίῳ θερίσομεν μὴ ἐκλυ-
όμενοι.
10 Ἄρα οὖν ὡς καιρὸν ἔχομεν, ἐργαζώμεθα τὸ ἀγαθὸν πρὸς πάντας,
μάλιστα δὲ πρὸς τοὺς οἰκείους τῆς πίστεως.

11 Ἴδετε πηλίκοις ὑμῖν γράμμασιν ἔγραψα τῇ ἐμῇ χειρί.

12 Ὅσοι θέλουσιν εὐπροσωπῆσαι ἐν σαρκί, οὗτοι ἀναγκάζουσιν ὑμᾶς
περιτέμνεσθαι, μόνον ἵνα μὴ τῷ σταυρῷ τοῦ Χριστοῦ διώκονται. 13 οὐδὲ
γὰρ οἱ περιτετμημένοι αὐτοὶ νόμον φυλάσσουσιν ἀλλὰ θέλουσιν ὑμᾶς περι-
τέμνεσθαι, ἵνα ἐν τῇ ὑμετέρᾳ σαρκὶ καυχήσωνται.
14 Ἐμοὶ δὲ μὴ γένοιτο καυχᾶσθαι εἰ μὴ ἐν τῷ σταυρῷ τοῦ κυρίου ἡμῶν
Ἰησοῦ Χριστοῦ, δι’ οὗ ἐμοὶ κόσμος ἐσταύρωται κἀγὼ τῷ κόσμῳ. 15 ἐν γὰρ
Χριστῷ Ἰησοῦ οὔτε περιτομή τι ἰσχύει οὔτε ἀκροβυστία ἀλλὰ καινὴ κτίσις.
16 καὶ ὅσοι τῷ κανόνι τούτῳ στοιχήσουσιν, εἰρήνη ἐπ’ αὐτοὺς καὶ ἔλεος καὶ
ἐπὶ τὸν Ἰσραὴλ τοῦ θεοῦ.
17 Τοῦ λοιποῦ κόπους μοι μηδεὶς παρεχέτω· ἐγὼ γὰρ τὰ στίγματα τοῦ
κυρίου Ἰησοῦ ἐν τῷ σώματί μου βαστάζω.

6:1  πειρασθείς L 226 1315 1352 1735 2125 2147 2401

6:2  βάρει 1735 2147 2401
6:4  αὐτοῦ 1270 2423
6:7 σπέρει 1734 2147
6:9  θερίσωμεν L 927 1315 1448 2125 2147
6:9  ἐκκλθόμενοι 2125 2401
6:10  ἐργαζόμεθα L 876 1270 1315 1352 1735 2125 2147
6:12  διώκωνται DC 226 547 1315 1734 1854 2401 2423 … Chrys–1611 Ψ RP ¦ txt L
131 876 910 927 1270 1352 1448 1735 2125 2147
6:13  περιτεμνόμενοι DC 131 547 876 910 1315 1448 2125 2147 … Chrys–hark–1611
Ψ ¦ txt L 226 927 1270 1352 1734 1735 1854 2401 2423 RP
6:13  καυχήσονται DC 910 1315 1352 1735 2147
6:15  5 2 ... Chrys–1611 Ψ
272 Appendix B

18 Ἡ χάρις τοῦ κυρίου ἡμῶν Ἰησοῦ Χριστοῦ μετὰ τοῦ πνεύματος ὑμῶν,
ἀδελφοί· ἀμήν.
 Bibliography

Aland, Barbara, Kurt Aland, Johannes Karavidopoulos, Carlo M. Martini, and Bruce M.
Metzger, eds. Novum Testamentum Graece. Nestle-Aland 27th ed. Stuttgart: German
Bible Society, 1993.
–. The Greek New Testament. United Bible Societies 4th ed. Stuttgart: United Bible So-
cieties, 1994.
Aland, Barbara, Kurt Aland, Gerd Mink, and Klaus Wachtel, eds. Novum Testamentum
Graece: Editio Critica Maior. Vol. 4.1: James. Stuttgart: Deutsche Bibelgesellschaft,
1997.
–. Novum Testamentum Graece: Editio Critica Maior. Vol. 4.2: The Letters of Peter.
Stuttgart: Deutsche Bibelgesellschaft, 2000.
–. Novum Testamentum Graece: Editio Critica Maior. Vol. 4.3: The First Letter of John.
Stuttgart: Deutsche Bibelgesellschaft, 2003.
–. Novum Testamentum Graece: Editio Critica Maior. Vol. 4.4: The Second and Third
Letter of John, The Letter of Jude. Stuttgart: Deutsche Bibelgesellschaft, 2005.
Aland, Barbara, and Andreas Juckel. Das Neue Testament in syrischer Überlieferung. Vol.
2.2. Die paulischen Briefe. Berlin: de Gruyter, 1995.
Aland, Kurt, and Barbara Aland. The Text of the New Testament: An Introduction to the
Critical Editions and to the Theory and Practice of Modern Textual Criticism. 2d ed.
Trans. Erroll F. Rhodes. Grand Rapids, Mich.: Eerdmans, 1989.
Allen, W. Sidney. Vox Graeca: The Pronunciation of Classical Greek. 3d ed. Cambridge:
Cambridge University Press, 1987.
Alroy, John. “Continuous Track Analysis: A New Phylogenetic and Biogeographic Meth-
od.” Systematic Biology 44 (1995): 152–172.
Anderson, Amy S. The Textual Tradition of the Gospels: Family 1 in Matthew. New Testa-
ment Tools and Studies 32. Leiden: Brill, 2004.
Baarda, T. “Gal. 5.1a: ᾗ ἐλευθερίᾳ . . . Over de ‘Westerse Tekst’ en de Tekst van Mar-
cion.” Pages 173–193 in Christologische perspectieven: Exegetische en hermeneutische
studies. C. J. den Heyer, ed. Kampen: Kok, 1992.
Bakker, Stéphanie. “Adjective Ordering in Herodotus: A Pragmatic Explanation.” Pages
188–210 in The Language of Literature: Linguistic Approaches to Classical Texts.
Rutger Allan and Michel Buijs, eds. Amsterdam Studies in Classical Philology 13.
Leiden: Brill, 2007.
–. The Noun Phrase in Ancient Greek: A Functional Analysis of the Order and Articu-
lation of NP Constituents in Herodotus. Amsterdam Studies in Classical Philology 15.
Leiden: Brill, 2009.
Barrett, Charles Kingsley. “The Allegory of Abraham, Sarah, and Hagar in the Argument
of Galatians.” Pages 1–16 in Rechtfertigung: Festschrift für Ernst Käsemann zum 70.
Geburtstag. Ed. Johannes Friedrich et al. Tübingen: J. C. B. Mohr (Paul Siebeck),
1976.
274 Bibliography

Barrow, John D. The Constants of Nature: The Numbers that Encode the Deepest Secrets
of the Universe. New York: Vintage, 2004.
Bauer, Walter, ed. Griechisch-Deutsches Wörterbuch zu den Schriften des Neuen Testa-
ments und der übrigen urchristichen Literatur. 5th ed.; Berlin: Töpelmann, 1958.
Bauer, Walter, William F. Arndt, F. Wilbur Gingrich, and Frederick W. Danker, eds. A
Greek-English Lexicon of the New Testament and Other Early Christian Literature. 2d
ed.; Chicago: University of Chicago Press, 1979.
Bauer, Walter, Frederick W. Danker, William F. Arndt, and F. Wilbur Gingrich, eds. A
Greek-English Lexicon of the New Testament and Other Early Christian Literature. 3rd
ed. Chicago: The University of Chicago Press, 2000.
Baumert, Norbert. Der weg des Trauens: Übersetzung und Auslegung des Briefes an die
Galater und des Briefes an die Philipper. Würzburg: Echter, 2009.
Bédier, Joseph. “La tradition manuscrite du Lai de l’ombre: Réflexions sur l’art d’éditer les
anciens textes.” Romania 54 (1928): 161–196.
Bénevot, Maurice. The Tradition of Manuscripts: A Study in the Transmission of St. Cypri-
an’s Treatises. Oxford: University Press, 1961. Repr. Westport, Conn.: Greenwood
Press, 1979.
Bentley, Richard. Epistola ad Joannem Millium. Toronto: University of Toronto Press,
1962. Photogr. repr. of The Works of Richard Bentley, D.D. Ed. Alexander Dyce. Vol.
2. London: Macpherson, 1836.
Berényi, Gabriella. “Gal 2,20: a Pre-Pauline or a Pauline Text?” Biblica 64 (1984):490–
537.
Betz, Hans Dieter. Galatians: A Commentary on Paul’s Letter to the Churches in Galatia.
Hermeneia. Philadelphia: Fortress, 1979.
Birdsall, J. Neville. “The Text and Scholia of the Codex von der Goltz and its Allies, and
Their Bearing upon the Texts of the Works of Origen, Especially the Commentary on
Roman,” Pages 81–86 in J. Neville Birdsall, ed., Collected Papers in Greek and Geor-
gian Textual Criticism (Piscataway, N.J.: Gorgias, 2006).
Black, Stephanie L. Sentence Conjunctions in the Gospel of Matthew: καί, δέ, τότε, γάρ,
οὖν and Asyndeton in Narrative Discourse. Journal for the Study of the New Testament
Supplement Series 216. Studies in New Testament Greek 9. London: Sheffield Aca-
demic Press, 2002.
Blass, Friedrich, Albert Debrunner, and Friedrich Rehkopf, eds. Grammatik des
neutestamentlichen Griechisch. 14th ed. Göttingen: Vandenhoeck & Ruprecht, 1976.
Blass, Friedrich, Albert Debrunner, and Robert W. Funk, eds. A Greek Grammar of the
New Testament and Other Early Christian Literature. Chicago: The University of Chi-
cago Press, 1961.
Bligh, John. Galatians: A Discussion of St Paul’s Epistle. Householder Commentaries 1.
London: St Paul Publications, 1969.
Boer, Martinus C. de. Galatians: A Commentary. New Testament Library. Louisville, Ky.:
Westminster John Knox, 2011.
Bonnard, Pierre. L’Épître de Saint Paul aux Galates. 2d ed. Commentaire du Nouveau
Testament 9. Neuchâtel: Delachaux et Niestlé, 1972.
Borse, Udo. Der Brief an die Galater. Regensburger Neues Testament. Regensburg: Pustet,
1984.
–. “Ἀραβία.” Page 149 in Exegetical Dictionary of the New Testament. Eds. Horst Balz &
Gerhard Schneider. Vol. 1. Grand Rapids, Mich.: Eerdmans, 1990.
Bortone, Pietro. Greek Prepositions: From Antiquity to the Present. Oxford: Oxford Uni-
versity Press, 2010.
 Bibliography 275

Bowers, Fredson. Bibliography and Textual Criticism. Oxford: Clarendon, 1964.

Breytenbach, Cilliers. Versöhnung: Eine Studie zur paulinischen Soteriologie. WMANT
60. Neukirchener-Vluyn: Neukichener, 1989.
Brown, Andrew J. Opera omnia Desiderii Erasmi Roterodami: recognita et adnotatione
critica instructa notisque illustrate. Ordo 6. Tome 3. Amsterdam: Elsevier, 2004.
Brown, Raymond E. Jesus, God and Man: Modern Biblical Reflections. Milwaukee:
Bruce, 1967.
Burer, Michael H., W. Hall Harris III, and Daniel B. Wallace, eds. New English Trans-
lation: New Testament. Dallas: Net Bible Press, 2004.
Burton, Ernest de Witt. A Critical and Exegetical Commentary on the Epistle to the Gala-
tians. International Critical Commentary. Edinburgh: T. & T. Clark, 1921.
Buscemi, Alfio Marcello. L’uso delle preposizioni nella lettera ai Galati. Analecta (Stu-
dium Biblicum Franciscanum) 17. Jerusalem: Franciscan Printing Press, 1987.
–. Lettera ai Galati: Commentario esegetico. Analecta (Studium Biblicum Franciscanum)
63. Jerusalem: Franciscan Printing Press, 2004.
Buth, Randall. “On Levinsohn’s ‘Development Units’,” Selected Technical Articles Rela-
ted to Translation 5 (October 1981): 53–56.
Buttmann, Alexander. A Grammar of the New Testament Greek. Trans. J. G. Thayer. An-
dover, Mass.: Draper, 1873.
Buttmann, Philipp. Novum Testamentum Graece: ad fidem potissimum codicis Vaticani B
recensuit, varias lectiones codicis B, Textus Recepti, editionum Griesbachii Lachmanni
Tischendorfii integras adiecit. 3d rev. ed. Leipzig: Teubner, 1865.
Cameron, H. Don. “The Upside-Down Cladogram: Problems in Manuscript Affiliation.”
Pages 227–242 in Biological Metaphor and Cladistic Classification: An Interdiscipli-
nary Perspective. Ed. Henry M. Hoenigswald & Linda F. Wiener. Philadelphia: Univer-
sity of Pennsylvania Press, 1987.
Campbell, Douglas A. The Deliverance of God: An Apocalyptic Rereading of Justification
in Paul. Grand Rapids, Mich.: Eerdmans, 2009.
–. “Galatians 5.11: Evidence of an Early Law-observant Mission by Paul?” New Testa-
ment Studies 57 (2011): 325–347.
Carlson, Stephen C. “Clement of Alexandria on the ‘Order’ of the Gospels,” New Testa-
ment Studies 47 (2001): 118–125.
–. “The Origin(s) of the ‘Caesarean’ Text.” Paper presented at the annual meeting of the
Society of Biblical Literature, San Antonio, 20 Nov. 2004.
–. “‘For Sinai is a Mountain in Arabia’: A Note on the Text of Galatians 4,25.” Zeitschrift
für die neutestamentliche Wissenschaft 105 (2014): 80–101.
Clabeaux, John J. A Lost Edition of the Letters of Paul: A Reassessment of the Text of the
Pauline Corpus Attested by Marcion. Catholic Biblical Quarterly Monograph Series 21.
Washington, D.C.: The Catholic Biblical Association of America, 1989.
Clark, Kenneth W. Eight American Praxapostoloi. Chicago: University of Chicago Press,
1941.
Colwell, Ernest C. Studies in Methodology in Textual Criticism of the New Testament. New
Testament Tools and Studies 9. Grand Rapids, Mich.: Eerdmans, 1969.
–. “Method in Establishing the Nature of Text-Types of New Testament Manuscripts.”
Pages 45–55 in Colwell, Studies. Originally published as “The Original of Texttypes of
New Testament Manuscripts.” Pages 128–132 in Early Christian Origins: Studies in
Honor of Harold R. Willoughby. Ed. Allen Wikgren. Chicago: Quadrangle, 1961.
–. “Genealogical Method: Its Achievements and its Limitations.” Pages 63–83 in Colwell,
Studies. Originally published in Journal of Biblical Literature 66 (1947): 109–133.
276 Bibliography

–. “Hort Redivivus: A Plea and a Program.” Pages 148–171 in Colwell, Studies. Ori-
ginally published in Transitions in Biblical Scholarship. Ed. J. Coert Rylaarsdam.
Essays in Divinity 6. Chicago: University of Chicago Press.
Colwell, Ernest C. and Ernest W. Tune. “Method in Establishing Quantitative Relation-
ships Between Text-Types of New Testament Manuscripts.” Pages 56–62 in Colwell,
Studies. Originally published as “The Quantitative Relationships Between MS Text-
Types.” Pages 25–32 in Biblical and Patristic Studies in Memory of Robert Pierce
Casey. Eds. J. N. Birdsall and R. W. Thomson. Freiburg im Breisgan: Publication
(Vetus Latina Institut), 1963.
Comfort, Philip W. Early Manuscripts & Modern Translations of the New Testament.
Wheaton, Ill.: Tyndale House, 1990.
Comfort, Philip W., and David P. Barrett. The Text of the Earliest New Testament Greek
Manuscripts: New and Complete Transcriptions with Photographs. 2d ed. Wheaton,
Ill.: Tyndale House, 2001.
Conzelmann, Hans. 1 Corinthians: A Commentary on the First Epistle to the Corinthians.
Hermeneia. James W. Leitch, trans. Philadelphia: Fortress, 1975.
Davies, G. I. “Hagar, El-Heğra and the Location of Mount Sinai.” Vetus Testamentum 22
(1972): 152–163.
Denniston, J. D. The Greek Particles. 2d ed. Oxford: Clarendon Press, 1954.
Devine, A. M., and Laurence D. Stephens. The Prosody of Greek Speech. New York: Ox-
ford University Press, 1994.
Di Mattei, Steven. “Paul’s Allegory of the Two Covenants (Gal 4.21–31) in Light of First-
Century Hellenistic Rhetoric and Jewish Hermeneutics.” New Testament Studies 52
(2006): 102–122.
Dickerman, Allan W. “Generalizing Phylogenetic Parsimony from the Tree to the Forest,”
Systematic Biology 47 (1998): 414–426.
Dik, Helma. Word Order in Ancient Greek: A Pragmatic Account of Word Order Variation
in Herodotus. Amsterdam Studies in Classical Philology 5. Amsterdam: J. C. Gieben,
1995.
–. Word Order in Greek Tragic Dialogue. Oxford: Oxford University Press, 2007.
Dover, Kenneth J. Greek Word Order. Cambridge: Cambridge University Press, 1960.
Dunn, James D. G. The Epistle to the Galatians. Black’s New Testament Commentaries.
London: Black, 1993.
–. “Once More, ΠΙΣΤΙΣ ΧΡΙΣΤΟΥ.” Pages 249–271 in Hays, Faith. Orig. in E.H. Lover-
ing, Jr., ed., Society of Biblical Literature 1991 Seminar Papers (Atlanta: Scholars
Press, 1991); and E. Elizabeth Johnson and David M. Hay, eds., Pauline Theology Vol.
4: Looking Back, Pressing On. (Atlanta: Scholars Press, 1997), 61–81.
Eadie, John. A Commentary on the Greek Text of the Epistle of Paul to the Galatians.
Edinburgh: T. & T. Clark, 1879.
Ehrman, Bart D. The Orthodox Corruption of Scripture: The Effect of Early Christological
Controversies on the Text of the New Testament. New York: Oxford University Press,
1993.
–. “The Text as Window: New Testament Manuscripts and the Social History of Early
Christianity.” Pages 361–379 in Ehrman and Holmes, Text.
–. The Orthodox Corruption of Scripture: The Effect of Early Christological Controver-
sies on the Text of the New Testament. 2d ed. Oxford: Oxford University Press, 2011.
Ehrman, Bart D., and Michael W. Holmes, eds. The Text of the New Testament in Contem-
porary Research: Essays on the Status Quaestionis. Studies and Documents 46. Grand
Rapids, Mich.: Eerdmans, 1995.
 Bibliography 277

Elliger, Karl, and Wilhelm Rudolph, eds. Biblia Hebraica Stuttgartensia. 4th rev. ed.
Stuttgart: Deutsche Bibelgesellschaft, 1990.
Elliott, J. Keith. “Thoroughgoing Eclecticism in New Testament Textual Criticism.” Pages
321–335 in Ehrman and Holmes, Text.
–. A Bibliography of Greek New Testament Manuscripts. Society of New Testament Stu-
dies Monograph Series 109. 2d ed. Cambridge: University Press, 2000.
Elliott, Susan M. “Choose Your Mother, Choose Your Master: Galatians 4:21–5:1 in the
Shadow of the Anatolian Mother of the Gods,” Journal of Biblical Literature 118
(1999): 661–683.
Epp, Eldon Jay. The Theological Tendency of Codex Bezae Cantabrigiensis in Acts. Soci-
ety for New Testament Studies Monograph Series 3. Cambridge: Cambridge University
Press, 1966.
–. “The Twentieth Century Interlude in New Testament Textual Criticism.” Journal of
Biblical Literature 93 (1974): 386–414. Repr. Epp & Fee, Studies, 85–108.
–. “A Continuing Interlude in New Testament Textual Criticism?” Harvard Theological
Review 73 (1980): 131–151. Repr. Epp & Fee, Studies, 109–123.
–. “The Multivalence of the Term ‘Original Text’ in New Testament Textual Criticism.”
Harvard Theological Review 92 (1999): 245–281.
–. “Text-Critical, Exegetical, and Socio-Cultural Factors Affecting the Junia/Junias Vari-
ant in Romans 16,7.” Pages 227–291 in New Testament Textual Criticism and Exegesis.
A. Denaux, ed. Festschrift J. Delobel. BETL 161. Leuven: Leuven University Press,
2002.
–. Junia: The First Woman Apostle. Minneapolis: Fortress, 2005.
–. “It’s All about Variants: A Variant-Conscious Approach to New Testament Textual
Criticism.” Harvard Theological Review 100 (2007): 275–308.
Epp, Eldon Jay, and Gordon D. Fee, eds. Studies in the Theory and Method of New Testa-
ment Textual Criticism. Studies and Documents 45. Grand Rapids, Mich.: Eerdmans,
1993.
Eshbaugh, Howard. “Theological Variants in the Western Text of the Pauline Corpus.” Ph.
D. dissertation. Case Western Reserve University, 1975.
–. “Textual Variants and Theology: A Study of the Galatians Text of Papyrus 46.” Jour-
nal for the Study of the New Testament 3 (1979):60–72.
Farahian, Edmond. Le “je” paulinien: Étude pour mieux comprendre Gal. 2, 19–21. Ana-
lecta gregoriana 253. Rome: Editrice Pontificia Università Gregoriana, 1988.
Fee, Gordon D. Review of John J. Clabeaux, A Lost Edition of the Letters of Paul: A Re-
assessment of the Text of the Pauline Corpus Attested by Marcion. Catholic Biblical
Quarterly 53 (1991): 320.
–. “Rigorous or Reasoned Eclecticism–Which?” Pages 124–140 in Epp and Fee, Studies.
Originally published as pages 174–197 in Studies in New Testament Language and
Text: Essays in Honour of George D. Kilpatrick on the Occasion of His Sixty-Fifth
Birthday. Eds. J. K. Elliot. Leiden: Brill, 1976.
–. Pauline Christology: An Exegetical-Theological Study. Peabody, Mass.: Hendrickson,
2007.
Felsenstein, Joseph. Inferring Phylogenies. Sunderland, Mass.: Sinauer, 2004.
Fischer, Bonifatius. “Limitations of Latin in Representing Greek.” Pages 362–374 in The
Early Versions of the New Testament: Their Origin, Transmission, and Limitations.
Bruce M. Metzger. Oxford: Clarendon Press, 1977.
Fraenkel, Eduard. Kleine Beiträge zur klassischen Philologie. Vol. 1. Raccolta di studi e
testi 95. Rome: Edizioni di storia e letteratura, 1964.
278 Bibliography

Frede, Hermann Josef. Eine neuer Paulustext und Kommentar. II: Die Texte. Vetus Latina
8. Freiburg: Herder, 1974.
Froger, J. La critique des textes et son automatisation. Initiation aux nouveautés de la
science 7. Paris: Dunod, 1968.
Gamble, Harry Y. Books and Readers in the Early Church: A History of Early Christian
Texts. New Haven, Conn.: Yale University Press, 1995.
Gignac, Francis T. A Grammar of the Greek Papyri of the Roman and Byzantine Periods.
Vol. 1: Phonology. Milan; Instituto Editoriale Cisalpino - La Goliardica, 1976.
Goldstein, David Michael. “Wackernagel’s Law in Fifth-Century Greek.” Ph.D. dis-
sertation. University of California, Berkeley, 2010.
Greetham, David C. Textual Scholarship: An Introduction. Garland Reference Library of
the Humanities 1417. Corrected reprint. New York: Garland, 1994.
Greetham, David C., ed. Scholarly Editing: A Guide to Research. New York: Modern Lan-
guage Association of America, 1995.
Greg, W. W. “The Rationale of Copy-Text.” Studies in Bibliography 3 (1950): 19–36.
Grossouw, W. K. “Some Observations on the Text of the Pauline Epistles.” Pages 3–11 in
Mélanges offerts à Mademoiselle Christine Mohrmann. L. J. Engels et al., eds. Utrecht:
Spectrum, 1963.
Haines-Eitzen, Kim. Guardians of Letters: Literacy, Power, and the Transmitters of Early
Christian Literature. New York: Oxford University Press, 2000.
Hall, Robert G. “The Rhetorical Outline for Galatians: A Reconsideration.” Journal of Bib-
lical Literature 106 (1987): 277–287.
Harrison, James R. Paul’s Language of Grace in its Graeco-Roman Context. Wissen-
schaftliche Untersuchen zum Neuen Testament 2d. ser., 172. Tübingen: Mohr Siebeck,
2003.
Harvey, W. Wigan. Sancti Irenaei episcopi Lugdunensis: libros quinque adversus haere-
ses. 2 vols. Cambridge: Cambridge University Press, 1857.
Hay, David M. “Paul’s Indifference to Authority.” Journal of Biblical Literature 88
(1969): 36–44.
Hays, Richard B. The Faith of Jesus Christ: The Narrative Substructure of Galatians 3:1–
4:11. 2d ed. Grand Rapids, Mich.: Eerdmans, 2002.
–. “Πίστις and Pauline Christology: What Is at Stake?” Pages 272–297 in Hays, Faith.
Orig. in E.H. Lovering, Jr., ed., Society of Biblical Literature 1991 Seminar Papers
(Atlanta: Scholars Press, 1991); and E. Elizabeth Johnson and David M. Hay, eds.,
Pauline Theology Vol. 4: Looking Back, Pressing On. (Atlanta: Scholars Press, 1997),
35–60.
Head, Peter M. “Galatians 2.20: ‘I live by faith in God and Christ ...’.” Evangelical Textual
Criticism (March 15, 2006). Online: http://evangelicaltextualcriticism.blogspot.com
/2006/03/galatians-220–i-live-by-faith-in-god.html
Heckert, Jacob K. Discourse Function of Conjoiners in the Pastoral Epistles. Dallas: Sum-
mer Institute of Linguistics, 1996.
Hemphill, Wesley L. Codex Coxianus of the Homilies of Chrysostom on Ephesians and his
Commentary on Galatians. Thesis presented to the faculty of the graduate school of the
University of Pennsylvania in partial fulfillment of the requirements for the degree of
Ph.D. Norwood, Mass.: Norwood, 1916.
Hennig, Willi. Phylogenetic Systematics. English trans, D. Dwight Davis & Rainer
Zangerl. Urbana, Ill.: University of Illinois Press, 1966. Repr. 1999.
Hester, James D. “The Rhetorical Structure of Galatians 1:11–2:14.” Journal of Biblical
Literature 103 (1984): 223–233.
 Bibliography 279

–. “The Use and Influence of Rhetoric in Galatians 2:1–14.” Theologische Zeitschrift 42

(1986): 386–408.
Hodges, Zane C. and Arthur L. Farstad. The Greek New Testament According to the Majo-
rity Text. 2d ed. Nashville: Nelson, 1985.
Holmes, Michael W. “Reasoned Eclecticism in New Testament Textual Criticism.” Pages
336–360 in The Text of the New Testament in Contemporary Research: Essays on the
Status Quaestionis. Bart D. Ehrman and Michael W. Holmes, eds. Studies and Docu-
ments 46. Grand Rapids, Mich.: Eerdmans, 1995.
–. “Reasoned Eclecticism and the Text of Romans.” Pages 187–202 in Romans and the
People of God: Essays in Honor of Gordon D. Fee on the Occasion of His 65th Birth-
day. Eds. Sven K. Soderlund and N. T. Wright. Grand Rapids, Mich.: Eerdmans, 1999.
–. “The Case for Reasoned Eclecticism.” Pages 77–100 in Rethinking New Testament
Textual Criticism. Ed. David Alan Black. Grand Rapids, Mich.: Baker, 2002.
–. “Polycarp’s Letter to the Philippians and the Writings that later formed the New Testa-
ment.” Pages 187–227 in The Reception of the New Testament in the Apostolic Fathers.
Eds. Andrew F. Gregory and Christopher M. Tuckett. Oxford: Oxford University Press,
2005.
–. “The Text of Epistles Sixty Years After: An Assessment of Günther Zuntz’s Contribu-
tion to Text-Critical Methodology and History.” Pages 89–113 in Transmission and Re-
ception: New Testament Text-Critical and Exegetical Studies. Eds. J. W. Childers and
D. C. Parker. Texts and Studies 3d ser. 4. Piscataway, N.J.: Gorgias, 2006.
–. “The Text of  46: Evidence of the Earliest ‘Commentary’ on Romans?” Pages 189–206
in New Testament Manuscripts: Their Texts and Their World. Thomas J. Kraus and To-
bias Nicklas, eds. Texts and Editions for New Testament Study 2. Leiden: Brill, 2006.
–. “Working with an Open-Textual Tradition: Challenges in Theory and Practice.” Pages
65–78 in The Textual History of the Greek New Testament: Changing Views in Contem-
porary Research. Eds. Klaus Wachtel and Michael W. Holmes. Text-Critical Studies 8.
Atlanta: Society of Biblical Literature, 2011.
–. “From ‘Original Text’ to ‘Initial Text’: The Traditional Goal of New Testament Tex-
tual Criticism in Contemporary Discussion.” In The Text of the New Testament in Con-
temporary Research: Essays on the Status Quaestionis. Bart D. Ehrman and Michael W.
Holmes, eds. 2d ed. Leiden: Brill, 2012. Preprint.
Holmes, Michael W., ed. The Apostolic Fathers: Greek Texts and English Translations. 3d
ed. Grand Rapids, Mich.: Baker, 2007.
–. ed. The Greek New Testament: SBL Edition. Atlanta: Society of Biblical Literature,
2010.
Holsten, Carl. Das Evangelium des Paulus. Vol. 1. Berlin: Reimer, 1880.
Horn, Laurence R., and Gregory Ward, eds. The Handbook of Pragmatics. Blackwell
Handbooks in Linguistics 16. Malden, Mass.: Blackwell, 2004.
Horrocks, Geoffrey. Greek: A History of the Language and its Speakers. Longman Lingui-
stics Library. London: Longman, 1997.
Housman, Alfred E. M. Manilii Astronomicon, Liber Primus. London: Grant Richards,
1903.
Hurtado, Larry W. Text-Critical Methodology and the Pre-Caesarean Text: Codex W in the
Gospel of Mark. Studies and Documents 43. Grand Rapids, Mich.: Eerdmans, 1981.
–. “Beyond the Interlude? Developments and Directions in New Testament Textual Crit-
icism.” Pages 26–48 in Studies in the Early Text of the Gospels and Acts. D. G. K.
Taylor, ed. Atlanta: Society of Biblical Literature, 1999.
280 Bibliography

Joubert, Stephan. Paul as Benefactor: Reciprocity, Strategy and Theological Reflection in

Paul’s Collection. Wissenschaftliche Untersuchen zum Neuen Testament 2d. ser., 124.
Tübingen: Mohr Siebeck, 2000.
Kerkeslager, Allen. “Jewish Pilgrimage and Jewish Identity in Hellenistic and Early Ro-
man Egypt.” Pages 99–225 in Pilgrimage and Holy Space in Late Antique Egypt. Ed.
David Frankfurter. Leiden: Brill, 1998.
Kitching, Ian J., Peter L. Forey, Christopher J. Humphries, and David M. Williams. Cla-
distics: The Theory and Practice of Parsimony Analysis. The Systematics Association
Publication 11. Oxford: Oxford University Press, 2000.
Knox, John. Chapters in a Life of Paul. New York: Abingdon-Cokesbury, 1950.
Krans, Jan. Beyond What is Written: Erasmus and Beza as Conjectural Critics of the New
Testament. New Testament Tools and Studies 35. Leiden: Brill, 2006.
Krentz, Edgar. The Historical-Critical Method. Philadelphia: Fortress, 1975.
Lachmann, Karl. Novum Testamentum Graece et Latine. Vol. 2. Berlin: Reimer, 1850.
Lagrange, M.-J. Saint Paul Épître aux Galates. Études bibliques. Paris: Gabalda, 1950.
Lake, Kirsopp. Codex 1 of the Gospels and its Allies. Text and Studies 7. Cambridge: Cam-
bridge University Press, 1902.
Lake, Kirsopp, and Robert P. Blake. “The Text of the Gospels and the Koridethi Codex,’
Harvard Theological Review 16 (1923): 267–286.
Lake, Silva. Family Π and the Codex Alexandrinus: The Text According to Mark. Studies
and Documents 5. London: Christophers, 1936. Repr. Salt Lake City: University of
Utah Press, 1965.
Lambrecht, Knud. Information Structure and Sentence Form: Topic, Focus, and the Men-
tal Representation of Discourse Referents. Cambridge Studies in Linguistics 71. Cam-
bridge: Cambridge University Press, 1994.
Lee, III, Arthur R. “Numerical Taxonomy Revisited: John Griffith, Cladistic Analysis and
St. Augustine’s Quaestiones in Heptateuchum.” Studia Patristica 20 (1989): 24–32.
Originally delivered at the 10th International Conference of Patristic Studies, Oxford,
1987.
Légasse, Simon. L’épître de Paul aux Galates. Lectio Divina 9. Paris: Cerf, 2000.
Legendre, P., and V. Makarenkov. “Reconstruction of Biogeographic and Evolutionary
Networks Using Reticulograms.” Systematic Biology 51 (2002): 199–216.
Levinsohn, Stephen H. “Phrase Order and the Article in Galatians: A Functional Sentence
Perspective Approach.” Occasional Papers in Translation and Textlinguistics 3 (1989):
44–64.
–. “Some Constraints on Discourse Development in the Pastoral Epistles.” Pages 316–333
in Discourse Analysis and the New Testament: Approaches and Results. Eds. Stanley E.
Porter and Jeffrey T. Reed. Journal for the Study of the New Testament Supplement
Series 170. Studies in New Testament Greek 4. Sheffield: Sheffield Academic Press,
1999.
–. Discourse Features of New Testament Greek: A Coursebook on the Information Struc-
ture of New Testament Greek. 2d ed. Dallas: SIL International, 2000.
–. “Self-Instruction Materials on Non-Narrative Discourse.” SIL International, 2008.
Online: http://www.sil.org/~levinsohns/NonNarr.pdf Accessed June 5, 2010.
–. “Los rasgos discursivos comparativos aplicados a la traducción de Gálatas.” Dated
2009. Online: http://www.recursosteologicos.org/Documents/Galatas_traduccion_dis-
cursivo.pdf Accessed June 1, 2010.
 Bibliography 281

–. “‘Therefore’ or ‘Wherefore’: What’s the Difference?” Paper presented at the annual

meeting of the Society for Biblical Literature. San Francisco, November 2011. Online:
http://www.sil.org/~levinsohns/InferentialsPaper.pdf Accessed February 6, 2012.
Lightfoot, J. B. St. Paul’s Epistle to the Galatians. Repr. Peabody, Mass.: Hendrickson,
1995.
Longenecker, Richard N. Galatians. Word Biblical Commentary 41. Dallas: Word Books,
1990.
Luck, Georg. “Textual Criticism Today.” American Journal of Philology 102 (1981): 164–
194.
Lyon, R. W. “A Re-Examination of Codex Ephraemi Rescriptus.” New Testament Studies
5 (1959): 260–272.
Maas, Paul. Textual Criticism. Trans. Barbara Flower. Oxford: Clarendon Press, 1958.
Marshall, M. H. B. Verbs, Nouns, and Postpositives in Attic Prose. Scottish Classical Stu-
dies 3. Edinburgh: Scottish Academic Press, 1987.
Martini, Carlo M. Novum Testamentum e codice vaticano graeco 1209 (Codex B): Tertia
vice photoypice expressum. Codices e vaticanis selecti 30. Vatican City: Bibliotheca
apostolica vaticana, 1968.
Martyn, J. Louis. Galatians: A New Translation with Introduction and Commentary.
Anchor Bible 33A. New York: Doubleday, 1997.
Matera, Frank J. Galatians. Sacra Pagina 9. Collegeville, Minn.: Glazier, 1992.
Meiser, Martin. Galater. Novum Testamentum Patristicum 9. Göttingen: Vandenhoeck &
Ruprecht, 2007.
Mendoza Magallón, Pedro. «Estar crucificado juntamente con Cristo»: el nuevo status del
creyente en Cristo. Estudio exegético-teológico de Gal 2,15–21 y Rom 6,5–11. Tesi
Gregoriana Serie Teologica 122. Rome: Editrice Pontificia Università Gregoriana,
2005.
Metzger, Bruce M. A Textual Commentary on the Greek New Testament. Corr. ed., 1975.
London: United Bible Societies, 1971.
–. The Early Versions of the New Testament: Their Origin, Transmission, and Limitations.
Oxford: Clarendon Press, 1977.
–. The Text of the New Testament, its Transmission, Corruption, and Restoration. 3d ed.
New York: Oxford University Press, 1992.
–. A Textual Commentary on the Greek New Testament. 2d ed. Stuttgart: Deutsche Bibel-
gesellschaft, 1994.
Meyer, Heinrich August Wilhelm. Critical and Exegetical Handbook to the Epistle to the
Galatians. Trans. G. H. Venables. 2d ed. Edinburgh: T. & T. Clark, 1904.
Migne, J.-P. Joannis Chrysostomi opera omnia quae exstant. Patrologiae Graecae. Vol. 61.
Paris, 1862.
Mink, Gerd. “Eine umfassende Genealogie der neutestamentliche Überlieferung.” New
Testament Studies 39 (1993): 481–499.
–. “Editing and Genealogical Studies: The New Testament.” Literary and Linguistic
Computing 15 (2000): 51–56.
–. “Was verändert sich in der Textkritik durch die Beachtung genealogischer Kohärenz?”
Pages 39–68 in Recent Developments in Textual Criticism, New Testament, Other Early
Christian and Jewish Literature: Papers Read at a Noster Conference In Münster,
January 4–6, 2001. W. Weren and D.-A. Koch, eds. Studies in Theology and Religion
8. Assen: van Gorcum, 2003.
282 Bibliography

–. “Problems of a Highly Contaminated Tradition: The New Testament–Stemmata of

Variants as a Source of a Genealogy for Witnesses.” Pages 13–85 in Studies in Stemma-
tology II. Piet van Reenen, ed. Philadelphia: Benjamins, 2004.
–. “The Coherence-Based Genealogical Method–What is it All About?” Online at http://
www.uni-muenster.de/INTF/Genealogical_method.html (Accessed Feb. 13, 2009).
Moulton, J. H. and G. Milligan. Vocabulary of the Greek Testament. London: Hodder &
Stoughton, 1930. Repr. Peabody, Mass.: Hendrickson, 1997.
Mussner, Franz. Der Galaterbrief. Herders theologischer Kommentar zum neuen Testa-
ment 9. Freiburg: Herder, 1974.
Naber, S. A. “ϒΠΕΡ ΤΑ ΕΣΚΑΜΜΕΝΑ,” Mnemosyne new series 6 (1878): 85–104.
Nelson, Gareth. “Reticulation in Cladograms.” Pages 105–111 in Advances in Cladistics 2.
Norman I. Platnick & V. A. Funk, eds. New York: Columbia University Press, 1983.
Nes, Jermo van. “‘Faith(fulness) of the Son of God’? Galatians 2:20b Reconsidered.” No-
vum Testamentum 55 (2013): 127–139.
O’Hara, Robert J., and Peter M. W. Robinson. “Computer-Assisted Methods of Stemmatic
Analysis.” Pages 53–74 in The Canterbury Tales Project: Occasional Papers, 1993.
Vol. 1. Online: http://www.canterburytalesproject.org/pubs/op1–cladistics.pdf
O’Neill, J. C. The Recovery of Paul’s Letter to the Galatians. London: SPCK, 1972.
–. “‘For this Hagar is Mount Sinai in Arabia’ (Galatians 4.25).” Pages 210–219 in The
Old Testament in the New Testament: Essays in Honour of J.L. North. Ed. Steve
Moyise. Journal for the Study of the New Testament Supplement Series 189. Sheffield:
Sheffield Academic Press, 2000.
Oepke, Albrecht. Der Brief des Paulus an die Galater. Theologischer Handkommentar
zum Neuen Testament 9. 2d ed.; Berlin: Evangelische, 1957.
Omason, Roger L. A Textual Guide to the Greek New Testament: An Adaptation of BRUCE
M. METZGER ’s Textual Commentary for the Needs of Translators. Stuttgart: Deutsche
Bibelgesellschaft, 2006.
Palmer, Humphrey. The Logic of Gospel Criticism: An Account of the Methods and Argu-
ments Used by Textual, Documentary, Source, and Form Critics of the New Testament.
London: Macmillan, 1968.
Parker, David C. The Living Text of the Gospels. Cambridge: Cambridge University Press,
1997.
–. “Textual Criticism and Theology.” Expository Times 118 (2007): 583–589.
–. An Introduction to the New Testament Manuscripts and their Texts. Cambridge: Cam-
bridge University Press, 2008.
Petzer, J. H. “Author’s Style and the Textual Criticism of the New Testament.” Neotesta-
mentica 24 (1990): 185–197.
Platnick, Nelson I., and H. Don Cameron. “Cladistic Methods in Textual, Linguistic, and
Phylogenetic Analysis,” Systematic Zoology 26 (1977): 380–385.
Poirier, John C. “Living Text or Exquisite Corpse?” Expository Times 119 (2008): 437–
439.
Porter, Stanley E. “When and How was the Pauline Canon Complied? An Assessment of
Theories” Pages 95–127 in The Pauline Canon. Ed. Stanley E. Porter. Pauline Studies
1. Leiden: Brill, 2004.
Quentin, Henri. Essais de critique textuelle (ecdotique). Paris: Picard, 1926.
Rahlfs, Alfred, and Robert Hanhart, eds. Septuaginta. Rev. ed. Stuttgart: Deutsche Bibel-
gesellschaft, 2006.
Reichardt, Alexander, ed. Der Codex Bœnernianus: Der Briefe des Aposteks Paulus (Msc.
Dresd. A 145h). Leipzig: Hiersemann, 1909.
 Bibliography 283

Reneham, Robert. Greek Textual Criticism: A Reader. Cambridge, Mass.: Harvard Univer-
sity Press, 1969.
Richards, E. Randolph. Paul and First-Century Letter Writing: Secretaries, Composition
and Collection. Downer’s Grove, Ill.: InterVarsity, 2004.
Riches, John. Galatians Through the Centuries. Blackwell Biblical Commentaries. Mal-
den, Mass.: Blackwell, 2008.
Ridderbos, Herman. The Epistle of Paul to the Churches of Galatia. New International
Commentary on the New Testament. Grand Rapids, Mich.: Eerdmans, 1953.
Robertson, A. T. A Grammar of the Greek New Testament in the Light of Historical Re-
search. 4th ed. New York: Hodder & Stoughton, 1923.
Robinson, Maurice A. and William G. Pierpont. The New Testament in the Original Greek
According to the Byzantine / Majority Textform. Atlanta: Original Word, 1991.
–. The New Testament in the Original Greek: Byzantine Textform. Southborough, Mass.:
Chilton, 2005.
Robinson, Peter M. W., and Robert J. O’Hara. “Report on the Textual Criticism Challenge
1991.” Bryn Mawr Classical Review 3 (1992): 331–337.
Rohde, Joachim. Der Brief des Paulus an die Galater. Theologischer Handkommentar zum
Neuen Testament 9. Rev. ed. Berlin: Evangelische, 1988.
Romaniuk, Kazimeirz. L’amour du Père et du Fils dans la sotériologie de saint Paul.
Analecta biblica 15. Rome: Pontifical Biblical Institute, 1961.
Royse, James R. Scribal Habits in Early Greek New Testament Papyri. New Testament
Tools, Studies, and Documents 36. Leiden: Brill, 2008.
Runge, Steven E. A Discourse Grammar of the Greek New Testament: A Practical Intro-
duction for Teaching & Exegesis. Lexham Bible Reference Series. Bellingham, Wash.:
Logos Research Systems, 2010.
Schäfer, Karl Th. “Der griechisch-lateinische Text des Galaterbriefes in der Handschriften-
gruppe D E F G.” Pages 41–70 in Scientia Sacra: Theologische Festgabe zugeeignt
seiner Eminenz dem hochwürdigsten Herrn Karl Joseph Kardinal Schulte, Erzbischof
von Köln zum fünfundzwanzigsten Jahrestage der Bischofsweihe 19. März 1935. Ed.
Carl Feckes. Cologne: Bachem, 1935.
Schlier, Heinrich. Der Brief an die Galater. Kritisch-exegetischer Kommentar über das
Neue Testament 7. 10th ed. Göttingen: Vandenhoeck & Ruprecht, 1949.
Schmid, Ulrich. Marcion und sein Apostolos: Rekonstruktion und historische Einordnung
der marcionitischen Paulusbriefausgabe. Arbeiten zur neutestamentlichen Text-
forschung 25. Berlin: de Gruyter, 1995.
–. “Scribes and Variants–Sociology and Typology.” Pages 1–23 in Textual Variation:
Theological and Social Tendencies? Papers from the Fifth Birmingham Colloquium on
the Textual Criticism of the New Testament. Eds. H. A. G. Houghton and D. C. Parker.
Test and Studies 3d ser. 6. Piscataway, N.J.: Gorgias, 2008.
Schott, Heinrich August. Epistolae Pauli ad Thessalonicenses et Galatas. Commentarii in
Epistolas Novi Testamenti 1. Leipzig: Barth, 1834.
Scrivener, Frederick Henry. An Exact Transcript of the Codex Augiensis, a Græco-Latin
Manuscript of S. Paul’s Epistles, Deposited in the Library of Trinity College, Cam-
bridge, to Which is Added a Full Collation of Fifty Manuscripts Containing Various
Portions of the Greek New Testament in the Libraries of Cambridge, Parham, Lei-
cester, Oxford, Lambeth, the British Museum, &c. Cambridge: Deighton, Bell, 1859.
Sellin, Gerhard. “Hagar und Sara. Religionsgeschichtliche Hintergründe der Schrift-
allegorese Gal 4,21–31.” Pages 59–84 in Das Urchristentum in seiner literarischen
Geschichte: Festschrift für Jürgen Becker zum 65. Geburtstag. Eds. Ulrich Mell and
284 Bibliography

Ulrich B. Müller. Beihefte zur Zeitschrift für die neutestamentliche Wissenschaft und
die Kunde der älteren Kirche 100. Berlin: Walter de Gruyter, 1999.
Sieffert, Friedrich. Der Brief an die Galater. Kritisch-exegetischer Kommentar über das
Neue Testament 7. 9th ed. Göttingen: Vandenhoeck & Ruprecht, 1899.
Silva, Moisés. “Text and Language in the Pauline Corpus: With Special Reference to the
Use of Conjunctions in Galatians,” Neotestamentica 24 (1990): 273–281.
–. Interpreting Galatians. 2d ed. Grand Rapids, Mich.: Baker, 2001.
Skeat, Theodore. “The Codex Sinaticus, the Codex Vaticanus, and Constantine.” Journal
of Theological Studies, n.s., 50 (1999): 583–625.
Smyth, Herbert Weir. Greek Grammar. Cambridge, Mass.: Harvard University Press,
1920. Rev. ed. Gordon M. Messing, 1956.
Soden, Hermann Freiherr von. Die Schriften des neuen Testaments in ihrer ältesten er-
reichbaren Textgestalt hergestellt auf Grund ihrer Textgeschichte. Göttingen: Vanden-
hoeck und Ruprecht, 1911.
Spencer, Matthew, Klaus Wachtel, and Christopher J. Howe. “The Greek Vorlage of the
Syra Harclensis: A Comparative Study on Method in Exploring Textual Genealogy.”
TC: A Journal of Biblical Textual Criticism (2002). Online: http://purl.org/TC
–. “Representing Multiple Pathways of Textual Flow in the Greek Manuscripts of the
Letter of James Using Reduced Median Networks.” Computers and the Humanities 38
(2004): 1–14.
Steinhauser, Michael G. “Gal 4,25a: Evidence of Targumic Tradition in Gal 4,21–31?”
Biblica 70 (1989): 234–240.
Streeter, B. H. The Four Gospels: A Study of Origins Treating of the Manuscript Tradition,
Sources, Authorship, & Dates. 2d ed. London: Macmillan, 1926.
–. “Codices 157, 1071 and the Caesarean Text.” Pages 149–150 in Quantulacumque. Ro-
bert P. Casey, Silva Lake, and Agnes K. Lake, eds. London: Christophers, 1937.
Stuart, Douglas. “Exegesis.” Pages 682–688 in vol. 2 of The Anchor Bible Dictionary.
Edited by David Noel Freedman. 6 vols. New York: Doubleday, 1992.
Swanson, Reuben J. New Testament Greek Manuscripts: Galatians. Wheaton: Ill.: Tyndale
House, 1999.
–. New Testament Greek Manuscripts: Romans. Wheaton: Ill.: Tyndale House, 2001.
–. New Testament Greek Manuscripts: 1 Corinthians. Wheaton: Ill.: Tyndale House, 2003.
–. New Testament Greek Manuscripts: 2 Corinthians. Wheaton: Ill.: Tyndale House, 2005.
Tanselle, G. Thomas. Textual Criticism and Scholarly Editing. Charlottesville, Va.: Uni-
versity Press of Virginia, 1990.
Tasker, R. V. G. The Greek New Testament Being the Text Translated in the New English
Bible 1961. Oxford: Oxford University Press, 1964.
Taylor, Ann. “The Distribution of Object Clitics in Koiné Greek.” Pages 285–315 in Indo-
European Perspectives. Mark R. V. Sothern, ed. Journal of Indo-European Studies
Monograph 43. Washington, D.C.: Institute for the Study of Man, 2002.
Thompson, E. M. Facsimile of Codex Alexandrinus: New Testament and Clementine Epi-
stles. Vol. 4. London: Trustees of the British Museum, 1879.
Thornton, T. C. G. “The Meaning of καὶ περὶ ἁμαρτίας in Romans viii. 3,” Journal of
Theological Studies 22 (1971):515–517.
Thorpe, James. Principles of Textual Criticism. San Marino, Calif.: The Huntington Libra-
ry, 1972.
Timpanaro, Sebastiano. The Genesis of Lachmann’s Method. Trans. Glenn W. Most. Chi-
cago: University of Chicago Press, 2005.
 Bibliography 285

Tischendorf, Konstantin. Codex Ephraemi Syri Rescriptus sive fragmenta Novi Testamenti
e codice graeco parisiensi celeberrimo quinti ut videtur post Christum seculi. Leipzig:
Tauchnitz, 1843.
–. Codex Claromontanus sive epistulae Pauli omnes graece et latine ex codice parisiensi
celeberrimo nomine Claromontani plerumque dicto sexti ut videtur post Christum sae-
culi. Leipzig: Brockhaus, 1852.
–. Novum Testamentum Graece: ad antiquissimos testes denuo recensuit apparatum criti-
cum omni studio perfectum apposuit commentationem isagogicam. 2 vols. 8th ed., criti-
ca maior. Leipzig: Giesecke & Devrient, 1869.
Tobin, Thomas H. Paul’s Rhetoric in its Contexts: The Argument of Romans. Peabody,
Mass.: Hendrickson, 2004.
Tregelles, Samuel Prideaux. An Account of the Printed Text of the Greek New Testament:
With Remarks on its Revision upon Critical Principles Together with a Collation of the
Critical Texts of Griesbach, Scholz, Lachmann, and Tischendorf, with that in Common
Use. London: Bagster, 1854.
–. The Greek New Testament, Edited from Ancient Authorities, with their Various Read-
ings in Full, and the Latin Version of Jerome. Vol. 4. Romans to 2 Thessalonians. Lon-
don: Bagster, 1869.
Turner, Nigel. A Grammar of New Testament Greek: Syntax. Vol. 3. Edinburgh: T. & T.
Clark, 1963.
Van Voorst, Robert E. “Why is There No Thanksgiving Period in Galatians? An Assess-
ment of an Exegetical Commonplace.” Journal of Biblical Literature 129 (2010): 153–
172.
Vogels, H. J. “Der Codex Claromontanus der paulinischen Briefe.” Pages 274–299 in
Amicitiae Corolla: A Volume of Essays Presented to James Rendel Harris, D. Litt., on
the Occasion of his Eightieth Birthday. Ed. H. G. Wood. London: University of London
Press, 1933.
Wachtel, Klaus. Der byzantinsiche Text der katholischen Briefe: Eine Untersuchung zur
Entstehung der Koine des Neuen Testaments. Arbeiten zur neutestamentlichen Text-
forschung 24; Berlin: de Gruyter, 1995.
–. “Reconstructing the Initial Text of the Editio Critica Maior of the New Testament
Using the Coherence-Based Genealogical Method,” oral presentation reported by Paul
Foster, “Recent Developments and Future Directions in New Testament Textual Crit-
icism: Report on a Conference at the University of Edinburgh,” Journal for the Study of
the New Testament 29 (2006), 229–235.
–. “Towards a Redefinition of External Criteria: The Role of Coherence in Assessing the
Origin of Variants.” Pages 109–127 in Textual Variation: Theological and Social Ten-
dencies. H. A. G. Houghton and D. C. Parker, eds. Texts and Studies 6. Piscataway,
N.J.: Gorgias, 2008.
Wackernagel, Jacob. Kleine Schriften. Göttingen: Vandenhoeck & Ruprecht, 1953.
Wallace, Daniel B. Greek Grammar Beyond the Basics: An Exegetical Syntax of the New
Testament. Grand Rapids, Mich.: Zondervan, 1996.
–. “Sharp Redivivus? A Reexamination of the Granville Sharp Rule.” Online: http://
bible.org/article/sharp-redivivus-reexamination-granville-sharp-rule. Posted: 30 Jun.
2004. Accessed: 28 Nov. 2009.
–. Granville Sharp’s Canon and Its Kin: Semantics and Significance. Studies in Biblical
Greek 14; New York: Peter Lang, 2009.
Wechsler, Andreas. Geschichtsbild und Apostelstreit: Eine forschungsgeschichtliche und
exegetische Studie über den antiochenischen Zwischenfall (Gal 2,11–14). Beihefte zur
286 Bibliography

Zeitschrift für die neutestamentliche Wissenschaft und die Kunde der älteren Kirche 62.
Berlin: Walter de Gruyter, 1991.
Weiss, Bernhard. Textkritik der paulinische Briefe. Texte und Unterschungen zur
Geschichte der altchristlichen Literatur 14,3. Leipzig: Hinrichs, 1896.
–. Das Neue Testament: Handausgabe. Vol. 2. Die paulinische Briefe und der Hebräer-
brief im berichtigten Text mit kurzer Erläuterung zum Handgebrauch bei der Schrift-
lektüre. 2d ed. Leipzig: Hinrichs, 1902.
Weitzman, Michael P. “The Analysis of Open Traditions.” Studies in Bibliography 38
(1985): 82–120.
West, Martin L. Textual Criticism and Editorial Technique Applicable to Greek and Latin
Texts. Stuttgart: Teubner, 1973.
Westcott, B. F., and Hort, F. J. A. Introduction to the New Testament in the Original
Greek. Repr. Peabody, Mass.: Hendrickson, 1988.
Winer, George Benedict. A Grammar of the New Testament Diction: Intended as an Intro-
duction to the Critical Study of the Greek New Testament. Trans. Edward Masson. 4th
ed. Edinburgh: T. &. T. Clark, 1863.
Wordsworth, John, and Henry Julian White. Nouum Testamentum Domini Nostri Iesu
Christi Latine secundum editionem Sancti Hieronymi. Part 2. Fascicle 4. Oxford:
Clarendon, 1934.
Wright, N. T. “Paul, Arabia, and Elijah (Galatians 1:17).” Journal of Biblical Literature
115 (1996): 683–692.
Yoon, David I. “The Antioch Incident and a Textual Variant: ‘ΗΛΘΟΝ’ or ‘ΗΛΘΕΝ’ in
Galatians 2:12.” Expository Times 125 (2014): 423–439.
Zahn, Theodor. Der Brief des Paulus an die Galater. Kommentar zum Neuen Testament 9.
Leipzig: Deichert, 1905.
Zimmer, Friedrich. “Zur Textkritik des Galaterbriefes.” Zeitschrift für wissenschaftliche
Theologie 24 (1881): 481–494.
–. “Zur Textkritik des Galaterbriefes (Cap. III.).” Zeitschrift für wissenschaftliche Theo-
logie 25 (1882): 327–343.
–. Exegetische Probleme des Hebräer- und Galaterbriefs. Neutestamentliche Studien 1.
Hilburghausen: Gadow, 1882.
–. “Zur Textkritik des Galaterbriefes (Cap. IV-VI).” Zeitschrift für wissenschaftliche The-
ologie 26 (1883): 294–308.
Zuntz, Günther The Text of the Epistles: A Disquisition upon the Corpus Paulinum. The
Schweich Lectures of the British Academy 1946. London: Oxford University Press,
1953.
–. “The Text of the Epistles.” Pages 252–268 in Opuscula Selecta: Classica, Hellenistica,
Christiana. Ed. Günther Zuntz. Manchester: Manchester University Press, 1972.
 Index of Ancient Sources

Septuagint
Genesis 26:46 165
13:14–17 111 27:34 165
17:3–8 111
21:10 127, 199 Numbers
24:7–8 111 3:1 165
28:6 165
Exodus
19:11 165 Deuteronomy
19:16 165 27:26 189, 235, 239
19:18 165
19:20 165 2 Esdras
19:23 165 19:13 165
24:46 165
31:18 165 Psalms
34:2 165 143:2 234–235, 239
34:4 165
34:32 165 Habakkuk
2:4 190
Leviticus
7:38 165 Isaiah
18:5 235, 239 2:15 19
19:18 134, 202 54:1 167, 198
25:1 165

New Testament
Matthew 7:21–22 204
5:43 202 13:30 163
15:19 204
16:22 127 Luke
25:20 207 10:27 202
27:16 128 11:11 118

Mark John
5:31–16:8 58 4:10 160
288 Index of Ancient Sources

5:13 118 5:21 92, 135

7:53–8:11 25–27 5:23 94
8:19 105, 160 6:23 92
9:28 118 7:6 195
13:33 125 7:17 195
15:19 160 7:25 92, 135
8:3 98, 143–144
Acts 8:7 105
15:10 200 8:11 106
16:12 36 8:15 94
18:15 206 8:16 125
20:8 124 8:17 94, 125, 222
8:23 157
Romans 8:29 98
1:1 135 8:30 146–147
1:3 92–93, 98 8:32 98
1:4 92, 98, 135 8:33 106
1:6 135, 147–148 8:38–39 104
1:7 100, 135 8:39 92
1:8 135 9:7 125
1:9 98 9:8 125
1:29 130 9:11 147
2:11 153 9:12 146
2:16 135 9:14 147
3:3 99, 251 9:20 106
3:20 234, 239 9:22 106
3:21 195 9:24 146
3:22 135 9:29 160
3:23 148 10:5 235, 239
3:24 148 11:6 148
4:3 100 11:25 162
4:10 239 11:29 147
4:11–12 100 11:30 195
4:16 237 12:16 94
4:17 147 12:20 118
4:20 112 13:8 134
4:23 99 13:9 202
4:23–24 100 14:15 143
4:24 92 15:6 92, 135
4:25 96 15:14 206
5:1 92, 135 15:16 135
5:2 146 15:17 145
5:6 143 15:23 95, 195
5:7 143 15:25 195
5:8 143 15:30 92, 135, 201
5:10 98 16:7 7–8
5:11 92, 135, 246 16:18 92
5:15 97, 135, 148 16:20 92, 146
5:17 135 16:25 135
 Index of Ancient Sources 289

16:27 135 11:14 105

11:26 162
1 Corinthians 11:31 160
1:1 135 12:18 195
1:2 92, 135 12:20 195
1:3 92–93, 100, 135 12:31 196, 209
1:4 148 13:5 245
1:7 92, 135 13:6 246
1:8 92, 135, 245 13:11 113
1:9 92, 135, 146–147, 13:13 195
222 14:6 195
1:10 92, 135 14:34–35 8
1:12 135 14:39 245
1:13 143 15:1 118, 120
1:17 132 15:3 143–144
1:18 245 15:10 148
1:19 98 15:12 230
1:21 106 15:20 195
1:25 132 15:23 135
1:26 147 15:25 162
2:4 245 15:28 98
2:8 160 15:31 92, 246
2:13 94 15:45 145
3:2 105 15:50 105
3:7 104–106 15:57 92, 135
3:12 245
3:23 135 2 Corinthians
4:14 125 1:1 135
5:1 105 1:2 92–93, 100, 135,
5:4 92 148
5:11 195 1:3 92, 135
6:9–10 104–105 1:19 98
6:11 135 1:20 112
6:14 246 2:8 191
7:5 245 4:4 106
7:14 125, 195 4:6 106
7:15 146–147 5:12 153
7:17 146–147 5:14 143
7:18 146 5:15 143
7:20 146 6:13 125
7:21 146 8:1 118, 120
7:22 146 8:7 148
7:24 146 8:9 92, 135, 146
8:2–3 245 8:11 195
8:8 104, 246 8:19 148
9:1 92 8:22 195
9:9 145 10:7 135, 156
10:30 148 11:6 156
11:11 104 12:2 95
290 Index of Ancient Sources

12:9 224 1:14 48, 64, 69, 71

12:10 224 1:15 64–65, 69, 71, 83,
12:13 135 105–107, 112, 136,
12:14 125 142, 146, 147–148,
13:2 205 251–252
13:13 146 1:15–16 146, 166
1:16 64, 69, 71, 98, 142
Galatians 1:17 48, 64–65, 69, 71,
1:1 63, 68, 70, 104, 135 93–94, 137, 167,
1:1–5 92 211, 213, 247–248,
1:1–6:8 47 250–251
1:2 10–11 1:18 64–65, 69, 71, 94,
1:3 63, 65, 68, 70, 91– 95–96, 104, 137,
93, 99–100, 135– 185, 188, 209, 211,
136, 248, 251 213, 220, 228, 247–
1:4 51, 63, 68–70, 72, 248
142, 143–145, 184, 1:18–19 132
202, 209, 220, 228, 1:19 65, 69, 71, 107–
251 108, 137, 139, 186,
1:4–5 99 209, 246
1:6 63, 69–70, 106, 1:21 65, 69, 71
135, 142, 145–149, 1:22 65, 69, 71
155, 251–252 1:24 186, 209
1:6–7 148 2:1 94–96, 186, 209,
1:6–9 120 211, 214, 220, 228
1:6–10 119 2:1–14 121
1:7 63, 69–70, 151 2:2 94, 170
1:8 53, 62–63, 65, 69– 2:3 105
70, 142, 149–152, 2:4 49, 136, 214, 219,
185, 209, 211, 214, 221, 228
251 2:5 50, 169–170
1:8–9 62 2:6 118, 136, 142, 152–
1:9 63, 69–70, 120, 154, 186, 209, 234,
152, 222 239, 251–252
1:10 106, 119–121, 160, 2:7 50, 142
185, 229, 233, 251 2:7–8 131
1:10–13 120 2:8 106, 142, 187, 209
1:11 49, 62, 64, 69, 71– 2:9 123, 131–132, 137,
72, 117–121, 136, 142, 187, 188, 208–
142, 185, 209, 251 209
1:11–12 118 2:10 187–188, 209
1:11–2:14 118 2:11 121–123, 188, 209,
1:11–2:21 119 220, 229, 233
1:12 10, 64–65, 69, 71, 2:11–14 1
103–105, 135–136, 2:12 1, 108, 121–123,
229, 233, 251 137, 212, 214, 247,
1:12–13 120 249–250
1:12–2:14 155 2:13 48, 94, 108–109,
1:13 118 123, 137, 142, 251
 Index of Ancient Sources 291

2:14 107–108, 155, 169– 3:19 50, 111, 142, 162–

170, 188, 209, 220, 163, 192, 193, 208–
229–230, 231, 233, 209
234, 239 3:20 111–112, 117–118
2:15 155–157 3:21 50, 110–112, 113,
2:15–21 155 137, 142, 159–161,
2:16 49, 92, 97, 99, 109, 193, 209, 231, 233,
135–137, 142, 154– 246, 251
157, 215, 219, 234, 3:22 51, 135–136, 193
239, 251 3:23 118
2:18 221 3:23–24 192
2:19 10, 99, 101, 153, 3:24 113, 136–137, 193,
227 209, 246
2:19–21 96 3:26 110, 132, 193–194
2:20 10, 51, 96–101, 3:27 136
137, 140, 143, 193, 3:28 52, 105, 136, 155,
202, 212, 214, 248, 194
251 3:29 48, 135–136, 147,
2:20–21 99 194, 198, 209, 222,
2:21 101 228, 239
3:1 188, 209, 221, 228, 4:1 50
230, 233 4:3 124, 137, 212, 214,
3:1–4:31 155 248
3:2 189, 209 4:4 98
3:3 54, 82, 142, 158 4:5 51
3:5 106 4:6 98, 132, 137, 194,
3:6 100 208–209, 222, 228,
3:7 54, 109–110, 137, 246
189, 209 4:7 136, 222, 228
3:8 118 4:8 133, 137, 194, 208–
3:8–9 100 209, 223, 228
3:10 50–51, 189, 209, 4:9 54, 83, 142, 195,
215, 235, 239 209, 230
3:11 51, 113, 189, 190, 4:10 195, 208–209
209 4:10–11 54
3:12 219, 235, 239 4:12 10, 54
3:13 51, 143, 215, 235, 4:13 118, 195, 196–197,
239 209, 223
3:14 50–51, 132, 137, 4:14 50, 215, 219, 223,
142, 190, 208–209 228
3:15 192 4:15 231, 233, 236, 239
3:15–17 191–192 4:16 196, 209
3:16 111–112, 190, 195, 4:17 162, 196, 197, 209,
197, 209, 235, 239 232
3:17 95, 191, 192, 208– 4:18 54, 82, 117–118,
209, 230, 233, 236, 142, 161–162, 197,
239 209, 231–232, 233
3:18 51, 112–113, 118
292 Index of Ancient Sources

4:19 124–125, 137, 142, 5:16 128–129

162–163, 212, 214, 5:16–26 201
248, 251 5:17 48, 50, 63, 102,
4:20 189 117–118, 128–129,
4:21 48, 197, 208–209 137, 139, 142, 200,
4:21–31 209, 224 203, 206, 209, 213,
4:21–5:1 165 214, 225–226, 227–
4:23 113–114, 137, 142, 228, 247
198, 209, 236, 239 5:18 48, 54, 118, 129
4:24 118, 164, 166–169, 5:19 118, 129, 131, 204,
198 208–209, 226
4:25 117–118, 142, 163– 5:19–21 205, 214
169, 198, 209, 237, 5:20 51, 130, 226, 228
239, 250 5:21 129–131, 137, 204–
4:26 50, 126, 164, 167, 205, 208–209, 213,
212, 237, 239 247
4:27 50, 167, 198–199, 5:22–23 205
209 5:23 50, 205, 209
4:28 125–126, 137, 212, 5:24 134–136, 137, 142,
214 147, 205, 213, 214,
4:29 126 233, 251–252
4:30 126–127, 137–138, 5:25 142, 205, 209
199, 209, 251 5:26 114, 137
4:31 125–126, 142, 199, 6:1 206, 209
212, 214, 216–217, 6:2 102–103, 137, 213–
224 214, 248
5:1 199–200, 208–209, 6:3 206, 209, 227, 228
216–217, 219, 224, 6:4 115, 137, 246
232, 233, 238 6:5 117–118
5:2 10, 50 6:7 142, 154, 203, 206,
5:3 48, 200, 209 209, 227
5:4 232, 233 6:8 50–51, 207, 209
5:5 118, 142 6:9 82, 142, 171, 218,
5:6 104, 136, 207, 209, 219
227–228, 238–239 6:10 10, 82, 142, 171,
5:7 142, 169–171, 221, 172
228, 252 6:11 11, 115, 137, 207,
5:8 106, 146 209, 246
5:10 10, 50, 54, 122, 6:12 83, 116, 136–137,
142, 200, 203, 206, 142, 227, 228
209, 225, 227 6:13 116, 137, 238, 239
5:11 10, 50, 200, 209 6:14 50, 92, 135, 218–
5:12 142, 201, 209 219
5:13 118, 134, 146, 201, 6:15 54, 104, 207, 209,
202, 209 227, 228, 238, 239
5:14 133–134, 137, 202, 6:16 50, 207–208, 209
208–209, 218, 219, 6:17 10, 83, 136, 208–
225, 228 209, 219–220
5:15 203, 209 6:18 92, 135, 146, 219
 Index of Ancient Sources 293

Ephesians 3:16 94, 148

1:2 92, 100 3:18 8
1:3 92 3:20 125
1:17 92 3:21 125
2:2 64, 94 4:6 148
2:3 125
2:5 148 1 Thessalonians
2:8 148 1:1 91–93
2:13 195 1:2 100
3:1 135 1:3 92, 135
3:11 92, 135 1:8 100
4:13 98 1:10 98
4:19 98 2:5 154
4:22–24 8 2:5–6 104
5:1 125 2:7 125
5:2 98 2:11 125
5:8 125, 195 2:12 146–147
5:20 92 2:19 92
5:25 92 3:3 206
6:1 125 3:11 92
6:4 125 3:13 92
6:23 92, 100 4:7 146–147
6:24 92 5:9 92, 135
5:10 144
Philippians 5:11 224
1:1 135 5:23 92, 135
1:2 92, 100, 135 5:24 106, 146–147
1:6 106, 135 5:28 92, 135, 146
1:8 135
1:11 135 2 Thessalonians
1:19 117, 135 1:1 92
2:13 106 1:2 92, 100
2:15 125 1:8 92
2:21 135 1:12 92
3:8 135 2:1 92
3:12 135 2:14 92, 146
3:14 147 2:16 92, 148
3:15 156 3:4 91, 249
3:18 195 3:18 92, 146
4:23 146
4:25 135 1 Timothy
1:2 92, 100
Colossians 1:12 92
1:12 91, 245, 249 1:14 92
1:22 195 2:8–15 8
1:26 195 5:21 100
2:6 135 6:3 92
3:8 195 6:14 92
3:15 246
294 Index of Ancient Sources

2 Timothy 10:25 94
1:2 100 10:26 143
1:8 92 11:4 245
4:1 100 11:15 64
11:39 245–246
Titus 12:1 245
1:4 100 12:4 94
2:13 100 12:9 64
13:5 246
Philemon 13:6 246
1 135 13:11 94
3 92, 100, 135
9 135, 195 1 Peter
10 125 1:12 118
11 195 3:18 143
25 135, 146
1 John
Hebrews 2:1 125
1:3 246 2:2 143
1:6 94 2:12 125
2:8 245 2:28 125
3:6 245 3:7 125
5:1 245–246 3:18 125
6:11 94 4:4 125
7:2 245 5:21 125
7:27 94
8:10 245 Revelation
9:11 64 14:13 118
10:12 143

Other Ancient Authors

Clement of Alexandria Irenaeus
Stromata Adversus haereses
1.26.167 192 3.7.2 192–193

Epiphanius Origen
Panarion Contra Celsum
42.11.8 129, 133–134, 204 2.3.8 197
42.12.3 129, 133–134, 204 4.44.24 197

Hermas Plato
Shepherd Gorgias
93.5 98 514e 158
 Index of Ancient Sources 295

Republic 5.3.6 187

336e 158 5.3.11 132
5.4.5 133, 194–195, 223
Polycarp 5.4.8 191, 224
5.4.9 199–200
Philippians
5.4.12 133–134, 202
3.3 237
Theodoret
Tertullian
In Galatas (PG 82)
Adversus Marcionem
488A 223
5.2.5 149
 Index of Modern Authors

Aland, B. 4–6, 13, 17, 25, 35, 38, 40, 126, 130–131, 135, 143, 145, 147,
46–47, 52, 82, 127 149–150, 153–154, 156, 160–161,
Aland, K. 4–6, 13, 17, 25, 35, 38, 40, 165, 169–172, 184, 190–192, 195,
46–47, 127 216, 218, 221, 224, 226, 230–231,
Allen, W.S. 83, 126, 158, 162, 171–172, 236–237
201, 218 Buscemi, A.M. 91, 94, 97, 100, 103,
Alroy, J. 42, 65–66 105, 107, 109–110, 112–115, 118–
Anderson, A.S. 4, 76 119, 124, 126–128, 130–131, 135,
149, 151, 154–155, 157, 162, 170,
Baarda, T. 199 185–186, 188–189, 192, 194, 196,
Bakker, S. 95, 97–99, 111, 184 198–199, 201, 204, 206–207, 216,
Barrett, C.K. 166 218–219, 221–223, 226–227, 235–
Barrett, D.P. 48, 145 238
Barrow, J.D. 73 Buth, R. 119, 155
Baumert, N. 101 Buttmann, P. 18
Bédier, J. 42, 75
Bénevot, M. 32, 78 Cameron, H.D. 60
Bentley, R. 167–168 Campbell, D.A. 97, 100, 156, 193, 200
Berényi, G. 98 Carlson, S.C. 76, 89, 188, 252
Betz, H.D. 92, 94, 99–100, 105, 107– Cavalli–Sforza, Luca 60
108, 112–113, 116, 122, 125–126, Chrysostom, J. 53, 82
128, 130, 133, 135, 143, 145–147, Clabeaux, J.J. 113–114, 132
154–156, 161, 164–166, 170, 172, Clark, K.W. 52–53
186, 188–189, 204–205, 207, 216, Clarke, K.D. 47
221–223, 227, 229, 235, 238 Colwell, E.C. 28, 33, 35–36, 38, 55, 58–
Birdsall, J.N. 243 60, 75, 144
Black, S.L. 119, 128, 155, 199 Comfort, P.W. 48, 93, 145, 149
Blake, R.P. 57 Conzelmann, H. 143
Bligh, J. 99
Bonnard, P. 216 Davies, G.I. 165
Borse, U. 166, 216 de Boer, M.C. 146–147, 192, 216
Bortone, P. 144 Denniston, J.D. 104
Bowers, F. 24–25 Dickerman, A.W. 75–76
Breytenbach, C. 144 Dik, H. 159, 216
Brown, A.J. 18 Di Mattei, S. 197
Brown, R.E. 100–101 Dunn, J.D.G. 97, 165, 216
Bruce, F.F. 121, 216, 218
Burton, E.W. 91, 94, 96–98, 100, 103– Eadie 93–94, 120, 126, 130, 153, 156,
105, 107, 109, 111–118, 120, 124, 162
298 Index of Modern Authors

Edwards, A. 60 Humphries, C.J. 59

Ehrman, B.D. 4–6, 15, 99–101, 116, Hurtado, L.W. 5, 77
181–182, 191, 193, 208
Elliger, K. 19 Joubert, S. 147
Elliott, J.K. 34, 38 Juckel, A 52, 82
Elliott, S.M. 165
Epp, E.J. 4–8, 35, 182, 187 Karavidopoulos, J. 5
Eshbaugh, H. 99–100, 134, 187, 191, Kerkeslager, A. 166, 168–169, 198
193, 202 Kitching, I.J. 59
Knox, J. 4
Farahian, E. 96, 100 Krentz, E. 3
Farris, J.S. 60
Farstad, A.L. 22–27 Lachmann, K. 26, 54, 89
Fee, G.D. 35, 97, 101, 114 Lake, K. 55–57
Felsenstein, J. 59–60, 73–74 Lake, S. 57
Fischer, B. 190 Lee, III, A.R. 60
Forey, P.L. 59 Legendre, P. 75
Fraenkel, E. 160, 217 Légasse, S. 94, 112, 118–119, 125–126,
Frede, H.J. 52 130, 135, 148, 151–152, 156–157,
Froger, J. 28–29, 62 164–165, 172, 216
Legg, S.C.E. 46
Gamble, H.Y. 10–11 Levinsohn, S.H. 89, 108, 110, 114, 118–
Gignac, F.T. 105 119, 128, 150, 153, 155, 170, 186,
Goldstein, D.M. 160, 188, 217 189, 199, 216–218, 223–224, 227,
Greetham, D.C. 2, 15 230
Greg, W.W. 18 Lightfoot, J.B. 94, 101, 104, 106–107,
Gregory, C.R. 46 114, 116–117, 120–127, 130–131,
136, 143, 145, 147, 151, 153, 156,
Haines–Eitzen, K. 4 162, 165
Hall, R.G. 119 Longenecker, R.N. 94, 97, 100, 103–
Hanhart, R. 127 104, 107, 109–112, 115–116, 118–
Harrison, J.R. 147 120, 124–126, 128, 130, 133, 135,
Harvey, W.W. 105, 129 143–144, 146–147, 150, 152, 155–
Hay, D.M. 153 156, 161–162, 164–166, 170–172,
Hays, R.B. 97 195, 204, 215–216, 221–224, 234–
Head, P.M. 101 236
Hemphill, W.L. 53 Luck, G. 17
Hennig, W. 60 Lyon, R.W. 49–50
Hester, J.D. 118, 120–122
Hodges, Z.C. 22–27 Maas, P. 3, 27, 30–33, 111, 168
Holmes, M.W. 3, 7–9, 22, 30–31, 34– Magallón, M. 100
36, 89, 142, 169, 237, 244 Makarenkov, V. 75
Holsten, C. 167 Marcus, J. 187
Horrocks, G. 201, 218 Martini, C.M. 5, 49
Hort, F.J.A. 3, 14, 18, 20, 25, 35–36, Martyn, J.L. 119–120, 126, 128, 145–
45–46, 55–56, 84–91, 96, 121–123, 146, 148, 156, 164–166, 202, 216
243–244, 249, 253 Matera, F. 91, 93, 97, 100, 103–104,
Housman, A.E. 21 107, 110, 119, 122–123, 126, 135,
Howe, C.J. 29, 61, 72, 75
 Index of Modern Authors 299

143, 147, 151, 156–158, 164–166, Royse, J.R. 15, 48, 63–64, 90, 94, 105–
170, 216 106, 112, 118, 120, 130–131, 145,
Meiser, M. 192, 221, 223 148, 155, 157, 164, 169, 245
Metzger, B.M. 5, 34–36, 38, 80, 90–91, Rudolph, W. 19
93, 100–101, 103, 105–107, 111– Runge, S.E. 119, 128, 155, 217
112, 120, 122–123, 126–127, 130–
132, 135, 145, 148, 151, 164–165, Schäfer, K.T. 163
185, 187–188, 191, 212, 219–220, Schlier, H. 125, 135, 155–156, 159,
222–223, 237 162, 172, 216
Meyer, H.A.W. 94, 101, 103, 107, 112, Schmid, U. 13, 52, 129, 132–134, 145,
114, 120, 123, 127, 130, 148, 153, 149, 163, 183, 187, 191, 194–195,
157, 165 200, 202, 204, 208, 223–224
Migne, J.–P. 53 Schott, H. 167–168
Milligan, G. 108 Scrivener, F.H. 50, 52
Mink, G. 6–7, 39–41, 43, 47 Sellin, G. 164
Moulton, J.H. 108 Sieffert, F. 94, 103, 107, 110, 112, 120,
Murphy, H. 59 124–126, 130, 154, 156–157, 160,
Mussner, F. 94, 103, 107, 112, 119, 126, 169
128, 130, 135, 148, 156, 164, 172, Sihler, A.L. 162
216, 223 Silva, Moisés 117, 119–120
Skeat, T.C. 24
Naber, S.A. 167 Smyth, H.W. 98, 104, 110, 126, 147
Nelson, G. 75 Sneath, P. 59–60
Sokal, R. 59–60
O’Hara, R.J. 29, 61 Spencer, M. 29, 61, 72, 75
O’Neill, J.C. 17, 96, 167–168 Steinhauser, M.G. 164
Streeter, B.H. 57–58, 77
Palmer, H. 29 Stuart, D. 3
Parker, D.C. 3, 48, 243 Swanson, R.J. 47–53, 144
Petzer, J.H. 14, 17
Pierpont, W.G. 22–26, 89 Tanselle, G.T. 16, 19
Platnick, N.I. 60 Tasker, R.V.G. 93, 121, 123, 148
Poirier, J.C. 3 Thompson, E.M. 48
Porter, S.E. 11 Thornton, T.C.G. 143
Thorpe, J. 16
Quentin, H. 28 Timpanaro, S. 18, 20, 26
Tischendorf, K. 46, 49–52, 89
Rahlfs, A. 127 Tobin, T.H. 97
Reichardt, A. 51 Tov, E. 7
Reneham, R. 33 Tregelles, S.P. 89
Richards, E.R. 10–11 Tune, E.W. 28, 59–60
Riches, J. 184 Turner, N. 99
Ridderbos, H. 164
Robertson, A.T. 197, 218 van Nes, J. 99
Robinson, M.A. 22–26, 89 Van Voorst, R.E. 92
Robinson, P.M.W. 29, 61 Vogels, H.J. 163
Rohde, J. 103, 118, 120, 125–126, 130, von Soden, H. 22–23, 89
147, 156, 162, 216
Romaniuk, K. 100–101
300 Index of Modern Authors

Wachtel, K. 6, 29, 40, 47, 61, 72, 75, Williams, P.J. 234
242 Winer, G.B. 104, 127, 130
Wackernagel, J. 160, 217 Wordsworth, J. 52
Wallace, D.B. 65, 98–99 Wright, N.T. 166
Wechsler, A. 121
Weiss, B. 63, 89, 92–96, 101–104, 106– Yoon, D.I. 123
107, 109, 112–117, 119, 123–126,
130, 144, 150, 152–153, 159, 162, Zahn, T. 93–95, 103–105, 109, 112,
172 118, 120, 125–126, 129–130, 147,
Weitzman, M.P. 78, 81 155–157, 159, 162, 169, 171
West, M.L. 27, 29–32 Zimmer, F. 94–95, 102, 104, 110, 117,
Westcott, B.F. 3, 18, 36, 46, 55–56, 84, 120–121, 124–125, 151, 154, 161,
89, 244 163, 171–172, 235
White, H.J. 52 Zuntz, G. 36, 38, 45, 64, 81, 85, 90–91,
Wikgren, A. 107 117, 120, 126, 135, 237, 243–250,
Williams, D.M. 59 253
 Index of Subjects

accidental coincidence 55, 57, 61, 139, 222, 228, 230, 233, 238–239, 241,
187, 198, 236, 244 248
accusative 114, 185 anti-Marcionite 190–192, 228
– of respect 162 anti-pagan 133, 194, 209
adultery 131, 204–205, 226, 228 anti-Patripassionist 101, 182, 212, 214
adversative conjunction 119, 164 anti-Separationist 116, 182, 193, 208–
age of printing 1, 18, 23 209, 219, 251
agreement in error 28, 57, 59, 76, 83– aorist 103, 113, 127, 204
84, 245–246, see also common error – imperative 103
principle – subjunctive 126–127, 201, 214, 251
Alexandrian text 35–36, 81, 85, 211, apodosis 160–161, 205, 231
228, 243 apparatus 5–6, 92, 102, 158, 162, 171,
– corrections 85 250–251
– secondary 35, 81, 246–247 apposition 106, 134, 165, 222, 228
allegory 114, 166–167, 209 archetypal text 8
alternative reading 12, 121 archetype 16, 27, 29–30, 32–33, 36, 41,
ambiguity 116, 157, 184, 188, 190, 194, 44, 56, 62, 72, 78, 121, 138–139,
199, 204, 209, 215–217, 219–220, 169, 241–246
223, 225, 227, 233, 239–240 article 83, 92, 97–100, 106–107, 110–
– syntactic 184, 220 111, 135, 152–154, 160–161, 164,
anarthrous noun 145, 153–154, 170 168–171, 186, 189–190, 197–199,
anathema 149–151, 185 215, 217–219, 231–234, 236, 250,
ancestor 12, 23, 27, 30, 41–43, 54, 62, 253
65–66, 74–77, 81, 83, 85, 96, 110, assimilation 57, 93–94, 96, 104, 109,
112, 131, 137, 139, 182–184, 195, 120, 129, 132, 151, 190, 194, 211–
199, 203, 213–214, 220, 228, 241– 212, 214, 220, see also scribal
243, 245–246 harmonization
Andreas group 26 asyndeton 155–158, 165, 190, 195, 197,
anomaly 32–33 199, 216–217, 219, 229, 233, 238
anthropological focus 97 Athos Codex Ψ 82, 184, 214, 220, 242,
Antioch incident 1–2, 122–123, 250 246
anti-adoptionistic 182 attention–getting vocative 118
anti-Demiurgic 209 Atticism 115, 196, 219, 235, 239
anti-docetic 182 attributive 147, 164–165
anti-dualist 228 Augustine 1
anti-Gnostic 220, 228 author 3–4, 7, 14–17, 19, 32–33, 39, 43,
anti-Judaic 182, 187, 191–193, 195– 62, 70, 111, 143, 151, 154
197, 200, 202–203, 208–209, 221– authorial original 8, 12, 16, 30
302 Index of Subjects

authorial reading 16, 24, 26–27, 30, 32, 208–209, 213–214, 217, 219, 222–
37, 39, 44–45, 62, 70–72, 87, 109, 223, 225, 227, 229–230, 232–233,
136–139, 161 239, 251–252
authorial text 7–8, 14, 33, 43–44, 86, – pre-existence of 191
88, 169, 249, 252
authorial variant 15–17, 32 Christological theme, focus 5, 91–93,
authority 2, 8, 21–22, 58 97, 182, 202, 208–209, 251
– presumptive 19 Christology 6, 100
autograph 16, 27, 33, 36, 121, 138 Chrysostom, John 53, 82, 190, 214, 220,
auxiliary principle 60 228, 234, 242
circumcision 108, 116–117, 123, 146,
background material 122, 150 200, 238
base (of stemma) 72, 86–89, 136, 140 cladistics 44, 54, 60–61, 65, 67, 72–75,
basic witnesses 89, 105, 126, 129, 138, 80, 82, 84, 87, 136, 210, 241, 244,
171 246, 252
best–found stemma 79–83 – algorithm 72, 88, 136, 241, 244, 252
best reading 21–22, 32, 138 – analysis 73, 80, 246
best text 21 clarity 152, 161, 233, 237, 248
Biblia Hebraica Stuttgartensia 18 classical form 124, 200, 221, 225–226,
bipolarity assumption 78, 81 see also Atticism
bootstrap test 74, 82–84, 228 classically trained scribes 102
bracketed readings 89, 142, 251–252 Codex Sinaiticus (ℵ) 48, 56, 242
burden of proof 19 Codex Leningradensis 19
Byzantine text 23–24, 35–36, 45, 53, Codex Vaticanus (B) 18, 36, 49, 56,
56–57, 75, 80–84, 87, 93, 114, 117, 245–246
161, 182, 184–185, 188, 192, 210– Codex W 58
211, 214, 220, 233, 237, 239, 242– coherence 40, 43–44, 89, 103, 107–110,
243, 246, 248–250 112, 124, 132, 138, 141. 148, 232
– manuscripts 22, 53, 56, 83, 127, 235, Coherence–Based Genealogical Method
238 (CBGM) 39–44, 54, 89
– prototype 83–84, 233–234, 236, cohesive 112
238–239, 243, 264–272 coincidence 76, 223, 229, 233, see also
– readings 53, 56–57, 77, 81, 160, accidental coincidence
247–248 – change 225
– omission 107, 230, 234
Caesarean text 55, 57–58, 77 collation 4, 12, 16, 22–23, 35, 44, 47,
Canterbury Tales 61 52, 57, 62, 70, 80, 241, 244
category 38, 40, 43, 58 collection 10–12, 16–17, 30, 37, 58–59,
Catholic Epistles 6, 40, 46 77, 81, 169, 242, 244
Center for the Study of New Testament common ancestor 12, 27, 30, 81, 83, 85,
Manuscripts 52 96, 110, 112, 131, 139, 182–184,
Cephas 1, 2, 95, 104, 107–108, 121– 199, 213–214, 220, 228, 241–243,
123, 131–132, 155–156, 185, 187– 245–246
188, 209, 220, 229–230, 250–251, common descent 83, 241
see also Peter common-error principle 27, 28–30, 44,
Christ 91–93, 97–101, 103, 109–110, 54–57, 60
113, 116, 119–120, 135–136, 140, community of error 55–58
143–144, 146–149, 151, 155–156, complementary infinitive 105
160, 167, 182, 185, 191–194, 202,
 Index of Subjects 303

computational biologists 29, 44, 60, 74, disclosure formula 118

241 discontinuity 119, 128, 155, 217
computational techniques 244 discourse topic see topic
computer 29, 44, 60–61, 72–75, 78, 80, divinatio 27, 33
82, 241, 252 divinity of Christ 98
conflation 232, 249 documentary approaches 34
conjecture 17, 20, 32–33, 140, 250 double contract construction 154
conjectural emendation 33, 36, 38, 44, double–factorial formula 73
96
conjunctive particle 128, 250 earliest attainable text 7–8
connective particle 185, 213–215, 229, easier reading 14, 110, 113, 121, 133,
237–238, 251 142
Constantine 23 Eastern branch 45, 88–91, 93, 95–96,
contamination 12, 16, 29–33, 37, 39, 102–103, 105, 107, 109, 112, 116–
41–42, 44, 53–56, 66, 74–78, 81, 117, 121, 124–126, 128–129, 134,
84–85, 90, 210, 220, 225, 228, 236, 136–139, 141–142, 145, 152, 159,
239, 244, 248–250, 252 161, 171, 182, 199, 204, 210–213,
contemporary relevance 207 235, 239, 242, 248, 250
contextually contingent 151, 203 eclecticism 32–45, 54, 88, 90
Continuous Track Analysis 65–66 Editio Critica Maior 6, 40, 47
copies 1, 10–12, 16, 19, 23–25, 27, 30, edition 1–2, 4–7, 9, 12, 15, 18, 20, 22–
75, 81 24, 34, 46–50, 52–53, 56, 58, 70–72,
copy–text 18–22, 24, 26–27, 34, 37, 43, 89, 142, 169, 250–253
54 emendatio ope codicum 20
corrections 11, 12, 14, 47–51, 53, 57, emendatio ope ingenii 20
62, 85, 133, 233 emendations 18–21, 24, 27, 33, 36, 38,
correctors 48–50 44, 96, 250
cost 67–68, 70, 76, 80, 84, see also emphasis 6, 22, 44, 89, 91–92, 94–96,
reticulation cost 107–108, 110, 133, 150–154, 160–
crasis 187, 209 161, 170, 182, 186, 189, 194, 196,
critical text 2, 8, 21, 32, 34, 37–38, 40, 200, 203, 206–207, 209, 219, 223,
45, 87–88, 101, 110, 140–141, 154, 225, 227, 230, 234, 236
158–159, 171, 173–180, 181, 216, – emphatic constituents 89, 160, 227
250–252 – emphatic reading 133
Cyprian’s text 78 errors 19–20, 27–30, 32–33, 44, 49–50,
52–60, 62, 76, 78, 83–84, 90, 101,
dative 114, 185, 189, 216–217, 232–233 121, 140, 193, 209, 218, 242, 245–
decontextualized readings 151, 185 247
de-emphasis 186,225, 230, 234 examinatio 27, 32, 34
deletion 136, 197, 233 exemplar 12–13, 15–16, 19, 23, 27, 30,
demiurge 184, 209, 220, 239 32, 42, 43, 54, 62, 75, 84, 94, 154,
deponent 161 159, 163, 222, 245, 247
descendant 12, 14, 16, 19, 23, 25, 27, explicitness 98, 105–106, 123, 126, 132,
30, 32, 41–42, 69, 75, 77, 139–140, 144, 146–148, 150, 153, 155, 157,
214, 217, 243, 246 190–192, 201–202, 212, 214, 229,
dictation 11 248, 250
different authors 17 external evidence 13–14, 16, 18, 21, 26,
diminutive 124–125, 212, 214 34, 37–40, 43–44, 61, 70, 83, 89,
diplomatic edition 20 102, 105, 107, 109–110, 112, 114,
304 Index of Subjects

117, 121, 124–126, 128–129, 134, Hebraisms 198

138, 140–141, 143, 145, 148, 152, Hebrews, Epistle to 36, 245, 247
154–155, 157–159, 161–163, 168, Hellenistic 102, 124, 198, 203, 206,
172, 182 221, 226
heuristic search 73–74
faith in God 99–100 highlighting 95, 114, 150, 219, 223
faithfulness of Christ 97, 99, 156, 193, historians 1, 4, 33, 43, 44
251 historically contingent 149, 151
family 1 55–57 historically significant variants 250, 252
first attributive position 147 history of the text 5, 6, 8–9, 12–13, 15,
formula 93 20, 22, 26–29, 31, 33, 39–40, 43–45,
– disclosure 118 54, 56, 59, 61, 70, 74, 76, 80, 84, 86,
– greeting 91–92 88, 141, 172, 181–182, 185, 241,
fragmentary 31, 183 243–244, 247, 250, 252
fronting 95, 150, 160, 196, 206, 227 homoeoteleuton 106, 130
future indicative 103, 126–127, 139, homophonous see orthographic variant
201, 214 hyparchetype 32, 41
hyperbaton 110, 147, 164
genealogy 25–26, 28, 33, 37, 40, 41, hypertree 75–76
54–55, 86, 243 hypothetical ancestors 41–42, 54, 66–
– coherence 40, 43, 103, 107–110, 67, 75, 77, 183
124, 132, 138, 148, 162 hypothetical textual history 9–10, 13,
– independence 23 15, 22, 70
– method 55
– tree 73–74 identifiability 35, 169–170, 189, 218
genitive idiom 152–153, 199, 227
– epexegetical 147 inadvertent omission, see accidental
– objective 99 omission
– of content 147 incarnation 191
– of possession 146 independent 16, 23, 26, 30, 37, 57, 60,
– subjective 97, 101, 251 87, 112, 118, 139, 157, 186–187,
global stemma 41–43 196, 198, 219, 225, 233, 236, 242,
gloss 168, 198, 250 249
gnomic 202, 225 independent change see accidental
Gnostic 184, see also anti–Gnostic coincidence
grace 91, 118, 145–147, 252 intentional interference 219
Greek–Latin diglots 80, 85, 93, 102, inferential particle 224, 228, 232, 238
131, 182, 184, 201, 208 information structure 219
greeting formula 91–92 initial letter collection 12
initial text 7–8, 40–43, 54
handwritten 11 inner branch 87–90, 103, 117, 131
harder reading 14, 16, 90, 104, 117, Institute for New Testament Textual
120–121, 124–125, 131–132, 154, Research see Muenster Institute
165, 170 instrumental 146, 189
Harklean 82, 103, 117, 214, 228, 234, internal evidence 8, 13–18, 21, 24, 26,
236, 242 30–35, 37–40, 43–45, 54, 62, 70,
Harklensis Syriac versions 52 86–87, 90, 94–96, 102, 104–105,
harmonization see scribal harmonization 107, 109–110, 112, 114–118, 121,
harsher reading 158 125–126, 128, 130–131, 133, 136,
 Index of Subjects 305

140–141, 145, 148, 152, 154, 157– l’orientation 28, 62

159, 162, 170–171, 181, 250 LXX 138–139, 209, 235
International Greek New Testament
Project (IGNTP) 46 Majority–text 22–27, 34, 37, 43, 54, 83
interpolation 17, 167–168, 237 Marcion 52–53, 81, 86–87, 129, 131,
intonation unit 217 138,182–184, 190–192, 195, 199,
intrinsic probabilities 14–15, 21, 33, 35, 202–203, 208, 220, 228, 241–242
90–91, 100–101, 111–112, 123, 127, marginal note 167, 169, see also gloss
135, 144, 151, 166 marginal readings 15–16, 89, 142
itacism 226 maximum parsimony 61–62, 67, 70, 74,
78, 244
James 1, 107, 122–123, 131–132, 187, mechanical error 90, 193, 209
250 meta-comment 118, see also disclosure
Jerome 1, 80, 221 formula
Jerusalem 94–96, 104, 123, 167, 169, microfilm 46, 52
201, 237, 251 minuscule manuscripts 18, 35, 246
– apostles 94 minuscule 1739 246
Jewish Christianity 187 mood 103, 172, 213–214, 219, 221
Junia 7–8 Muenster Institute 6–7, 39, 46–47, 51–
52
Koine Greek 191, 235 multiple originals 16
Koine text 22–23, 127 murder 129–130, 204–205, 213–214
Koridethi Codex (Θ) 57
Lachmann, Karl 26, 41, 54–55, 5960, 89 negative particle 104, 223, 227
Latin version 27, 35, 51–52, 80, 161, Nestle–Aland edition 1, 4–5, 34, 47–48,
182–183, 201, 208, 241 53, 62, 70–72, 87–89, 92, 96–98,
Latinism 190, 248 138, 142, 161, 250–253
Law, the 111, 113, 116, 133–134, 153– Neutral text 55–56, 86–88, 138, 249
155, 160–161, 167, 189–191, 193, neutral witnesses 85
197, 200, 202, 208–209, 222–223, New Testament Virtual Manuscript
225, see also Torah Room (NT.VMR) 47, 51–52
leap 15, 130, 133, 184, 186, 189, 196, nominative, pendent 206
200, 224, see also parablepsis non-classical constructions 209, 214,
l’enchaînement 28, 62 231
lectio difficilior potior 14, 43, see also nonsense spellings 54
harder reading noun phrase 98, 110–111, 160, 236
Leitfehler 241, 245 normality 24
letter collection 12, 16–17, 30, 37, 169,
244 objective genitive 99
Levitical love commandment 134, 202, Occam’s razor 61
209 Old Latin versions 35, 80, 110, 112,
linguistics 4, 8, 250, 253 182, 185, 199, 208, 241, 247
linkage pattern 28–29, 62, 87, 141 omission 83, 89–90, 102, 106–108, 110,
local genealogy 40–41 112, 114–115, 120, 127, 131, 133–
local stemma 35, 41, 43 134, 136, 145, 151–152, 157, 159,
locative 146–147, 188–189, 204, 232 164, 166, 169, 185–186, 190, 193,
longer reading 90, 105–107, 112, 116, 195, 200, 215, 218–219, 224, 230,
134–136, 145–149, 166, 168, 251– 232–235, 242–243
252, see also shorter reading
306 Index of Subjects

– accidental 90, 102, 106, 112, 114– – biography 123

115, 120, 133, 145, 159, 166, 200, – usage 125, 143
219, 224 Pauline Corpus 46, 84, 93, 98, 140, 244,
– coincidental 107, 169, 190, 200, 230, 247, 252
233–234 “pay as you go” system 76
– small 112 pendent nominative 206
open tradition 31, 78 perfect tense 113, 202, 225
orientation 28, 62, 140–141 – participle 116, 238
oriented stemma 66, 72, 86,136, 138– Pericope Adulterae 25–26
141, 242 Peshitta 52, 81, 214, 228, 242
Origen 57, 197, 243 Peter 122, 131–132, 155, 185, 187–188,
original context 39 209, 220, 228–230, 233, see also
original form 19–20 Cephas
original reading 27–28, 32, 38, 55, 57– phenetics 59
58, 165 photograph 51
original text 3–5, 7–8, 16, 26, 43, 54, – black and white 48–49
56, 72 – color 49, 51
orthodox corruption 5, 181 – facsimile 50
orthographic change 49, 83, 162, 209, – microfilm 52
218 phylogeny 60
orthographic variant 54, 63, 82–83, 126, pistis Christou debate 251
139, 142, 201, 226, 231, 233 plural 121–126, 135, 155–156, 163,
orthography 20 203–204, 206, 212, 214, 222, 226
overall similarity 28, 243 poles, opposite 78
Polycarp 237
P46–B cluster 68, 71–72, 86, 88–91, 95, potential ancestor 41–43
103, 105, 107–110, 128–129, 136, pragmatics 57, 89, 216, 250, 253
138–142, 145, 159, 182–183, 210, pre-existence of Christ 191
229, 242, 246, 250 present imperative 103
paleographers 21 presumptive authority 19
papyrus 3–4, 11, 31, 35, 46, 71, 244, presupposed material 9, 150–151
248 pretense 108–109
parablepsis 90, 102, 106, 110, 120, 133, primitive error 140
159, 165–166 prolepsis 118, 187
parallel construction 207 – proleptic subject 156
parallelism, parallels 100, 108, 112, proto-Alexandrian group 35, 80, 248
120, 130, 134, 139, 144, 162–163, proto-Orthodox 116
222, 225–226, 228, 230–231, 233, proverbial 154, 234
239 Pseudo-Clementine 187
parenthesis 120, 129 punctuation 20
parchment 3, 11 punctuation after the fact 159
parsimony 60, 66, 67
passage 17, 38, 96, 111–112, 128, 133, Q group 26
144, 170, 193, 230 qualitative effect 145
patristic sources 35 quantitative method 28, 59
pattern of readings 6–7, 27, 32, 58, 141 Queen of the Cursives 51, 242
Paul
– aggrandizement of 212, 231, 233 radical eclecticism 38
– apostleship 104 reading in the main text 16, 89
 Index of Subjects 307

reading in the margin 12, 89 – new 118–119, 155–156, 199, 216–

reasoned eclecticism 34–36, 43–45, 90 217, 224, 232
recensio 27, 34, 140 – thanksgiving 92
recension 12, 56, 249 Septuagint 127, 138, 189, 198–199, 235
reduced median networks 75 serviceable authorial text 43–44
redundant 166 shape of the stemma 28–30, 53, 76, 78,
resumptive 120–121 241
– pronoun 102, 206 shape and linkage pattern 28–29, 62, 87,
reticulating link 75–76, 239 141
reticulation 75, 210 shorter reading 90, 105–107, 112, 130,
reticulation cost 76, 80, 84 134–135, 145–146, 148, 166, 196,
Revelation 25–27 245, 252
revision 17 – criterion 90
Romans 8, 201 Sinai 164–169, 198, 237, 250
singular, as opposed to plural 102, 106,
salutation 91 122–123, 155, 203, 206, 212, 214,
scribal activities 210, 243, 248 226, 245
scribal assimilation 92–93, 96, 120 singular reading 53, 65, 68, 71–72, 80,
scribal behavior 144, 239–240 118, 142, 164, 245–246
scribal character 245 skip, see parablepsis
scribal conjecture 54 Son of God 96–98, 100–101, 140, 214,
scribal corruption 33, 86 251
scribal change 72 split 24, 26, 28, 30, 42, 62, 95–96, 102,
scribal error 20, 32–33, 55, 62, 90, 246 107, 109, 116, 129, 136, 141–143,
scribal freedom 249 145, 152, 159, 171, 242
scribal habits 142, 148, 245 statistical methods 244
scribal harmonization 90, 94, 103, 106, stemma 25–35, 37, 39–45, 53, 55–56,
113–114, 120, 122, 124, 126, 127, 61–62, 65–86, 87–91, 131, 136–141,
130–132, 138, 142, 144, 151, 163, 181–183, 203–204, 210, 214, 228,
171–172, 189, 193–194, 199, 205– 233, 239, 241, 243–246, 248, 250,
207, 209, 212–215, 218–220, 222, 252
227–228, 235, 237–242 stemmatic archetype 138–139
scribal improvements 124 stemmatics 25–37, 39, 41, 44, 54–55,
scribal interference 37 59–61, 80, 84, 89, 103, 138–140,
scribal modifications 165 181, 241, 252
scribal motivation 144 stereotyped formula 93
scribal readings 24, 26, 39, 44, 62, 136, stream 30, 38, 206, 208, 211, 241–244,
141 248
scribal slips 245 “Streeter Problem” 77
scribal tendencies 15, 136, 211, 214 strengthening 107–108, 119, 134, 139,
scribal variants 12,16–17, 39, 70 186, 195, 204, 209, 213–214, 225–
scribes 3–5, 8, 10, 11–19, 27, 30, 39 227, 230, 233, 240, 251
scriptural quotation 4, 134 style 14, 17, 32, 120, 151
second century 11, 83, 126, 162, 169, subjective genitive 97, 101, 251
171, 201, 241–242, 248, 253 subjunctive 126–127, 171–172, 201,
second position 160, 217, 231 214, 251
secretary 10 substemma 41
section sub–singular reading 246
– major 118–120, 224, 251
308 Index of Subjects

superfluous wording 90, 111, 167–168, unoriented stemma 29, 44, 62, 65–71,
250 73, 78–80, 84–88, 138
supersessionist 192–193, 230 variant 4–7, 11–17, 22, 28–29, 32–35,
Swanson’s transcriptions 47, 50–53 38–41, 43, 45, 53–54, 58–59, 61, 66,
Syriac version 52, 56, 81, 187, 210–211, 70, 74, 76–78, 80, 82–84, 89–90, 95,
214, 220 106, 109, 117, 123, 126–127, 133,
Syrian text 55–56, 82, 85–86, 211, 242, 139–142, 144, 149–151, 157–159,
248 164, 168–169, 181–182, 184, 197–
template 216 198, 202–203, 206, 209, 213, 218–
tenacity of the tradition 38–39 219, 221, 223, 226–229, 233, 237,
text-flow diagram 42 239, 242, 244, 247, 250–251, 253
text-type 35–40, 43, 58–59, 210 variant reading, see variant
text-type eclecticism 37 variation unit 6, 31, 34–35, 39–41, 43,
textual flow 40–43, 54 46, 53–54, 59, 62–63, 65, 67–68,
textual history, see history of the text 70–72, 76–77, 80, 89, 91, 96, 102–
Textus Receptus 56–59 103, 107, 114, 117, 131, 135, 137–
thanksgiving section 92 138, 141–143, 149, 152, 162, 171,
Theodoret of Cyrus 222 211, 233, 238, 241–243, 245, 251
theologians 1, 8 version 35, 47, 89
theological effects 181–182, 184, 190– voice 5, 10, 212, 214, 221
192, 193, 208, 219–220, 228, 233 Vulgate 52, 80–81, 182–183, 198, 247–
– motivated change 214 248
– significant variants 182, 208, 239, Wackernagel’s Law 160
241, 251–252 wax tablets 11
theological interests 116 Western text 35–36, 45, 55, 57–58, 75,
thoroughgoing eclecticism 34, 38 80–81, 85, 88–91, 93, 95–96, 102–
topic 25, 94, 115, 118, 150, 155, 181, 103, 105, 107–110, 112–117, 121,
208, 216–217 124–126, 128–129, 131–134, 136–
Torah 1, 111, 153, 161, 202, 208–209, 141, 143, 145, 149, 152, 154, 157–
see also Law, the 159, 161–163, 168–169, 171–172,
Torah observant 230 182–187, 189–190, 192, 195–199,
traditional formulation 144 201, 203–206, 208–210, 219–220,
transcription 47–53 222, 225–233, 235, 239, 241–243,
transcriptional probabilities 14–15, 21, 247–250, 253
33, 35, 90, 92–93, 101, 107, 112, – prototype 183–185, 197, 199, 220
115, 124, 144, 148, 152, 157, 165 – readings 55, 90, 189, 192, 196, 220,
translation 1–2, 4, 22, 81, 159 228, 242, 247–250
transposition 110, 133, 136, 184, 186– – of Acts 182, 187
187, 189, 195, 200, 203, 205, 207, word order 68, 95–96, 110, 133, 150,
211, 216–217, 220, 223, 225, 227, 154, 165, 189, 194, 196, 207, 216–
230–231, 234–236 217, 225, 227, 250
– variant 39, 95, 150, 187, 189, 209,
uncial 35, 58, 83, 233 219, 228, 231, 251, 253
uncircumcised 123, 146, 187 work (ancient) 3–4, 16, 19–20, 23, 29,
United Bible Societies 34 39, 53, 70