¹ Journ. Roy. Geogr. Soc. XII (1842), 143 sqq. Quoted as (a) in n. 3 below.

² BCH 1877, 338–346, Annuario III (1921), 263 sqq. Hula and Szanto also passed through Caunus in 1894, but confined their activities to ‘einige Collationen’ (Sitzb. Alad. Wien cxxxii, 33). Philippson in 1904 went direct from Marmaris to Köycegiz by the west side of the lake without visiting Caunus, (Petermanns Mitt. Ergänzungsheft 183, 85–6).Google Scholar See also Robert, Ét. Anat. 503–4Google Scholar and ATL I, 499.

³ They are: (a) Hoskyn, loc. cit. (cf. Pros. Imp. Rom ². II 121, no. 519), Omitted in IGR. I checked the text in 1950; there is no doubt of the name but the lapicide went astray among the succession of similar letters and wrote at first ΛΛ in place of Μ. The stone is some 20 yards north-east of the building mentioned in (d) below, (b) Lebas-Waddington no. 512, Collignon, op. cit. 345, Maiuri, op. cit. 267, SEG II, no. 534, Repeated in the same words on a neighbouring tomb. The demotic Κυσηρεύς has not reappeared in the new inscriptions, (c) Collignon, op. cit. 346, republished here as No. 19 below, (d) Collignon, op. cit. 343, fragment of ‘dedication’ found in a ‘Hellenic’ building on the ridge north of the theatre (see p. 14), (e) Maiuri, loc. cit., SEG II no. 535, on a tomb, I have not seen (d) and (e).

⁴ Maiuri observes of this tomb: ‘nella tomba maggiore … a grossi pilastri e capitelli quadtangolari di forma inusitata, i loculi sono scavati nello spazio degli intercolonni all' altezza dei capitelli’, regarding it evidently as a finished work.

⁵ Collignon, speaking of these tombs, observes rightly (op. cit. 344: there are in fact exceptions, but the statement is true in the main) that they are arranged in two horizontal rows, the upper row temple-tombs comparable with those of Lycia, the lower row simple chambers with plain doors. Bürchner in RE s.v. ‘Kaunos’, quoting Collignon, says the tombs have two storeys like the tombs in Lycia. In fact, there is no tomb of this type a t Caunus; Bürchner's statement is presumably due to a misreading of Collignon. The error is unfortunate, as tending to confuse the question of the relation between the rock-tombs of Caunus and of Lycia; on this, see Maiuri, op. cit.

⁶ A fine specimen, but badly ruined, at B² in Fig. 3.

⁷ The descriptions given by Collignon (from Duchesne's notes) and by Maiuri go onlyas far as the point where the wall turns at right-angles to the east (J in Fig. 3); Hoskyn appears to have seen the whole wall, but says merely that the masonry is Hellenic near the port, Cyclopean on the hills.

⁸ Maiuri's 4.60 m. (cf. Collignon's ‘4 mètres au moins’) applies only to the part of the wall he is there describing.

⁹ Fancy polygonal work of this kind dates elsewhere in Asia Minor to late Hellenistic or Roman times, e.g. at Cnidus (Scranton, , Greek Walls, 166Google Scholar) and Balbura (Petersen, –Luschan, von, Reisen II, 183Google Scholar, and Pl. XXIX).

¹⁰ Ps.-Scylax 99, (ca. 350 B.C.)

¹¹ This view is the outcome of a discussion by correspondence with Dr. Scranton, for whose willing help I am very grateful.

¹² The fragmentary inscription (d) in n. 3 above was thought by Collignon to be the dedication of the temple. [τέ]κνων is more suggestive of an epitaph. (Another epitaph, No. 40, was found close to this building, where there are many re-used stones.)

¹³ The small lake is called by Collignon and Maiuri Ana Göl, but this name is not now used locally; the name Alagöl is given to the larger lake to the south-west. The lakes are correctly named on the GS map (1946) Fethiye sheet, is understood by Maiuri to mean an ‘enclosed’ or ‘sheltered’ harbour: I take it to have its normal meaning ‘closed with a chain’.

¹⁴ A sluggish and marshy stream called Sariöz flows down the eastern side of the plain into Solungur Lake, but has no connexion with the lake of Köyceǧiz and does little in the way of drainage.

¹⁵ The dalyan or fishery itself, in the form of a barrage of posts and wire netting across the river, was originally opposite the village, but in 1947–48 a new and more up-to-date structure was erected further down stream, close under the acropolis hill. For the possibility of a dalyan in antiquity see No. 20 below.

¹⁶ Fig. 10 was very kindly drawn for me by Mrs. Michael Gough from photographs taken on the hill north of the city.

¹⁷ So Maiuri: ‘una cittá eminentemente marittima’. Since the recess on the north-east, on the side opposite to the harbour, must also have been sea, the acropolis hill formed a peninsula joined to the mainland by a narrow isthmus; such sites are particularly common on the west coast of Asia Minor: cf. Lebedus, Aerae, Myonnesus, and the site (Marathesium?) near Kuşadasι.

¹⁸ Collignon, op. cit. 342, ‘le Kalbis était navigable á son entrée Les navires avaient en effet à franchir une certaine portion du fleuve avant d'entrer au port’; ib. 340, ‘le port était alimenté par le fleuve.’ ATL I, 532 n. 4, ‘the river was navigable up to Kaunos.’ Even Maiuri accepts that this was the case in Strabo's time. LS ⁹ s.v. IV translate ‘channel of entrance to a harbour’, quoting Strabo xvii, 801 and Peripl. Mar. Erythr. 37; in neither of these passages is any harbour in question, and the translation looks like a reminiscence of Strabo xiv, 651 (Caunus), which however is not quoted.

¹⁹ This is a question for the geologists. M. X. de Planhol kindly informs me that, so far as he can judge from a rapid visit, the above suggestion is possible, or even probable.

²⁰ The dalyan is opened for an hour or so several times a week to let boats through; for the rest of the time priority is given to the fish.

²¹ Strabo xiv, 651, Dio Chrys. xxxii, 92: re-emphasised by Maiuri, loc. cit.

²² Cic. de div. II, 40, 84. It is, of course, possible that the figs were called Caunian because they were shipped from Caunus, being grown elsewhere, just as the figs of the Maeander valley are to-day called Smyrna figs. Ramsay, , Asianic Elements, 128Google Scholar, even appears to suggest that the Caunian figs were these same figs from the Maeander: ‘it is to be noticed that Caunus was the harbour of export, not Smyrna as at present’. On the face of it, this seems improbable in the extreme.

²³ E.g. Imbros, , Carbasyanda, , and Pasanda, in ATL IGoogle Scholar, Pyrnos, Tarbelos, and Pasada on the maps of Kiepert and Philippson.

²⁴ Fraser-Bean, , Rhodian Peraea 57.Google Scholar

²⁵ It must be remembered that the intervening ground between Kiziltepe and Caunus was in antiquity sea; there appears no particular reason for a Caunian fort at this point. For the spelling of Carpasyanda see No. 5 below.

²⁶ Or Yangi: k and g are virtually interchangeable in the peasants' speech.

²⁷ I presume that this is the same site as that described by Cousin in BCH 1900, 43: ‘à l'endroit appelé Yangi s'étagent, du bord du lac jusqu’ au sommet de la colline qui surplombe, des ruines qui avaient été visitéés par mon ami M. Deschamps. En haut, quelques pierres taillées mêlées à d'autres; en bas, beaucoup de ruines anciennes, mais aucune ne mérite qu' on s'y arrête. Il est possible que ce soient les restes de murailles unissant la ville et le lac’. The name Yangi is evidently the same, but the hill I visited could not be said to overhang the lake, nor do I know of any site in this neighbourhood to which the term ‘ville’ could be applied. The contemptible ruins ‘en bas’ might be the medieval castle, but the rest of the description does less than justice to the site. Hula and Szanto (Sitzb. Akad. Wien cxxxii, 33) also thought poorly of Yanki: ‘das von dort [Yüksekkum, i.e. Köycegiz] besuchte Castell Yangi bot nichts Wesentliches.’

²⁸ Called Evlemez Daǧ on the 1946 GS map, but the local name is ölemez, ‘Immortal’.

²⁹ 937 m. on the GS map; my aneroid registered 865 m.

³⁰ The fort is in fact discernible (once one knows it is there) from the southern part of the lake.

^30a It is possible that Caunus later made an attempt to free herself from Hecatomnid control. It has been inferred from TAM I, 45 that Pixodarus was assisted in a war with Caunus by Xanthus, Tlos, Pinara, and Cadyanda; but the text is very fragmentary—no mention of a war is actually preserved, and the inference depends solely on the fact that Caunus is named (if indeed the name is correctly read) separately from the four Lycian cities. The conflict in question cannot be considered an established fact; Kalinka in TAM, loc. cit. implies that Herodotus' mention of the Caunians as holding ‘eine selbständige Stellung’ between Carians and Lycians supports the view that they maintained their independence of the Hecatomnids; this point is now, in view of the present Nos. 3 and 4, no longer valid.

³¹ Paton-Hicks, Inscr. of Cos 109, no. 53: For the date see the note ad loc. This ‘King of Caunus’ is surprising and rouses suspicion; for a quite different interpretation of this text see below on No. 37.

³² Diod. Sic. xx, 27 (309 B.C.): On the question whether this is the same exploit recorded by Polyaenus III, 16 see below n. 34.

³³ Livy xxxiii, 20: illam alterant curam non omiserunt (sc. Rhodii), tuendae libertatis civitatum sociarum Ptolemaei, quibus bellum ab Antiocho imminebat. nam alias auxiliis juverunt, alias providendo ac praemonendo conatus hostis; causaque libertatis fuerunt Cauniis, Myndiis, Halicarnassensibus Samiis que.

³⁴ Holleaux, (BCH 1893, 64Google Scholar = Ét. d'Ép. et d'Hist. Gr. I, 420) referred this stratagem to the Ptolemaic capture in 309 B.C., observing that Polyaenus' account may explain Diodorus' words concerning the capture of the Persicon. But the Persicon (one of the two parts of the acropolis hill, probably the lower western part) is close above the city; the impression one receives is rather that the soldiers in Polyaenus were brought down to the city from the long wall running over the hill to the north. Indeed, since Philocles was attacking by land it seems certain that this must be the wall in question; the acropolis hill was in antiquity almost entirely surrounded by sea. In 309, on the other hand, Ptolemy came by sea, and even then did not attack the acropoleis until he had taken the city (n. 32 above). This last detail is in fact fatal to the identity of the two events; how could the traitorous sitophylaces distribute rations to the Caunian defenders when the πόλις was in the enemy's hands? Unless the details of Polyaenus' narrative be rejected as mythical, it seems that Philocles must have taken Caunus from Lysimachus after 285 B.C. If Beloch's belief that Philocles was originally Demetrius' general is correct (though recent opinion tends to be unfavourable), this view becomes a necessity. If Holleaux is right, the occasion after 286 B.C. on which Caunus became Ptolemaic is quite unrecorded.—I have suggested that the Persicon was the lower, western part of the acropolis hill. I suppose that Ptolemy, having the lower city in his hands, would proceed to attack the main portion of the acropolis immediately above; when this fell, the defenders of the western hill, isolated and no longer defensible, might very naturally surrender of their own accord.

³⁵ xxx, 31: more accurately, he records that the Rhodian envoy to the Senate in 166 B.C. made the statement. Its truth is doubted by Magie, , Roman Rule, 880, n. 73Google Scholar; on this, and on the whole question of the Rhodian acquisition of Caunus, see Fraser–Bean, Rhodian Peraea.

³⁶ In any case, there could hardly be any question after Apamea of their buying it ‘from the generals of Ptolemy’.

³⁷ Cic. ad Q.F. 1, 1, 33. Strabo xiv, 652 seems also to refer to this occasion.

³⁸ J. W. Cohoon on Dio Chrys. xxxi, 124 infers from Dio's words there that it was rejected; this overlooks the period of freedom that intervened; see below.

³⁹ A likely occasion for the loss of Caunus would be the Rhodian resistance to Cassius in 42 B.C., as is suggested by Jones, , CERP, 77.Google Scholar

⁴⁰ Dio Chrys. xxxi, 125 (early years of Vespasian): (sc. the Caunians) (sc. the Rhodians) This period of freedom and its subsequent loss are now confirmed by the new inscriptions Nos. 37 and 38.

⁴¹ A search revealed another piece of the lintel, but this bore no writing.

⁴² Reading from left to right in both cases and restoring Μ rather than Ν for the first letter of l. 2 at Tasyaka. If this is correct, the left-to-right reading is established, contrary to the suggestion implied in Robert's n. 2 on p. 18.

⁴³ I have a squeeze of the Taşyaka inscription, which is at the disposal of anyone interested.

⁴⁴ Robert, , Sinuri I, 99Google Scholar, no. 76, cf. Ét. Anat. 572.

⁴⁵ is the obvious suggestion, and it is in fact easy to imagine that one recognises on the squeeze of No. 4 the traces of a round letter in the second place and of two upright strokes in the fourth place; but no reliance can be placed on this.

⁴⁶ Was the dedication for some reason displeasing to the dynasts? Is it conceivable, if the erased word was in fact that Mausolus' anti-Athenian policy of 357 B.C. and following years was the cause? I note, if only as a coincidence, an exactly similar erasure, though of very different date, at Assus, (Papers of the Amer. Sch. at Athens I, 58Google Scholar, no. XXIX): The editor restores for which, if the drawing is reliable, there is certainly not room.

⁴⁷ There seems no reason at all why the name of the deity should be written smaller than the rest, but I see no other likely way of filling the erasure. If the Caunians had in error added a title, such as βασιλέα, which the Hecatomnids never used, the erasure would be explained; but the position of the word at the end is unnatural, nor can I find a likely title of the required length. The ethnic Μυλασέα is even less satis-factory. The artist's signature might well be written small, but would hardly follow without interval on

⁴⁸ I saw the stone again in 1952 in a neighbouring field, and arranged to have it carried for safe keeping to the fishery.

⁴⁹ See Robert, , Ét. Ép. et Phil. 211.Google Scholar

⁵⁰ See Robert, , Ét. Anat. 486Google Scholar; it is among the commonest names at Caunus.

⁵¹ At least eleven: the inscription may be incomplete at the top, where the name of the deity may also have been written.

⁵² ATL I, Register and Gazetteer s. vv.

⁵³ Ét. Anat. 504 n. 2; cf. ATL I, 532.

⁵⁴ Strabo xiv, 651 has: In ATL I, 532 n. 4 this Pisilis is boldly identified with Pasanda. μεταξύ must apparently mean ‘between Calynda and Caunus-cum-Calbis’; the reader might well understand ‘between Caunus and the Calbis’.

⁵⁵ Collignon regards this as certain: ‘il n'y a pas de doute sur le nom antique de cette imposante forteresse; c'est à coup sûr Imbros, l'acropole des Kauniens’ (BCH 1877, 339). There are in fact remains of a small fort on the summit of the acropolis hill, shown in Fig. 20; it is remarkable that neither Collignon nor Maiuri advances these as the ruins of Strabo's φρούριον. They seem indeed to have been overlooked.

⁵⁶ In ATL the question is made to depend on the value to be attached to the epithet ‘snowy’. Since Nicander (Theriaca ad fin.) can apply the same epithet to Claros, that value may be thought to be nil.

⁵⁷ xx, 27, quoted in n. 32 above. Note that they are both neuter adjectives with the article; that is, they are forts (φρούρια). There is no room for a third φρούριου here.

⁵⁸ For the identification of Tarbelus I have no evidence to offer. There are several possibilities. (1) The hill north of the city over which runs the long wall. (2) If Imbrus is the name of the fort, Tarbelus may be Ölemez Daǧ itself. (3) Kιzιltepe is hardly attractive, as in antiquity it lay about two miles from Caunus along the coast. (4) The acropolis hill is also a possibility, if Quintus is thinking of where the man dwelt rather than where Caunus lay. (This is not inconsistent with Diodorus: may quite possibly have been on the hill Tarbelus.)

⁵⁹ Stadiasmus 251–2, among several unknown places between and Telmessus. Cf Steph. Byz. It is marked conjecturally on Kiepert's map.

⁶⁰ Compare the isolated Rhodian enclave around Daedala (Fraser, –Bean, , Rhodian Peraea 55Google Scholar).

⁶¹ Discussed by Robert, , BCH LXX (1946), 518 n. 2Google Scholar, and with reference to the present inscription Hellenica VII, 189–90. For the possibility of reading or in I. v. Magn. 59 see the discussion in Hellenica, loc. cit. Robert suggests in BCH, loc. cit., on numismatical evidence, a possible identification with the rather dubious city of Astyra, described by Stephanus as (Holstenius attractively). The site of this Carian Astyra, if it really existed, is equally unknown; but if Stephanus' Φοινίκη is Strabo's it must have been on the Loryma peninsula, in the heart of the incorporated Rhodian Peraea. This is no place for a Caunian deme.—Tod, in JHS LXXII (1952), 50Google Scholar, referring to Robert's mention of the present Nos. 5 and 6 in Hellenica, loc. cit., says: ‘two inscriptions from Caunus refer to Ptolemais in Caria (Lebedus)’. This was written without knowledge of the actual text of the inscriptions; Lebedus is, of course, an even more absurd situation for a deme of Caunus than the Loryma peninsula.

⁶² I am indebted to Prof. L. Robert for the decipherment from my squeeze of these two lines; a subsequent charcoal reading of the stone entirely confirmed it.

⁶³ On the other hand, Sultaniye is close to the shore of the lake, and it is perfectly possible that the stone may have been transported from Dalyan. I could learn of no other ancient remains at Sultaniye.

⁶⁴ I quote a few examples. In IG II², 2358, a list of eranistae, out of ninety-two names four have demotics added. In IG XII, 1, 1442 = SGDI 3761, among twenty-nine names one demotic is added. In Fraser-Bean, Rhodian Peraea 30, no. 17, out of twenty-five names wholly or partially legible, four have a demotic. In IG XII.1. 4 about one in four has a demotic. The only case I have found which seems to afford a parallel to the present is IG II², 1335 (Piraeus), a list of eranistae, of whom thirty-six have a demotic, thirteen have an ethnic, and four have no toponymic.

⁶⁵ GIBM 795 = SGDI 3510 = Michel 1005. Sums range from 300 to 5 dr., but the bottom is missing.

⁶⁶ I take it as certain that No. 6 is considerably later than No. 5. There seem to be no indications of early date in the style of the script. Many of the letters have distinct apices, particularly tan, which has a noticeable downward turn at each end of the cross-stroke.

⁶⁷ See Fraser, –Bean, , Rhodian Peraea 127–8.Google Scholar

⁶⁸ This need not necessarily mean that the present inscription dates to a period of Rhodian control over Caunus; the demes, once detached, might have continued independent after that control came to an end.

⁶⁹ Such is not the case in the parallel document from Cnidus, in which the ten foreigners comprise a Libyan, a Phrygian, a Thracian, two natives of Myndus and one each of Aradus, Soli, Selge, Seleucia, and Samos.

⁷⁰ Not included among Stephanus' twenty-five places of the name.

⁷¹ See Sterrett, , Epigraphical Journey, p. 26Google Scholar, Robert, , Ét. Anat. 332 n. 1Google Scholar; in Hierocles 689, 2 Apollonias is next to Tabae.

⁷² Can it be a simple genitive in place of the usual ἐκ + genitive? Is there a parallel for such a phenomenon?

⁷³ SGDI, 4275, found at Muǧla.

⁷⁴ For the date see SGDI, loc. cit.

⁷⁵ VIII, 87–8, where they furnish one ship to the Persian fleet at Salamis. It is further clear from I, 172 that they were neighbours of Caunus.

⁷⁶ Until 425 B.C., when the Caunian tribute was raised to ten talents.

⁷⁷ P. Cair. Zen. 59341 a, b (247 B.C.), cf. 59340. Zeno was himself a Caunian and had family connections with Calynda (59341 c).

⁷⁸ BCH XLV (1921), 6, where it appears in the singular form Jones, , CERP 50Google Scholar, says Calynda had been absorbed by Caunus by the end of the third century, but the theori would hardly visit a town which was actually incorporated in another city.

⁷⁹ Polyb. xxxi, 5. Our Calyndian has the Rhodian name Δαμόνικος, but the significance of this is dubious; P. Cair. Zen. 59341 b is addressed to a certain Damonicus, who may also have been a Calyndian.

⁸⁰ BMC Lycia xlvi, 48.

⁸¹ Hdt. I, 172; Strabo xiv, 651; Pliny, , NH V, 103Google Scholar; Ptolemy, V, 3, 2. Robert, , Ét Anat. 493 n. 2Google Scholar says it is ‘tout à fait incertain’; this is perhaps unduly pessimistic.

⁸² I visited this spot in 1946 and again in 1950, and was much impressed by its suitability as the site of Calynda. It is now easily accessible, as the new Muǧla–Fethiye road passes the foot of the hill. It is called Kozpinar, from the abundant spring which issues by a coffee-house at the roadside, 4 km. from Dalaman çiftliǧi towards Fethiye. The site as a whole is more impressive than I had realised from the description in JHS. At the top of the hill is the enclosure described by Davies; the style of its wall may be seen in Fig. 22. On the south-east side are two towers of more regular masonry. This wall is now surmounted by another of medieval date. In the north-east corner of this enclosure is a tower or small fort 12.25 by 8.00 m. in area, of good regular Hellenistic masonry, with vertical draft-lines at the corners. In the north-east side of this tower is a door 1·06 m. wide, with corbelled arch (Fig. 23); the wall is here 1·57 m. thick. But a much greater area than this was walled in. Lower down the hillside considerable stretches of a good polygonal ring-wall are preserved, especially on the east side, where it stands to a height of 2–3 m. for a length of over 200 yards; the thickness is 2·10 m. This wall (Fig. 24), though its blocks are more or less polygonal, shows a marked tendency to regular coursing; and ‘coursed polygonal’ work seems confined to the early Hellenistic period. (Scranton, , Greek Walls, 52Google Scholar, 69, 165–6. Dr. Scranton, in a private letter, kindly informs me that this wall at Kozpιnar does in fact fall within the category of ‘coursed polygonal’ as he uses the term in his work.) On the west side of the hill, at a lower level, is another stretch of wall 2·50 m. thick, rather more massive and perhaps of earlier date, but apparently part of the same ring-wall. On the north is a vertical precipice in which are the pigeon-hole tombs mentioned by Davies; even above this precipice fragments of wall are visible, as they are also on the south. The ring-wall, after encircling the hill on the west, runs up to form an angle on the south-east at the top of the hill; in this angle is the inner enclosure described above. The whole area thus enclosed is strewn with immense quantities of building-blocks, many of them carefully squared; numerous foundations of houses, rock-cut and otherwise, are to be seen, and on the east side is a doorway cut in the natural rock. The peasants have dug in several places for treasure, and great tales were told by my guide of the gold and silver that had been found on the hill; about 1939 or 1940 a gold dagger is said to have been dug up, having a female figure on one side and on the other a boar and a snake. My informant spoke only from hearsay; the dagger was apparently sold to an antique-dealer. I saw no remains of public buildings, and in particular nothing that was evidently of Roman date.—In the west face of a low hill called Aladaǧ near Dalaman Çiftliǧi is a singlerock-cut temple-tomb similar to those at Caunus; it is uninscribed, but presumably belongs to the city at Kozpιnar.

⁸³ Among half a dozen coins brought to me at Dalaman was a bronze of Cassander; the others were rubbish.

⁸⁴ 60 stades in Strabo, 6 miles = 9·6 km. on the modern map. Pliny, NH V, 103Google Scholar has: Crya fugitivorum, flumen Axon, oppidum Calynda, amnis Indus. Between Crya on the gulf of Fethiye and the Kargιn Çayι no stream at all enters the sea; since there is no ancient site between the Kargιn Çayι and the Dalaman river (amnis Indus), we may well suppose that Calynda was on the Axon; if so, this testimony and that of Strabo combined lead precisely to the site at Kozpιnar.

⁸⁵ See Jones, , CERP, 107–8Google Scholar: Crya, Calynda, and Symbra are all in the same case, and were probably not far removed from each other geographically.

⁸⁶ A certain is a member of a Rhodian χοινόν including numerous foreigners in IG XII.1. 127, l. 77 (ca. first century B.C.).

⁸⁷ Another unusual sign for drachma in No. 22 below.

⁸⁸ Two small points in the text: (1) in l. 8, should perhaps be rather cf. No. 8, l. 10; (2) in l. 10, for is anomalous: on further examination of the squeeze I fancy the lapicide wrote nu and corrected it to kappa.

⁸⁹ In general, the stones at Caunus have been very little disturbed; there has been no wholesale removal of blocks for building purposes.

⁹⁰ IG XII.i. 43, 47, 66, 846 (= SGDI 3779, 3792, 3802, 4200), Blinkenberg, Lindos II 197 g, 246, 281 a.

⁹¹ Lindos II 232, 235, 236.

⁹² IG XII. 3. 103 (Nisyrus), Lindos II 234.

⁹³ Holleaux, in Rev. Phil. XVII (1893), 176Google Scholar (= Ét. d'Ép. et d'Hist. Gr. I, 387), arguing from recurrence of names, concluded that the Epicharmi might be working together about the beginning or towards the middle of the first century B.C. This seems certainly too late, and inspires doubts as to the validity of this method of argumentation.

⁹⁴ The final upsilon was visible in 1946 on a separate fragment, since lost.

⁹⁵ The victory recorded in IG II² 2971 (late fourth century) is now taken to refer to the Delia at Tanagra: see the note ad loc. and in Syll. ³ 319. For an epigraphical reference to the hippodrome see Roussel, , Délos Col. Ath. 157.Google Scholar

⁹⁶ Roussel, , Délos Col. Ath. 208–210Google Scholar, cf. Laidlaw, , Hist. of Delos, 48.Google Scholar

⁹⁷ List in Magie, , Roman Rule, 1613.Google Scholar

⁹⁸ It seems hardly possible that we are meant to understand two separate festivals, at each of which Polyxenus won one victory.

⁹⁹ As Dr. P. M. Fraser pointed out to me.

¹⁰⁰ For the general absence of foreigners, at least from distant cities, at Caunus see above p. 25.

¹⁰¹ For example: IG XIV, 1771, ibid. 1911, in fr(ontem) p(edes) XVI., in ag(rum) p(edes) XVII; Judeich, Altertümer von Hierapolis no. 116, cf. ibid. nos. 159, 262, IG XII. 1. 736.

¹⁰² The Caunian formulae are: (1) genitive alone (Nos. 18, 19, 43); (2) ( etc.) (Nos. 41², 44) 45 and (b) in n. 3); (3) (Nos. 18a, 39, 40 and (e) in n. 3) Nos. 41¹ and 42 are somewhat different.

¹⁰³ I should add that this wall is clearly much older than the date of the inscription, so that we should need to suppose Euphanes contributed to a repair, of which I saw no evident signs.

¹⁰⁴ LS ⁹ s.v., quoting IG XII. 3. 1120 (Melos). Cf. SGDI 3802 = IG XII. 1. 66. On the other hand, in IG XII. 1. 1029 (= SGDI 4317) (Carpathus), it has evidently the usual meaning ‘nurse, foster-mother’; Φύσις5 is clearly a slave-name. The difference of spelling may correspond to a difference of meaning, since this particular interchange of vowels is otherwise unknown in Rhodian, epigraphy (SGDI 4317Google Scholar, cf. Thumb-Kieckers, , Handb. der Gr. Dialekte, 187).Google Scholar