Speaking of Nothing

Author(s): Keith S. Donnellan

Source: The Philosophical Review, Vol. 83, No. 1 (Jan., 1974), pp. 3-31
Published by: Duke University Press on behalf of Philosophical Review
Stable URL: http://www.jstor.org/stable/2183871
Accessed: 09/09/2008 10:49

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=duke.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

http://www.jstor.org
 SPEAKING OF NOTHING

RUSSELL tells us in "On Denoting" to test our logical

theories by their "capacity for dealing with puzzles."2 In
this paper I raise the question of how a theory of reference, one of
recent origin, might handle one of the major puzzles Russell
mentioned. The theory of reference that I have in mind-and one
I subscribe to-I will call "the historical explanation theory." (It,
or ones similar to it in important respects, has also been called the
"causal theory." For various reasons, I prefer a different title.3)
Among a number of puzzles mentioned by Russell, two stand
out as more important than the others. One is the well-known
problem of identity statements with which Frege begins his article,
"On Sense and Reference,"4 the question of how a statement of
the form "a is identical to b," when true, can differ in "cognitive
value" from a corresponding statement of the apparently trivial
form, "a is identical to a." The second puzzle is the topic of this
paper. In a large number of situations speakers apparently refer
to the nonexistent. The most obvious example of this is, perhaps,
the use of singular terms in negative existence statements-for
example, "The discoverer of the philosopher's stone does not
exist" or "Robin Hood did not exist." The problem is, of course,
well known and ancient in origin: such statements seem to refer
to something only to say about it that it does not exist. How can
one say something about what does not exist? For a few philos-
ophers, to be sure, these questions have led to attempts to provide
the referent. But in general such attempts have been met with
suspicion. Russell certainly thought it a merit of his theory of

1 Earlier versions of this paper were read at a number of meetings and

colloquia and several important changes have resulted from those discussions.
I am particularly grateful for detailed comments by Tyler Burge.
2 Reprinted in Logic and Knowledge, ed. by Marsh (2956), p. 47.
3 Why I am reluctant to use the word "causal" may become somewhat
clearer further on, but the main reason is that I want to avoid a seeming
commitment to all the links in the referential chain being causal.
4Translations from the Philosophical Writings of Gottlob Frege, ed. by Geach
and Black (I 952).

3
 KEI TH S. D ONNELLAN

definite descriptions (and his fully developed views on singular

expressions) that such apparent references to the nonexistent were
explained without having to entertain the idea of referents of
singular terms that are nonexistent.
Where the singular terms involved are definite descriptions,
"On Denoting" provided a solution to the two puzzles mentioned
that was at once a break-through in the treatment of these expres-
sions and satisfying in the coherent explanation it gave. Russell's
fully developed theory of singular terms, perhaps best represented
in "Lectures on Logical Atomism,' 5 extends the proposed solution
to ordinary proper names, for these turn out to be concealed
definite descriptions. The view of "On Denoting" now could be
made to cover most of the uses of singular terms in language as we
actually speak it and, moreover, seemed to meet the test of solving
the various puzzles about reference. But the fully developed view
also introduced a category of singular expressions that were
acknowledged to be rarely, if ever, found in everyday speech-
what Russell called names in "the strict logical sense" or
''genuine' names.
Genuine names and the motivation for giving pride of place to
such exotic singular terms have special interest for the historical
explanation theory, because while its treatment of ordinary
singular expressions is radically different from Russell's it has some
similarities to his characterization of genuine names.
The question posed, then, is how the historical explanation
theory of reference can handle the puzzle that Russell's view has
no difficulty with, the problem of apparent reference to the non-
existent.
I cannot in this paper plead the full case for the historical
explanation theory, though I shall try to give its main features;
so it may be best to consider it an exercise in the hypothetical:
if the theory is correct what follows concerning apparent reference
to the nonexistent ?6

5Reprinted in Logic and Knowledge.

6 If we divide the theory into its negative aspects (see sec. III) and its
positive (see sec. IV), what the theory denies and the reasons for doing so
have been, perhaps, better delineated in the literature than the content of the
positive theory. (This is certainly true of my own contributions.) My papers

4
 SPEAKING OF NOTHING

I. THREE KINDS OF APPARENT REFERENCE TO THE

NONEXISTENT

We need to keep distinct three situations in which apparent

reference to the nonexistent occurs. The differences are important
in their own right, but I need to call attention to them because
one kind of situation will be excluded from consideration in this
paper.
I will, in the first place, distinguish what I will call "discourse
about fiction" from "discourse about actuality"; and, secondly,
within the latter category, the use of "predicative" statements
from the use of "existence" statements.7 What is to be excluded
from consideration here is an account of discourse about fiction.
(This is not, of course, to say that such an account is not in the
end needed.)
Under "discourse about fiction" I mean to include those occa-
sions on which it is a presupposition of the discourse that fictional,
mythological, legendary (and so forth) persons, places, or things
are under discussion. I believe, for example, that said with the
right intention, the following sentences would express true propo-
sitions: "The Green Hornet's car was called 'Black Beauty,' "
"Snow White lived with seven dwarves," and "To reach the
underworld, one must first cross the River Styx." (By the "right
intention" I mean that the speaker wishes to be taken as talking
about fiction, mythology, or legend.) At the same time I also
believe it is true that neither the Green Hornet, his car, Snow
White, nor the River Styx exists or ever has existed. These two
beliefs, however, are entirely consistent. And therein lies the puzzle:
dealing with various parts of the theory as I see it are: "Reference and Definite
Descriptions," Philosophical Review, LXXV (i966), 28I-304; "Putting Humpty
Dumpty Together Again," Philosophical Review, LXXVII (i968), 203-2I5;
"Proper Names and Identifying Descriptions," Synthkse (1970), reprinted in
Davidson and Harman (eds.), Semantics of Natural Language (Dordrecht, I972).
By others, Saul Kripke's paper, "Naming and Necessity," in Semantics of
Natural Language, is the most important in that it gives not only arguments
for the negative aspects of the theory, but also a positive account (that,
however, I do not altogether agree with).
I The terminology, of course, is for convenience and not supposed to reflect
a prejudgment that existence cannot be in some sense a genuine predicate.

5
 KEITH S. DONNELLAN

how can there be true propositions that apparently involve

predicating properties of what does not exist?
Discourse about actuality carries the presupposition that the
speaker is talking about people, places, or things that occur in the
history of our world. A puzzle arises when the speaker is unfor-
tunate enough to use a singular expression, intending to attribute
a property to something, but fails, in his use of that expression, to
refer to anything. This very likely occurred, for example, some
years ago following the publication of The Horn Papers,8 that
purported to contain the diary of one Jacob Horn and that would,
if genuine, have shed light on the colonial history of Washington
County, Pennsylvania. Many people believed them to be genuine,
but, on the evidence, it seems likely that they are not and that
Jacob Horn did not exist. There must have been many believers,
however, who made statements using the name "Jacob Horn"
with the intention of predicating various properties of a historical
figure. For example, someone might well have said, "Jacob Horn
wrote about Augusta Town and now we know where it was
located." It would have been some sort of inconsistency-exactly
what kind is another question-for such a speaker then to affirm
the nonexistence ofJacob Horn. This contrasts with discourse about
fiction there one can, for example, consistently deny the existence
of Snow White while also stating that she enraptured a prince.9
The puzzle about predicative statements, as I shall call them,
in discourse about actuality with a singular expression and no
referent is more subtle philosophically than the puzzle about
fictional discourse. There is not the same possibility of stating
something true. Nor can the speaker with consistency acknowledge
the nonexistence of what he speaks about. To see how statements

8 See Middleton and Adair, "The Mystery of the Horn Papers," William
and Mary Quarterly, 3rd series, IV (I947); reprinted in Winks (ed.), The
Historian as Detective (New York, i968).
9 The denial of Snow White's existence, it should be noted, is in discourse
about actuality, while the statement that she enraptured a prince is in discourse
about fiction. (If the question of existence arose in discourse about fiction alone,
Snow White existed, whereas Hamlet's father's ghost, again presuming we are
talking about fiction, probably did not.) This does not disturb the point:
no such contrast can be made out for Jacob Horn; if Jacob Horn did not
exist then there are no true predicative statements to be made about him.

6
 SPEAKING OF NOTHING

such as those made by believers in the authenticity of The Horn

Papers can puzzle a philosopher requires the ability to see a
difficulty in how one can even speak and be understoodwhen using
a singular expression with no referent.
The difference between discourse about fiction and discourse
about reality, it is important to keep in mind, is a matter of pre-
suppositions about the intent of the speech act. It is not that in the
one fictional characters are involved and in the other real people,
places, and things. A not too well-informed person might have
taken (at least the first part of) the movie Doctor Strangelovefor a
documentary. His statement, "Doctor Strangelove, the top mili-
tary scientist in the United States, is a psychopath," would then
be a bit of discourse about reality, even though Doctor Strangelove
is, in fact, fictional. On the other hand, this very same sentence
used by someone having seen the whole movie would probably be
a comment on the movie, a bit of discourse about fiction.
While I will need often to consider predicative statements about
actuality, the problem I want to concentrate on concerns "exis-
tence" statements-those that have either the form "S does not
exist" or the form "S exists," where "S" is a singular expression.
Negative existence statements, unlike predicative statements, are
true when there is no referent for the singular expression. If I speak
the truth in saying, "Jacob Horn does not exist," I would be
apparently referring to what does not exist. But even more para-
doxically, the truth of what I say depends directly upon the non-
existence of a referent for "Jacob Horn." Moreover, this is
discourse about reality; I do not, clearly, intend to talk about a
fictional character. Negative existence statements, of all those
mentioned, bring apparent reference to the nonexistent into
sharpest focus.
It is of some importance to mention the difference between
denying the existence of something altogether and denying its
present existence or its existence at some point or during some
period in time. To begin with we certainly want to distinguish
between
(i) Napoleon no longer exists,
and
(2) Napoleon does not and never did exist.

7
 KEITH S. DONNELLAN

The first statement is both true and not an apparent reference to

the nonexistent in the sense we want. (I) contains a reference to
Napoleon in the same way that "Socrates was snub-nosed"
contains a reference to Socrates. (I) should, it seems, be put into
the class of predicative statements, despite the fact that existence
is involved. On the other hand, (2) is a paradigm of the kind of
statement that generates the problem of apparent reference to the
nonexistent.
What shall we say, however, about a statement such as

(3) Santa Claus does not exist?

Often, I believe, it expresses a statement of the same form as (2),
an absolute denial of existence, not confined to any one period in
time. But suppose, for example, that someone is unsure whether
Jacob Horn ever did exist, but is certain that he does not now
exist. He might express this by saying, "Jacob Horn does not
exist." So, perhaps, sentences of the surface form of (3) are
ambiguous. But what is the other meaning that they might have?
The sentence given, in the imagined circumstances, seems to me
to be equivalent to

(4) Jacob Horn does not now exist.

This is neither the absolute denial of Jacob Horn's existence nor
the predicative assertion thatJacob Horn no longer exists. (4), I
believe, amounts to the disjunction of the two: "Either Jacob
Horn does not and never did exist or he did exist and does no
longer." In which case, the dichotomy illustrated by (I) and (2)
is still maintained. In what follows, however, it is the absolute
denial of existence that will be of concern and any examples of the
form "N does not exist" should be construed in that way.

II. A THEORY GONE WRONG WITH INTERESTING

MOTIVES-RUSSELL

Russell's theory of singular terms holds interest for the historical

explanation theory, not only because of obvious oppositions on
some key issues-several more recent discussions would serve that

8
 BOOK REVIEWS

presupposition of the possibility of evidentjudging or evidence (Evidenz)

that judging in which what is judged is in some mode "itself given"
and which grounds the possibility of truth.
These and other presuppositionsof formal logic and mathematics are
taken by these disciplines "naively" as something already given, about
which logicians need not concern themselves. If, however, "logic" is
to fulfill its responsibility as unifying and foundational for science
as a whole, it must undertake a radical criticism of these presuppositions
One of the merits of FTL is the manner in which Husserl links the
themes of his own explication of formal logic to these new themes, so
as to strengthen both systematically and in detail his claim that formal
logic, the logic of the analytic a priori, is indissolubly bound to a
transcendental logic of the synthetic a priori, and thus to trans-
cendental phenomenology.
If FTL is not a good introduction to Husserl, Part II nonetheless
contains some of the clearest repudiations of several common and
elementary misunderstandings of his views. The most prominent of
these is his repudiation of psychologism in all its forms. Bachelard's
preface has much to say about the theme of antipsychologism in
Husserl, and makes no apologies for "taking Husserl's antipsychologism
as a clue in characterizing the major traits of Husserl's conceptions"
(p. xxxi). Phenomenology does not deal with the subject's "impressions"
and its laws are in no case empirical laws governing "psychic occur-
rences" (FTL, p. 154). Even Vaihinger's "fictionalism" and phenome-
nalism are rejected by Husserl precisely as forms of psychologism. The
phenomenalist, for example, invents "sense-data" as simple psychologi-
cal facts, by means of which he attempts to account for the "being-
sense" "physical object." For such forms of "positivism," says Husserl,
physicalthings become reducedto empiricallyregularcomplexesof psychic
Data ("sensations"); . . . It is not merely a false doctrine, completely blind to
the essentialphenomenologicalfacts; it is also countersensical,becauseof its
failureto see that even [sense-data]have theirmode of being, their mannerof
evidence,their mannerof being unities of multiplicities,and thereforecarry
with themthe sameproblemthat was to be theorizedaway by meansof them.
[FTL, p. I66].
The phenomenalist wishes to reduce the being and evidence of "physical
objects" to those of "sense-data." But he both fails to solve his own
problem and creates further ones for himself. For he loses the "being-
sense of physical things in his reduction to sense data, and gains a
whole set of further questions as to the sense of these newly invented
entities and the intentionality in which they are constituted.

270
 KEITH S. DONNELLAN

always have descriptive content. The question is, does this mean
that perhaps ordinary singular expressions may fulfill the function
that Russell thought only "genuine" names, with all their pecu-
liarities, could? And, if so, how can the historical explanation
view deal with the puzzles about reference?
What was the motivation for introducing "genuine" names?
Russell often talks in ways that can seem nonsensical-that, for
example, when a definite description such as "the author of
Waverly" is involved, the denotation of the definite description,
Scott in this case, is not a "constituent" of the proposition ex-
pressed. The implied contrast is that if "Scott" is a genuine name
and were there in place of the definite description "the author of
Waverly" then Scott would be a constituent. But it certainly
sounds queer at first glance to find a flesh-and-blood person in a
proposition!
Russell's analysis of statements containing definite descriptions
and, by extension, ordinary proper names, shows, he believed, that
such statements are not really about, do not really mention, the
denotation of the description or the referent of the name. Russell
emphasizes this again and again. "Genuine" names, on the other
hand, can somehow perform the feat of really mentioning a
particular individual. To try to put much weight on such terms as
"about" would lead us, I think, into a morass. What it is for a
statement to be about an individual, if that requires any attempt
to define aboutness,is a question better avoided if we are ever to get
on with the problem. (After all, Russell himself recognized a
well-defined relationship that a statement containing a definite
description can have to some particular individual-its denotation.
It would be a delicate task to show either that in no sense of
"about" is such a statement about the denotation of the definite
description or that there is some clear sense of "about" in which
it is not.)
But I believe we can say something useful about the reasons
Russell had for talking in this way. On his theory of definite
descriptions the singular expression, the definite description, is
really a device that introduces quantifiers and converts what
might seem at first sight a simple proposition about an individual
into a general proposition. "The 0 is f" expresses the same propo-

IO
 SPEAKING OF NOTHING

sition as "There is a S and there is at most one b and all O's are
if's"; and the latter clearly would express a general proposition
about the world. Ordinary proper names, of course, function on
his view in the same way, since they are in reality concealed
definite descriptions. Now if we contrast these singular expressions
with ones, if there are any, that do not introduce quantifiers, that
when put as the subject of a simple subject-predicate sentence do
not make the sentence express a general proposition, then I think
there is a strong temptation to say that only the second kind of
singular term can be used to really mention an individual.
Russell clearly believed that there must be the possibility, at
least, of singular terms that do not introduce quantifiers; that
seems in large part to be his reason for believing in "genuine"
names. Whether or not there is some argument that shows the
necessity of such singular terms, I believe that prior to theory the
natural view is that they occur often in ordinary speech. So if one
says, for example, "Socrates is snub-nosed," the natural view
seems to me to be that the singular expression "Socrates" is simply
a device used by the speaker to pick out what he wants to talk
about while the rest of the sentence expresses what property he
wishes to attribute to that individual. This can be made somewhat
more precise by saying, first, that the natural view is that in using
such simple sentences containing singular terms we are not saying
something general about the world-that is, not saying something
that would be correctly analyzed with the aid of quantifiers; and,
second, that in such cases the speaker could, in all probability,
have said the same thing, expressed the same proposition, with the
aid of other and different singular expressions, so long as they are
being used to refer to the same individual. To illustrate the latter
point with a different example: if, at the same moment in time,
one person were to say, "Smith is happy," a second "You are
happy," a third "My son is happy," and a fourth "I am happy,"
and if in each case the singular expression refers to the same
person, then all four have expressed the same proposition, have
agreed with each other.
What I see as the natural pre-theoretical view might be cap-
tured as a certain way of representing what proposition is ex-
pressed. For example, the sentence "Socrates is snub-nosed" might

II
 KEITH S. DONNELLAN

be represented as an ordered pair consisting of Socrates-the actual

man, of course, not his name-and the predicate (or property,
perhaps), being snub-nosed. (More complicated sentences, in-
volving relations and more than one singular expression of this
sort would be represented as ordered triplets, and so forth.) Now
if someone were to say to Socrates, "You are snub-nosed," or
Socrates were to say about himself, "I am snub-nosed," the
proposition expressed would, in each case, be represented by the
same ordered pair-propositional identity, given the same predi-
cate, would be a function simply of what individual is referred to.
This way of representing propositions would, I think, meet with
at least provisional approval by Russell, but only if it were
restricted to those propositions expressed by statements containing
"genuine" names. We might even say that the manner of repre-
sentation gives a respectable sense in which an individual might
be a constituent of a proposition. But my examples of statements
for which this representation was suggested would, on Russell's
view, be incorrect just because they involve singular terms from
ordinary language. For Russell, they would be examples of
sentences that express complex general propositions and, whatever
our view of the nature of propositions, I do not think we would
want propositional identity for general propositions to be a func-
tion of the individuals that happen to make the propositions true
or false.
Russell pays the price, I believe, of giving up the natural view
of many uses of ordinary singular terms, a price he is willing to
pay-chiefly, perhaps, because he thus can dissolve puzzles about
reference. The special properties of "genuine" names, on the other
hand, are supposed to rescue them. The "natural" view, on the
other hand, seems to generate Russell's budget of puzzles, in
particular the one which is the concern of this paper. If I say,
"Socrates is snub-nosed," the proposition I express is represented
as containing Socrates. If I say, instead, "Jacob Horn does not
exist," the "natural" view seems to lead to the unwonted con-
clusion that even if what I say is true, Jacob Horn, though non-
existent, must have some reality. Else what proposition am I
expressing? The "natural" view thus seems to land us with the
Meinongian population explosion.

12
 SPEAKING OF NOTHING

Russell, of course, avoids this problem easily. Since the proper

name "Jacob Horn" would, for him, be a concealed definite
description, to say "Jacob Horn does not exist" is not to refer to
some individual in order to say something about him, but merely
to assert that a particular class of things, perhaps the class of
writers of diaries about certain events in early Pennsylvania
history, is either empty or contains more than one member. (So
a singular nonexistence statement of this kind is on all fours with
statements such as "There are no flying horses" or "There is more
than one living ex-President." It does not mention a particular
individual any more than these do.)
The issue has importance for the historical explanation view
because it denies that many singular terms in ordinary language,
in particular proper names, are concealed descriptions of the sort
that Russell had in mind. "Homer," for example, is not a con-
cealed description such as "the author of the Homeric Poems," to
use Russell's own example. The question is, does the historical
explanation view, if correct, support what I have called the
"natural" view? In the next section this question will be con-
sidered.

III. THE HISTORICAL EXPLANATION VIEW: NEGATIVE ASPECT

I now want to begin to lay out the bare bones of the theory of
reference I want to discuss. As I have said, I will not here argue
for its correctness nor will I try to fill in all the gaps.
Russell and the majority of philosophers in contemporary times
who have discussed (ordinary) proper names have held that by
one mechanism or another they are surrogates for descriptions. For
Russell, as I have mentioned, they are simply abbreviations for
definite descriptions; for others-for example, Searle"1-they are
correlated with a set of descriptions and what one is saying in, say,
a simple subject-predicate sentence employing a proper name is
that whatever best fits these descriptions has whatever property is
designated by the predicate. The descriptions, both on Russell's

1 See references in n. 1o.

13
 KEITH S. DONNELLAN

view and on the looser view of Searle and others, which the proper
name masks, are thought of as obtained from the people who use
them-roughly speaking, by what they would answer to the
question, "To whom (what) are you referring?" This view of
ordinary proper names embodies what I have called the "principle
of identifying descriptions."12 The theory of reference I am con-
cerned with holds that the principle of identifying descriptions is
false.
What this means, to give an example, is that, supposing you
could obtain from me a set of descriptions of who it is that I
believe myself to refer to when I say, "Socrates was snub-nosed"-
perhaps such things as "the mentor of Plato," "the inventor of the
'Socratic method,' " "the philosopher who drank the hemlock,"
and so forth-it is theoretically possible that I am referring to
something about which no substantial number of these descrip-
tions is true or that although there is something that fits these
descriptions to whatever extent is required by the particular
variation of the principle, that is not in fact the referent of the
name I used.
On this theory, then, ordinary proper names are like Russell's
"genuine" names at least in so far as they do not conceal descrip-
tions in the way he thought. This is, I think, a virtue of the theory.
As David Kaplan has remarked, there was always something
implausible about the idea that a referent of a proper name is
determined by the currently associated descriptions.

IV. THE HISTORICAL EXPLANATION THEORY: POSITIVE ASPECT

The first tenet of the theory of reference I have been describing

was negative-the view that proper names must have a backing
of descriptions that serves to pick out their referents is false. The
second tenet is positive, but more tentative. How is the referent of
a proper name, then, to be determined? On Russell's view and
variants on it, the answer to this question would be simple: the
referent is that which fits the associated descriptions best, where

12 In "Proper Names and Identifying Descriptions," op. cit.

14
 SPEAKING OF NOTHING

"best" may be defined differently by different writers. As I see it,

one of the main reasons a backing of descriptions for proper names
is so attractive is that it furnishes a simple way of ascertaining
what a speaker is saying, and of determining whether what he says
is true or false (given that we are dealing only with assertions).
We find, so to speak, that thing in the world which uniquely fits
the descriptions and then see whether or not it has the properties
ascribed to it. If proper names do not have a backing of descrip-
tions, how do we decide whether or not, when someone says, for
example, "Russell wrote 'On Denoting,' " he has said something
true or false ?
Putting existence statements aside, when a speaker says some-
thing of the form, "N is p," where "N" is a name and "p" a
predicate, we can say that in general the truth conditions will have
the following form. What the speaker has said will be true if and
only if (a) there is some entity related in the appropriate way to
his use of "N" in this sentence-that is, he has referred to some
entity, and (b) that entity has the property designated by O. (I say
"in general" because there are difficulties for any theory of
reference about uses of names for fictional characters, "formal"
objects such as numbers, and so forth.) The question is, what is the
"appropriate relation" mentioned in condition (a) ? How, that is,
does an entity have to be related to the speaker's use of the name
"N" to be its referent? The principle of identifying descriptions,
were it only true, has a simple answer to this: the entity must have
(uniquely) the properties or some sufficient number of the prop-
erties designated by the "backing of descriptions" for this use of
the name "N." Roughly speaking, and on the most usual view,
it will be the entity that answers to the descriptions the speaker
would (ideally) give in answer to the question, "To whom are you
referring ?"
But even without the arguments that, I believe, show the
principle of identifying descriptions not only false, but implausible,
putting the matter in this general way is somewhat liberating. It
shows that what we need is some relation between the speech act
involving the name "'A" and an object in the world-the right
one, of course-but the relation supplied by the principle of
identifying descriptions is now only a candidate for that office.

'5
 KEITH S. DONNELLAN

But if the principle of identifying descriptions is false, what then

is the appropriate relation between an act of using a name and
some object such that the name was used to refer to that object?
The theory of reference I want to discuss has not as yet, so far as
I know, been developed in such a way as to give a completely
detailed answer. Yet there are positive things that can be said and
enough, I believe, both to contrast it with the principle of identi-
fying descriptions and to give us something like an answer to the
original question: how will it handle apparent reference to the
nonexistent in such statements as "Santa Claus does not exist"?
The main idea is that when a speaker uses a name intending to
refer to an individual and predicate something of it, successful
reference will occur when there is an individual that enters into
the historically correct explanation of who it is that the speaker
intended to predicate something of. That individual will then be
the referent and the statement made will be true or false depen-
ding upon whether it has the property designated by the predicate.
This statement of the positive thesis leaves a lot to be desired in the
way of precision, yet with some clarifying remarks I think it has
more content than might at first sight be supposed.
Suppose someone says, "Socrates was snub-nosed," and we ask
to whom he is referring. The central idea is that this calls for a
historical explanation; we search not for an individual who might
best fit the speaker's descriptions of the individual to whom he
takes himself to be referring (though his descriptions are usually
important data), but rather for an individual historically related
to his use of the name "Socrates" on this occasion. It might be that
an omniscient observer of history would see an individual related
to an author of dialogues, that one of the central characters of
these dialogues was modeled upon that individual, that these
dialogues have been handed down and that the speaker has read
translations of them, that the speaker's now predicating snub-
nosedness of something is explained by his having read those
translations. This is the sort of account that I have in mind by a
"historical explanation."
Several comments are in order here. First, it is not necessary, of
course, that the individual in question be snub-nosed; obviously
the speaker may have asserted something false about the referent

i6
 SPEAKING OF NO THING

of the name "Socrates." Second, if we take the set of descriptions

the speaker could give were we to ask him to whom he was
referring, the historical explanation as seen by our omniscient
observer may pick out an individual as the referent of the name
"Socrates" even though that individual is not correctly described
by the speaker's attempt at identification. For example, the
speaker may believe that Socrates-that is, the person he refers
to-was a philosopher who invented the Socratic method. But it is
clearly imaginable that our omniscient observer sees that while
the author of the dialogues did intend one of the characters to be
taken as a portrayal of a real person, he modestly attributed to
him a method that was his own brain child. And, in general, it
would be possible to have the historical connection with no end
to mistaken descriptions in the head of the speaker. The descrip-
tions the speaker gives, however, may play an important role,
though not the one given to them by the principle of identifying
descriptions. The omniscient observer may see, for example, that
the reason the speaker believes himself to be referring to someone
who invented a certain philosophical method is that his present
use of the name "Socrates" is connected with his having read
certain translations of these dialogues. Or, to take a slightly
different case, he may see that his descriptions come from a faulty
memory of those dialogues, and so forth. The question for the
omniscient observer is "What individual, if any, would the speaker
describe in this way even if perhaps mistakenly?"
I have used the notion of an omniscient observer of history and,
of course, we ordinary people cannot be expected to know in
detail the history behind the uses of names by those with whom we
converse. Nor do we often make the sort of historical inquiries
which would reveal those details. We often assume, for example,
that if another speaker's descriptions of the referent of a name he
has used more or less jibe with descriptions we would give of a
person, place, or thing that we believe ourselves to know about,
then he is referring to that. Also, for example, the context of the
use of a name may lead us to assume without question that the
speaker refers to someone with whom we are both acquainted.
But the historical explanation theory need not deny this or be
troubled by it. All it needs to hold is that the final test for reference

'7
 KEITH S. DONNELLAN

is the kind of historical connection I have described, that the

customary assumptions and use of indicators are in the end
dependent upon being fairly reliable guides to the existence of
such a connection.
What the historical explanation theory must attempt to estab-
lish is that when there is an absence of historical connection
between an individual and the use of a name by a speaker, then,
be the speaker's descriptions ever so correct about a certain
individual, that individual is not the referent; and, on the other
hand, that a certain historical connection between the use of a
name and an individual can make the individual the referent even
though the speaker's descriptions would not by themselves single
out the individual. This job must be accomplished by building up
examples in which these two points are made obvious. We might,
for instance, try to show that the historical connection is necessary
by constructing a situation in which, for instance, one person
begins by assuming that another is referring to a friend of his,
perhaps because the descriptions seem accurate, the context is
appropriate, and so forth, and who then discovers that it is
practically impossible for the speaker to have been acquainted
with or otherwise related to his friend. In such an event, surely
confidence that the speaker was referring to the friend would be
shaken despite the apparent accuracy of description or appropri-
ateness of context. But, as I have said, I cannot here undertake the
full defense of the historical explanation theory.
There are, however, two further points of clarification that
ought to be mentioned here. It should be obvious that I have only
provided an example of what counts as a historical explanation
rather than a formula for obtaining the referent of a particular use
of a name. Even in the illustration several individuals entered into
the account, only one of which was the referent. Of the individuals
who are in some way or other part of the historical explanation
of a use of a name, which is the referent? What kind of theory is
this if it does not give us the means to make this determination ?
In defense against this charge that the theory is excessively
vague, it is helpful, I think, to compare it with another philo-
sophical theory about a quite different problem. The causal theory
of perception can be taken as holding that an observer, 0, per-

I8
 SPEAKING OF NO THING

ceives an object, M, only if M causes 0 to have sense impressions.

The theory seems to me to have content and to be important,
whether or not it is correct. For one thing, if true it means that
certain other theories are mistaken. But the theory as stated does
not, obviously, allow us to say which among the various causal
factors involved in an observer having sense impressions is the
thing he perceives; nor does it tell us which ways of causing sense
impressions are relevant. Possibly no philosophical analysis can
determine this, although in any particular case we may be able
to say that this is or is not the right sort of causal connection.
Analogously, the historical explanation theory lacks this sort of
specificity. But for all that, if it is true, certain other theories, in
particular the identifying descriptions theory, will be wrong and
the theory does tell us something of importance.
Because there have sometimes been misunderstandings about
this, I think I- should point out that the history to which the
historical explanation theory alludes is not the history of the use
of a name. It is not the history of the use of, say, the name "Soc-
rates" that is important. Socrates may not have been, as far as
theory goes, called "Socrates"; corruption of names is just as
possible as corruption of information. (The history of such a
corruption, however, might enter into the historical explanation.)
Nor, I think, should the theory be construed as holding that the
historical connections end with some original "dubbing" of the
referent. It may be that people, places, and things usually receive
names by some such ceremony and that we generally use names
(or corruptions of them) as a result of such a ceremony, but it is
not a theoretical necessity that names enter our linguistic trans-
actions in this way.13
What the historical explanation does, then, is to provide the

13
That is to say, the first use of a name to refer to some particular individual
might be in an assertion about him, rather than any ceremony of giving the
individual that name. (In fact, my own name is an example: I discovered that
colleagues were pronouncing my last name differently than my parents do-so,
orally, they referred to me by a different name-and I let it stand. But I was
never dubbed by that new name. I am sure that the first use of it was either
an assertion, question, or whatever about me and not a kind of baptism.
And I think it is probable that whatever audience there was knew to whom
the speaker referred.)

I9
 KEITH S. DONNELLAN

relationship between the use of a referring expression and the

referent which the principle of identifying descriptions presupposes
could be provided only by some measure of correct descriptions of
the referent known to the speaker. I think there are counter-
examples to the principle of identifying descriptions14 and, of
course, if there are that defeats it straight off. Still a plausible, if
not clearly correct, alternative theory in this case also acts as an
objection. For one of the principle reasons that many philosophers
have for adopting the principle of identifying descriptions is
that they cannot see how there could be an appropriate relation
otherwise that would pick out the referent of (as the main example)
a proper name.
I have, in describing this theory of reference, talked about a
"historical explanation." I hope it is obvious that "historical" is
being used in the broadest sense possible; that all of what I have
said could just as well be applied to cases in which one refers, by
use of a name, say, to someone still extant, to someone who has
just gone out of the room, or to someone presently in one's
company. The "historical explanation," in other words, can
involve as brief an interval of time as one pleases.

V. A SOLUTION TO THE PUZZLE REJECTED

My problem, then, is to show how such a theory of reference

can deal with simple existence statements expressed by the use of
a proper name, the difficulty being that on this theory, proper
names do not have a backing of descriptions and, in general, they
function to refer via what I have called a historical connection
with some individual. But a true negative existence statement
expressed by using a name involves a name with no referent and
the corresponding positive existence statement, if false, will also.
But in other contexts, when a name is used and there is a failure
of reference, then no proposition has been expressed-certainly
no true proposition. If a child says, "Santa Claus will come
tonight," he cannot have spoken the truth, although, for various

14
See my "Proper Names and Identifying Descriptions," op. cit.

20
 SPEAKING OF NO THING

reasons, I think it better to say that he has not even expressed a

proposition
One apparently possible solution to the problem must be
rejected. Russell and others, as we have seen, thought of (ordinary)
proper names as concealed definite descriptions; he held a version,
that is, of the principle of identifying descriptions. Existential
statements involving ordinary proper names were therefore no
problem for him-they were really existential statements involving
definite descriptions and could be analyzed in accordance with his
theory of definite descriptions. The suggestion I want to look at is
that while our theory tells us that names in predicative statements
do not obey the principle of identifying descriptions and are not
concealed definite descriptions, existential statements may rep-
resent a special case. Thus, so this suggestion would run, "Santa
Claus" in "Santa Claus will come tonight" is not a concealed
definite description, but is one in the special context of "Santa
Claus does not exist" and "Santa Claus exists." This would, of
course, immediately solve our problem, but unfortunately it is not
a solution that our theory can accept. The difficulty is not that
names would be treated as functioning differently in different
contexts; in fact, as will become evident, my own view is that they
do behave differently in existence statements. Rather, the trouble
is that any theory that rejects the principle of identifying descrip-
tions for predicative statements must also reject it for existence
statements.
To simplify matters, let us restrict ourselves to Russell's version
of the principle of identifying descriptions in which a name simply
stands in place of some definite description. If we adopt the prin-
ciple for existence statements involving names, this will come to
saying that, for example, "Socrates did not exist" means the same
thing as (expresses the same proposition as) some other sentence
formed from this by replacing "Socrates" by a definite description
-perhaps, say, the sentence, "The Greek philosopher who was
convicted of corrupting the youth and drank hemlock did not

15
Given that this is a statement about reality and that proper names have
no descriptivecontent, then how are we to represent the proposition
expressed?

2I
 KEITH S. DONNELLAN

exist." But, now, on any view we must, I think, accept the

following:
(E) That Socrates did not exist entails that it is not true
that Socrates was snub-nosed.

Our theory tells us that the second occurrence of "Socrates"

in (E) is not a concealed definite description. But then neither can
the first occurrence be one. For if we take some definite description
such as the one suggested as'what the first occurrence of "Socrates"
stands for, rejection of the principle of identifying descriptions for
the second occurrence means that it could be true that Socrates
was snub-nosed even though no unique individual existed who
satisfied that description. That is to say, if "Socrates" in "Socrates
did not exist" is a concealed definite description, but is not.in
"Socrates was snub-nosed," then the antecedent of (E) could be
true while the consequent is false. Since we want to accept the
entailment expressed by (E) our theory cannot treat "Socrates"
as a concealed description in existential statements.
This solution not being open to us, we cannot on the other
hand go to the opposite extreme and handle existential statements
involving ordinary proper names in the way Russell did for what
he called names "in the strict logical sense." There simply are no
meaningful existential statements involving these "genuine"
names and so the problem does not arise about how to deal with
them. But, of course, we cannot countenance this about ordinary
proper names, for it does make sense to say, "Homer existed" or
"Santa Claus does not exist."

VI. TRUTH CONDITIONSAND BLOCKS"

What we need to do first is see what, on our theory of reference,

the truth conditions are going to look like for existence statements
involving names. In predicative statements, such as "Homer was
a great poet," if everything goes well, there will be some individual
related to this use of "Homer" "historically," as I have put it, and
the statement will be true if that individual had the property
expressed by the predicate and false otherwise. This, of course,

22
 SPEAKING OF NO THING

cannot be so for a negative existence statement such as "Homer

did not exist." This statement would be true, in fact, just in case
there is a failure of reference, not in the statement itself, but in
other possible or actual predicative statements involving the name.
That is, if there is no individual related historically in the right
way to the use of "Homer" in, say, the statement "Homer was a
great poet," no individual whose possession or nonpossession of
poetic genius makes this true or false, then we can truly state that
Homer did not exist.
Initially then the question comes to this: "What, on our theory,
constitutes a failure of reference in a predicative statement
involving a proper name?" (As we shall see there is more to the
matter than just this.) Since the positive part of our theory, the
part that attempts to say what successful reference to an individual
consists in, has been, perhaps because of the nature of things, left
more suggestive than in a rigorously formulated state, it cannot
be hoped that we shall do much better with failure of reference.
But we can say some things of a non-trivial nature.
Suppose a child who believed in Santa Claus now learns the
truth, the truth which he expresses by saying, "Santa Claus does
not exist." He comes to learn this, as usual, from cynical older
children; what has he learned? Our account is that he has learned
that when in the past he believed something, for example, which
he would have expressed by saying, "Santa Claus comes tonight,"
and would have thought himself in saying this to be referring to
someone, the historical explanation of this belief does not involve
any individual who could count as the referent of "Santa Claus";
rather it ends in a story given to him by his parents, a story told
to him as factual. I do not mean, of course, that the child would
or could express the knowledge he has in his new state of disillu-
sionment in this fashion-that would require him to know the
correct account of reference. But if we are approaching the correct
theory, then this is how we can state what he has discovered.
When the historical explanation of the use of a name (with the
intention to refer) ends in this way with events that preclude any
referent being identified, I will call it a "block" in the history. In
this example, the block is the introduction of the name into the
child's speech via a fiction told to him as reality by his parents.

23
 KEITH S. DONNELLAN

Blocks occur in other ways. For example, children often invent

imaginary companions whom they themselves come to speak of as
actual. The block in such a case would occur at the point at which
a name for the unreal companion gets introduced by the child
himself via his mistaken belief that there is a companion to name.
A somewhat different example would be this: suppose the Homeric
poems were not written by one person, but were a patchwork of
the writings of many people, combined, perhaps, with fragments
from an oral tradition. Suppose, further, that at some point in
time an ancient scholar for whatever reasons-he might have seen
a name attached to some written version of the poems and
supposed it to be the name of the author-attributed the poems to
a single person he called "Homer." If this were the historical
explanation of our saying, for example, "Homer wrote the Iliad,"
then the block occurs at the point at which this scholar enters the
picture.
On theories that subscribe to the principle of identifying des-
criptions, examples of failure of reference such as occur in this last
example would be treated as a failure to satisfy a uniqueness
condition. The reason that Homer would not have existed given
these circumstances is that no single individual satisfies the des-
criptions we associate with Homer (or satisfies a "sufficient"
number, according to certain views). But according to our theory
this is not the reason for failure of reference; it is rather that the
history of our use of the name, a history with which we may not be
familiar, does not end in the right way. One way to see that the
opposing account, though plausible, is wrong is to think of the
possibility of someone existing who doessatisfy the descriptions we
might supply of the referent of a name we use, but who has no
historical connection with us whatsoever. Suppose, for example,
that contrary to what we adults believe we know, there is, in fact,
a man with a long white beard and a belly like a bowl full ofjelly
who comes down chimneys on Christmas night to leave gifts (the
ones whose labels are missing about which parents worry because
they don't know to what aunt the child should write a thank-you
note). We must, of course, imagine that it is absolutely fortuitous
that our descriptions of Santa Claus happen to fit so accurately
this jolly creature. In that case I do not think that he is Santa

24
 SPEAKING OF NO THING

Claus. The fact that the story of Santa Claus, told to children as
fact, is historically an invention constitutes a block even if the
story happens to contain only descriptions that accurately fit
some person.

VII. A RULE FOR NEGATIVE EXISTENCE STATEMENTS

Using the technical, but admittedly not well-defined, notion of

a "block," we can now sketch the way the historical explanation
theory may treat negative existence statements involving names.
A similar treatment could then be given for positive existence
statements.
I will suggest a rule, using the notion of a block, that purports
to give the truth conditions for negative existence statements
containing a name. This rule, however, does not provide an
analysis of such statements; it does not tell us what such statements
mean or what proposition they express. This means that in this
case we are divorcing truth conditions from meaning.
With the deletion of some qualifications that would be needed
to make it strictly correct, the rule can be expressed as follows:
(R) If N is a proper name that has been used in predicative
statements with the intention to refer to some individual,
then FN does not exist" is true if and only if the history
of those uses ends in a block.
The rule as stated obviously requires some modifications. For
one thing we would need some way of distinguishing, for example,
the denial of the existence of Aristotle the philosopher, from
Aristotle the ship magnate. To accomplish this we must do two
things: first, find a means of collecting together the uses of
"Aristotle" in predicative statements that were, so to speak,
attempts to refer to the philosopher, separating them from a
similar collection of uses of the name that were attempts to refer
to the ship magnate, and do this without, of course, assuming that
any of these uses succeeds in referring. Second, we must be able
to relate a particular negative existence statement using the name
"Aristotle" to one such collection rather than any other.

25
 KEITH S. DONNELLAN

The way of amending Rule (R) that seems to me in keeping

with the historical explanation theory and to accomplish these
tasks is this. Certain uses of the name "Aristotle" in predicative
statements will have similar histories, histories that will distinguish
them from other uses of the name. Each use of the name will, of
course, have its own historical explanation, but these may, at a
certain point, join up. So, in tracing back several uses of the name
"Aristotle" by me and several uses by you, we may find a common
root in certain ancient writings and documents, while other uses
of the name by me or by you may have nothing in common with
the history of the first set of uses. It is possible that the histories
may join at what I have called a block. Another possibility,
however, is that although different uses of the name end in
different blocks, these blocks are themselves historically connected.
This might occur, for example, for the use by different children
of the name "Santa Claus." I have suggested that the block in this
example occurs where the parents tell the children a fiction as if it
were fact. The block, however, would be a different one for each
child. Still the blocks themselves are historically related in an
obvious way since the parents' deception is rooted in a common
tradition.
Still another possible source of difficulty with Rule (R) as
stated is that it makes use of prior instances of the name in
predicative statements. Is it possible meaningfully to assert "N
does not exist" when N has never been used in predicative
statements (about actuality) ? If it is, then Rule (R) would have
to be amended in some way, perhaps by talking of potential or
possible uses. But at the moment I am not sure how this would go
and I will not attempt it.
Even without worrying about the vagueness of the idea of a
"block," Rule (R) may look unexciting, but its consequences are
interesting. In the first place its form is completely antithetical to
the principle of identifying descriptions, for it has nothing to do
with whether an individual of a certain description existed or not.
Second, it does not involve our theory of reference in any diffi-
culties: there is the connection with the notion of historical
explanation and so it ties in neatly with the positive aspects of the
view, but it has no Meinongian implications, no overpopulation

26
 SPEAKING OF NO THING

with entities whose existence is being denied. This result is bought,

to be sure, at the price of making a name function differently in
existence statements as opposed to predicative statements. But, as
I have said, I think that this is not an unintuitive result.
While the above are important consequences of (R), what
interests me about (R) is that it gives the truth conditions for
statements that assert that some individual does not exist in terms
of a linguistic failure-the failure of a name to refer on account
of a "block." And it should occur to one that there may be
something wrong with this. How, it might be asked, can Homer's
existence or nonexistence be a matter of a fact about language, a
fact about the name "Homer"? One is reminded, at this point,
of a similar problem connected with the other puzzle about
reference mentioned at the beginning of this paper. In "On Sense
and Reference," immediately after propounding the puzzle about
identity statements, Frege mentions a solution that he had
formerly thought correct, but which he now repudiates just
because it seems to involve turning identity statements, which
apparently express facts about the world, into statements about a
particular language.
Rule (R), in so far as it is supposed to express truth conditions
for negative existence statements of a certain kind, seems objec-
tionable for the same reasons. The crux of the problem in both
cases seems to be this. We are inclined to say that the propositions
expressed by us as "The Evening Star is identical with the
Morning Star" and "Homer did not exist" can be the very same
propositions that someone else may express using entirely different
names. Therefore, how can we give a rule, such as (R), which
makes the truth conditions of what we say depend upon facts
about particular names?
The child who has become disillusioned expresses his new-found
knowledge by saying "Santa Claus doesn't exist." A French-
speaking child, with a similar history of being deceived by adults,
might express his discovery by saying, "Pere Noel n'existepas."
Although the names are different, I believe we should want to say
that the two children have learned the same fact and, on that
account, that they have expressed the same proposition. Yet if we
apply Rule (R) to each case it seems that the truth conditions

27
 KEITH S. DONNELLAN

must be different; they involve a block in the history of the use of

the name "Santa Claus" for the English-speaking child and a
block for the French-speaking child in the history of the use of the
different name, "Pere Noel."
Perhaps we can see the problem more clearly by looking for a
moment at predicative statements. If we consider a simple
(grammatically) subject-predicate statement, such as "Socrates is
bald," and think of this as divided into its referring element,
"Socrates," and its predicative element, "is bald," then if a certain
change in the predicative element-for example, from "is bald"
to "is short"-results in a change in the truth conditions for the
statement, we want to say that the result expresses a different
proposition. In general only interchange of synonymous predicates
will maintain the same truth conditions and the same proposition.
If referring expressions such as "Socrates" wereconcealed descrip-
tions-that is, introduced predicate elements into a statement-
then the same could be said about them: substituting a different
referring expression, unless it happened to conceal the same or
synonymous descriptions as the one it is substituted for, would
shift both the truth conditions and the proposition expressed (and,
in fact, this is the heart of Frege's way of avoiding his puzzle about
identity statements).
But our theory of reference denies that referring expressions
such as "Socrates" conceal descriptions or introduce predicate
elements. If we keep the predicative element the same and
substitute a different referring expression-say, "Plato" for
"Socrates"-then whether or not we have the same proposition
expressed depends solely upon whether or not the same thing is
referred to. And this in turn depends upon whether the historical
explanation of the use of these two expressions traces back to the
same individual. If you say "Henry is bald" and I say "George is
bald" we express the same proposition if the person you referred
to by using the name "Henry" and I by using the name "George"
are the same person. But what you say is true if and only if the
person you referred to-that is, the person historically con-
nected-when you used the name "Henry" has the property of
being bald; whereas what I say is true if and only if what I referred
to by using the name "George" has the property of being bald.

28
 SPEAKING OF NOTHING

The truth conditions are different because they must be stated in

terms of what is referred to by different expressions, in the one case
my use of the name "George" and in the other your use of the
name "Henry." Yet we may express the same proposition.
So with predicative statements involving proper names, given
the same predicate, sameness or difference of propositions comes
down to sameness or difference of the referent of the names. It
seems that if we try to state the truth conditions for a particular
use of such a statement, we are not going to arrive at what we
should like to call the proposition expressed. But although we thus
are separating truth conditions from propositions expressed, the
latter notion is still a fairly clear concept. It seems, however, that
we cannot in the same way preserve a clear notion of what
proposition is expressed for existence statements involving proper
names.
Our problem arose because we wanted on the one hand to make
it possible that one child saying "Santa Claus does not exist" may
express the same proposition as another who says "Pere Nodl
n'existepas." But, on the other hand, our explanation of the truth
conditions for such statements in Rule (R) made them different
for the two cases. We have seen, however, that if the historical
explanation theory is correct, a difference in truth conditions
without a shift in proposition expressed can occur in any case with
predicative statements. This can occur when there is a difference
in names used without a change in referent. So this seems to be a
general feature of the theory's treatment of names. When we turn
to negative existence statements and Rule (R), however, we cannot
give as-a criterion for propositional identity sameness of referent.
For, of course, if true, the name in such a statement has no
referent.
What we would like, still continuing with the example, is a
reason for saying that both children express the same proposition
that is at once in line with our theory and intuitively satisfying.
I want to suggest that we may find such a reason once more by
using the idea of a historical connection, that, in our example, it is
the blocks in the historical explanation of the use respectively of
the names "Santa Claus" and "Pere Nodl" that are themselves
historically connected. Once again, I do not have the resources to

29
 KEITH S. DONNELLAN

spell out a general principle for what this historical connection

must be, any more than I did with the notion of a block itself. Yet
in the example before us, and others- one can think of, our incli-
nation to say that people using different names express the same
negative existence proposition seems to be a matter of historical
connection between the blocks involved. In our example, it seems
to me that the reason we think both children express the same
proposition is that the story of Santa Claus and the story of Pere
Noel, the stories passed on to the two children as if they were
factual, have a common root. And if there were not this common
history, I think we should rather hold that the two children
believed similar, perhaps, but not identical falsehoods, for example,
when the one attributed gifts to Santa Claus and the other to Pere
Noe~l and that they expressed different truths when one said
"Santa Claus does not exist" and the other said "Pere Nodl
n'existe pas."

VIII. CONCLUDING REMARKS

If this discussion has been on the right track, then at least the
outline of a solution to some problems concerning nonexistence
statements is available to the historical explanation theory. One
point emerged in the course of the last sections. We can perhaps
point to criteria for saying when two existence statements invol-
ving names express the same proposition, but these criteria take a
different form from those for predicative statements involving
names. In particular, it cannot be a matter of sameness of referent.
For predicative statements we were able to suggest a way of
representing propositions, as ordered n-tuplets, but no obvious
way of representing propositions expressed by existence statements
suggests itself. This does not seem to me to count against the
theory, since the notion of a proposition is not, I think, a clear one
that has established use outside of a theory. The fact that the
representation suggested for predicative statements involving
proper names has no counterpart for existence statements,
however, may account in part for the fact that Russell took the
alternatives for proper names to be either a Meinongian view or

30
 SPEAKING OF NOTHING

a concealed descriptions view. For the representation of proposi-

tions suggested is, I think, essentially Russellian and either of these
views of ordinary proper names would allow him to apply it to
existence statements.
KEITH S. DONNELLAN

Universityof California, Los Angeles

3I