¹ 1, 21, 3 (Niobe); 3, 22, 4 (image of the Mother of the Gods); 5, 13, 7 (lake and tomb of Tantalus, throne of Pelops); 8, 2, 7 (Niobe).

² (5, 13, 7).

³ The founder of the ‘Smyrniote’ school was of course Texier, who found all the monuments, including the lake, on the hill above Bayraklı; more recent adherents are Weber, (Sipyle et ses Monuments, 1880)Google Scholar and Perrot-Chipiez. Leading, ‘Magnesians’ are Humann (AM XIII, 22 sqq.)Google Scholar, Hogarth, in CAH III, 504Google Scholar, Bürchner, in PW III A, 279Google Scholar, and Ramsay. Cadoux, Ancient Smyrna (full bibliography) is reticent, but seems also inclined to the latter view (p. 39, l. 10).

⁴ Reported in AM XII, 252–3, 271 sqq.Google Scholar The finds include three inscriptions, two naming Mother Plastene and the third mentioning ‘the temple and the goddess’, some remains of a building (which cannot, however, have been a temple), and some pieces of sculpture—all apparently of Roman date. The site is described as one hour east of Magnesia and about a quarter of an hour from the figure of Cybele.

⁵ Weber and Perrot-Chipiez argue for the identification with the hieron of Cybele; and in an article by M. Lionel Belhomme, published in Turkey as recently as 1940, this view is still accepted without question; Ramsay, and Schweisthal, (RA XVI)Google Scholar favour a fortress.

⁶ Now Professor Haspels, but with characteristic modesty she asks that her old title should be retained.

⁷ The district is not well shown on the maps, even on the new 1/200,000 G.S. map.

⁸ Some assistance is given by a fragmentary flight of two or three steps, W, noticed by Weber; they lead towards the sheer edge of the rock, and were presumably supplemented by a wooden ladder or the like.

^8a Scranton, , Greek Walls, 52, 69, 165Google Scholar. All the dated examples there quoted are early Hellenistic. [How far the principles for dating walls by their style, established by Dr. Scranton for Greece proper, may hold good for Asia Minor also, is a question not without importance for the history of Anatolia. ‘Coursed polygonal’ is an interesting case in point. Undated walls of similar type to that of Ada Tepe are not uncommon down the west coast, where fortification walls in general are mostly of Hellenistic date; there is accordingly some a priori likelihood that coursed polygonal may turn out to be characteristic of that period in Asia Minor as in Greece. What is needed is a few satisfactorily dated examples; and from this point of view the wall at Ada Tepe may have some importance. One such example is already to hand (Scranton op. cit. 166, no. 12); a second would greatly strengthen the case. Dr. Scranton kindly informs me that, so far as he can safely judge from a photograph which I sent him, this wall may fairly be classed as coursed polygonal as he understands the term—that is, as a genuine variant of normal polygonal, not merely a broken-down form of ashlar or trapezoidal— and that he would therefore expect it to be later than the middle of the fourth century. To use the mere style of this wall as an unsupported argument for the Hellenistic date of the remains at Ada Tepe would be, in the present state of the evidence, hazardous; but, when on the one hand the other evidence for a Hellenistic date is so strong (see especially Miss Haspels' report on the pottery below), and on the other hand walls of this type are elsewhere characteristically Hellenistic, there does appear to be a case for supposing that we have at Ada Tepe a second dated example of coursed polygonal, and a second step towards applying Dr. Scranton's classification to the walls of Anatolia. At Smyrna in particular, the a priori probability that any wall which is not archaic will prove to be Hellenistic, needs no emphasizing.—G. E. B.]

⁹ Ramsay, op. cit., 71, mentions a covering stone forming a roof overhead; this stone is no longer in position.

¹⁰ For example, the inner walls at Bel Kahve vary from 1·13 to 2·10 m.; the great land-wall at Iasus averages 1·85 to 1·90 m. Even in the city-walls of all but the largest cities of the Aegean coast of Asia Minor, an average thickness of more than 2·50 m. is, in our experience, exceptional; e.g. the wall of Colophon is 2·15 to 2·35 m. thick; at Alinda the western acropolis wall is 1·80 to 2–10 m.; the long wall at Caunus is in general 2·00 to 2·70 m. [The figure 4·60 m. given by Maiuri, in Ann III, 263 sqq.Google Scholar, for the wall at Caunus (cf. Collignon's ‘at least 4 m.’ in BCH I, 339Google Scholar), applies only to certain parts of the wall described by the former.]

¹¹ A mixture of ashlar and polygonal work in the same wall is not in itself proof of a difference in date; cf. the Hellenistic walls of Colophon (AM XI, 398 sqq.) and of Teos (BCH XLIX, 281 sqq.Google Scholar).

¹² Prof. W. J. McCallien has kindly sent us the following description of a specimen submitted to him: ‘a typical lightly decomposed porphyritic andesite, with hornblende and biotite.’

¹³ The coursing is most noticeable in the lower part, which was below ground before we set to work.

¹⁴ Noticed by Ramsay, JHS I, 72Google Scholar.

¹⁵ ‘Much decomposed flesh-coloured rusty-weathering lava, probably trachyte, with phenocrysts of felspar and quartz’: Prof. W. J. McCallien.

¹⁶ These traces seem, surprisingly, to have escaped the notice of previous observers.

¹⁷ But not, apparently, these actual blocks, whose rough under-surface indicates that they rested in the earth.

¹⁸ At a much later time, the cistern underwent certain alterations, (a) At the foot of the S.E. wall of the cistern, near the rounded northern end, a hole has been roughly hacked through the masonry (Fig. 10); it is 0·56 m. wide by 0·80 m. high. This hole is obviously no part of the original arrangement, and the idea suggests itself that the cistern was connected by an underground channel with the other cistern Q, on the terrace below, which is itself of late date and has a similar hole in its western side. As there is no apparent means of opening and closing the holes, the upper cistern would have been out of use at this period. Both holes were afterwards blocked up. (b) Later still, all the walls, including the polygonal wall, the natural rock-face and the hole just mentioned, were plastered over with lime mortar, most of which has now fallen away.

¹⁹ As to the ‘two fragments of Hellenic ware of the 5th or 4th century’ which were according to Ramsay, (JHS I, 73)Google Scholar picked up on Ada Tepe,. we have no information. Specimens of the pottery here described by Miss Haspels are in the Izmir Museum.

²⁰ For Bel Kahve see Weber Sipyle, Ramsey in JHS op. cit., Perrot Chipiez V. 55. The article promised by Seylaz, in ÖJh XXVIII Beibl. 121Google Scholar has not appeared. For Akça Kay a see Weber, in AM X, 213 sqq.Google Scholar; he remarks on its general resemblance to Ada Tepe, but does not abandon his belief that the latter is the hieron of Cybele. The mutual similarity of the three sites Ada Tepe, Bel Kahve and Akça Kay a has been independently observed by Dr. Ekrem Akurgal; see his interesting comments in Belleten 37 (Ocak 1946) 68–71Google Scholar.

²¹ For the identifications proposed in the neighbourhood of Magnesia, see Cadoux, Ancient Smyrna, 39Google Scholar, with bibliography.