¹ Scranton, Greek Walls 52, 69, 165Google Scholar; cf. JHS. LXVII (1947), 130Google Scholar, and LXXIII (1953), 26, n. 82.

² This, rather than Arvalı, is apparently the true form of the name.

³ BSA. LI (1956), 143Google Scholar.

⁴ AJA. IV (1888), 271Google Scholar, No. 2 = IGR. III, 397Google Scholar. I have not visited this cemetery; but IGR. III, 398Google Scholar, is now at the school in Kestel, much damaged and with the left part missing. It was recently taken from a bridge half-way on the road from Kestel to Kuşbaba.

⁵ The belief that writing and other marks on ancient stones indicate the position of buried treasure is still widespread in Anatolia, but more deeply rooted at Kestel than anywhere else in my experience. Time and again I was entreated to point out the right places to dig, and was offered a share of the treasure when it was found. In particular, dowel-holes with lead-channel are supposed to point in the direction of the hidden gold.

⁶ The term κοράσιον is also commonly, though not invariably, used of a slave-girl.

⁷ In this inscription the editors understand that a free man dedicated his two daughters, τὰς ἑαυτοῦ παῑδıς (sic), as hierodules of Meter Oreia; but it is surely more probable that παῖς here means “slave”.

⁸ A dedication at Kağlıcık to Ζεὺς Πλούτων (Smith, JHS. VIII, 249Google Scholar, No. 27 = Ramsay, CB. I, No. 110Google Scholar) has evidently no connexion with our sanctuary.

⁹ I was shown a bronze of Codrula at Zivint (below, p. 69, n. 57). For another possibility concerning Codrula see below, p. 79–80.

¹⁰ The photograph was given to me by my friend Ibrahim Şadi Balaban; his informant was not willing to reveal the exact location of the relief.

¹¹ No doubt in reality the base of a tomb or other monument: cf. Part I, p. 92.

¹² The colony was founded about 25 B.C., but our inscription cannot be earlier than the 2nd century. It looks as if πρώτη and primus may denote precedence in rank rather than in time.

¹³ The difference between the sums, three and sixteen myriads of denaria, might well correspond to the difference in size between Comama and Termessus, each being calculated to provide one denarius per annum to each citizen. But this cannot, of course, be relied upon.

¹⁴ The honorand repaired the baths at Cretopolis and this is thought worthy of record in an inscription of Comama. That Cretopolis was, in fact, his “second fatherland” (line 11), Comama being the first, is an evident possibility.

¹⁵ Ritter, Erdkunde XIX, 706Google Scholar, adding that Leake took this to be the site of Lysinoe. But in the passage quoted (Asia Minor 151) Leake conjectures, rightly as it now turns out, that Lysinoe (Lysinia) was on the Lake of Burdur.

¹⁶ Rott, Kleinas. Denkmäler 21, and 360, No. 51Google Scholar.

¹⁷ Rott also speaks only of a “Stätte”, not a “Stadt”.

¹⁸ In any case, Incirlihan is to be carefully distinguished from Incirli, the site of Hyia south of Bucak: see below, p. 80.

¹⁹ On this question see BMC Lycia xcix, Jones, CERP. 126Google Scholar.

²⁰ If, on the other hand, it is an error, it is the only one on the stone.

²¹ The grid in Fig. 2 may not be entirely accurate, but the general picture is not, I think, seriously distorted.

²² An extraordinary feature is the total omission of Sagalassus. Ptolemy has a Sagalassus in Lycia, close to Corydalla and Rhodiapolis; but this, as was pointed out by Jones, CERP. 406Google Scholar, n. 19, is surely an error for Acalissus. Some of Ptolemy's towns, notably Uranopolis, Milyas and Corbasa, are otherwise totally unknown.

²³ It is not, however, unique on the map of Turkey; there is a Fuğla Tepe in the Troad.

^23a For the rather unexpected corruption of Π to Σ compare (almost certainly) Σάσανδα for Πάσανδα in Diod. Sic. XIV 79Google Scholar; see Robert, Ét. Anat. 504Google Scholar, n 2, ATL. I, 532Google Scholar, n.3.

²⁴ Unlike the cities of the Lysis valley (except Olbasa: see Part I, p. 78), those on the east side of the central mountain-range are mostly fortified in the usual way; see above Kuşbaba and Kaynar Kalesi, and below Melli, Karaot (Sia) and Incirli (Hyia). On the other hand, Comama, Andeda and Sibidunda seem not to have been defended by any fortification-wall.

²⁵ My guide said he had seen one or two words of writing on these steps, but we could not find them. Possibly casual marks on the rock had been mistaken for writing, as often happens.

²⁶ Though Kakasbos occurs at Pogla; see No. 107 below.

²⁷ Apart from one or two at Termessus. Without entering into the question of what territory was properly included in the Milyas (on which see most recently Magie, Roman Rule 761–2Google Scholar), I use the term in the sense defined by Strabo 631, that is approximately the area covered by the present article. See Part I, p. 67.

²⁸ Robert, L.REG. LXIX (1956), 159Google Scholar, Bull. Epigr. No. 244, notes that “l'épigraphie funéraire de Myrina montre des formes lunaires dès l'époque Hellénistique”. None are cited by Welles, Roy. Corr. li.Google Scholar I add some other early examples which I have noted. In iv B.C.: Inscr. Lindos No. 41 (С), with Blinkenberg's note; and see IOSPE. I ¹, p. 40Google Scholar. In ii B.C.: Sardis VII, Nos. 21 (ϵ, С), 106 (С,ω, if correctly dated), and in northern Asia Minor, Stud Pont. III, p. 118Google Scholar. In i B.C.: I.v. Priene 142 (ω), Sardis VII, Nos. 125–7 (ϵ, С, ω). It seems that С made its appearance in Asia rather earlier than ϵ and ω.

²⁹ I am at a loss what to make of the peculiar second letter in line 19. In Pamphylian ψ is some kind of sibilant, but whether it may be so here also I do not know, as I cannot recognise the word to which it belongs.

^29a The absence of any texts of similar date from the neighbourhood makes a decision on these grounds highly precarious. I may recall the case of the Araxa inscription that I published in JHS. LXVIII (1948), 46Google Scholar, No. 11, in which the general quality of the script is not unlike the present. Opinions as to its date have varied considerably, but have tended on the whole to favour the early 2nd century.

³⁰ For this normal practice, c.f. IG. XII, 5, 840Google Scholar (Tenos): .

³¹ They might, for example, be part of the word ἀποσταλῶσιν. But the possibilities are hardly numerous.

³² would be possible, but not more than a possibility.

³³ Possibly the commune Milyadum mentioned by Cicero, Verr. I, 95Google Scholar; see Jones, CERP. 143–4Google Scholar.

^33a It is issued by “the people of Pogla” in line 29, by the “city and magistrates” in line 25.

³⁴ Prof. A. H. M. Jones, when I first showed him the text of No. 103, suggested to me that δῆμος may here denote a community of lower grade than a city; Hierocles, at a much later date, records a number.of δῆμοι in these parts, including Pogla itself (δήμου Σὠκλα). The commune Milyadum would then be a κοινὸν of these δῆμοι, and the old view could be maintained. But I feel sure that this cannot be right: in line 25 the eta is clear, necessitating τῆ[ι πόλει], and city-status must be admitted.

³⁵ The possibility of reading merely ἒτεσιν πολ[λοῑς] in No. 104, line 4, does not seem to have been considered.

³⁶ TAM. III, 2Google Scholar; OGI. 751 = Welles Roy. Corr. No. 54; see Jones, CERP. 131Google Scholar, Greek City 46, Magie, Roman Rule 264Google Scholar.

³⁷ Comama, Andeda, Verbe and Sibidunda; see the descriptions of these above and below.

³⁸ For the eagle on funeral monuments, cf. Sterrett, WE. 26Google Scholar, Λούκιος ἀνέστησε Τήλεφον … [καὶ] ἑαυτὸν ἀετὸν καὶ … τὸν π[ατέρα] ἀετὸν τειμῆς Χάριν ἀετὸν τειμῆς Χάριν discussed by Robert, L., Et. Anat. 394Google Scholar. Ramsay, Studies in the Eastern Provinces 278Google Scholar understood ἀετὸς here to mean “pediment”, but Robert is surely right in taking it to mean “eagle”. The eagle is common on funerary reliefs; sculpture in the round, as in the present case, is more exceptional. For a spreadeagle on a tomb at Cormasa (Eğneş) see Part I, p. 92.

³⁹ Ramsay, AM X (1885), 337Google Scholar, Άνδηδέων [;τῇ β];ουλῇ καὶ τῷ δήμῳ; Woodward, BSA. XVI (1909–1910), 123Google Scholar, No. 18, citizens of Andeda victors in the diaulos.

⁴⁰ Jones, CERP. 417Google Scholar, n. 37, observes that Hierocles' ΣΙΝΔΑΥΝΔΑ ΜΥΟΔΙΑ “perhaps represent Sibidunda and Andeda; the first seems to be a conflation of the two names”; but it appears rather to represent Σιβίδουνδα alone. Μυοδία is obscure and can hardly conceal Andeda.

⁴¹ I was shown at Zivint in 1958 a handsome specimen of the type BMC Lycia Pl. XXX, 12.

⁴² AJA. IV, 14Google Scholar, No. 6, at Andeda, and JHS. VIII, 254Google Scholar, Nos. 35 and 36, at Zivint.

⁴³ AM. X, 338, at AndedaGoogle Scholar.

⁴⁴ I examined the stone carefully and was satisfied that it has ΠΛ, not ΠΑ.

⁴⁵ I could learn of no site on the plain itself; it appears that the ancient towns, like the modern villages, were ranged along its edges at the foot of the bordering hills—a very natural arrangement.

⁴⁶ Inscriptions have been found at Garipçe, but it does not seem to have been an ancient site.

⁴⁷ They are of the usual Roman types ubiquitous in this region.

⁴⁸ It may be thought surprising that the territory of the little town at Bozburun should extend all across the plain to Belen. The boundary-stone may, of course, not be in situ; but wherever it stood, it is clear from the relative positions of the names on the stone that Pl— lay somewhere in the direction of Bozburun.

⁴⁹ To these another was added in BSA. XVII, 209Google Scholar, No. 6.

⁵⁰ The sceptic will note, however, that Woodward's inscriptions were mostly in cemeteries—three of them forty minutes distant—and that no new ones seem to have come to light in the forty-eight years since his visit. Alternatively, of course, I may have failed to find them.

⁵¹ This sounds like the “cemetery 40 minutes east of Yerten” in which Woodward found his No. 7; but the present stone appeared to have been there a long time, and I saw no sign of his stone.

⁵² Apart from corrupted forms in Hierocles and the Notitiae which might represent it: see Ramsay, CB. II, 753–4Google Scholar, and above, p. 65, n. 40.

⁵³ It is included in the Phrygian volume of BMC.; cf. Imhoof-Blumer, Kleinas. Münzen 289Google Scholar, and Ramsay loc. cit., followed by Ruge, in RE. s.v. (1921)Google Scholar.

⁵⁴ Head, BMC. Phrygia xcivGoogle Scholar; Jones, CERP 417Google Scholar, n. 37; Robert, Hellenica III, 68Google Scholar, n. 1.

⁵⁵ Klio XXIII, 248Google Scholar. He says, indeed, that Zivint is “proved” to be Sibidunda, but in what the proof consists is not clear.

⁵⁶ e.g. with Jones loc. cit. and Robert loc. cit.

⁵⁷ At Zivint also I was shown a bronze of Sibidunda with reverse type similar to BMC. Phrygia No. 9 and Imhoof-Blumer, Kleinas. Münzen 289Google Scholar, No. 1, but of Caracalla. This may be added to the total of evidence, though in itself, of course, it proves nothing (except that Sibidunda was probably somewhere in this region); I was shown at the same time bronzes of Attaleia, Isinda, Andeda and Codrula.

⁵⁸ This stone is said to be a single block now re-used in a fountain in Zivint village; it was pointed out to me later, but only a small part of it is visible.

⁵⁹ For the “Baufinanzkommission” under this name at Didyma see now Didyma II (1958), p. 62Google Scholar.

⁶⁰ For Mormonda, see most recently Bean, Anadolu Araştırmaları I ( = Jahrb. Kleinas. Forsch. III), p. 51Google Scholar.

⁶¹ .

⁶² loc. cit.: .

⁶³ The confusion of αὐτόν and ἑαυτόν is quite common, especially in the Roman period. In JHS. LXVIII (1948), 42Google Scholar, No. 2, τᾗ ἑαυτοῦ γυναικἰ…φιλοστοργίας καί εὐνοίας ἒνεκεν τῆς εἰς ἑατήν (Arsada i B.C.), I supposed an error of gender; I now believe rather that ἑατήν = αὐτήν. Similarly in TAM. II, 148Google Scholar, Μηνοδώρα…τὸν ἑαυτῆς ἄνδρα… φιλοστοργίας ἃνεκεν τῆς εἰς ἑαυτόν, where Kalinka writes ἑαυτ[ή]ν. See also Meisterhans-Schwyzer, Grammatik p. 154Google Scholar and No. 128 below.

^63a IGR. IV, 894–5Google Scholar; see Jones, CERP. 396Google Scholar, n. 77. For the same family at Attaleia see Belleten XXII (1958)Google Scholar, Nos. 11, 13, 14, 15, 16.

⁶⁴ A road, just passable for a jeep, leads from Dağ Nahiyesi to Karaot.

⁶⁵ Bérard speaks strangely of the site as on a summit of the range with a wide view as far as Aspendus. He calls it Duchemé, i.e. Döşeme, but this name is not now used; Döşemealtı is away to the south, beyond the Çubuk Boğazı pass.

⁶⁶ Ramsay had previously regarded it (HG 416) as a corruption of δήμου Ἴσβα. In his posthumous Social Basis 69 it is surprisingly explained as “the property of Mousis”. As we learn from Anderson's Foreword that this part of the book was left by Ramsay “complete in paged proof”, it appears that having found the true interpretation he subsequently abandoned it. Or perhaps he had forgotten it. In REG loc. cit. he says that Osia is well known from the Byzantine lists, of which the texts are unfortunately not available to me here.

⁶⁷ Published by me in Belleten XXII (1958), 39, No. 30Google Scholar.

⁶⁸ Ptolemy V, 12, 10, records a place Σίαι in Armenia. While speaking of Hierocles, I may perhaps take the opportunity here to put forward a suggestion which occurred to me too late for inclusion in my article on the Cibyratis and Caralitis in BSA. LI, 136 sqqGoogle Scholar. Hierocles' list of Lycian cities ends, after Balbura, with the enigmatic entry Κομιστάραος. Ramsay's interpretation of this as κώμη Μάσταυρα found some favour for a time; then Meritt, , in AJP. 1937, 385 sqq.Google Scholar, proposed to read it as Κομίστρατος and to recognise this town in the Athenian assessment list for 425 B.C. [Κουίστρ]ατος [ἐν Λυκία]ι. This view was attacked by Robert, in Ét. Ép. et Phil. 245 sqqGoogle Scholar. and was withdrawn in ATL. II, 86Google Scholar. Surely the solution is supplied by the inscription I published in BSA. loc. cit., No. 59. The stone is at Kozağacı and reveals the name of that interesting site as Toriaion or something similar (Τοριαειτῶν ή πεντακωμία). Should we not read in Hierocles κώμης Τοριάου or the like? The place is named immediately after Balbura and Kozağacı is known to have belonged to that city (BSA. loc. cit.); and its status as a κώμη is confirmed by the term πεντακωμία. In connexion with this same inscription, I should like also to refer to a suggestion recently made by L. Robert, before the weight of his reputation should mislead scholars into supposing it probable. The stone has the words (lines 8–9) Τοριαειτῶν ν ἡπετακωμία with the letters πεντ and κωμ in an erasure; I suggested that the original text was ἡ κολωνία (see my commentary ad loc.). Robert, in a critique of my article (REG. LXXII (1959)Google Scholar, Bull. Epigr. No. 415), expresses the opinion that the photograph of my squeeze shows an erasure in line 8 only, and suggests that the stone had originally τετρακωμία. By what right does he prefer his impression concerning the squeeze to my observations made in front of the stone? The facts concerning the erasure are as I stated them, and M. Robert was not justified in questioning them. But his proposed τετρακωμία is quite unacceptable for another reason. The words following Τοριαειτῶν ἡ as far as the end of the inscription are written in a different script, rougher and more widely spaced, with different letter-forms. If M. Robert had not observed this in his examination of the photograph, he might still have known it from the mention in my commentary. Now the letters κωμια in line 9 are in this inferior script, with alpha, mu and omega quite unlike those of the main text; they belong therefore to the addition, not to the original inscription. The suggestion τετρακωμία is irresponsible and wrong, and ought not to have been made.

⁶⁹ Called by earlier travellers Milli or Milli Başköy. It now form a tiny nahiye attached to Bucak.

⁷⁰ Probably owing to the peculiarly hard and gritty soil, almost like decomposed flint.

⁷¹ BCH. XVI (1892), 436 sqq.Google Scholar, Nos. 70–5.

⁷² His No. 73 ( = IGR. III, 388Google Scholar). The first line of this was wrongly copied by him: in place of [Κ]όμ[μοδον], is certain. I saw also two of the three fragments comprising his No. 76 ( = IGR. III, 385Google Scholar) and found that this, too, is wrongly copied: the stone has not but ; the building was dedicated to Antoninus Pius. Bérard's texts in general are none too reliable; cf. Belleten XVIII (1954), 482, No. 14Google Scholar.

⁷³ My photograph of this tomb was unfortunately damaged in the printing; such as it is, I shall be pleased to send it to anyone who is interested.

⁷⁴ e.g. by Kiepert (who had, indeed, previously proposed the identification on the strength of the name Milli, : FOA. IX, text p. 6, n. 82Google Scholar). Recent writers have been somewhat taciturn; e.g. Jones and Magie, in their accounts of the region Milyas, make no mention of the city. Similarly RE. s.v. “Milye”.

⁷⁵ Melli has no meaning in Turkish and Milli no probable meaning; millî means “national”. It may be added that Ptolemy's location (see Fig. 2) agrees quite well with the position of Melli.

⁷⁶ No exception need be made of the inscription published in Mon. Linc. XXIX (1933), 679Google Scholar, and mentioned by Jones, CERP. 556Google Scholar (fine payable τῷ Μι(λύ)ων…). It comes from the region of Phaselis, which is far removed from Melli.

⁷⁷ See the discussion in Magie, Roman Rule 761Google Scholar.

⁷⁸ Such unenclosed names are mostly those of rivers and tribes; Milyas is the only such name in this part of the world. Yet its position on Ptolemy's map agrees well with his figures for the city of Milyas and very poorly with his region of Milyas.

⁷⁹ Miss Levick saw this stone, but owing to its awkward position was unable to read the name.

⁸⁰ It is not, I suppose, excluded that the name should be Via (ὁ δῆμος Οὐιηνῶν, equally unknown), but this is much less likely; see above, p. 75, in connexion with ὁ δῆμος ὁ Σιηνῶν.

⁸¹ A mutilated inscription at Kağlıcık, (Sterrett EJ. 82Google Scholar = Ramsay, CB. I, No. 119Google Scholar) appears to be an ex-voto to Artemis: [Ἀρτ]έμιδι [εὐχήν].