¹ Notably: L. Duchesne and M. Collignon in 1876; results in BCH. I, 365–370Google Scholar (short notice only), III, 333–347, 478–482. J. R. S. Sterrett in 1884 and 1885; results in Papers of the American School at Athens II (quoted as EJ.), III (quoted as WE.). W. M. Ramsay in 1884 (with A. H. Smith) and 1886; results in AJA. II, 128–131Google Scholar; III, 366–8; IV, 6–21, 263–275 (Ramsay), JHS. 1887 (Smith). V. Bérard in 1892; results in BCH. XVI, 417–438Google Scholar. Jüthner and Heberdey in 1897; results in Wiener Studien 1902, 53–9Google Scholar. A. M. Woodward and A. H. Ormerod in 1910; results in BSA. XVI, 76–136Google Scholar; XVII, 205–214.

² Artemidorus' list of Pisidian towns, as given by Strabo 570, touches only the fringes of our region with Termessus, Ariassus and Cremna. Pliny gives no help, and Mela is totally silent.

³ Supplemented by the ungeographical Notitiae.

⁴ CRAI. 1948, 402Google Scholar; Anadolu I (1951), 60–1Google Scholar, brief notice only; the Roberts' discoveries are still unpublished.

⁵ Belleten XVIII, 469–510Google Scholar. J. and Robert, L., REG. LXIX (1956)Google Scholar, Bull. Épigr. no. 319, in a review of this article, complain that a certain number of these stones had already been seen by them (presumably in their original positions), and that I have anticipated their publication of them. Their annoyance is understandable, and I am sorry to have been the cause of it; I published the stones at the invitation of the Department of Antiquities in Ankara, and had no means of knowing that the Roberts had already seen some of them. However, the world will be relieved to learn that my publication has given the Roberts no cause to forgo their own. In the present article I have suppressed all inscriptions that I had reason to believe were previously seen by the Roberts; I have, however, for the sake of completeness, taken the liberty of referring briefly to the site of Malgasa which was discovered by them. If I have unwittingly offended again I can only express my regret. In the same critique the Roberts complain of the quality of the photographs, some of which were taken by me. In this connexion Prof. Robert has himself recently given us an admirable object-lesson. In Hellenica X, 278–9Google Scholar he publishes three short epigrams on a sarcophagus from Parium, with photographs Pll. XXXVII and XXXVIII. In these sixteen lines (disregarding three faults of accentuation) we find the following errors: in B3, ἥρπασεν νηλεόθυμος contra metrum; the photograph shows ἥρπασε: in D2 λαïνέῃ; the photograph shows ληινέῃ: in D5 ἴππος γάρ μ՚ ἔκτανε, with the note, “le poète a commis une faute de prosodie en condidérant comme long l'alpha de ἔκτανε”; the photograph shows ἔκτεινε. Whatever may be thought of these blunders, it is unlikely they would have been suspected without the photographs. But from a scholar whose own publications are no more accurate than this we are perhaps entitled to expect a less offensive tone in his criticism of others.

⁶ See AS. VIII, 127–156Google Scholar.

⁷ According to the recent Turkish map. The old GS map, however, gives them as 3,018 feet and 2,820 feet respectively.

⁸ As compared, that is, with the coastal lands of the west and south familiar to me. By comparison with the central Anatolian plateau J. Mellaart (AS. loc. cit.) is able to call them mild.

⁹ A Turkish analysis in 1937 gave the following result: in 1 litre of water, 7·3 gr. So₄, 5·3 gr. Cl, 0·6 gr. Mg, 0·6 gr. Ca, 6·9 gr. Na (Ardel, A., Istanbul Üniv. Coğrafya Enstitüsü Dergisi 1952–1953, p. 75)Google Scholar.

¹⁰ Late in the summer the water as it recedes leaves a whitish incrustation round the shores at the north end of the lake; Ramsay, CB. I, 299Google Scholar, says he once saw the inhabitants carrying this away, and that he learned later that it was saltpetre.

¹¹ In their criticism of my Belleten article in REG. loc. cit., the Roberts observe justly that the local residents are very ready to say that a stone comes from Ağlâsun, the great ruin-field of the district, when in fact they simply do not know. This is indeed very noticeable; even the written inventory kept by the Educational Officer, on which I was relying in my article, is proved to be guilty of this error in at least one case.

¹² Pace, Annuario VI–VII, 448, no. 167Google Scholar; Metzger, Catalogue 48, no. 22Google Scholar; Bean, Belleten XXII (1958), 69, no. 85Google Scholar.

¹³ Also at the Askerlik Şubesi is the stone carrying Ramsay, CB. IGoogle Scholar no. 174.

¹⁴ JHS. LXVIII (1948), 57Google Scholar; Belleten XVIII (1954), 470Google Scholar, Figs. 3 and 4.

¹⁵ Called Eski-Yere by Ramsay, and in fact so pronounced sometimes.

¹⁶ τρίτευμα might, for example, be a political division, like a τριττύς, and the altar be dedicated to the use of the community. But it is safer to reserve judgment.

¹⁷ Line 4, ΠΑΝΤΙΤѠ; line 5, ?

¹⁸ Ramsay, published in Rev. des Univ. du Midi I (1895), 353–362Google Scholar, a series of epitaphs from Sofular near Eğirdir, which he regarded as written in the Pisidian language mentioned by Strabo 631 as having been spoken at Cibyra. They afford, however, no help in understanding our present text. They are written wholly in Greek characters, and were taken by Ramsay to comprise only personal names, with perhaps one or two ethnics; Brandenstein, however, in RE. s.v. ‘Pisidien’, identifies a number of common nouns and verbs. I have not seen Shafer's, R. article ‘Pisidian“ in AJP. 71 (1950), 239–270Google Scholar.

¹⁹ I have indifferent photographs, taken in the dusk, of this stone and of No. 20, which are available to anyone who is interested.

²⁰ This, though accepted (with a query) on Kiepert's map, is very properly rejected by Jones, CERP. 417Google Scholar, who suggests that Limobrama, otherwise unknown, may be a corruption of δήμον Βραμα.

²¹ The form Buldur is not now heard, but was so formerly; it is used, for example, by Hamilton, Collignon and Sterrett and sometimes by Ramsay; cf. Arundell, Asia Minor II, 101Google Scholar. In Yerten-Yelten the converse change has occurred.

²² See below under Cormasa and Takina.

²³ In Hierocles the name is corrupted to Λυσήναρα.

²⁴ Imhoof-Blumer, Kleinas. Münzen 384Google Scholar, and so very clearly on two fine specimens in the collection of Herr von Aulock; in BMC. Lycia, etc. Pl. XXXVI, 12, the second iota is not visible to me.

²⁵ Variations in the terminations of ethnics occur, though not very frequently: see Robert, Hellenica II, 73Google Scholar: but the present case is different.

²⁶ I was told also that there used to be another similar stone in the village, but this seems to have disappeared.

²⁷ CIL. III, 7174–6Google Scholar.

²⁸ The alternative, that our stone shows three miles from Ilyas, cannot in my opinion be accepted. Ilyas, whose communications are with the north, is a most improbable caput for a road running to the south—just as Lysinia would be for a road running to the north.

²⁹ Including that recently published by Calder and myself (Supplement to AS. VII, 1957Google Scholar). Except for Lysinia, my own investigations came too late to be utilised.

³⁰ This would have the incidental advantage of avoiding the difficulty felt by Ramsay, HG. 57Google Scholar, namely that a Galatian road should pass for part of its way through the province of Asia, since Burdur Lake formed the boundary.

³¹ Murray's, Handy Classical Maps, Asia Minor Sheet (London 1903)Google Scholar.

³² CB. I, 322Google Scholar, n. 1.

³³ The suggestion put forward in BCH. I (1877), 371Google Scholar (cf. Hirschfeld, Act. Minor. Acad. Berol. 1879, 322)Google Scholar that Ilyas is Hierocles' Ἴλονζα, rests only in the similarity of names, and has found no favour. The name Ilyas is better explained by Ramsay, CB. I, 323–4Google Scholar.

³⁴ Sterrett, WE. 600–8Google Scholar; Ramsay, CB. I, 152–163Google Scholar.

³⁵ Instituted probably not by our Teimotheanus but by the Aur. Fonteius Teimotheus mentioned in other inscriptions at Kılıç.

³⁶ “Baris is attested at Kilidj”, quoting CB. I no. 153Google Scholar, where the victorious wrestler has the ethnic Βαρηνός. “Isbarta Sparta has been taken as site of Baris: it is ancient Saporda”. This appears to mean that Kılıç, is Baris, though the inscription does not of course prove this, any more than our present inscription proves that it is Apollonia. In HG. 406 Ramsay accepted the usual identification of Baris with Isparta, and in CB. I, 324–5Google Scholar he argues strongly against Radet's theory that Isparta is Saporda; the view adopted in Social Basis seems to derive from a suggestion in CB. I, 324Google Scholar, n. 3, where it is said to involve the placing of Eudoxiopolis at Ilyas; but in Social Basis Ilyas is Valentia. These perpetually shifting conceptions leave the reader quite bewildered.

³⁷ In CB. I, 336Google Scholar and Addenda 348–9 he says that Σεβαστοῦ in line 2 is also in erasure, replacing the word Κλαυδίου.

³⁸ I realise that in face of Ramsay's positive assertions this statement may appear startling; but those who have worked with Ramsay will confirm that on numerous occasions he persuaded himself that he had actually seen on the stone what he later felt ought to be there. A striking example in MAMA. VII, 58Google Scholar.

³⁹ By reason of the absence of Imp. from Nero's titles and the known period of Praesens' career: see Magie, Roman Rule 1419Google Scholar, n. 65.

⁴⁰ I know of no exact parallel to this nomenclature for Claudius; for the omission of the personal name cf. CIL. VI, 5539Google Scholar, X 5056 = Dessau 1786, 997: “Caesaris Augusti nudis vocabulis interdum etiam Claudius et Nero appellantur,” Dessau.

⁴¹ Notably by Broughton in TAPA. LXV (1934), 220 ff.Google Scholar, approved by Magie, , Roman Rule 1325–6Google Scholar.

⁴² e.g. [ἒτους.. Νέ]ρ[ωνος] as in IGR. III, 335Google Scholar, or, as Dittenberger prefers, [ἔπὶ Νέ]ρ[ωνος Κλ]αυδ [ί]ου Καίσαρος [Σεβαοτοῦ Γ]ερμανικοῦ [καὶ Λουκίον Ἀντιστίου Οὐέτερος ὑπάτων], i.e. A.D. 55.

⁴³ Except indeed that according to Ramsay erasure marks could be seen in Z2. In view of the absence of such erasure on X and Y (see above), I cannot but doubt whether these marks can really have existed. I have not myself seen Z.

⁴⁴ I was told that Y came from a spot about 1 km. to the north, where it had been used in a building; so its original position remains unknown.

⁴⁵ I say three copies, not four, because I feel no doubt that the stone I saw in Düver cemetery (X) is the same that Ramsay saw there. It is true that in Social Basis 234 he describes it as a “thick rounded cippus” standing about 3 feet out of the ground, a description which in no way applies to the stone that I saw; but it seems more likely that this is a lapse of memory at an interval of fifty years than that we have really four stones. The line-divisions and the lacunae agree too closely for coincidence. My guide thought he remembered having seen another inscribed stone in the cemetery, but a search failed to reveal this.

⁴⁶ CIG. 3956b: Ramsay, CB. IGoogle Scholar, no. 138. The name was at first misread as Λακινέων or Λακανέων: in fact, as Smith observed, Τακινέων is clear and certain.

⁴⁷ See the latest discussion in Magie, Roman Rule 1048Google Scholar, n. 39.

⁴⁸ See below, p. 94.

⁴⁹ Ptolemy has the sequence Themisonium—Phylacaeum—Sala—Gazena. If Ramsay is right in taking Gazena to be Takina, and if a hazardous conjecture is permissible, it is attractive to suppose that Ptolemy is following this road; Sala (or Sal⟨d⟩a) may then be equated with the modern Salda—or more exactly with the site at Kayadibi on the shore of Salda Lake. (This site was formerly identified with Keretapa-Diocaesareia, but Robert, , Villes d'Asie Mineure 105–121Google Scholar, has shown that this town lay further to the west at Kayser, now Yeşilyuva.) The Lydian (or Phrygian) Sala is out of place in Ptolemy's sequence. The site of Phylacaeum is quite unknown.

⁵⁰ Livy XXXVIII, 15. See below, p. 113 sqq. The corresponding passage of Polybius calls it Κύρμασα.

⁵¹ Even in 1879 Duchesne remarked on the spoliation of the site to build the village; and the same process is still continuing.

⁵² In CB. I, 328, he speaks of “the bridge site”Google Scholar.

⁵³ cf. IGR. III, 401Google Scholar (Comama), where the D's are written as deltas.

⁵⁴ The figure on the Table seems to me to be xii, though others (e.g. Magie, Roman Rule 1138Google Scholar) have read it as xv.

⁵⁵ Since Perge is just 12 m.p. from Attaleia, it seems likely that we have preserved the first and last stages of the stretch Themisonium-Perge. It is not certain that Themisonium was at Karahüyük: see Robert, Villes d'Asie Mineure 112Google Scholar, n. 4: but it was undoubtedly in that neighbourhood.

⁵⁶ Die Peutingersche Tafel (1916), p. 12Google Scholar.

⁵⁷ This is not, however, apparently, Miller's own view. On his map no. IX he shows the long road as passing further to the south, reaching Cormasa at Istanoz (Korkuteli), with a branch road from Hierapolis to Takina. Ramsay, in CB. I, 327Google Scholar, n. 1, placing Cormasa at Gâvur Ören, declares that the Peutinger Table “has made two roads into one; the road from Laodicea goes by Themisonium, Phylakaion, Cibyra, Lagbe, Isinda, Termessos, to Perga. The road from Apameia goes by Kilij, Ilias, Kormasa, Komama, Panemou Teichos, to Perga”. In Social Basis 239, on the other hand, it is suggested that Cormasa was at the site close to the north-west shore of Kestel Gölü (Kuşbaba) which in CB. I, 327Google Scholar, is identified with Colbasa; how this can be on the line of a road from Ilyas to Comama and Perge I do not understand. See further below, p. 114, n. 79.

⁵⁸ Ramsay, at first suspected they were identical; in CB. I, 327Google Scholar, n. 2 he abandons this view (though it survives to cause confusion on p. 339), but appears to revert to it in Social Basis 239: see the previous note. Scholars in general have accepted the two places as distinct.

⁵⁹ The only remaining uncertainty is with regard to Ptolemy's Corbasa. As Ptolemy also mentions Cormasa but not Colbasa, some have thought Corbasa may be identical with Colbasa; and Ramsay, in AM. X (1885), 342Google Scholar, takes all three to be one and the same. Ruge, however, in RE. s.v. Kolbasa, is prepared to accept Corbasa as a separate place. It is otherwise quite unattested.

⁶⁰ Some of the coins, inscribed Παλεοπολειτῶν θέμις, attest the celebration of games.

⁶¹ Ramsay, CB. I, 321–2Google Scholar identifies Palaeapolis with Alastos, an obscure place mentioned in two inscriptions (CB. I, 307Google Scholar, nos. 114–15) but otherwise unknown; he places it at Akören. In the course of a rapid visit to Akören (below, p. 103) I saw no indication of a city-site likely to have coined money or celebrated games. On Colbasa see further Part II.

⁶² Gebren, however, is the name of a village in the upper course of the Lysis, with a small ancient site; and its true form is Gebrem.

⁶³ Other inscriptions attributable to Cormasa are Duchesne, BCH. 1879, 481Google Scholar, at Eğneş, SterrettEJ. 115Google Scholar, no. 84, at Mussalar (i.e. Mürseller), Ramsay, AJA. IV, 21Google Scholar, at Elmacık (now at the gate of the school). None of these is of any great interest.

⁶⁴ A large mortar, hollowed out of a stone block (frequently an ancient block) and set up in the street for public use. A large wooden pestle is usually supplied.

⁶⁵ Also called Bebekler, and now officially Sertaç; but the old name Kemer is almost universally used.

⁶⁶ Ramsay's “tracery” (Social Basis loc. cit.) is a false recollection.

⁶⁷ It is hardly likely that this mark is intended for a tau, added later above and on its side, even though this would have the advantage of blocking the hiatus.

⁶⁸ Ramsay, CB. IGoogle Scholar, nos. 166, 168, 169. In no. 166, line 6, the stone has Αὐληλίῳ not Αὐρηλίῳ.

⁶⁹ Soğanlı village was at that time lower down the hillside; it moved up to the ancient site in 1926 because of a water shortage.

⁷⁰ BCH. III, 342Google Scholar, no. 16, erected by Κάλλιππος Μοσνω Καλλίππου with his sons Mosanos and Kallippos.

⁷¹ Ramsay at first suggested that Hadriane is a late name of Olbasa, which does not occur in the Byzantine lists. The absence of any earlier mention of Hadriane (whereas coins of Antoninus Pius, Marcus Aurelius and later continue to give the name Olbasa) led him to suppose that the name was taken not from the Emperor Hadrian but from some saint or bishop. This theory, to which our present inscription is clearly fatal, he later (CB. I, 284Google Scholar) amended so as to give the name Hadriana (i.e. τὰ Ἃδριανά) to the “Killanian or Milyadic” estates which he located in the upper Lysis valley and supposed to have been reorganised by Hadrian; the absence of Olbasa from the Notitiae is then explained by the supposition of a double bishopric (ὁ Ἁδριανῶν καὶ Ὀλβάσων), of which the second part only was named by Hierocles and the first part only by the copyists of the extant Notitiae. The present inscription, with its mention of Council and People, is equally destructive of this amended version also, and is indeed a warning against these arguments ex silentio.

⁷² See above, p. 73.

⁷³ This side of the stone was partly obscured by boarding, and I could not make it out to my satisfaction.

⁷⁴ Unless we prefer to write κλειδώσις for κλειδώσεις.

⁷⁵ Formerly called the Istanoz Su; the recent maps give it no name.

⁷⁶ “The site which he reached in perhaps two days from Xyline Kome must have been that of Giaour-Euren. This then must be Kormasa.” Manlius might well take two days (continentibus itineribus) from Xyline Kome to here, a distance of full six hours over difficult country, roadless in 189 B.C. (p. 326, n. 5).

⁷⁷ Manlius would need to descend into the Lysis valley at Kozluca; the country lying directly between Gâvur Ören and Hacılar is cut up by steep gullies and quite impassable for an army.

⁷⁸ Assuming that it was in fact Cormasa, not Darsa, that Manlius sacked. On Darsa see further below.

⁷⁹ Much later, in Social Basis 239, Ramsay proposed a variation of this theory which is far from being an improvement. Here he locates Darsa at Düver and Cormasa at a city on a hill near the west coast of a marshy lake into which the Istanoz Su flows; this is apparently the site at Kuşbaba above the lake of Kestel, which in CB. I, 327Google Scholar, is identified with Colbasa. (This seems to be a revival of his old idea that Cormasa and Colbasa are one and the same.) Manlius thus takes three days marching down the Taurus and on to Kuşbaba, then covers the distance from here to Düver in one day. This last day's march is utterly impossible, nor could Düver reasonably be called proxima urbs in relation to Kuşbaba. In any case, the new inscription from Boğaziçi (No. 42 above) shows that Cormasa was nowhere near the lake of Kestel. For the site at Kuşbaba see Part II.

⁸⁰ Or more strictly the line which this road used to follow before The present chaussée was constructed. On the map FOA. IX, however, his route is shown as passing further west than this road, cutting in impossible fashion (as Ramsay pointed out) across mountain ranges and valleys.

⁸¹ A chaussée along this line is now under construction but is not yet open to motor traffic.

⁸² On his map, however, Magie follows Kiepert in placing it south of Lake Kestel.

⁸³ In 1958 I climbed by an alternative route a little further to the west, but this is not passable for wheeled traffic or for the impedimenta of an army.

⁸⁴ At one point it crosses a col whose razor edge is now barely 4 m. wide, but was probably wider in antiquity.

⁸⁵ Perhaps from mere necessity; the route may well at that time have been unmitigated pine-forest. Even to-day there is no village between Aziziye and Kozluca. The expression continentibus itineribus has caused much trouble; Ramsay, in Social Basis 239Google Scholar is reduced to supposing (unnaturally, as it seems to me) that it means that Manlius made no halt for more than a night, but marched every day, throughout this portion of his journey.

⁸⁶ Not necessarily exactly; for example, a road branches from the Aziziye-Kozluca road and runs down to Akören. I have not followed this road and cannot speak for it, but it would suit Livy's narrative almost equally well.

⁸⁷ Ramsay himself supposed a conflation of two roads in the Table; see above, p. 94, n. 57. I could not myself identify the remains of Roman work in the bridge with any certainty; I noticed built into it a block carrying the ends of five lines of an inscription: - - ΩΙ, - - ΟΝ, - - ΟΝ, - - ΙΑ, - - ΟΙΣ.

⁸⁸ Or appears to say: it would, I suppose, be possible, though very awkward, to take the words Darsa proximo urbs erat as a parenthesis, and refer eam to Cormasa.

⁸⁹ I am informed by Turkish friends that Düver has in fact no connexion with duvar, “wall,” being the name of an influential derebey from whom a number of villages in Turkey have taken their name. No ancient wall is to be seen at Düver to-day.

⁹⁰ If a city of Darsa ever really existed, and is not a mere phantom in the text of Livy, it would be tempting to imagine that it was the early representative of Hadriani at Gâvur Ören. Dr. A. H. McDonald kindly informs me that the MSS. of Livy show no trace of uncertainty with regard to the reading Darsa.