22 THE ROAD-SYSTEM OF EASTERN ASIA MINOR

THE ROAD-SYSTEM OF EASTERN ASIA MINOR WITH THE

EVIDENCE OF BYZANTINE CAMPAIGNS.

[PLATE I.]

PART I.

O F late years a good deal of discussion has been devoted to the Road-
System of Cappadocia and the Tauros region in ancient times, and it might
seem at first sight superfluous to discuss the subject over again. But con-
clusions already reached must always be tested in the light of new facts; and
in the case before us several new facts have come to hand, which illuminate
our subject and enable us to introduce into it a considerable amount of
simplification. I propose, therefore, in the following paper to describe the
roads which traversed this part of the country and then to prove their
direction as well as their importance from the evidence of Byzantine cam-
paigns. This is the simplest order to follow, because one campaign generally
covers several routes and it would involve a sacrifice of clearness to break up
the campaigns into a series of disjecta membra.
At every period in the history of Asia Minor the most important roads
from the west converged towards Caesareia-Mazaka (Kaisariye), which in
later times became the metropolis of Cappadocia, and radiated thence towards
east and south. Sebasteia-Sivas forms another centre only second in import-
ance to Caesareia; and the entire road-system of Eastern Asia Minor is most
easily described and most clearly understood by taking these two cities as
the starting-points. I shall therefore begin with the roads leading East
and South from Caesareia and afterwards go on to those radiating from
Sebasteia-Sivas.

I. ROADS FROM CAESAREIA TO TEE EAST.

These are two in number: (1) what may be called the great Eastern
route by Herpa, Ariarathia, Tzamandos (Azizie), and Gurun to Melitene and
the east; and (2) the Roman road over Anti-Tauros by the "Kuru Tchai pass
and thence by Kokusos (Geuksun) and Arabissos (Yarpuz) to Melitene.
(1) The former of these two routes has been almost entirely over-
looked. Yet it was at al} times the great route to the east. It is
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 23
the Persian Royal road J : it existed in Roman times: and it is the
road to the east throughout the Byzantine period. The course of the
road is as follows. From Caesareia it goes over the plain to Arasaxa
(Zerezek) and after crossing the River Karmalas (Zamanti Su) proceeds by
Larissa2 to Herpa (Yere Getchen) on the main stream of the river which it
follows as far as Tzamandos (Azizie). The fortress Tzamandos (T£a/i«i>8o?),
which is mentioned several times during the tenth and eleventh centuries
after Christ,3 is placed by Prof. Ramsay {Hist. Geog. pp. 289 ff.) with the
greatest probability beside the modern Azizie, and the name is regarded as a
native Anatolian word, which survives in | h e modern name of the river
(Zamanti). At Azizie there is a " magnificent series of fountains which rise
from the hills that fringe the Karmalas-Zamanti" and flow down into that
river: and Prof. Ramsay supposes that the modern name Zamanti Su is
derived from the city beside these fountains, " the river being called ' the
water that comes from Tzamandos" just as the Hermos is now called Gediz
Su, ' the water that comes from Kadoi' (KaSou?, accus.), though both
Tzamandos and Kadoi were situated some distance below the actual source of
the river." i While Tzamandos is frequently referred to in the late centuries,
no mention is made of Ariarathia, which was situated at an important point
in the upper Karmalas valley on the Sebasteia-Kokusos road. In order
to account for this strange fact, Prof. Ramsay formerly conjectured (H. G. pp.
310, 289 f.) that Tzamandos and Ariarathia were to be identified, Tzamandos
being the native name which had been preserved in popular usage and passed
into official use about the ninth century of our era. 5 He would now,
however, modify this suggestion in view of a new piece of evidence. In an
Armenian Notitia JSpiscopatuum (a translation of a Greek original of ca.
1200), published by Mr. Conybeare in Byz. Zft. V. p. 127, we find Tchamanton
(obviously Tzamandos)6 and Ararathias "quae est in Dauthn (i.e. 'the
warm')" given as two distinct bishoprics under Caesareia. Now Dauthn
(see infra) is probably the pass leading by Kuru Tchai and Kokusos-Geuksun
into Kommagene; and consequently Ariarathia should be brought lower
down the Karmalas valley and located at, or very near, Herpa.7 The

1
See Ramsay, Cities and Bish. of PJirygia, and 1068 (Mich. Att. pp. 121-2, Skylit. 678),
vol. i. p. xiv. n., and the Excursus at the end see infra. The Armenian name is Dzamentav
of this paper. or Dzamentou, Arab. Samandou (St. Martin,
2
Larissa and Herpa must have been near each Mim. sur VArm(nie,l. p. 191).
other. Herpa (Strabo, pp. 537, 539) or Herpha * The quotation is from MS. notes of Prof.
(p. 663) was on the road from Caesareia to Ramsay's, to whose unfailing kindness I owe far
Melitene at the point where it crossed the more than can be actually specified in the
Karmalas (see Hist. Gcog. pp. 289, 272-3). preparation of this paper.
6
Larissa cannot be located with certainty, but For similar cases see pp. 279 n., 280 n.
e
lay on the direct road to Melitene, not far east Tfa/iavrhs in Mich. Att. 121.
7
of Arasaxa (H. O. pp. 272-3, and campaign of It is quite likely that Herpa is the older
1069 infra). It was given, along with Komana, name of the town, which was renamed Aria-
Tzamandos, etc., to the Armenian prince Gagik rathia after one of the Cappadocian kings,
in 1064. Ariarathes (ca. 350-36 B.C.): Herpa is not
3
By Const. Porph. De Them. p. 32 and De mentioned after beginning of first century B. c,
4dm- Imp. p. 228 ; in 976 (Kedr, ii. p. 423)
24 THE ROAD-SYSTEM OF EASTERN ASIA MINOR
Dazmentos of this Notitia is probably the same place as the fortress
Dasmenda mentioned by Strabo (<f>povptov airorofiov AacrfiivSa, p. 540) as
situated in Chamanene, " at the western extremity of the ridge which bounds
Cappadocia on the north " (H. G. p. 290).
After passing Tzamandos-Azizie, our road goes over the hills eastwards
to Gurun. The section Gurun-Caesareia just described was traversed by
the late Col. Stewart, and it will be useful to give his statistics (for which I
am indebted to Prof. Ramsay).
Miles
Gurun. *
192 Keupek Em-en, alt. 5994 ft.
3 Commenced ascent of Godilli Dagh.
1 3 | Crest of Pass, about 6,700 or 6,800 ft, due W. of Azizie.
5 Borandere vill.
4J Karagoz vill.
1J Ford of Zamanti.
i Kara Boghaz.
4 Azizie. Road over Godilli D. is bad ; the araba road goes round North end of Dagh.
18J Ekrek.
4J Karadai.
26 Kaisariye.

At or near Gurun the road passes through Lapara-Lykandos, which

Prof. Ramsay now places here and identifies with the Paulician city Lokana,
mentioned in Basil's march, 872 A.D.1 This localization is convincing and
suits admirably the description of the march of Bardas Skleros in 976 {infra).
The KXeia-ovpa of which Tzamandos and Lykandos are the limiting points
(Const. De Adm. Imp., p. 228) will then be the pass over Godilli Dagh.
Leaving Gurun, the road descends the Tokhma Su (Melas) to Taranta,
Pliny's Daranda-Dalanda, mod. Derende 2 ; and thence to Melitene (Malatia).
Somewhere in this vicinity was the pass (a-Tevo^wpla) Boukoulithos (fiovicov
Xt^o?) mentioned by Kedr. II. p. 421. From Melitene the road goes to the
Euphrates which it crosses at Tomisa, situated on the left bank at the
1
See Class. Review, April 1896, p. 140 and Malatia was taken by the Greeks ca. 693, the
137, and Map accompanying this paper. Moslems settled at Taranda after it had been
2
Ol. Rev. I.e. p. 137. The comparison ofthe captured in 702. It lies three marches distant
Arabic Tavanda (Biladhuri, p. 186, ed. M. de from Malatiyya, deep in the Greek country ;
Goeje), Armenian Daranda, Syriac Turanda (St. and was held by a garrison of about 2,000
Martin I.e. p. 190), makes assurance doubly troops from the Mesopotamian army, during
sure. Taranta was a place of importance, the summer only (Biladh. p. 185). When
especially in the first half ofthe eighth century, Omar II. became Caliph (end of 717 A.D.), 'he
when the district of Melitene was in Saracen removed the Moslem population from Taranda,
hands. It is mentioned by Theoph. p. 312 De for he feared for their safety, the enemy being
Boor (see infra); in 701 A.D. it was beseiged in so near ; the people however objected, and when
vain by Abd Allah ('AjSSeAas), p. 372. Some they were perforce removed, they would leave
further information about the fortress is given nothing for the enemy, breaking even their oil
by Biladhuri, for all references to whom I am and vinegar jars. The Caliph transported them
indebted to the kindness of Mr. Le Strange who to Malatiyya, leaving Taranda to ruin' (p.
has taken the trouble to translate for me his 186). In the following century it was a
chap, on Mesopotamian Fortresses in the ' Book Paulician stronghold and surrendered to Basil I.
of the Conquests' (ca. 869 A.D,). When in 872 A.D. (Theoph. Cont. 267).
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 25
extremity of Sophene, and then enters Anzitene, called in Byzantine times
Xav^ir, the military centre of which was the fortress Hanzit, frequently
mentioned by Armenian writers as Handzith and by Arabic geographers as
Hanzit (Hinzit), one of the Greek frontier fortresses near the Euphrates,
between Melitene and Samosata. Anzitene-Hanzit is generally placed further
east, but the evidence seems clearly to show that it denoted the country
between the Euphrates (starting from about Tomisa where Sophene ends),
the Murad Tchai (as far at least as Palu-Romanopolis), and the sources of
the Tigris.1 A little to the south-west of Kharput (XdpTrore, Kedr. II. 419)
the road forks. One section goes to Kharput and thence by Arsamosata
(Arab. Shamshat or Shimshat) on the Murad Tchai to Palu (Romanopolis):
but no doubt there was an alternative route by mod. Kizin to Palu.
Arsamosata-Shamshat has been located by Mr. Le Strange from the de-
scription of Ibn Serapion (see p. 57) who says, ' the Nahr Salkit ( = Peri
Tchai) falls into the Arsanas (Murad Tchai) one mile below the city of
Shamshat, near a mountain that is over the city and closes it in' (p. 314,
cf. pp, 45, 63). Combine this with Ptolemy's fierajjv TOV Hv<f>p. ical T&V
TOV TiypiSo*; 7rr)ya>v . . . . 'ApaafiocraTa K.T.X. (V. 1 3 , 18-19) a n d t h e
inference is clear that the city is to be placed one mile from the junction of
Peri Tchai and Murad Tchai on the left (south) bank of the latter river.2
It formed at one time a Theme in the Byzantine Empire (' Atr/Moa-drov3 defia,
Const. De Adm. Imp., p. 226). Romanopolis, which derives its name from
Romanus I. Lecapenus, has been identified with the highest probability by
Prof. H. Gelzer (Qeorg. Cypr., pp. 176-7) with Palu, Armen. Balu. The
KKeiaovpa, therefore, mentioned by Const. I.e. p. 226 (TO Be Xapgir
Kal f) 'PrnfiavoTT. ickeio-ovpa), lies on the road between Palu and Kharput

1
Only the most important evidence can be preserving the geographical order from south-
given here. Ptolemy (v. 13, 19) places east to north-west. The Arabic writers all agree
'AyfiTj)^ jUETafi TOV Evrppirovs Kal TUV TOS in placing the fortress Hanzit close to the
TtypiSos vrry&v, including amongst its towns Euphrates between Malatya and Sumaysat
"AefTjTa and Arsamosata (below). In Byzantine (Samosata), on a tributary of the Euphrates,
and Arabic times it clearly denotes the district says Ibn Serapion, which 'passes the city of
indicated above. It is always connected with Hanzit and the province thereof and then falls
the KXeio-oSpo Romanopolis-Palu: before into the Euphrates (ed. Le Strange, p. 54, cf.
Romanus I. it was attached to Melitene (TO n. on p. 49). The Euphrates cMpasse la ville de
XavflT Ka\ r) 'Vujiavoir. K\*iaovpa TUV MeAiTTjvia- Hanzyt, puts tourne vers Vouest, arrive A
T&V iirijpxov, Const. De Adm. Imp. p. 226), and Sumaysat...(Ibn Khordadbeh, Trans, p. 177).
was assigned by him along with Kamacha, Space forbids further quotations. The fort
Keltzine (Acilisene), etc. to the newly-eonstituted then, should apparently be looked for near the
Theme of Mesopotamia (p. 227). Nicephorus, Euphrates, west of Kizin : the position of the
De Velit. Bell. p. 250 (quoted at end of Pt. I.), is ' province ' Anzitene-Hanzit seems clear.
2
quite explicit: the trans-Euphrates passes into Ibn Serapion's description is confirmed by
Saracen territory are those crossing ' the (Tauros) Ibn Khordadbeh and Yakut (cf. Le Strange, I.e.
mountains which separate Chanzit from the p. 57).
3
enemy's country as far as Romanopolis.' All This is a reproduction of the Armenian
the passages feom Syriac and Arabic writers form Aamusat, as Xav0r is of Handzith (Prof.
collected by Gelzer, Geo. Gyp. p. 178 f., confirm Gelzer on Geo. Gyp. p. 172). The Theme of
this localization : e.g. Faustus Byzant. v. 16 which Arsamosata was the central fortress
gives as conquered in succession Arzanene, apparently extended north of Murad Tchai
Sophauene, Ingilene (about Egil), and Anzitene, (Arsanias).
26 THE ROAD-SYSTEM OF EASTERN ASIA MINOR.
or Kizin. Crossing the river at Palu, the road then follows the
right bank of the Murad Tchai to Akhlat (XXidr, see campaign of 1069
infra) on the lake of Van. The other section of this Eastern road passes to
the south of Kharput by way of Kizin to Amida (Diarbekr) whence it
follows the valley (left bank) of the Tigris to Nineveh, then crosses the K.
Zab and proceeds to Arbela (Erbil). This I believe to be the line followed
by the Persian Royal Road from Tomisa (see Excursus).
This great and direct line of communication between West and East
is the route generally taken in Byzantine Expeditions against Persia, and
the section Tomisa-Caesareia will be seen to be the favourite route for
Turkish raids into Asia Minor. Its direction is fixed by Theoph. p. 312, ed.
De Boor, where the return of Heraclius from his second expedition against
the Persians is described. On March 1, 626 A.D., before leaving the Lake
of Van on his homeward journey, Heraclius held a consultation with his
troops as to the route which he should take. The choice lay between two
routes, (1) one leading eVt Tdpavrov, i.e. Taranta-Derende, and (2) another
eVt rt)v TWV 2,vpi(ov yrjv. The latter, which was unanimously chosen
because it was better supplied with provisions, although the more difficult of
the two, is the route leading over the Eastern Tauros (near Van), across
the Tigris, and then by Martyropolis (Meiafarkin) and Amida (Diarbekr) to
Samosata.1 The alternative route 'by way of Derende' went along the
right bank of the Murad Tchai (Arsanias) to Palu where it crossed the
river and descended by Arsamosata-Shamshat to Kharput, and then joined
the 'Eastern road' through Tomisa, Melitene, Derende, &c, to Caesareia.
Some other marches which concern this route alone may be added here.
In starting for his second expedition (624 A.D.), Heraclius probably took this
same road. It has been generally supposed that he went to Armenia by
sea; but it is pointed out by E. Gerland (Die Pers. Fddzilge des Kaisers
Herahleios in Byz. Zft. III. p. 345 ff.) that while Byzantine authors are silent
on the point, the Armenian historian Sebe'os states that Heraclius marched
from Constantinople to Caesareia in Cappadocia and thence to Armenia.
This is obviously the correct account. Caesareia would be the most con-
venient airXrjKTov at which his forces could concentrate for an expedition to
the East; and from Caesareia he then marched to Erzerum and the Araxes
valley. He thus chose the same route as Philippicus, the general of Maurice,
had done in 585-6 : for it is stated that he also marched to Armenia by way
of Caesareia. Finally, it is most probable that Heraclius returned by this
way in 628 A.D. after his third expedition.
A very important march is that of Bardas Skleros in 976 A.D. (Kedr. II.
pp. 419—423). Skleros, who was appointed governor of Mesopotamia by
Basil II., revolted against the king and proceeded to invade Asia Minor.
After laying up stores at Kharput and obtaining assistance from the Emirs of
Amida and Martyropolis, he began to march towards Caesareia. A
detachment sent to reconnoitre fell in with a division of the Imperial troops

1
See the description in Theoph. p. 313.
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 27
at the pass Boukou-lithos {ev nvi aTevo^mpia), and suffered defeat. After
some delay Skleros started himself and in three days1 reached Lapara,' now
called Lykandos,' where he met and defeated the Emperor's forces, and
proceeded thence to Tzamandos, a populous and wealthy city situated on a
steep rock (airoKpijpvq) irerpa)? Lykandos and Tzamandos, therefore, are
both on the direct route to Caesareia.
For other campaigns see Part II.
(2) The Roman Road.—The other route from Caesareia to the East is
that followed by the Roman military road, viz. Kuru Tchai—Sirica (Kemer)
—Kokusos (Geuksun)—Arabissos (Yarpuz)—Melitene. The direction of this
road has been established by the discovery of a series of milestones (several
of them in situ). A large number of these was found by Mr. D. G. Hogarth
and Mr. J. A. R. Munro in 1891, and the whole subject is treated in a
complete and admirable paper by Mr. Hogarth in Mod. and Anc. Boads in
East. Asia Minor (R. G. S. vol. iii.), part ii. pp. 38—78. Only a brief
description, therefore, is required here. From Oaesareia the road follows the
route just described [no. (1)] to Arasaxa-Zerezek where it branches off to
Muhajir on the Karmalas-Zamanti and then crosses Anti-Tauros by the Kuru
Tchai pass through the modern village Tass and Coduzabala to Sirica-
Kemer.3 Coduzabala, which the Antonine Itinerary gives as a station both
on the Caesareia-Kokusos and the Sebasteia-Kokusos roads, should probably
be placed on the Kuru Tchai pass at the junction of these two roads (see
Map). Sirica, placed by Prof. Ramsay on the Saros.six4 miles east from
Komana-Shahr (JET. G. p. 312), i.e. at Kemer, probably corresponds to the
Serikha of the Armenian Notitia following Tchamanton (Tzamandos): for, as
Prof. Ramsay remarks, it naturally follows' Tzamandos which was situated in
the same region. From Sirica the road goes nearly due South along the base
of Bimboa Dagh to Kokusos-Geuksun, and then strikes North-East along the
Geuk Su to Arabissos-Yarpuz, after which it crosses the Khurman Su at
Izgin and the Sogutli Irmak near Ahazli and thence passes over the hills in
a nearly direct line by Osdara, Dandaxina, and Arga-Arca5 to Melitene.
The latter section of this road from Arabissos is fully described in H. G.
pp. 273-4.
1
Measured apparently from about the their numbers increased Tephrike was added.
Euphrates, though this is not precisely stated. The first city would be in, or close to, Saracen
2
Tzamandos was situated on the hill above territory : but as the sect grew in strength and
the modern Azizie, which occupies the lower became to a certain extent an independent state,
slopes. they would have to find sites for their new cities
3
Cf. B. G. p. 271. outside Saracen territory, i.e. further north.
4
' One too many,' Hogarth I.e. p. 51. Now Amara (see infra iv. (2) J) is north of
6
With Arga it would be possible to identify Argaous and Tephrike is north of Amara.
the Paulician fortress Argaous, which occurs in Argaous therefore might be Arga. But it is far
the marches of Basil I. in 872 ('Apyaoie, more probable that it should be identified with
Theoph. Cont. p. 270) and of Eomanus IV. in modern Argovan, about twenty-five miles almost
1068 ('Apyaov, loan. Skylit. 670). Kedr. II. due north of Malatia. This suits the line of
p. 154 tells us that Argaous and Amara were both marches (infra) far better and is in itself a
the first cities founded by the Paulicians with more natural position for the first Paulician
the aid of the Emir of Melitene; and that when city.
28 THE ROAD-SYSTEM OP EASTERN ASIA MINOR.
The evidence of the milestones shows that the military road was built or
reconstructed (restituit) by Sept. Severus, i.e. not earlier than the end of the
second century after Christ, but a road of some kind may have previously
existed along this line. That there was a trade route from Ephesus to the
East as early as 100 B.C. is certain. This KOIVT) 080? is described by Strabo
(p. 663) on the authority of Artemidorus. Up to Caesareia the description is
full and clear. But what line did the section Caesareia-Euphrates take ?
Strabo merely says evrevffev B' em TOV JLixftpdrrjv fie^pi To/itcrmv ^copiov rrj<s
"Zm^rjjnj^ Bia 'Hpcfiwv TroXt'^ij? %t\ioi Terpatcocrioi rerrapaKOVTa. THp<f>ai,
elsewhere THpTra (pp. 537, 539), is Herpa on the Karmalas-Zamanti (see
H. G. p. 289). The route indicated therefore is evidently that by Herpa-
Azizie-Gurun-Derende-Malatia-Tomisa (opposite Isoghli): for the Roman
Road did not go by Herpa but branched off at Arasaxa, and the other route
is the natural line for a trade-route to the East and the easiest way to the
Euphrates. I take this, then, as a proof of the importance of the Caesareia-
Derende-Tomisa route in the early Roman period. The line of the Roman
military road was probably determined by different considerations—viz. to
connect with Germaniceia-Marash and the Syrian frontier, as well as with the
frontier at Melitene. In the Byzantine period an army marching from the
West towards Melitene never takes this route.

II. B0AB8 FROM CAESAREIA TO THE SOUTH.

(1) To Germaniceia-Marash. An army marching to Germaniceia would
follow the Roman road by Arasaxa-Zerezek and over Anti-Tauros by the
Kuru Tchai pass to Sirica-Remer. At Sirica, the road to Germaniceia forks,
and two routes were possible: (a) by Maroga (Maragos) and Tanadaris (Tanir)
to Arabissos and thence by the well-known pass to Germaniceia and Adata.
See H. G. pp. 271-2. This was the route almost always taken by Byzantine
armies, and it is hardly necessary to quote campaigns in proof.1
(6) The alternative route still followed the Roman road to Kokusos-
Geuksun. From that point there are three modern routes to Germaniceia, but
only two of these are known to have been used in ancient times : (a) the Ayer
Bel pass, which crosses the Geuk Su, ascends Ayer Bel, and passing Kalli-
polis and Padasia (at Temelilik) crosses the Pyramos-Jihun to Germaniceia.
This road is marked in the map in H. G. p. 266, and mentioned on p. 276.
It was taken by Basil in 877, and Romanus IV. in 1068 (infra). (/S) The
other route, the most difficult of all, is the pass by Geben and thence along
the Kursulu Su, round Dolaman Dagh to the Jihun (see Map). This was the
road followed by the Crusaders in 1097 (below).
Germaniceia is a great centre from which roads radiate in all directions,
and it is just this fact which accounts for its strategical importance. Leading
towards the East there is a road over the Ak Su past Adata (which lay to
1
For the importance of Arabissos, see H. G. pp. 277, 311, 2§0, etc. From Tanadaris—Tapir
there is also a direct road to Sebasteia (iv, (2) b infra),
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 29
1
the south of Inekli ), Katamana, Nisus, and Tharse to the Euphrates which
it crosses at Samosata. See II. G. p. 279. Towards the South there are two
roads. One leads straight to Antioch along Mt. Amanos, and was followed
e.g. by the Crusaders in 1097. The other goes by Doliche-Duluk to Aintab
and thence to the Euphrates at Zeugma-Birejik (see H. 0. p. 279) or south-
wards to Aleppo (Xa\e7r). This latter route was frequently taken in Byzan-
tine marches into Syria (infra).
This route by Kuru Tchai and Kokusos to Kommagene was, called TO
Aaovffd, or at least it passed through the district which went by that name.
In the difficult but important passage of Niceph. de Velitat. Belli p. 250
(quoted below, p. 32), which summarises the Passes from Byzantine
territory over Tauros into the Saracen country, the Anti-Tauros region is
denoted by TO Karjo-ovv KOI TOV (?) AaovOd. By TO ~Karjcrovv is probably
meant the district traversed by the passes over Anti-Tauros to Sis (see (2)
infra). The word AaovOd seems to have been at last explained by Mr.
Conybeare's Armenian Notitia, which gives as separate bishoprics under
Oaesareia Tchamanton-Tzamandos and Ararathias in Dauthn. As already
mentioned, therefore, Ariarathia must be placed at or near Herpa, and the
pass crossing AaovOd—Dauthn is the road from Caesareia by Kuru Tchai.

(2) To Sision-Sis. There are two roads to Sis, both indicated in the map
in H. G. p. 266 2 : (a) from Caesareia by mod. Tomarze to Sebagena-
Seuagen (or Suwagen) on the Karmalas-Zamanti and thence by the Gez Bel
pass over Anti-Tauros to Hadjin. Between the point where this road leaves
the Karmalas valley and Hadjin, probably near the mod. village Urumlu, is
to be placed the Kaisos mentioned in Basil's march 877 A.D. (Theoph. Cont.
279, Kedr. II. p. 214, infra). Kaisos should probably be connected with
Kabissos (/? = F, cf. H. G. p. 312 n.) given in Not. I. as a bishopric of Cilicia
Secunda, and by Ptolemy as Kabassos in Kataonia,3 and also with the
TO Karjaovp of Nicephorus, I.e. Karjaovv is the district of which the fort
Kaisos-Kabissos is the centre. The name of the fort would be exteaded
to cover a district beyond its actual vicinity, just as Sebasteia, or Koloneia,
gives its name to the whole Theme. The passes therefore which cross TO
Karjaovv are the pass which we are describing and the following more
westerly pass to Sis; and the whole expression TO Karj. /cal TOV (?) AaovOd
will denote 'the Anti-Tauros region crossed by the passes leading over
Tauros.' From Hadjin the road leads across the Saros-Sihun (here called
the Geuk Su) and thence over Mt. Tauros to Sis.
(b) The alternative route branches off from (a) on the north side of
Mt. Argaios to Ferakhtin or Frakhtin on the Karmalas, thence to Kiskisos-
Kisken and across Anti-Tauros by Enderessi Yaila to the Saros, after which
it crosses Mt. Tauros to Sis.
(3) The two passes on the west of mount Argaios leading from
Caesareia to the south are of great importance. See H. G. pp. 350 ff.
1 2 3
Cl. Review, I.e. pp. 138 f. Cf. pp. 271, 281, 291. See H. G. pp. 386, 451.
30 THE ROAD-SYSTEM OF EASTERN ASIA MINOR.
(a) The less important of the two is the difficult road which runs nearly due
south by Develi-Kara-Hissar to Podandos-Bozanti and through the Cilician
Gates to Tarsus. This pass was called " Karydion" (H. G. p. 351). (b) The other
pass " Maurianon " by way of Tyana and Loulon was the regular route across
Tauros into Cilicia. It coincides with (a) nearly as far as Develi Kara Hissar
and then branches off to the place now called Zengibar Kalesi, half an hour
west of Develi, "a striking mediaeval castle on a lofty two-peaked hill." This
is the absolutely impregnable fort which the Crusaders in 1097 passed by
without attempting to take (see infra). It is not named by the historians of
the first Crusade, but Prof. Ramsay points out to me that it was Kyzistra, as
is proved beyond all doubt by a passage in Chamich's history. In 1079
Gagik, the exiled king of Armenia, marched from Tarsus in the direction of
Caesareia to annoy the Greeks and on arriving " on the plains of Arzias, near
the fort of Kyzistra," allowed himself to be led into an ambush and was
imprisoned in the fort, which was impregnable. The Armenian chiefs laid
siege to the place but could not take it, and when the body of the murdered
King was suspended from the walls before their eyes, they retired, convinced
that nothing could be done against his murderers. From Kyzistra the road
proceeds to Tyana (Kizli-Hissar, three miles south of the mod. village Bor)
and thence by Loulon to Podandos where it rejoins (a) and passes through
the Pylae Ciliciae to Tarsus.
From Tyana there is another route to Herakleia-Kybistra (Eregli) and
thence either through the Cilician Gates or westwards to Barata, where roads
diverge to Iconium and over the Isaurian mountains. These routes occur in
the marches of Romamis and the Turks in 1069 and of the Crusaders in
1097 (infra).

III. PASSES FROM MELITENE INTO KOMMAGENE.

The consideration of these will complete the list of Tauros-passes.
There are at least two, and probably three, roads over Tauros from Melitene
into Kommagene, indicated by Niceph. l.c.as those crossing ret, (sc.oprj) Trapaicel-
fieva MeXirrfvqv re ical ra KaXovSia. The word KaXov&ia is explained by a
reference in Biladhuri (I.e. p. 187) who says that the fortress Kalaudhiyya
was destroyed by the Greeks under Constantino Copron. in 751 A.D., after the
capture and sack of Malatiyya (Malatia). Kakovhia is therefore the
Graecized form of the Arabic name for Claudias. This fortress was situated
on the Euphrates near Melitene and not south of Samosata, as is sometimes
supposed. This is confirmed by Amm. Marcell. xviii. 7. The Persian King
Sapor, marching into Asia Minor by way of Nisibis and Constantina, halted
at the latter town where he learned that the Euphrates had risen high and
could not be crossed by a ford; and consequently he decided to turn north-
wards (fkcti in dexterum, lattis) and, taking a more circuitous road through a
fertile district, to make for the two fortresses Barzala and Laudias (Claudias),
where the Euphrates " tenuis prope originem et angustus, nullisque adhuc
aquis advenis adolescens, facile penetrari poterit, ut vadosus." "Prope
 THE ROAD-SYSTEM OP EASTERN ASIA MINOR. 31
originem " is of course an absurd exaggeration, but the passage indicates that
Claudias was far up the river near Melitene, as is shown by the fact that the
Roman troops on receiving intelligence of this movement prepared to hasten
to Samosata and, after crossing the river there and breaking down the bridges
at Zeugma and Capersana, to repel the Persian advance. Now it has already
been mentioned (§ 6) that a detachment under the tribunes was engaged in
fortifying the western bank of the Euphrates " castellis et praeacutis sudibus
omnique praesidiorum genere." All these facts, combined with the words of
Niceph., seem to prove the existence of a pass leading south along the river
by Claudias, Barzalo, and other places of uncertain situation to Samosata (as
marked in the map in //. 0. p. 266).
The other two passes are better known. (1) One leads direct to
Germaniceia up the course of the Sultan Su past Sozopetra-Zapetra (Arabic
Zibatra, situated at Viran Sheher, four miles off the road towards the rightx)
and over Tauros by Surghi, Erkenek, Pavrelu, Inekli on the Ak Su, and past
the ' famous' fortress Adata (Al-Hadath) to Germaniceia.
(2) The other pass follows this same route as far as Surghi and thence
turns south-eastwards to Perre (Hisn Mansur, mod. Adiaman) and Samosata-
Samsat (as in map in H. G.). It was traversed in 872 A.D. by a detachment
of Basil's array despatched from his base camp on the Zarnuk south-west of
Melitene. This column after passing through T« areva TJ}? 6SOV captured
Zapetra, laid waste the adjacent country, and took Samosata: whence they
returned to the Zarnuk.

IV. BOAJDS RADIATING FROM BEBABTE1A.

Almost all these roads join one or other of the routes already described.
They are all Roman roads except Sebasteia-Tephrike (3), which is not known
to be Roman.
(1) Sebasteia-Caesareia.—This road is of considerable importance as
affording a direct route from Sivas (on the great military road of the Byzantine
period) to Isauria or to the Cilician Gates; and as such it was used e.g. by
Romanus IV. in 1069 when he wished to reach the Cilician passes without
loss of time in order to intercept the rapid retreat of .the light Turkish
horsemen. The road is clearly marked in Prof. Ramsay's map (p. 266) and
described on p. 270. It runs parallel to the course of the Halys through
Malandara, Armaxa, and Aipolioi to Caesareia. Aipolioi is the Aepolion of
the Armen. Not., and the name is preserved in mod. Palas.2

' (2) Sebasteia-Germaniceia.—There are two routes:—

(a) Sebasteia-Tzamandos-Ariarathia and thence over Kuru Tchai by
Kokusos to Germaniceia. The change in the position of Ariarathia
necessitated by the Armen. Not. and the recognition of Kuru Tchai as the
1
See 01. Rev. I.e. p. 138 f. as in modem Greek.'—Prof. Ramsay in MS.
' ' lJalas = PdiroKlovs, I being pronounced as y, notes.
32 THE ROAD-SYSTEM OF EASTERN ASIA MINOR.
great Anti-Tauros pass will involve a modification of the route laid down in
//. G. p. 274. The road will now run by Tonosa-Tunus, Karmalis on the
Karmalas-Zamanti (Viran Sheher, H. G. p. 289), Tzamandos-Azizie,
Ariarathia-Herpa (at Yere Getchen), and thence by Kuru Tchai and
Coduzabala to Sirica and Kokusos-Geuksun. At Tzamandos it joins the
eastern route Caesareia-Azizie-Tomisa [I. (1)] which it follows to Ariarathia-
Herpa and there leads it into route II. (1) to Germaniceia.
(5) Sebasteia-Gurun-Arabissos-Germaniceia. See H. G. pp. 274-5-
This was an important Roman road, used also in Byzantine times as affording
a direct route north to Sebasteia from Germaniceia by the famous Arabissos
pass [II. (1) a]. From Sebasteia it runs to Blandi (near Ulash) and thence
to Euspoena-lspa (at Deliklitash) on a branch of the Tokhma Su, the course
of which it follows to Gauraina (Gurun) and Lykandos-Lokana where it
crosses the eastern route. About midway between Euspoena and Gurun,
near Manjilik, is probably to be placed the Paulician fortress Abara-Amara.
The two names are obviously to be identified. Amara (Kedr. II. 154) was
one of the first two Paulician cities, the other being Argaous-Argovan (see
note on Arga, supra, p. 27). Abara was one of the Paulician forts captured by
Basil I. in 872 in his march from Tephrike by Taranta-Derende to Melitene
(Gont. 267, Kedr. 207), and it was given along v ith Sebasteia, Larissa, and
other cities to Senakerim, prince of Asprakania, by Basil II. in 1021 (Kedr.
II. 464). It was a rovpfia of the Theme Sebasteia and became a KXeicrovpa
under Romanus IV. (Const. JDe Adm. Imp. p. 228). These statements seem
to leave no doubt that it was situated on the pass between Sivas and Gurun,
near Manjilik (see Map). From Gurun the road goes to Tanadaris-Tanir
where it leads into the Arabissos-Germaniceia pass.
(3) Sebasteia-Tephrike (Devrik).—This road leading to Tephrike and
thence to Zimara (Zimarra) on the Euphrates was of great importance during
the Paulician revolt in the ninth century. It is probable that there was
also a road of some kind from Euspoena joining a road from Tephrike at
mod. Kangal and thence following the course of the Kuru Tchai to Melitene.
On this road Aranga-Arani was perhaps situated (H. G. p. 275).1 This is
the direction in which Basil I. marched after withdrawing from Melitene in
872, capturing Argaous-Argovan and several other Paulician forts in this
district.

Before I go on to give some proof of the lines laid down for these
roads by an investigation of Byzantine campaigns which passed over them
it will be useful to quote and endeavour to explain the passage of Niceph.
Be Vel. Bell. p. 250, which summarises the majority of the routes described
above. The words are Si ota? yap 6Sov BiekOeiv fiovXrjOSxriv (sc. the
Saracens), airo re T<av iv Ziekevicelq KKeicovprnv ical TOV T&V 'A.vaTo\iK(bv
1
This might possibly be the 'Aragines in fjoiicov \l8os near the Euphrates (Kedr. ii. 421,
Pharakn ' of the Armen. Notitia. Is it possible supra, p. 24) ? Prof. Ramsay, however, think
that Pharakn ( = ' t h e sheepfold') is the pass that Pharakn=Everek at base of Mt. Argaios.
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 33
Oefiaros, Ka6a ra lavpiica opr] rr/v re HiXi/ciav Bioplt^ovai K.a7nra8o/ciav re
KOI AvKavBov Trpos TOVTOLS Be (sc. Bi1 o'ia<; 6Bov B i e\ 6 el v f3ov\.) Kal ra
(sc. T. oprf) Trapaicetfieva Tepfiavliceidv re Kal "ABarav * Kal TO Karicrovv Kal
Kal rov (?) Aaovdd MeXirijvrjv re ical ra KaXovBia' Kal TO, iripaOev rod
Ev^/jarou vora/iov Biopi^ovra rijp re rov Xav^iJTi Xeyofievrjv %cbpap Kal rrjv
iroXefiiav a%pi 'PwfiavovTroXeW ev oKoif roi<; roiovroi? Oefiacri, Si oXa<; av
6Sov viro(TTpecjiovre<; SieXOelv 7r/ao? rtjv ISiav f3ov\t)da><rt,v, K.r.\.
The importance of this passage lies in the fact that Nicephorus is
speaking from personal knowledge of the country gained during the wars
against the Saracens. But the passage is exceedingly difficult, and
especially the first clause (atro. . . . AvKavSov), which can hardly be right as
it stands (if the words are taken strictly) : for the passes between Seleukeia
and the Anatolic Theme cannot be the passes leading from Cilicia over
Tauros into the Themes of Cappadocia and Lykandos! The meaning must
be (1) the passes from the Anatolic Theme over Isauria to Seleukeia, and
(2) those on the west of Mount Argaios, especially the Oilician Gates, and
probably also the passes to Sis; (3) the passes across the Tauros Moun-
tains overlying the district of Germaniceia and Adata on the one hand, and
the Anti-Tauros region (Kaesoun2 and Daoutha) on the other, i.e. the passes
to Germaniceia; (4) the passes across the Tauros Mountains overlying
Melitene and Kaloudia (Claudias), i.e. the passes from Melitene into Kornma-
gene: and (5) the passes beyond the Euphrates leading from the district
between Tomisa and Romanopolis-Palu (KavQr) into Saracen territory. Cf.
Const.'s words, TO 8e XavQr Kal f] 'VwjAavoir. KXeicrovpa (Be Adm. Imp. p. 226).

PART II.

CAMPAIGNS IN THE CAPPADOCIAN DISTRICT SHOWING THE

ROUTES DESCRIBED.

Hcraclius's March in 626 (Theoph. pp. 312-313). Routes traversed :

Samosata—Germaniceia [under II. (1) ] ; (?) Germaniceia- Arabissos—Sebas-
teia [II. (1) a and IV. (2) a].
After reaching Samosata by way of Martyropolis and Amida, Heraclius
took the direct road to Germaniceia, passing Adata on the way. Theo-
phanes' description of his subsequent route is confused; and it seems best
to accept Prof. Ramsay's correction "ASara (for "ABava) and the slight
transposition which makes the sentence read Trepdo-a*} rrjv "ABara et's Yepfi.

1
Perhaps taken as a fem. sing., but ordin- should strictly be included under rb Kay. /col rb
arily TV'ASOTO. 'ASorai' in Bonn ed. is clearly Aooufla but Niceph. is evidently thinking of the
wrong. passes leading from the Anti-Tauros region
2
From this passage alone it would be natural generally across Tauros to Germaniceia and
to connect Kaesoun with modern Khesun in Adata. The Sis passes ought to come under
Kommagene, south of Besne : but see above II. those leading from the Theme Lykandos into
(2) and campaign of 877 infra. The Sis passes Cilicia.
H.S.—VOL. XVII. D
 34 THE ROAD-SYSTEM OF EASTERN ASIA MINOR.
acf>i,K€To Kai iraXiv rbv Tavpov virep/3a<; fj\6e TT/JO? rbv Sapov.1 The route
taken by Heraclius will then be the Arabissos pass (which was the
ordinary route) to the Saros which he crossed by a bridge, a solid structure
with TrpoTTvpyia capable of defence, such as we might expect to find on this
road. While he lay encamped there, he was overtaken by the Persian
general, Shahrbaraz, who had reached the Euphrates before him, and broken
down the bridge of boats at Samosata, but had failed to intercept his
retreat. The Persians succeeded in bringing on a battle but were defeated.
Heraclius then continued his march to Sebasteia, when he went into winter
quarters.

Basil's Campaign in 876-72 (Theoph. Cont. p. 277 ff., Kedr. II.

213 ff.). Routes traversed: Caesareia—Kokusos—Germaniceia. [II. (1) ¥\;
and the passes towards Sis [II. (2) a & b].
To understand Basil's movements in 877, it is necessary to observe that his
march into Kommagene in that year was not a mere isolated expedition, but
part of a concerted scheme to drive the Saracens out of the whole Tauros region.
Operations were being simultaneously carried on in the north against the
remnants of the Paulician community, in the south-west against Loulon and
Tarsus, and by Basil himself against the country between Caesareia and
Adata. The Arabs did not really conquer this region: they merely held it by
strong garrisons in the various fortresses, levying imposts on the Greek
inhabitants; and consequently the capture of these strongholds would mean
the recovery of the whole country. This was Basil's object. In 876 the
fortress Loulon, commanding the Tyana-Tarsus pass and therefore occupying
a very important strategic position (Cont. 277), which ' through the
negligence of preceding Emperors had been captured by the Saracens and
fortified and garrisoned by them on account of its natural strength,' was
recovered by Basil's generals. This was followed by the surrender of the
fort Melouos, on the Laranda-Kelenderis pass. About the same time the
Paulician town Katabatala,3 to which the Paulician refugees had retired
after the fall of Tephrike (873), was taken and sacked. Next spring
(877) these successes were followed up by an expedition against the inter-
mediate country (between Caesareia and Kommagene) undertaken by Basil
himself, while his generals continued the war in the vicinity of Tarsus and
against the Paulicians in the north (see n. 37). Starting from Caesareia,
Basil sent forth a detachment to pioneer the way and followed himself with
the main body. The detachment captured the forts Psilo-kastellon (Cont.
1
See Class. Meview, I.e. p. 140 note, and most improbable that he would cross (l)Amanos,
H. G. p. 311. If the text of Theoph. is right, (2) Tauros by Cilician Gates, and thence by a
the description is obviously very bad. It may most difficult route come round to Sivas.
2
be answered that he meant to say ' on his way The date is 877, not 880, for Sima, who
over Tauros (Amanos, of. Mich. Att. 120, Skyl. submitted to Basil, was killed by Tulun of
677) he reached Germaniceia, and passing Adana Egypt in 878 (Weil, Gesch. der Khal. ii
came to the Saros.' If so, the text requires 473 n.).
8
much alteration, for the proper order is Ger- Kedr. calls it Kameia.
maniceia—Amanos—Saros—Adana ; and it is
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 35
Xylo-kastron, Kedr.) and Paramo-kastellon (Phyro-kastron, Kedr.). Then
the fortress Phalakron voluntarily surrendered. These forts lay between
Oaesareia and the Karmalas-Zamanti on the routes leading into the two Sis
passes and the pass over Anti-Tauros towards Kokusos, Basil's plan obviously
being to secure all the passes as he advanced southwards.1 Basil then
crossed2 the Onopniktes (Karmalas) and the Emir of Anazarbos (Anazarbe)
along with the troops from Melitene fled before his advance (eVet tfyyi^e,
Kedr.), thus leaving him free to secure the passes beyond the Karmalas.
This was effected by the capture of Kalsos [or Katasama; Kasama, Kedr.],
Robam [Kedr. gives Karba], Endelechone3 or Andala [Ardala, Kedr.], and
Eremo-sykSa (or -sykaea, Cont.); and thereupon Simas ' the son4 of Tael,'
who held the passes of the Tauros (i.e. Anti-Tauros) and harassed the Roman
borders, submitted to Basil, who then crossed the River Saros and continued
his advance towards Koukousos (Kokusos).
These fortresses are again to be looked for on the passes leading to Sis.5
Kai'sos and Robam cannot be identified with Ibn Khordadbeh's Kaisoum
and Ra'Mn (De Goeje's Trans, p. 70) which are frontier fortresses of Meso-
potamia and identical with the Armenian Khesoun, modern Khesun (south
of Besne), and the Armenian Rhaban, south-west of Khesun, between that
town and Marash (St. Martin, Mim. sur I'Armenie, I. p. 194). But Kai'sos
should probably be connected with Kabissos of Not. I, and Ptolemy's Kabassos
in Kataonia, and Niceph's. TO Karjcrovp (see above), TO Karjcrovv then is the
district of Kaisos-Kabissos, which is itself to be placed on the more easterly
of the two passes over Anti-Tauros towards Sis, nearly opposite to Kiskisos-
Kisken on the western pass.
Arrived at Kokusos-Geuksun, Basil set fire to the woods round the
town and then plunged into the defiles of Tauros, cutting his way through
the pathless forests,6 and cheering on his men by his personal exertions,
past Kallipolis and Padasia to Germaniceia. The Arabs remained within
their walls, not venturing to offer him battle, but as the siege of the town
was hopeless he passed on to Adata (Adapa in Kedr., TT for T), which he
besieged in vain. He then devastated the adjacent country and captured the
•jroKt'xyi.ov Geronta (? Geron). After another attempt on Adata, he retired
cautiously in fear of an ambush, and after receiving the submission of
1
Phalakron may be Frakhtin (Forakhtin) on It might be suggested that the curious name
the western pass to Sis; the -tin is the Arabic Onopniktes is a popular word expressing the
word Din widely adopted in Turkish [W. M. difficulty of fording the river (ovos, vvlyeiy).
3
E.]. Psilo-, Xylo-, etc. are all Graecized Cont. says TTJS 'PojOtyi ijroi 'EvS^Kexivris T\
forms. iripBriiris yiyovsv, S/»o 8^ xal ri rijs 'Av$<l\cv :
2
Cont. and Kedr. do not precisely say so ; probably to be changed to rrjs 'ApSd\ov ijrot
after enumerating all the forts they say vaguely 'E v 5 e A.
thv 'OvoirvlKTriv Aey6/ievov iroTafibv Kai rbv t&pov * Simas was not ' son of Tael' but his sur-
$iairepi.<ras, knowing only that the forts were in name was Tawil, i.e. ' t h e tall' (Weil, I.e. ii.
this district somewhere. The Saracen army p. 473 n.).
5
would not take to flight, nor would Simas, ' who Endelechone — Andala may perhaps be
held the Tauros passes ' (infra), submit before Enderessi on the western pass.
6
Basil had reached the Karmalas, as their language Which shows that this was not the ordinary
would imply. route to Germaniceia.
D 2
36 THE ROAD-SYSTEM OF EASTERN ASIA MINOR.
Abdelomel, 6? TWV exela-e TOTTCOV iicvplevae, returned across Mount Argaeos to
Caesareia. Here he received the news of his generals' victories,1 which
were soon confirmed by the arrival of the prisoners from the district of
Koloneia and from Loulon,—they are said to be Saracens and Kurds
(Kou/OTol) from Tarsus and the Paulician fortresses,—and after slaughtering
them all he returned by Midaion, where his army went into winter quarters,
to Constantinople.
Campaign of JRomanus IV. in 1068 (Mich. Attal. pp. 104 ff., Skyl. pp.
668 ff.). Routes traversed: Caesareia-Lykandos [I. (1)] Sebasteia-Kokusos-
Germaniceia-Aleppo [IV. (2) a and II. (1) b]; through Cilicia by Cilician
Gates to Constantinople [II. (3) b~\; the Turks traverse I. (1) to Amorion.
This is the first of a series of campaigns against the marauding bands of
Seljuks, whose ever-increasing raids made Asia Minor insecure from end to
end. In the spring of 1068 Romanus set out from Constantinople with the
intention of operating first of all against the Turks on the northern frontier.
He advanced through Bithynia and Phrygia, i.e. by the military road passing
through Dorylaion and Sebasteia, and when he had got as far (apparently)
as Basilika Therma, the Turks made a feint of retiring before him, and he
resolved to march southwards into Syria against the Saracens of Aleppo
(XoAeTr) who, in concert with the Turks, were constantly attacking Antioch
as the first step in a scheme for the reconquest of the whole of Syria. He
therefore left the road leading ' straight to Sebasteia and Koloneia' and
marched southwards obviously by the road leading to Caesareia and thence
by the eastern route [I. (1)] to Lykandos, where he intended to remain
during the hot season and then advance into Syria in the autumn. While
encamped here, he received intelligence that the Turks had made a sudden
raid on Neocaesareia-Niksar and were returning again loaded with their
spoils. Without losing a moment, he marched rapidly northwards again Bia
aTpairmv hvcrftarwv towards Sebasteia, and as he approached the town, he
ordered the main body of his army under Andronikos to proceed thither,
while he himself with the cavalry hurried over the hills between Argaous
and Tephrike2 in pursuit of the rapidly retreating Turks. This means that
he marched along the Gurun-Sivas road [IV. (2) b] about as far as Abara-
Amara and then struck right across the hill-country towards Tephrike and
the north-east. By this cross-march {KUTO, TO iyicdpo-iov) he succeeded in
overtaking the marauders and compelled them to relinquish their plunder
and prisoners. He then rejoined his army at Sebasteia (beginning of October)
and after a halt of three days marched south again by the defiles of Kokusos
(Bia rwv rrj<; KovKovaov avXmvav) to Germaniceia. Evidently, therefore, he

1
This means that the war begun the year the order of the words does not prove that
before in the south-west and north was being Argaous is north of Tephrike : he has just said
carried on at the same time as Basil's expedition ; eu0u KoAax/ei'as Kal 2eflaffTe£os. The site assigned
next year (878) Abdallah, Emir of Tarsus, was to Argaous {supra) at Argovan suits this passage
decisively defeated at Podandos. well.
2
Tijs TE Te<f>piKrjs KCtl Tfjs'Ap7ao0 (Skyl. 670) :
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 37
took the route by Tzamandos and the Kuru Tchai pass. Before reaching
Germaniceia, i.e. probably from Kokusos, he despatched a large division to
Melitene to guard the frontier [route I. (2)] and prevent Apsinalios or
Ausinalios [ = Afschin (Weil, I.e. iii. p. 112, n. 2)], the Turkish commander in
these parts, from raiding across the Euphrates. So inefficiently was the
command executed that a band of Turks actually passed Melitene before the
very eyes of the garrison and fell upon Romanus' foraging parties, some of
which they cut off. They must therefore have crossed into Kommagene by
the Melitene-Germaniceia pass [III. (1)]. Romanus, after leaving German-
iceia entered the district (Oefta) called by the Armenians TeXoi^ [Doliche,1
the Dolouk of Ibn Khordadbeh (p. 70), mod. Duluk, two hours north-west of
Aintab], and thence passed on to Aleppo, at this time under the Emir
Mahmud (Ma%/«oimo9). After ravaging the country around without
attacking the town itself, Romanus marched against Hierapolis (Membidj)
which he captured and fortified as an outpost to guard the Syrian
frontier. While he was engaged in besieging the Acropolis, Mahmud made
an unexpected advance from Aleppo and inflicted a serious defeat on the
troops set to guard the part of the town already captured: but the disaster
was avenged after the final capture of the town by a night attack on the
Saracen camp, which freed the Emperor from further molestation. Placing
Membidj under a o-Tparrjyos, he advanced to Azas,2 which he failed to take,
then entered the country of Ausonitis where he burned Katma 3 (a fort of
the Emir of Aleppo) and encamped at TerchSla (Tarchola). Shortly after
this he entered Byzantine territory and marching in the direction of Antioch
captured by the way a town Artach (near Antioch), which was evacuated by
its Saracen inhabitants. At this point he determined, in consideration of the
exhaustion of his troops, not to proceed to Antioch but turned towards north-
west and crossing Mt. Amanos by the Syrian Gates (Beilan pass, al
tckeicrovpai Si' wv r) KoiXrj "Evpia T^? Kt\t/«'a? %(opl%6rai) reached Alexandros
(Alexandretta). Thence he marched by the road which skirts the Amanos
range (rbv Tavpov), until he emerged into the plain of Issos; whence he
marched through Cilicia and the Cilician Gates to Podandos. Just as he
was entering Typsarion or Gytarion (Skyl.) which Prof. Ramsay with great
probability locates at the point where the Tarsus-Tyana and Tarsus-Caesareia
(' Maurianon' and ' Karydion') passes forked, ho received reports of the
mismanagement of the general sent to Melitene to guard the frontier,
who had allowed the Turks to cross the Euphrates and pass along the
' Eastern road' [I. (1)] by Caesareia to Amorion, which they took and
plundered. They had left their camp at a place called Chalceus (ry TOV
Xa\/eew? TOTro9e<ria) near Tzamandos, where the Roman general had his
troops stationed; but so far from suffering any inconvenience from his
1
Ao\t%h becomes Dolouk and then again in Chesney. I t is called "A(dCiov, two days' march
Greek Te\oiix ! It is mentioned both as *6\ts from Berroia (Aleppo)," in Kedr. ii. 492.
3
and as 04fm in Kedr. ii. 494. Modern Kutma, nearer Antioch. The
2
' Azaz is about twenty miles north by west description of the march is very accurate.
of Aleppo,' Finlay, i. 472 quoting from Col.
38 THE ROAD-SYSTEM OF EASTERN" ASIA MINOR,
presence there, the Turks on their return had actually defeated him and shut
him up in the fort. Consequently the Emperor finding himself unable to
pursue them returned direct to Constantinople.

Bomanus' campaign in 1069 (Mich. Attal. 122 ff., Skyl. 678 ff.). Routes
traversed: Caesareia—Melitene, over Euphrates, [ I (1)] and thence north to
Acilisene; Koloneia — Sebasteia — Caesareia — Herakleia [IV. (1)'and II.
(3)6]; Melitene — Caesareia — Iconium [ I . (1) and II. (3) b]; Iconium—
Seleukeia—by Syrian Gates to Aleppo.
In 1069 Komanus undertook a second campaign against the Seljuks.
After quelling the rebellion of Crispin, a Norman noble in his service, he
arrived with a large force at Caesareia and continued his march eastwards to
Larissa, where he heard that a Turkish horde was engaged in pillaging the
country in the. vicinity. A detachment despatched against them was driven
back in rout, and Romanus then moved onwards towards Melitene. While
he was engaged in pitching his camp, the Turks suddenly appeared and,
occupying the higher ground, proceeded to attack the Byzantine army in the
plain below, but were defeated. Romanus allowed them to retreat without
molestation and when he followed them three days afterwards they crossed
the Euphrates and encamped there, waiting till he should return home.
When he had advanced within less than two days' march of Melitene, he
thought of returning again and abandoning a wearisome and fruitless
pursuit, merely leaving a force to guard the frontier; but he
finally.determined to cross the Euphrates and march against XXidr, mod.
Akhlat, on Lake Van, hoping by the capture of the town to secure the
Armenian frontier and arrest the ruinous incursions of the Turks. Accord-
ingly he advanced by Melitene and crossed the river (T^? Trpocrcorepco
<f>epovar)$ ^yjraro eco<s rbv Eu<£/>. hiairepaicoOel? K.T.X.), compelling t h e Turks
to retreat inland (el? rh a-<f>irepa). The line of march is thus the ' Eastern
road' [ I. (1)]. The direct route from this point to Akhlat went by Kharput
and Romanopolis—Palu and thence through difficult country to Van (supra).
This route he followed for a short distance (a>? jap ev0v$ TT}<; ' Ptofiavotr.
ekavvcov i(f>aivero, eg rjs f) 777309 TO XXIUT icd6o&o<i Sea cnevayirSiv iviyiverai,
fj,eTaaTpeyjra<; TTJP yvmfir/v . .) and then suddenly halted iv ftaffei Toirqi, where
he divided his army and placed the stronger division under the command of
Philaretos for the defence of the frontier, while he himself turned north-
wards,1 preferring a cooler climate. After passing over rough and mountain-
ous country, he reached a place called Anthias, a fertile and well-watered
spot amidst high mountains. It should be looked for in the watershed south
of Mezur Dagh. Thence he proceeded to cross " Mount Tauros, called by the
inhabitants Moufot/pos," i.e. Mezur Dagh (Arabic Jabal Mazur), and passing
a second time over the Euphrates entered KeXecrlvrj (Acilisene, Skyl.
Ke\.T&vij), which is accurately described as separated from Mezur Dagh by
the river. While encamped here he received intelligence that Philaretos had

1
The crossing of Murad Tehai is not mentioned, but must be assumed.
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 39
been defeated by the Turks and the routed troops soon arrived at his camp,
fleeing by way of Anthias and Mezur Dagh. The Turks pursued for some
distance, but finding the country impracticable for light horsemen, they
turned back, crossed the Euphrates above Melitene, and over-ran Cappadocia
in their usual manner, making for the populous t u t defenceless city of
Iconium (Konia). Eomanus rallied his forces and determined to intercept
their return. His first plan was to "lead his army through the town of
Keramon to the banks of the Euphrates as far as Melitene," but it was
pointed out that this route ran through a deserted and pillaged district where
supplies would be hard to obtain, and that time would be wasted in traversing
ground where it was necessary to march in single file. The route indicated is
not clear, but apparently it crossed Mezur Dagh towards Murad Tchai and
Melitene.1 In any case he abandoned this idea and marched through
Koloneia and the Armeniac Theme to Sebasteia, i.e. by the road Satala—
Koloneia—Nicopolis—Sebasteia. At Sebasteia he learned that the Turks
were marching through Lycaonia and Pisidia on their way to Iconium2 and
so he advanced to Herakleia—Kybistra (Eregli), i.e. by the road through
Caesareia [IV. (1)] and thence by Tyana to Herakleia [II. (3) b]. Hearing
at this point that the Turks had sacked Iconium and were returning, he
despatched a detachment to Cilicia to effect a junction with Katatourios, the
governor ("duke") of Antioch, whom he requested to secure the passes east
of Mopsuestia (Missis). The Turks marched Sid, rS>v T^? SeXeuKet'a? 6p&v
and, as they emerged into the plain of Tarsus, they were attacked by the
Armenian inhabitants but escaped, with the loss of their booty, through
Cilicia. Being informed by Greek captives that a force was awaiting them at
Mopsuestia, they avoided the town and after a short halt at Blatilibas
(Baltolibas, Skyl.) hurriedly crossed Amanos (TO %apf3avSiKbp opos) by the
Syrian Gates to Aleppo. Romanus learning of their escape at Claudiopolis,
whither he had advanced to meet them, left a force to operate against other
Turkish bands and returned to Constantinople.

The First Crusade in 1097.3 Routes : Nicaea—Dorylaion—Iconium—

Herakleia (Eregli); thence to Tarsus—Adana—Syrian Gates—Antioch
[under II. (3) b etc.]; Herakleia—Caesareia—Kokusos—Germaniceia—
Antioch [II. (3) b and II. (1) &].
After the capture of Nicaea, the Crusaders proceeded by Dorylaion and
Iconium to Herakleia—Kybistra, which was evacuated by the Turks on their
approach. At this point the army divided. Baldwin and Tancred with their
own following marched southward by Podandos and the Cilician Gates to
Tarsus, which they captured without difficulty; whereupon Adana (Addana,

1
Keramon can hardly be connected with rb [II. (3) b].
3
Kepanlatov on the Zarnuk, the most easterly I have followed the accounts of the Latin
tributary of Tokhma Su (Melas) flowing past writers in Migne's Patrol. Lai. vols. clvi.
Melitene (Theoph. Cont. 268). (Guibert) and civ. (Rob. Mou., Tudebodius,
2
The Turks therefore took their favourite etc.).
route by Caesareia [I. (1)] and thence to Iconium
40 THE ROAD-SYSTEM OP EASTERN ASIA MINOR.
Bad. Cadom.; Athena, Guibert 728 etc.) and Mopsuestia (called Mamistra,
Mamysta, Manustra) voluntarily surrendered. From Mopsuestia Baldwin
marched (through Amanus Gates) across the Euphrates to Edessa, while
Tancred proceeded by the Syrian Gates to Antioch. Leaving Herakleia, the
larger portion of the Frankish army under Raymond, Bohemond, and Godfrey
took a longer route. They entered the ' Armenian country,' and marching by
Tyana towards Caesareia reached a certain fortress on an impregnable site,
which they made no attempt to take. This fort is not Tyana (in the plain) but
Kyzistra (see on II. (3) b), mod. Zengibar Kalesi, half an hour west of Develi
Kara Hissar, " a striking mediaeval castle on a lofty two-peaked h i l l . . . .
which has been (prob. not correctly) identified with Nora, where Eumenes
defied Antigonus in 320 B. C." (Hogarth in Sir C. Wilson's Handbook p. 163).
After passing Kyzistra they reached Caesareia, where they were welcomed by
the inhabitants. They thus took the route traversed by Romanus in 1069
(in the opposite direction). At Caesareia they turned again towards Antioch.
Marching no doubt by the Kuru Tchai pass, they arrived first at Plastentia,
a town situated in a beautiful and fertile country (multae pulchritudinis et
situs uberrimi civitatem), which had been besieged in vain by the Turks for
three weeks before the arrival of the Crusaders, who were received with open
gates. Plastentia 1 is evidently the Armen. Ablastha, Syriac Ablestin,
which has usually been identified with Albistan, but should apparently be
placed in the upper Saros valley. Thence they moved onwards to Coxon
(Guibert, 730; Coxan, Tudebod. 776 ; Cosor, Bob. Mon. 695), i.e. Geuksun—
Kokusos, which was at that time in a very flourishing condition.2 From
Geuksun they marched towards Marash by a route so exceedingly difficult
that it calls forth from the monk Tudebodius such choice epithets as diabolica,
exsecrata montanea. It is described as a narrow path (arctus et nimis scrupeus
calles praeruptus, Guibert), so broken and steep that everybody alike had to go
on foot and it was impossible to pass by the man in front. It is evident,
then, that they did not take the pass traversed by Basil in 877 and by
Romanus in 1068, i.e. the Ayer Bel pass by Kallipolis and Padasia, which by
general testimony is by far the easiest road through the eastern Tauros (see
the interesting account by Hogarth, Mod. and Anc. Roads in East. Asia Minor,
p. 20), but the route by Geben along Kursulu Su and round Dolaman Dagh
to the Jihun—Pyramos and thence to Marash. In the description of this
pass in Sir C. Wilson's Handbook, Mr. Hogarth says, " it crosses the spurs of
Dolaman Dagh by a very difficult rocky path. The descent to the Kursulu
Su which has run, with several falls, through a deep chasm, is very steep, and
there is an equally bad ascent, the path being in places only a foot wide " (p.
271). Compare'' the words used by Robert, the monk of Reims, who gives a
vivid account of the soldiers' despair, ' Semita non mnplius quam unius pedis
spatio dilatabatur' (p. 695).3

1 3
The name is given by Baldric. Von Moltke, quoted by Hogarth in Mod.
3
' In qua erat maxima ubertas atque stipata and Anc. Roads, etc. p. 20, describes the road
omnibus bonis quae nobis erant necessaria,' from Marash to Geuksun as difficult. This may
Tudeb. I.e. : so Guibert, etc, have been the route he took.
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 41
After emerging from this ' exsecrata montanea,' the Crusaders reached
Marash (Marasim, Guibert and Rob. Mon.; Marusim, Tudeb.), where they were
hospitably received, and after a day's rest proceeded towards Antioch.

Basil's campaign of 872 * (Theoph. Cont. 267 ff., Kedr. 207 ff). Routes :
Military Road to Tephrike [IV. (3)]; thence by IV. (2)'b to Gurun and [by
I. (1)] to Derende and over the hills to R. Zarnuk (west of Melitene).
In 872 Basil marched against the Paulicians by the Dorylaion—
Sebasteia road to Tephrike. Failing to take the town, he captured several
of their fortresses, the most important being Abara—Amara on the Sivas—
Derende road (supra). As he marched southwards along this road, Taranta—
Derende submitted and its submission was followed by the surrender of
Lokana—Lykandos. From Derende he then crossed the hill-country between
the Tokhma Su (Melas) and the Sultan Su (Arab. Karakis) to a position on
the river Zarnuk (supra), south-west of Melitene, sending a detachment
against Zapetra (Viran Sheher) and Samosata—Samsat. When this detach-
ment returned, he marched on Melitene. The Emir's forces sallied out to meet
him but were defeated and shut up within their walls. It was hopeless,
however, to attempt to besiege the strongly fortified town and Basil marched
northwards again through the Paulician territory by way of Argaous—
Argovan, which he captured. Several other forts were taken in the country
between Argaous and Arauraca (which seems to be the place meant by
Ararach—Rachat), and Basil then returned home. Tephrike was taken and
the Paulician community crushed in the following year (873).

EXCURSUS.
THE ROYAL ROAD.

BEFORE discussing the line of the Royal Road from Caesareia eastwards,
it is well to have realised the importance of the route by Herpa, Tzaman-
dos-Azizie, Melitene, and over the Euphrates at Tomisa throughout the
Byzantine period and apparently also in the last two centuries B.C., as
reported by Strabo on the authority of Artemidorus (supra on I. 2).
After passing Tomisa, the road to Persia would naturally turn south by
Amida-Diarbekr and along the left (north) bank of the Tigris,—much in the
line assigned to it in this part by Kiepert. The distance from the first
crossing of the Ilalys to the Euphrates by this road will be found to cor-
respond approximately to the 119J parasangs (3585 stadia) which Herodotus'
Itinerary (V. 52) gives as the whole distance for Kappadokia and Kilikia
(to the Euphrates).
Why then should this line for the Royal Road be doubted ? Largely

1
Discussed in Class. Rev., I.e. pp. 136 ff., and only summarised here in the briefest possible
manner.
42 THE ROAD-SYSTEM OF EASTERN ASIA MINOR.
because of the so-called ' Kilikian question' in Hdt.'s account of the road,
i.e. the extraordinary fact that while Kappadokia is crossed only in twenty-
eight stages (104 par. or 3120 stadia) the large district of ' Kilikia,'
extending to the Euphrates, requires only three stages (15J par. or 465
stadia). Now considering the large size of Hdt.'s ' Kilikia' which extends
on one side to the Halys (I. 72), on another to Euphrates (V. 52), and also
down to the Cyprian Sea (V. 49), the shortness of the distance across
Kilikia reasonably excites suspicion. It is possible then that the distances
are wrongly distributed between the two districts. This might be due (a) to
corruption in the text; for it is admitted that the text of the Itinerary
is corrupt at least in one place (de la Barre's emendation) and probably in
another (Stein's transposition of the three Armenian rivers to Matiene).
Or (b) it might be due to misconceptions on the part of Hdt. His knowledge
of the Royal Road is derived not from ctyt? but from some unknown authority.
But in one point Hdt. has misconceived his authority. The Sigal TTVXCU and
Siga <f>v\a,KTijpia passed by the traveller on the borders of Kappadokia and
Kilikia must almost certainly refer to the guard at the Cilician Gates. Hdt.
therefore conceived the road to pass through the Gates into maritime Kilikia
either because he confused the guard at the Kilikian frontier with the guard
at the Cilician Gates or because he has p u t ' together two separate and un-
connected facts: he has put the guard of the Cilician Gates on the Royal
Road, and he has connected the " Royal Road" therefore with maritime
Cilicia (V. 49) whereas it crossed Cappadocian Kilikia (V. 52)' (Ramsay,
Git. and Bish. of Fhrygia, I. p. xiv. n.). Such an initial error would lead to
other distortions of the facts before him, in order to bring them into harmony
with the first misconception. We are familiar with the manner in which
modern writers, more scientific than Hdt., often strain facts to make them
fit into a theory. But apart from this supposition as to the &<,%a\ wvXac,
Hdt., while very likely retaining the whole distance (119J par.), may have
modified the Kilikian distances1 to suit his own ideas of ' Kilikia,' which of
course he would believe to be right! ' Kilikia' with Hdt. is no very
definite region: it is the ' land inhabited by the Kilikians' as Assyria is that
inhabited by the Assyrians, and Egypt by the Egyptians (II. 17),—a con-
venient cloak for ignorance. Apparently it is made to extend to the Halys
and Euphrates, just because these were the two great dividing lines in
Eastern Asia Minor of which he knew, though his knowledge was vague
enough. But is ' Kilikia,' after all, a large district in his conception ? The
distance between the Halys, the Euphrates, and the Cyprian Sea must have
been for him exceedingly small. The source of the Halys must have been
near the Euphrates, for it divides Lower Asia eic daXao-arj? T»)? avrlov K.virpov '
69 TW JLv%eivov TTOVTOV (I. 72); and it is only five days' journey across this
narrow isthmits (!). Need we be surprised then that, with conceptions like
these to accommodate, 'Kilikia' is crossed in three stages of 15 J parasangs ?
The Kilikia of Hdt.'s authority—if Kilikia was mentioned by him—may
1
i.e. if his authority mentioned Kilikia.
 THE ROAD-SYSTEM OF EASTERN ASIA MINOR. 43
have extended to Halys and Euphrates, and he may have given 119J par. as
the whole distance for Kappadokia and Kilikia from the first crossing of the
Halys to the Euphrates. If so, his conception of Kilikia differed entirely
from the Kilikia of the old well-marked and natural division between
Kappadok (the country between Tauros and Euxine, Euphrates and Halys),
Kilik (the sea-board country south of Tauros and west of Amanos) and
Kumukh (Kommagene): according to which the Royal Road would not pass
through Kilikia at all (on any theory now held).1 We may note that this
older division is reflected in Hdt. e.g. v. 49, vii. 91, and that the inclusion of
' Posideion on the borders of Syria' (iii. 91), i.e. of the strip of coast fringing
Amanos, is consistent with the older conception and constitutes no argument
for the inclusion of Kommagene in Kilikia.
From all these considerations it would seem that an undue importance
has been attached to the ' Kilikian question' in discussions on the course of
the Royal Road. Various solutions of this question are possible: and we
must look outside Hdt. for evidence as to the line of the road.
An ingenious theory, which endeavours to explain the three Kilikian
stages, has been lately put forward by Mr. Hogarth and accepted by others
(see Macan's Hdt. iv.—vi., vol. II. pp. 299 ff.). This theory brings the road
from Pteria either in a direct line to the head of the Tokhma Su, and thence
by Derende to Melitene, or by a de'tour to Caesareia—Mazaka and east to
Melitene (as advocated in this paper): but instead of crossing at Tomisa the
road is made to turn south from Isoli and run up the basin of the Gerger
Tchai by Kiakhta to Samosata, where it crosses the Euphrates and runs
across the desert south of Mount Amasius to Nisibis and thence to Nineveh,
&c. The difficulties of such a route over Tauros to Samosata and then
through the desert to Nisibis, when an easier and more direct route is open,
do not predispose one in its favour. What are its advantages ? (1) It claims
to solve the ' Kilikian question' by making the distance between the spine of
Tauros (the frontier of Kappadokia and Kilikia) to Samosata represent the
three Kilikian stages of Hdt. Obviously this solution is reached only by an
arbitrary interpretation and limitation of Hdt.'s ' Kilikia,' which makes
it include Kommagene while denying that it extends north of Tauros. But
if Hdt. says that Kilikia extends to the Euphrates, he also says that it
extends beyond the Halys, o? peei Sta K«Xt/w'a? (i. 72): and the inclusion of
Posideion (iii. 91) does not support the extension beyond Amanos to
Kommagene. The reconciliation with Hdt., therefore, disappears. (2) It
claims to be supported by Strabo's account (p. 663) of the KOCVT) 0809 to the
east. The account, however, after Tomisa, is far from clear. At this point
there is a break in the description, where Strabo cites the authority of
Eratosthenes as confirming Artemidorus' account of the subsequent route to
India and refers to Polybius; we note a vagueness and a lack of sequence in
the following words as compared with the description of the Ephesus-

1
If Hdt.'s authority was an official document, should we not expect it to be based upon
this division ?
44 THE ROAD-SYSTEM OF EASTERN ASIA MINOR.
Caesareia section; and it looks as if Strabo had mixed up or fused together
two separate routes, one crossing the Euphrates at Tomisa and another
' beginning at Samosata' (which is not described and may simply have joined
the former road at Amida, so that it would be possible to make a ditour by
Samosata). Anyhow the description is not at all clear. The road goes ' to
the Euphrates as far as Tomisa in Sophene.' Mr. Hogarth explains that the
meaning is that the road ' touched Euphrates opposite to Tomisa but did not
cross the river.' But Strabo does not say this : for surely his words ought
plainly to mean that the road crosses to Tomisa. Then he goes on: T« B'
eV ev6ela<; TOVTOIS [Tofiicrois ? and the dat. ?] /tte^pt T»J? 'lvBucfj<; ra avra
KeiTai ical irapa T<£ 'Aprefi. airep ical irapa, TW 'EparoaOevei, . . . apteral
Be [subject? 1 ] airb %afiocrdTa>v . . . eh Be %a/j,oa. airo TO>V b'pcov TT}<;
Ka7nraBoKia<; TCOV irepl 16/Mcra virepdevn TOV Tavpov GTaBtovi elptjKe 450.
The last statement is incorrect: it is about 650 stadia. Mr. Hogarth explains
the discrepancy by supposing that' Strabo reckons from the spine of Taurus
on the right bank lower down than Tomisa, which is not in Cappadocia at
all': it must at least be admitted that a.7rb TWV opwv TJ}? K.ainr. T&V
irepl Tofuaa is a singular way of expressing it.
But in any case, even if the description were quite clear, we have to
remember that this was a trade route and that the Royal Road was not a
trade-route but a road for administrative purposes, a road for couriers. The
line of a later trade-route would be determined by different considerations ;
thus, for example, the Royal Road along the upper Hermos is so difficult
that it could never have been chosen as a caravan-route. Lastly, the Roman
bridge at Kiakhta need only show the importance of this district in a scheme
of frontier defence and the road, if it existed, would be used for this
purpose.2 It is hard therefore to see that this route affords any evidence for
the line of the Persian Royal Road.
J. G. C. ANDERSON.

1
I t ought to be ret 8' iir' eb8. TOUTOIS, etc. cation of the west bank of the Euphrates in
2
Cf. the importance attached to the fortifi- Amm. Marcell. xviii. 7 (supra iii.).