^{page 171 note 1} The outlines of the sketch map Plate 37 are drawn from Spratt's map in Archaeologia XLIX at p. 358. Additional sites have been inserted; and a few of the locations have been readjusted slightly to fit our observations better, without, however, any guarantee of greater general accuracy. Turkish place-names are given in the official form and spelling, and those rendered wholly in Turkish are all in current use at the present day. Names no longer in official use which are to be identified with modern settlements marked on the sketch map are Avlana (=Mesudiye), Batir (=Hizirşah), and Elaki, Alleyadah, Alaköy, Eli(=Reşadiye).

The first section of this article has been written by Cook, the second by Bean, and the third jointly. The drawings on Figs. 3 and 4 have been done out by Miss E. A. B. Petty, those on Figs, 1 and 5 and Plate 37 by Mrs. J. M. Cook.

Abbreviations.

BMI = The Collection of Ancient Greek Inscriptions in the British Museum.

SGDI = Collitz and Bechtel, Sammlung der griechischen Dialekt-Inschriften.

ATL = Meritt, Wade-Gery and McGregor, The Athenian Tribute Lists.

Michel = Michel, Recueil d'Inscriptions grecques.

^{page 171 note 2} This description is based where possible on observations made in the course of journeys in the Cnidia in 1949 and 1950, and supplemented at points by reference to the accounts of journeys by Spratt (Archaeologia XLIX 345 ff.), Newton (Halicarnassus, Cnidus and Branchidae), and Maiuri (Annuario IV–V 397 ff.). Especially valuable information about remains and reported discoveries in the vicinity of Datça is contained in an unpublished report rendered by Dawkins after a visit of investigation in 1907; by his courtesy and that of the Managing Committee of the British School at Athens we have drawn freely on this report. For the geology of the Cnidia cf. Philippson, Reisen u. Forschungen im westlichen Kleinasien V 71 ff.

^{pagr 171 note 3} Spratt, Archaeologia XLIX 356 f. His identification with Bybassus has been followed by Kiepert and Philippson. Spratt also refers to ruins on the Hacilar islets which shelter the bay, and lays down on his map mediaeval ruins on an islet farther to the west.

^{page 173 note 4} Annuario IV–V 403, fig. 8.

^{page 173 note 5} Fragments of Doric limestone column drums with twenty flutes (one piece with diameter 0·59 m.) and a Doric frieze block 0·4 m. high comprising a triglyph and a metope; the proportions imply a building of Hellenistic or Roman date. Another Doric column drum of diameter 0·47 m., also with twenty flutes, lies outside the enclosure on the south, and pieces of other Doric columns were remarked by Maiuri (Ann. IV–V 403 f.).

^{page 173 note 6} Ann. IV–V 404 f. (fig. 10 upside down).

^{page 173 note 7} Ht. of figured scene 0·38 m. The male figure is ithyphallic; his left hand grips the woman's chin, while his right seems to be held above the penis. The woman has a short retroussé nose and full chin; her right forearm appears to spring from just under the breast, and the indistinct outlines of the body suggest a chiton drawn up by the left hand on her flank. The style verges on the ripe archaic, perhaps rather after the middle of the sixth century.

^{page 173 note 8} Op. cit. 356; Maiuri Ann. IV–V 402. In Fig. 1a the distances are paced but the bearings of all walls were plotted by compass.

^{page 173 note 9} Ann. IV–V 401 f. (fortification at Uzún-gheri).

^{page 173 note 10} One face illustrated Ann. IV–V 402, fig. 7.

^{page 173 note 11} R. L. Scranton, Greek Walls 85 ff.

^{page 174 note 12} The pottery gathered by us included fragments of relief pithoi (Fig. 2), a sherd from an East Greek Orientalising Plate with a trace of lotus and lotus bud ornament, fragments of archaic and classical wine-amphorae, fragments of r.f. bell kraters, and of late fifth- and fourth-century lamps, black-glazed bowls and cups (selected fragments now stored at the Museum in Smyrna). Prof. J. L. Myres picked up two fragments of archaic plastic aryballoi, now in the Ashmolean.

^{page 174 note 13} Dawkins reported ancient walls on the island which he took to be part of a perimeter, but we could not confirm this.

^{page 174 note 14} Ann. IV–V 401 (‘Fortilizio’ at Burgaz).

^{page 174 note 15} Loc. cit. figs. 5–6.

^{page 175 note 16} This is apparently the lion from Cnidus in Berlin, Kurze Beschreibung, Skulpturen (1922) no. 1724.

^{page 175 note 17} In 1907 Dawkins noted a ‘complex of just visible walls’, and in a letter of 1911 Polemikos mentioned an ‘interior wall of the temple 10 m. deep’.

^{page 175 note 18} Brit. Mus. 93. 11–13. 2–3 (bought from John Kalesperis); BMC Sculpt. I 150 f., B320–321 (‘from Cnidus’); cf. Richter, G. M. A.Kouroi 119 f.Google Scholar, figs. 114, 129–31 (p. 100, the statuette apparently regarded as Cnidian). ‘According to Kalisperis it is a temple of the Dioscuri, and he talks of a statue found there with an inscription proving this’ (Dawkins, who visited the site in company with Kalisperis).

^{page 175 note 19} (1) ‘Small stone figures of men and women, generally carrying animals or other objects’ (e.g., ram, lyre, unidentified animal) in a style ‘archaic but soft’. (2) ‘Stone figures of animals, notably birds’. (3) ‘Terracotta figures of all sorts, men, women, animals, birds’, etc. (4) Paste or faience figurines pierced for suspension, including a hawk and a naked female figure apparently of the type Lindos I nos. 1282–88, pl. 56. (5) ‘Bronze pins, fibulae and rings’. (6) ‘A small vase of Corinthian ware’ and similar sherds. Roof tiles also are mentioned.

^{page 175 note 20} For other small dedications recorded from Datça cf. Blinkenberg, Lindos I 26Google Scholar, id. Knidia 204 f. Istanbul Mus.: terracottas, Mendel, Cat. figurines grecques (1908) 582 f.Google Scholar; limestone figures, Lindos I 444, SCE IV ii 332, fig. 52; faience, Lindos I nos. 1272, 1278. Berlin Mus.: terracottas, ib. 485 fig. 55, 504 nos. 1966, 1982, 1944; limestone figures, ib. 414, no. 1604; pottery, ib. 299, Mus. Inv. nos. 3677A and 3677A bis. British Mus.: two handmade terracotta bull figurines 59.12–26. 631–2; paste, vase fragment 93. 11–13. 6; faience, figurine 94. 7–17. 2 (cf. AA 1894, 177); pottery, painted Plate fragments 93. 11–13. 4 and 94. 7–17. 1 (RA 1894, 266), 93. 11–13. 5 (ib. 27 fig. 14), now JdI 1942, 129 figs. 1–2. For the limestone statuettes cf. also Lindos I 26, 402, 437, 440, 446, 456. Archaic painted pottery is singularly scarce on the Cnidian peninsula. Dawkins noted Corinthian but not East Greek painted wares. Ionic b.f. has not been remarked, and the handful of Orientalising fragments mentioned here are insufficient to support Schefold's contention that the Cnidians played a leading part in the production of East Greek pottery (JdI 1942, 129 f.).

^{page 175 note 21} Inv. no. 8 (pres. ht. 0·115 m.) resembles the Brit. Mus. statuette from the Burgaz temple site (see footnote 18), but is nearly half as large again in scale and is more firmly modelled in early archaic Greek style. Inv. no. 9 (pres. ht. 0·1 m.) is badly rubbed, but seems to be a male figure holding a flower (cf. the series BMC Sculpt. I 158 ff.).

^{page 176 note 22} On the slope of the earthy escarpment above the right bank of the stream there runs a clearly defined shelf, with a corresponding embankment crossing the mouths of the gullies which come off the Plateau; occasional heavy, roughly worked blocks lie at the foot of the slope, and the shelf is fringed by loose stones in great number which could well come from the backing of a plundered wall. South of Reşadiye the escarpment turns and follows the line of the road towards the Iskele, with the shelf still clearly defined for about a kilometre, after which it is lost in the more hilly terrain above the Iskele.

^{page 176 note 23} Kiepert (followed by Philippson) named this hill ‘Chios’ and located it too far to the south. It appears on the left in the background in the photograph Philippson Reisen V fig. 11 (= Sudhoff Kos und Knidos, fig. 33).

^{page 176 note 24} Tiles of both late fifth or fourth century and late Hellenistic–Roman type, an Attic black-glazed foot of middle-late fifth century date, late amphora toes and fragments of moulded lamps.

^{page 176 note 25} The surface finds were bits of red wall-plaster, pre-Hellenistic tile and amphora fragments, and a bit of a fifth-century Attic skyphos foot.

^{page 177 note 26} In spite of the lack of excavation a considerable number of relief pithoi are recorded from Datça. Cf. Berlin Inv. no. 3351 and 30150A (both Neugebauer, Führer 11); fragments in Athens, (AM XXI 229 ff.Google Scholar, cf. BCH LXXIV 162, 166); fragments reported in the Khaviaras collection in Syme, (Ann. IV–V 402 n. 1);Google Scholar fragments, coming from a number of pithoi, recently presented to the Rhodes Museum (BCH LXXIV 139, 147, 166, 168 f.)Google Scholar; other pieces mentioned by Sir John Myres. Cf. also p. 174, n. 12. The majority of those recorded correspond closely to Ialysan types (cf. Feytmans, BCH LXXIV 144Google Scholar), though displaying a variety of motives that cannot be matched on Ialysan pithoi found in Rhodes. A further discussion by Miss Feytmans is to appear in BCH.

^{page 178 note 27} Cf. Feytmans, BCH LXXIV 180.Google Scholar

^{page 178 note 28} Ibid. 180; cf. esp. fragments from Datça now in Rhodes ibid. 168 f. and pl. 28.

^{page 178 note 29} Cf. Feytmans loc. cit. 173 f., 177.

^{page 178 note 30} Spratt lays down tombs by the shore; Lieut. Smith noted some ruins, probably Byzantine (Newton, , Halicarnassus II 526Google Scholar); an old wall is reported at Kizilağaç on a height NE of the bay.

^{page 179 note 31} The left side is not visible, and the possibility that the piece was a shield of an altar cannot be excluded (cf. JdI 1911, 65 ff.); but the thickness from front to back is against this. In spite of this thickness it may possibly have been the top of a stele (cf. Johansen, Attic Grave-Reliefs, 74 fig. 31, 90 fig. 41).

^{page 179 note 32} In the Tekir gravefield 6·6 m., overall width on the viaduct near Çesmeköy ca. 7·6 m., at the Inn 4·5–5 m.

^{page 179 note 33} With animals the time would normally be rather more. Newton's estimate of eight and a half hours from Tekir to Datça, (Halicarnassus II 525)Google Scholar would allow nearly ten hours for Tekir to Burgaz.

^{page 179 note 34} Ann. IV–V 402 n. 2. Philippson's convincing surmise that the arm of the Datça Water passing south of Yarik Daǧ (which undoubtedly comes from further back than Spratt indicated) rises under the second saddle, has apparently led him to an unnoticed duplication of Spratt's ancient fortress; nothing is in fact known locally of a fortification in the hills south of the upper Datça Water. The valley is seen from the north-east in Philippson Reisen V fig. 11, where Yank Daǧ appears in the centre, with Maltepe on the left and the high Triopian ridge on the right.

^{page 179 note 35} It seems to have taken the present-day pack route over the second saddle where the new carriage road makes a detour to the south.

^{page 179 note 36} Newton, , Halicarnassus I pl. 49.Google Scholar

^{page 180 note 37} Cf. Wace, Mycenae 27.

^{page 180 note 38} Halicarnassus II 523 f.; Ann. IV–V 297 ff., fig. 4.

^{page 180 note 39} The difference between the two halves of the viaduct remarked by Maiuri is not confined to the height of the courses but extends to the filling. The east pier, with blocks 0·29 m. high, was laid in regular courses through the thickness of the wall, and still stands almost intact on the upstream face though it is the more exposed to the force of the torrent; whereas the corresponding part on the west, with blocks apparently 0·54–0·59 m. high, seems to have had an earth filling in the core between the two faces and has almost entirely disappeared; the walls of the west approach way are roughly built in irregular masonry towards the abutment on the bank. These differences could most simply be explained as due to two separate local authorities or contractors. Most probably the base of the triangular aperture is given by the two great water-worn sill blocks (totalling 3 m. width) on the upstream edge of the opening; in any case Maiuri's figure of 4·5 m. is overestimated. On either bank the ancient route entered a cutting before passing onto the viaduct; and from this it is clear that the roadway was hardly above the level of the highest preserved courses and that its height over the sill in the torrent bed did not greatly exceed 6 m.; the triangle may therefore have been closed at the top by horizontal lintel slabs.

^{page 181 note 40} Halicarnassus II 522 (‘intended to support the land from the undermining action of the stream’).

^{page 181 note 41} Newton, , Halicarnassus II 522 f.Google Scholar, ‘remains of a Genoese or Turkish castle, called Assar Kalessi, consisting of rough walls built with mortar’. The villagers have now no other name for it than Yaziköykalesi.

^{page 181 note 42} Halicarnassus II 471 ff.

^{page 181 note 43} Déscription de l'Asie mineure pls. 162–4.

^{page 181 note 44} Halicarnassus II 480 ff.; the lekythos there illustrated is of a type which goes back through the fourth century, but its form appears in Newton's drawing to be more developed than the examples from Olynthos (Robinson, D. M., Olynthus V pl. 172Google Scholar) and it should perhaps be dated after the middle of the fourth century. Krischen, F., RM LIX 173Google Scholar, has dated the tomb down beyond the possible limits of this vase form.

^{page 182 note 45} Ann. IV–V 397, figs. 1–3. Newton, , Halicarnassus II 524Google Scholar (followed by Maiuri), remarks a chapel and a pear-shaped cistern inside the fortress. Kiepert on his map has transferred the name Assar to this fortification, and in consequence Philippson has designated it mediaeval by confusion with the castle south of Yaziköy (p. 181).

^{page 182 note 46} Cf. loc. cit. fig. 3, where the gate is illustrated; the masonry of the curtain does not show clearly.

^{page 182 note 47} These figures, together with a marble statuette-base, are now numbered 10–25 in the inventory of the collection in the Reşadiye School. All the statuettes are headless. Those of Aphrodite include a poor half-draped Anadyomene (no. 14), fragments of three figures of the ‘Euploia’ type (nos. 17–19, cf. Plate 39, c–d), and two pieces which probably belong to a single statuette of the Cnidian Aphrodite (nos. 20–21, Plate 39, e). No. 17 (ht., without head, 0·34 m.) is similar to the BM statuette Smith and Porcher, Discoveries at Cyrene pl. 71, but without dolphin or rudder. Nos. 18 (Plate 39, d) and 19 (Plate 39, c) come from two closely allied statuettes, with the difference that only no. 18 has the female attendant below the Priapic figure; apparently in both statuettes the left foot of the goddess was supported by a nude kneeling male figure at the front edge of the base; the drapery of the Priapic figures is rucked up around the rigid penis. Ht. of no. 18 (without head) 0·52 m. (right arm preserved to wrist, not shown in Plate 39). For the kneeling attendant figure cf. the marble statuette Reinach, , Rep. Stat. II 349Google Scholar, a bronze in the Biscari Collection and coins of Aphrodisias (cf. Bernoulli, Aphrodite 334); the archaistic female attendant is not elsewhere associated with statuettes of this class but with draped or half-draped Aphrodites (cf. AZ 1881, pl. 7, Reinach, , Rep. Stat. I 341Google Scholar). The popularity of this type at Cnidus, and elsewhere in the Dodecanese (cf. Furtwängler, Sabouroff Coll. on pl. 37), suggests that the original (assumed by Lippold, Kopien 150, to be of statuette format) may have belonged to Cnidus or a neighbouring city. The Cnidia fragments no. 20 (pres. ht. 0·32 m., surface of water jar broken away) and no. 21 (pres. ht. of body 0·26 m.) (Plate 39, e) show the same meagre proportions as nos. 18–19 and are perhaps attributable to a peculiarity of local taste. The complete statuette of the Cnidia found in Carpathus, (Memorie II, pl. 35)Google Scholar looks like a product of the same workshop (in no. 20 the fingers of the left hand are less bent and there is no real indication of the very shallow plinth which shows under the water jar on the Carpathus statuette); the two statuettes seem to reflect a model like Torlonia 106 (Blinkenberg, Knidia, pl. 8; cf. the identification of the true Praxitelean type, Rizzo, Prassitele 50 f. Lacroix, L., Reproductions de statues sur les monnaies grecques 312Google Scholar; against this Lippold, , PhW 1934, 52 f.Google Scholar).

^{page 183 note 48} Cf. the group of statuettes from the ‘Temple of Aphrodite’ at Cyrene, Smith and Porcher Discoveries at Cyrene 76 f.

^{page 183 note 49} The latter consist of the underground stone-built burial vault and the stone sarcophagus with masks at the corners noticed by Newton, (Halicarnassus II 524)Google Scholar, and a grey limestone sarcophagus 2·6 m. long with Corinthian pilasters on the sides.

^{page 183 note 50} Halicarnassus II 521.

^{page 183 note 51} Halicarnassus II 521. Newton also notes a ‘square Greek tomb of isodomous masonry about 16 feet each way’ to the west of this, which does not closely correspond to the remains observed by us.

^{page 183 note 52} To the right of the road on the southern saddle is a massively walled area in heavy polygonal 22 by 14 m., while about 100 m. to the NE of this we found fragments of inscribed blocks and the limestone foundations of an oblong building 26 by 12 m. divided into two almost equal compartments. On this route also, apparently, Newton marked tombs (Halicarnassus I pl. 49), but it seems to have been further to the east that he noted tombs associated with an old road 20 feet wide (op. cit. II 519 f.).

^{page 184 note 53} Antiquities of Ionia III; Halicarnassus II 345 ff.

^{page 184 note 54} The name Deve Burnu, applied by the cartographers to the cape SE of Tekir, seems to have passed out of use; modern maps call this cape Arslanci Burnu (‘Lion Point’).

^{page 184 note 55} Newton, remark, Halicarnassus II 347Google Scholar, that in bad weather the small craft which ply the coasting trade find great difficulty in doubling Cape Kriò, is equally applicable today, and in summer Turkish and Rhodian caiques are constantly to be seen in the south harbour waiting for the north wind to moderate.

^{page 184 note 56} The west breakwater of the south harbour is laid in 100 feet of water at the end.

^{page 184 note 57} Scranton, Greek Walls 176, implies an earlier date for the corner of the square tower (on the west side of the north harbour mouth) shown in Benndorf and Niemann, Reisen pl. 5, but has apparently been misled by the inverted topographical directions (ibid. 16) which make it appear detached from the main system of fortification; Benndorf's error is repeated by Sudhoff, Kos und Knidos 323, fig. 30. Von Gerkan, Griechische Städteanlagen 117 f., argues from the regularity of the lay-out of the mainland city that this quarter cannot have been occupied before the fourth century B.C., and therefore assumes that the original settlement was on the Cape Kriò promontory.

^{page 184 note 58} Bent, , JHS IX 82.Google Scholar

^{page 184 note 59} Cook and three members of the Smyrna excavation party spent two days on the site in August 1949 and paid special attention to the abundant surface pottery. Apart from odd fragments from wine amphorae of types which were current through the fourth century, the earliest sherds recognised were two fragments of black-glazed Plates with cheap rouletting, of types dating to the later fourth century; some of these sherds were found high up on the mainland terraces and so suggest that the earliest Hellenistic habitation was already widely extended within the city area. It may be inferred from the terms in which he referred to the archaic finds from Salzmann and Biliotti's excavations in Rhodes, which he saw on the conclusion of his work at Tekir, (Travels and Discoveries in the Levant II 266 f.)Google Scholar, that Newton had found nothing comparable, and likewise from his silence that he found no b.f. or r.f. pottery. The place and circumstances of the discovery of the late fifth-century black-glazed ribbed jug (p. 208, no. vii, BCH XXXVI 532 f.) are not known, but it is likely to have come from a grave. Hanfmann's claim that a continuous development of pottery from Mycenaean to archaic may be observed at Cnidus, (AJA 1948, 146)Google Scholar is as yet unfounded; it perhaps arises from the same confusion that led him to regard Cnidus as one of the Sporades (ibid. 139).

^{page 184 note 60} Cf. BMC Coins, Caria, 87 no. 24 (early fourth century B.C., found by Gell in 1812).

^{page 185 note 61} BMC Lamps nos. 164, 170–71, and 226. Newton himself concluded that the sanctuary of Demeter was not founded before the middle of the fourth century; and his notices of lamps (Halicarnassus II 378 f., 387, 393–6, 445, 463–4) do not suggest the finding of archaic open types.

^{page 185 note 62} Archaeologia XLIX 348 f.

^{page 185 note 63} Cf. BCH XXXVI 530. The rock here, Spratt observed, is a hard dark green shining serpentine, on whose adamantine nature his geological hammer made but small impression. Spratt noted traces of ancient scarping of the rock faces under the col, but we were unable in the minutes at our disposal to confirm this observation, and the recent planing of a route for wheeled vehicles along the crest has modified the appearance of the ridge. What does not emerge clearly in Spratt's description is the fact that a trough not less than 30 m. broad and lying practically at sea-level runs inland from Bencik cove on the south coast, so that the rocky neck, whose height Spratt estimated (perhaps rather conservatively) at 50 feet, is only about 250 m. across at the base. Myres recalls that in 1893 the south end of the canal was conspicuous with mud, reeds and snakes; we picked up a few coarse potsherds on the ground in the trough, one fragment being of the same ware as the archaic relief pithoi. Before Spratt's survey the Bencik isthmus was not known, and consequently the site of the canal was located at the Datça isthmus; but the width of the latter at the low point near its east end is, as Spratt observed, more than double Herodotus' five stades, and since the ground rises uniformly from both coasts would involve the removal of between five and ten times as much spoil as the cutting of the col at the Balikaşiran, quite apart from the length of the carry out of the cutting. The trough at the Balikaşiran may be in part artificial, and in fact the narrowing of the land front by flooding the trough would suggest itself as a temporary measure at least during the breaking down of the col.

^{page 186 note 1} 0·055 m. to the top edge of the stone; two lines occupy about 0·044 m.

^{page 186 note 2} Austin, R. P., Stoichedon Style 93.Google Scholar

^{page 186 note 3} For example, in l. 4, from θ to μ measures on centres 0·10 m. (average 0·02 m.), from μ to τ only 0·084 m. (average 0·0168 m.).

^{page 186 note 4} See Austin op. cit. 35–7.

^{page 186 note 5} The fact that a chequer was not used may be thought to render it somewhat unlikely that iota would be included in the same space with another letter in the first line of the inscription, before the stoichos is set up. In place of [ἒδοξε Κνιδὶοις] a 27-letter formula could no doubt be found: e.g. ἒδοξε τᾶι ὲκκλησὶαι τᾶι Κνιδὶων on the analogy of SGDI 3657 from neighbouring Calymna. But we feel this to be improbable, apart from questions it would raise as to the date of the Cnidian democracy.

^{page 186 note 6} See the photograph of the upper portion in AM 1911, 97.

^{page 187 note 7} References in SIG loc. cit.

^{page 187 note 8} Cf. Hill, Historical Greek Coins, 62 ff.

^{page 187 note 9} It is tempting to suggest that this alliance may have a connection with the Theban attempt to wrest the control of the Aegaean from Athens in 364 B.C.

^{page 187 note 10} SIG ³ 126 = Michel 500 = Tod, Greek Historical Inscriptions 106.

^{page 187 note 11} There is also a faint suggestion of an oblique stroke after the kappa, but this may be illusory.

^{page 187 note 12} The change of formula as between l. 1 and l. 9 is not a serious difficulty; cf. SGDI 3567 quoted above.

^{page 188 note 13} The earliest example is apparently SGDI 4118 = IG XII. 1. 694= SIG ³ 339, which is not far removed in date from our present inscription: see below.

^{page 188 note 14} For δωρεὰν cf. δωριὰν ἒδωκαν Διονυ[σὶωι - -]Γὸρτυνς - - [καὶ οὶ - -]ὲν ᾿ΑFλῶνι Fολκὶοντες ὰτὲλειαν κτλ (Mus. ital. 2,231 no. 83–4).

^{page 188 note 15} Cf. Dittenberger, , Hermes 1906, 187Google Scholar: ‘in der Ableitung von Ortsnamen ist mir nur eine einzige Spur davon [i.e. short vowel before -της] bekannt, nämlich bei Harpokration Κολωνέτας’

^{page 188 note 16} The following paragraph was written in consultation with Mr. P. M. Fraser, to whose special knowledge of Rhodian affairs we are indebted.

^{page 188 note 17} κτοὶναι were territorial divisions of the Rhodian population before the synoecism, similar to the later demes. They survived under the deme-system after the synoecism, but apparently only for religious purposes, and no new ones seem to have been instituted after that date.

^{page 192 note 18} Before the dental: see Robert's, remarks in Hellenica IX 47 n. 2.Google Scholar

^{page 192 note 19} εὶδι<ο>γυναικὸς barbarous compound’, Hirschfeld ad loc. Have we in fact a case of a stock epithet merging into the noun?

^{page 193 note 20} E.g. θεᾷ Λητοῦς TAM III 380; ἀνθυπἀτου Σιλβανῷ Sardis VII 1, 139, and often in late inscriptions.

^{page 193 note 21} Νείκης was not written: the eta is close to the edge.

^{page 193 note 22} Archaic Cnidian: (1) dedication by Euarchus, BMI 1033 (see below p. 205); (2) Cnidian treasury at Delphi, FdD III i, 150, pl. V; (3) graffiti on kylikes, Petrie, Naukratis i, pl. 33, nos. 237, 239, 354Google Scholar, Blinkenberg, Lindos i, no. 2806. All these are earlier than our text. □ for eta seems confined to Cnidian; C for omicron is found also in Melian, but not in the earliest example, and it may possibly have been taken from Cnidian. In the present paragraph we are indebted to the special knowledge and courteous help of Miss L. H. Jeffery; we offer her our best thanks.

^{page 193 note 23} ὲργαοτὴρια are in fact occasionally mentioned in connection with tombs, e.g. Sterrett, Wolfe Exp. 518, Sardis VII 1, 163, where they stand on the ground which included the tomb. Our present case is not apparently of this kind.

^{page 193 note 24} Or on the ‘windy height’, [ὰ]νεμὸεντι. The epithet νεμὸεις is not quoted, but the factory would hardly be situated on the windy summit of the acropolis hill.

^{page 194 note 25} Cf. (among many examples) ἀνὲθηκεν δὲ μ᾿ Ε[ὒ]πολος αὒτἒι (Inscr. de Délos 17), ὲπεν δὶκαιον (Roehl IGA 489).

^{page 195 note 26} The text might conceivably be construed to mean that the rent shall consist of the entire revenue from the sacrifices, whatever this might be. But this would destroy any incentive on the part of the temenos officials to dispose of their wares to advantage.

^{page 195 note 27} For the δερματικὸν at Athens see IG II² 1496; and Harpocration s.v.: δερματικὸν ἂν εῖη λὲγεν ὸ ῥὴτωρ (sc. Lycurgus) ὸ ὲκ τῶν δερματὶων τῶν πιπρασκομὲνων περιγινὸμενον ὰργὺριον

^{page 199 note 28} A similar cornice-block in a fountain at Aǧlâsun (Sagalassus) is inscribed: - - ιος ἐκ τῶν ὶδὶων ὲποὶησε καὶ τοὺς Φανοὺς ὺπὲρ υὶοῦ ᾿Αττὰλου ἀνὲθηκεν

^{page 201 note 29} We owe this suggestion to Prof. R. Syme.

^{page 202 note 1} ‘Dioscuri’, Emecik, and perhaps by the Horos Liraenos. Cf. pp. 172, 175, 178.

^{page 202 note 2} ‘The plain and valley of Datcha is very fertile, having fine groves of olives and valonia, and of almonds and other fruit trees: with abundance of water, if properly utilized for irrigation; and thus greatly impressed me from its park-like scenery, also, as a locality with a very promising future; for several springs rise up in the plain, besides the mountain streams from the high mountains of the interior; and they have wood, and some timber growing upon them’ (Archaeologia XLIX 356; cf. Homer, Od. V 63 ff.Google Scholar).

^{Page 202 note 3} See most recently the doubts and uncertainties expressed in ATL I 504, 562. How and Wells, Commentary on Herodotus, ad. loc. observe: ‘The whole section is a model of confusion. It is to be noticed that H., as a Halicarnassian, knows the Cnidian territory minutely’—a sorry judgment on Herodotus' powers of description.

^{Page 202 note 4} This point is argued, without reference to the site of Cnidus, in Fraser-Bean The Rhodian Peraea. We add only that if Cnidus is at Burgaz, the improbability of Bybassus being at Emecik is much increased.

^{Page 203 note 5} Note also the preposition in ὲς τὴν ῆπειρον τελευτᾶ̣ ‘passes into the mainland and so ends’—not, for example, πρὸς τῆς ὴπεὶρου

^{Page 203 note 6} The question of Triopion is discussed in detail below p. 209. We think it improbable that in the fifth century there was a usable harbour or anchorage at Tekir, since the small peninsula (Cap Kriò, now Deveboynu) was an island until the foundation of the city (cf. p. 204). If there was, the six Spartan ships doubtless used it; but more likely they were stationed in the bay of Palamutbükü (see the discussion below, p. 209). In either case, if the fugitive sailors started soon after midday, they would arrive at the city at Burgaz shortly after dark; the distance from Tekir is between eight and nine hours, from Palamutbükü something over seven.

^{Page 203 note 7} Cnidus is described by Thucydides as unwalled (ὰτεὶχιστοσ) We understand this to mean that the fortifications at Dalacak were partially dismantled or in disrepair, not that they were non-existent; this seems indicated by the phrase ὰμεινον φραξαμὲνων αὐτῶν ὺπὸ νὺκτα We conceive that the Athenians landed on the open beaches or through the gaps in the sea-wall north and south of Dalacak and attacked the acropolis wall. Thucydides gives no indication of their numbers.

^{Page 203 note 8} We note also, without wishing to stress it unduly, that Ps-Scylax 99 (ca. 350 B.C.), going south, names Triopion before Cnidus.

^{Page 204 note 9} The same is mentioned by Aelian, V.H. II 33Google Scholar (quoted by Stephanus s.v. Χερρὁνησος), who speaks of Χερρονὴσιοι οὶ ἀπὸ Κνὶδου

^{Page 204 note 10} The true explanation is not our business here, and is not made easier by the uncertainty whether the inscription had ἐν Κνὶδῳ οΓ ἀπὸ Κνὶδου: but an example may suggest a possible line of interpretation. Several ancient authors (Thuc. III 88, Strabo VI 275, Diod. Sic. V 9, Paus. X 11, 3 and X 16, 7) mention the band of Cnidian emigrants who finally settled, about 580 B.C., in the Lipari islands; they frequently fought the Tyrrhenian pirates, and sent offerings to Delphi from their victories. Similar emigrants established on a Chersonese (most probably the Carian, i.e. ‘Bybassian’, later Rhodian) might well use the language quoted above, at least in the form given by Aelian. In this connection we may mention the late archaic Chersonesian coins with Cnidian lion's head on the obverse (Head HN ² 614), attributed by E. Meyer in RE s.v. ‘Peraia’ 566, together with the dedication at Olympia, to the Carian Χερρονὴσιοι of the Athenian tribute lists. May the ox-head on the reverse be a punning device., .Βου(βασσὶα)? (There is a confusion in ATL I 562 n. 1: these coins are not the same as those attributed by Seltman to the Thracian Chersonese.)

^{Page 204 note 11} Cf. Dümmler, , AM XXI (1896), 229 f. n. 2Google Scholar; Chaviaras, , BCH 1912, 529CrossRefGoogle Scholar; Hasluck, , BAS XVIII 211–2.Google Scholar Bürchner in RE s.v. ‘Stadia’ rejects the identification on phonetic and accentual grounds: the name in the Notitiae is Σταδὶα or Σταδεὶα not Στὰδια: yet immediately afterwards he is prepared to accept a view to which the same objections, if valid, ought to apply.

^{Page 204 note 12} Pegusa remains in the air, but is at least perfectly appropriate as an alternative name for Stadia = Datça: see the description above, p. 202 n. 2. The location of Triopion is discussed below.

^{page 205 note 13} The date of the move is discussed below, p. 210.

^{page 205 note 14} We shall find ourselves in this section inevitably confronted with the unenviable task of attempting to date the inscriptions on little or no evidence beyond the lettering. Such an attempt must always be precarious in a region which has yielded few, if any, precisely datable texts. The reader will of course form his own judgment in each case; our own views may be thought to be prejudiced.

^{page 206 note 15} Schede's no. 10 is Hellenistic.

^{page 206 note 16} The majuscule texts given in BMI are as good as can be expected from type, but a much more accurate idea of the script may be obtained from the fascimiles in Newton, Halicarnassus. In some of the most important cases we give here photographs of squeezes taken by kind permission of the British Museum.

^{page 206 note 17} In the last line of this decree surely restore [καταλ]ὺη[ι] rather than [ἀγνε]ὺη[ται]

^{page 207 note 18} The importance of the punctuation marks is not easy to assess. In the first two lines, where each metrical line has a line on the stone, the dots are absent; the following trochaic lines, however, are too long for the stone, so that in this particular case there is a special reason for the punctuation: having marked the first two lines in the usual way, it was natural to mark the others in the only way possible. The use of οἱ for Τοί is hardly significant, since the Doric dialect is abandoned in l. 3.

^{page 207 note 19} A dedication of this sort is hardly suitable for a cult-statue; and the injunction προπολεύειν is better discharged by a figure of a woman in attendance than by an image of the goddess herself. For a new discussion of the problem of the statue of Kore see Ashmole, JHS LXXI 25 ff.Google Scholar, where the indication of two Cnidian types of Kore, one being earlier than the seated Demeter, is of special interest in connection with the removal to Tekir. ἒφησε in l. 3 appears to mean ‘commanded’. This use of φημί is exceedingly rare: LS⁹ quote one example (and a dubious second) with the dative and infinitive. The accusative in our text is very remarkable, and it is tempting to suggest that the aorist of ἐφὶημι was intended; but whether as a peculiarity of local dialect or merely by an error of the lapicide we do not venture to say.

^{page 208 note 20} We cannot agree with the translation given by Hirschfeld of this epigram. We understand: ‘If you are going, stranger, to the precinct of the gracious hero Antigonus, but little of the journey remains [not ‘of the road to Cnidus’]; but that little you will accomplish by traversing the short path that leads uphill on my left hand, not forgetting to give me greeting; and if the Muses give you any good gift, make to the gods an offering from your repertoire [not ‘a careful offering’]’.

^{page 209 note 21} Hence, perhaps, the name Triopion, ‘facing three ways’, with reference to the triangular shape of this part of the peninsula; cf. Trinacria. The eponymous Triopas may be safely discarded. (In Paus. X 11, 1 he is oikistes of Cnidus.) Leake, , Antiq. of Ionia III 3Google Scholar, suggested that the name may allude to ‘the triple summit which the promontory, under some aspects, presents to those who sail by it’.

^{page 209 note 22} Tekir is not an attractive alternative; all the arguments against it as an early site for Cnidus (p. 184) apply equally as a site for Triopion. It is hardly conceivable that Tekir could be preferred to Betçe in early times.

^{page 209 note 23} There is accordingly no necessity to look for Triopian Apollo at Tekir, as Thucydides' words in VIII 35 might seem to suggest.

^{page 209 note 24} For the fortified hills near Yaziköy see above p. 181.

^{page 209 note 25} The sanctuary itself was perhaps on a hill or knoll: cf. Τρὶοπον κολὼναν in Theocr. XVII 68.

^{page 210 note 26} Cf. RE s.v. ‘Asklepios’, 1655.

^{page 210 note 27} The rude fortification, with tiles of c. fourth-century date, by the road in the valley west of Reşadiye (p. 176), can best be explained as a relic of this period of stasis on the peninsula.

^{page 210 note 28} Cf. dedications to the Anaktes and to the associated deities at the Demeter sanctuary at Tekir, , BMI IV 1Google Scholar, nos. 804, 806, 810 f.

^{page 210 note 29} BCH XXXIV 425, no. 1. Another dedication to Apollo Karneios in honour of Theupompos was copied by Hamilton at Tekir (Hamilton, W. J., Researches in Asia Minor II 459 no. 287Google Scholar).

^{page 210 note 30} Cf. the confused citation in Stephanus s.v. ‘Chios’. Kiepert located it near Datça, see p. 209, n. 23.

^{page 210 note 31} ATL I 565 f.

^{page 210 note 32} Von Gerkan's argument that, being incorporated in the street-grid, this sanctuary must be coeval with the city on the mainland at Tekir (Griechische Städteanlagen 118) is not entirely convincing, since the building blocks are not likely initially to have extended so far out.

^{page 210 note 33} Ashmole, , JHS LXXI 13 ff.Google Scholar The series of terracotta figurines dedicated in the sanctuary seems also to start about the same date.

^{page 211 note 34} SIG ³ 290.

^{page 211 note 35} Cf. BCH LXXIV 325, on account of the striking similarity to the inscriptions of the Thessalian Monument.

^{page 211 note 36} Cf. Koldewey, Antike Bauresten der I. Lesbos 3 ff.

^{page 211 note 37} Admiralty Chart no. 1531; Ann. IV–V 265 ff. The earlier settlement (Palaia Myndos) was not on this site.

^{page 211 note 38} Cf. Rizzo, Prassitele 48; Lippold, PhW 1934, 53.Google Scholar

^{page 211 note 39} Diog. L. VIII 88; Plut. adv. Coloten 1126d.

^{page 211 note 40} Antiquities of Ionia III 10; cf. Halliday, Greek Questions of Plutarch 47 f. on QG 4.

^{page 212 note 41} Cf. RE s.v. ‘Eudoxos’ 931. It might in fact be argued that the unctioning of prostatai in the decrees for Iphiadas and Parmenon (pp. 186–87) implies the existence of a democratic constitution before 360 B.C.

^{page 212 note 42} II 119.

^{page 212 note 43} XVII 807 δεὶκνυται γὰρ σκοπὴ τις πρὸ τῆς ῾Ηλὶου πὸλεως καθὰπερ πρὸ τῆς Κνὶδου

^{page 212 note 44} cf. Pickard-Cambridge, , CAH VI 210.Google Scholar

^{page 212 note 45} Since the above statement was written Mr. P. E. Corbett has examined the two fragments of glazed plates with rouletting referred to on p. 184 n. 59 and assigned one to the third quarter, and the other to the last quarter of the fourth century.