Old Smyrna: the Iron Age fortifications and associated remains on the city perimeter

R. V. Nicholls

doi:10.1017/S0068245400013058

Old Smyrna: the Iron Age fortifications and associated remains on the city perimeter

Published online by Cambridge University Press: 20 December 2013

R. V. Nicholls

Article contents

Extract
References

Get access

Rights & Permissions

Extract

Traces of fortifications around the area apparently once occupied by the city of Old Smyrna were observed by Louis Fauvel, and our first detailed description of them is that of Prokesch von Osten, who accompanied him there on a second visit in 1825. As we shall see later, it seems likely, though proof is no longer possible, that most of the circuit wall around the tell, as well as that on the low spur to the west of it on which the modern village now stands, as described by Prokesch, may have belonged to the defences of the classical city. Nothing today survives of these above ground, owing to extensive stone-plundering in the interval; and it is to be feared that the fate of much of this rather exposed classical enceinte has been to provide masonry either for the houses of the modern village or for the terrace walls which today encircle the tell.

The plundering of this outermost circuit probably left the earlier ones inside it rather more exposed to view. I have not been able to verify which of the city walls it was that was photographed by Keil in 1911, but when Franz and Helene Miltner excavated here in 1930 a part of the late-seventh-century B.C. circuit was visible on the east side of the city. Here they cleared about 80 metres of its face, for the most part to no great depth, then picked up its line again with a small probe some 20 metres farther north. Two further small trenches seem to have located more of this late-seventh-century wall-line south-south-west of their long cut, in addition to traces of yet other circuits. Besides this they report sinking two shafts into the mound dominating the north-west corner of the tell and making two small probes in occupation levels within the city itself.

Type: Articles
Information: Annual of the British School at Athens , Volume 53 , November 1959 , pp. 35 - 137

DOI: https://doi.org/10.1017/S0068245400013058 [Opens in a new window]
Copyright: Copyright © The Council, British School at Athens 1959

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

¹ The grid references are to the plan of the tell, Plate 74. The way in which they are made is explained on p. 37.

² Jahrbücher der Literatur lxviii (1834), Anzeige-Blatt 55 ff. Quoted by Walter, O., ÖJh xxi–xxii Beiblatt 226–7.Google Scholar

³ pp. 92 ff., 128, 137.

⁴ The remains of the circuit wall on the spur projecting from the hill-side to the north-west of the tell as described by Prokesch have since vanished under the expanding modern village. The area in question is shown on the plan of the city environs, Plate 1, and the possible limits of these fortifications are there suggested, but no certainty is at all possible. These traces were probably visible to Texier, who seems to have represented them, though not very intelligibly, on his plan, Description de l' Asie Mineure, pl. 129. They were not recorded by Hirschfeld, in 1871 (Abh. der kön. Ak. der Wissenschaften zu Berlin, phil.-hist. Kl. 1872, 74 ff., pl. iv)Google Scholar and had presumably already disappeared, but traces of a wall-line that may have linked this defence system with that of the tell were seen by Conze, in 1898 (AA 1899, 15–16)Google Scholar and these may have survived until considerably later (p. 135 below).

⁵ F., and Miltner, H., ÖJh xxvii Beiblatt 162Google Scholar, take this to be part of the wall whose outer ace they cleared on the east side of the city and which our subsequent investigations have dated to the late seventh century B.C.

⁶ F. and H. Miltner, op. cit. 162 ff., figs. 81–83; Miltner, H., Eski İzmir 16Google Scholar, figs. 17–18.

⁷ ÖJh xxvii Beiblatt 166–7, fig. 84. Some uncertainty remains about the identification of the walls revealed by these most southerly probes of the Miltners. The position is discussed at p. 56 n. 99. Cf. also p. 64. Apart from these two which involve special difficulties, the positions of the Miltners' trenches have been approximately marked from their indications on our plan of the tell, Plate 74. Their positions on the grid are as follows: long trench: Fxx–Jxxiii; trench to the north of it: Exix; probes in the mound: Gviii–ix, GHix; probes inside the city: JKx–xi, LMx–xi.

⁸ Op. cit. 167–8; Miltner, H., Eski İzmir 16–17.Google Scholar

⁹ ÖJh xxvii Beiblatt 167.

¹⁰ These will be dealt with in the forthcoming publication of the domestic architecture in these areas.

¹¹ p. 88 n. 245.

¹² p. 71 n. 157.

¹³ pp. 88 ff, Sections ε–ε′, ζ–ζ′.

¹⁴ It seems desirable to enter a plea for the establishment of a competent international laboratory for the purpose in the Aegean area itself, where local conditions might be better apprehended and Customs complications avoided.

¹⁵ Plate 1.

¹⁶ Plate 74, the large folded colour plan at the rear. The clear overall picture afforded by the use of colour has not been achieved without some simplification; also, in the case of the houses and temple (but not of the fortifications) there has been a slight fluidity in the division between later Geometric (green) and pre-Alyattan archaic (red), in a few instances structures of the beginning of the seventh century B.C. being, for better differentiation, shown in green, where-as the division is otherwise made at c. 700 B.C.

¹⁷ Thus H4.×5 gives the position of the modern well in the square more loosely referred to as Hx.

¹⁸ The only surface indications were of the outerface of the seventh-century B.C. circuit at Section λ–λ′and at Fig. 10, k towards the south-east of the city and of a small portion of the Middle Geometric tower at the north-east gate.

¹⁹ Sections κ–κ′ and λ–λ′.

²⁰ Ancient India iii. 143, fig. 3.

²¹ Section λ–λ′, Plate 7.

²² Plate 7, Fig. 2.

²³ The slope down to the east and south in Monochrome Levels 3–5 (see the forthcoming publication of the prehistoric remains), on the other hand, is an exception which may argue a further period in which the bounds of the tell were neglected.

²⁴ See the plan of the environs of the city, Plate 1, where it is marked as the ‘River Meles‘, an identification that can probably not be substantiated (p. 23), although it doubtless played no mean role in the life of the earlier city of Smyrna.

²⁵ p. 12 n. 13 and p. 61 below. Cf. also p. 68 n. 143.

²⁶ Because of the batter on its faces it would probably have been distinctly narrower than this at its top.

²⁷ pp. 102, 121.

²⁸ Plate 10 a.

²⁹ See below, pp. 43 and 103.

³⁰ See above, p. 39. Cf. also pp. 66, 121 n. 13, 122.

³¹ On the chronological aspect see also pp. 122 ff.

³² Here, at a height of 5 m. above sea-level, it was about 8¾ m. thick.

³³ Its possible structural role is considered on p. 114.

³⁴ Plate 10b and c.

³⁵ p. 47.

³⁶ p. 103.

³⁷ p. 40, Plate 10a.

³⁸ p. 124. A selective publication of the imported pottery from Corinth is given by J. K. Anderson on pp. 138 ff.

³⁹ p. 124.

⁴⁰ p. 53.

⁴¹ Plate 11a, from a photograph by J. M. Cook, shows some of the burials cleared. On the change in child burial customs to which they seem to testify see p. 126 below.

⁴² Called ‘Area A’ on Fig. 2.

⁴³ The serial numbers shown on Figs. 3, 4, and 5 are those assigned in the field cataloguing of the child burial vases of the λ–λ′ section. To avoid confusion they have been retained here.

⁴⁴ However, this does not necessarily imply that this cemetery is the consequence of some single calamity such as a plague. The custom here instituted of burying children on the city perimeter thereafter continued unabated until the site was abandoned in the fourth century B.C.

⁴⁵ Plate 11b.

⁴⁶ Plate 11c.

⁴⁷ Plate 11d, where it was necessary to insert a strut to support the skull.

⁴⁸ Plate 11e.

⁴⁹ Burials c, f, i, j, and l.

⁵⁰ See Fig. 3.

⁵¹ Pots 7, 8, 11, 12, 17, and 18.

⁵² Pots 9 (Burial b) and 14 (Burial e).

⁵³ Plate 12b.

⁵⁴ p. 48. See also Plate 7.

⁵⁵ pp. 52 f.

⁵⁶ Cf. pp. 39 f. above.

⁵⁷ Plate 12a.

⁵⁸ Their dating and orientation seem quite to preclude the possibility that they had any connexion with the earlier projecting Wall 2 construction to the south (p. 53).

⁵⁹ pp. 52 f.

⁶⁰ Plate 12c and d. The photograph showing the upper part of the face (Plate 12d) belongs to an earlier stage in the excavations, the topmost preserved course having been removed by stone-plunderers in 1949. In the small area excavated the masonry seemed to approximate more or less to ashlar. However, the Miltners, who cleared a great deal more of this course (ÖJh xxvii Beiblatt 162 ff., figs. 82–83)Google Scholar, much of which no longer survives, saw it clearly as polygonal. It was stepped back slightly (between 0·07 and 0·12 m. according to the Miltners) from the face of the wall below. Of the smaller masonry that they described above it we found no trace. Presumably, whether in stepped ashlar blocks, as to the north (pp. 66 ff.), or in polygonal levelled off with smaller masonry, as here retailed by the Miltners, the effect was doubtless much the same of achieving a level stepped-back crowning to the socle. One may question, however, whether the smaller masonry might not be of subsequent date.

⁶¹ The position of these burials is also indicated on the section (λ–λ′, Plate 7), the pithos, n (the amphora, m, was just beside it), at around 2½ m., and the vase fragments, o, at about 4 m. above sea-level. The amphora found by the Miltners just outside the wall face in this area (op. cit. 165 n. 49) is perhaps also to be interpreted as a burial vase.

⁶² On the section (Plate 7) the burials have been labelled ‘post-sack’ and the wall ‘late seventh-century’. It is probably safer to regard the exact date of both as uncertain.

⁶³ A complicating factor in this regard is that the limits of the Miltners' trench cannot be detected in the dissolved mudbrick at this point. However, it seems doubtful if they dug even as deep as this mudbrick debris just here. Certainly, of the o vases, the almost complete oinochoe must be below the bottom of their probe.

⁶⁴ p. 103 below.

⁶⁵ The Miltners (op. cit. 162–6) apparently interpreted the projection of the socle face below the crowning as being similar to the projection sometimes found at the foundation of Lesbian and polygonal walls (e.g. Larisa am Hermos i. pls. 3b, 33) which, before being applied to such more monumental work, seems to have been developed from the protruding footing of house-walls (cf. pp. 93 and 98 and Plate 20c). But here we have to deal with a far deeper type of construction. Also, however the question of ground-level outside the face be resolved, it is at least clear that a considerable part of the projecting face below was exposed down into the sixth century B.C.

⁶⁶ p. 53.

⁶⁷ Fig. 2 (where marked area ‘C’) and Fig. 5.

⁶⁸ The dating evidence as a whole is considered on pp. 135 f.

⁶⁹ Plate 10b bottom right. These walls are marked with the levels 4·5 and 3·86 on the plan, Fig. 2.

⁷⁰ Plate 7 (where it is shown in elevation), Plate 10b left.

⁷¹ More of this structure may have been uncovered to the south at Fig. 2, e. See below.

⁷² e.g. the west edge of the Wall 1 Platform Fill has yet to be excavated.

⁷³ e.g. see p. 114.

⁷⁴ op. cit. 162 ff.

⁷⁵ The dating of the shallower part of the face at the south-east corner is much confused by plunder pits, some of them perhaps ancient and some certainly modern. See also p. 127 below.

⁷⁶ Plate 13a.

⁷⁷ Fig. 2, d, p. 50 above.

⁷⁸ At 3·2 to 3·2 m. above sea-level towards the south-east.

⁷⁹ On Fig. 2 marked with the level 4·14.

⁸⁰ Plate 13 b (viewed from the west).

⁸¹ Its full width was not cleared.

⁸² p. 52 above.

⁸³ Their thickness was about 0·09 m.

⁸⁴ pp. 53 ff.

⁸⁵ Plate 74, J-Kxxi–xxii.

⁸⁶ pp. 52 f. above.

⁸⁷ p. 124. Cf. also p. 43 above.

⁸⁸ p. 43. Cf. p. 103.

⁸⁹ Plate 13c and d. The reader is also referred to the forthcoming publication of similar pieces from the temple area at Old Smyrna.

⁹⁰ JHS Archaeological Reports 1954, 21; BCH lxxix. 288.

⁹¹ The Smyrna material is to be assembled in dealing with the column bases from the temple area mentioned above. So far as domestic architecture is concerned, it seems that houses of this ‘megaron-type’ were beyond the means of the ordinary eighth-century B.C. citizenry living in their thatched mud huts, but perhaps rather austere and simple for the tastes of their expansive seventh-century descendants. But such Smyrna examples as do occur seem to be of the seventh century B.C.

⁹² Fig. 9; Plate 13 d.

⁹³ JHS lxx. 13, fig. 8.

⁹⁴ Cf. p. 52 n. 75 above, p. 127 below.

⁹⁵ p. 52.

⁹⁶ pp. 47, 52 f.

⁹⁷ pp. 52 f.

⁹⁸ p. 50.

⁹⁹ Op. cit. 166–7, fig. 84. A completely certain identification of these wall traces is not possible from the Miltners' photograph. Presumably the main wall-line traced, their eigentliche Mauerflucht, is that of Wall 3 (or a subsequent rebuilding of it—cf. n. 94 above). It is uncertain whether the wall in front of it is the very top of Wall 2 passing on its wider circuit out beyond Wall 3 or Wall 4 built at this point very close in to Wall 3 or whether, indeed, the Miltners did not in fact uncover faint traces of all three walls. If they came upon part of Wall 4 here it would suggest that the negative evidence for the continuance of that wall adduced at Fig. 10, z may be due to subsequent plundering or failure to probe far enough in to the west (cf. p. 64). Unfortunately, the point shown in this photograph has not been relocated by excavation and, in view of the uncertainties obtaining and their wider implications, the Miltners' plan is not detailed enough to permit its plotting on our own plan of the tell (Plate 74) independently of that.

¹⁰⁰ p. 43. Width and thickness corresponded throughout, but a length of 0·57 as well as that of 0·51 m. occurred, e.g. behind the fountain-house at Fig. 10, l. See also p. 103, and p. 105 n. 108.

¹⁰¹ p. 40. Cf. also pp. 102 and 105.

¹⁰² pp. 53 ff.

¹⁰³ pp. 52 ff.

¹⁰⁵ pp. 52, 55.

¹⁰⁴ p. 57 above.

¹⁰⁶ p. 48 n. 60. See also pp. 98 f

¹⁰⁷ See Fig. 11, which shows the wall face in section at the edge of the plunder trench.

¹⁰⁸ p. 61.

¹⁰⁹ Fig. 12, p. 61.

¹¹⁰ See Figs. 10, 13.

¹¹¹ For such a wooden raft under the stone walls of a classical well at Smyrna see p. 11 n. 13.

¹¹² Plate 15 a.

¹¹³ Fig. 13.

¹¹⁴ p. 59. A view of the draw-basin from the passage is shown in Plate 15 b.

¹¹⁵ Plate 14b. For the purpose of this photograph the water-level has been lowered to approximately that obtaining in antiquity.

¹¹⁶ With the important proviso that our own establishment of modern sea level falls far short of perfection (p. 3).

¹¹⁷ The normal modern ground-water level in this area in summer is between 0·9 m. and a metre above sea-level, but it may drop considerably in the autumn, the lowest I have recorded being around 0·6 m. above sea-level in September. See further p. 12 on the progressive rise in sea-level from the second millennium B.C. to the present day at Old Smyrna and comparable phenomena elsewhere.

¹¹⁸ p. 56.

¹¹⁹ I take it that the knucklebones were not also dedications from the fact that they were found scattered in the overflow channel as well as in the draw-basin.

¹²⁰ Figs. 13, 14.

¹²¹ The fill seems to have been a simultaneous one antedating the abandonment of the city, though not by many years.

¹²² Plate 15c. The plundering of the façade shows clearly in Plate 14 a.

¹²³ Op. cit. 166–7. But here there is much uncertainty. See p. 56 n. 99.

¹²⁴ Plate 74, KLMxxiii.

¹²⁵ Plate 1 (where it is marked ‘6th century tumulus’); Texier, , Description de l'Asie Mineure pl. 131 F.Google Scholar I and II; Weber, , Le Sipylos et ses monuments pl. 1Google Scholar; Perrot-Chipiez v, fig. 16; F. and H. Miltner, op. cit. 149 ff., figs. 74, 76–77; Miltner, H., Eski İzmir 12 ff.Google Scholar, fig 12. No pottery dating this grave is recorded from the excavations of Texier or the Miltners, but on the criteria (by no means certain) of masonry style it should be one of the earliest of the Smyrna tumuli. However, since, as the forthcoming publication of the outlying cemeteries will show, our own excavations have adduced no certain evidence for the practice of tumulus-burial at Smyrna before the Lydian period, it seems unlikely to be earlier than the sixth century B.C. Also in favour of a sixth-century date is the fact that dovetail clamps seem to have been used to fasten the course above the apex of the vault (F. and H. Miltner, op. cit. 153). Such lead dovetail clamps occurred in the tomb of Alyattes at Sardis (Abh Ak Berlin 1858, 548, pl. iii. 4) but appear to have been unknown in the Smyrna temple buildings of the end of the seventh century B.C.

It would be rash to use such details as the mouldings shown on the tomb in Texier's drawings to establish a closer dating, by reason alike of the anterior and subsequent accounts of Prokesch von Osten and Hirschfeld (p. 36 nn. 2 and 4) and the complete lack of such elements or of a possible attachment for them on the monument as it appears today.

These vaulted structures may reflect an older Bronze Age tradition evidenced by the corbelled tomb-chambers at Isopata, Ras Shamra, and Mycenae and by the vaulted passages at Mycenae and Tiryns.

¹²⁶ Compare Fig. 13 and Plate 14a with F. and H. Miltner, op. cit. figs. 74, 76–77. In the ‘Tomb of Tantalus’ the inward curve of the side walls begins much lower down.

¹²⁷ Cf. p. 56 above.

¹²⁸ pp. 65 and 123.

¹²⁹ p. 41. See also p. 114 below.

¹³⁰ pp. 52, 55, 58.

¹³¹ Op. cit. 162 ff.; p. 36 above.

¹³² pp. 41, 47, 52.

¹³³ pp. 50 ff.

¹³⁴ Cf. the situation at the north-east gate, p. 72.

¹³⁵ Cf. pp. 39, 41, 121.

¹³⁶ There had been a little erosion in this area and a small part of this fill may have perished. It is just conceivable that a shallow dressed-stone ledge may also have been set under the inner mudbrick face at ground-level at the top of the stone fill, but if so it seems to have vanished completely. In view of the findings to the north (see below) it is possible that the lower part of the deep stone pack at this point may also have belonged to one of Wall 3's predecessors. However, the material of the fill seemed consistent throughout and no ready divisions suggested themselves.

¹³⁷ See also Plate 16 a.

¹³⁸ The other probes in this area have not been carried down to a sufficient depth to ascertain whether this pronounced ‘step’ in the Wall 3 outer face is a normal decorative feature in this part of the circuit or is an isolated phenomenon, serving, e.g., as a ledge under a drainage channel. The latter explanation, however, is insufficient in itself to account for the obvious weakness of the jointing above and below it at this point.

¹³⁹ See pp. 125 f. on Wall 2/3.

¹⁴⁰ ‘Tower’ seems the most convenient term. Actually the construction involved prior to Wall 4 seems to have consisted in the revetting of the end of the outer overlapping stub of wall in an ‘overlap-type’ gate (see pp. 116 f.).

¹⁴¹ See Section κ–κ′, Plate 8.

¹⁴² Cf. pp. 71 f. below.

¹⁴³ Ground-water on this side of the city is today distinctly higher than it is to the south (p. 61 n. 117) although this difference may not have obtained in antiquity with ready drainage into the estuary to the east. Today there are flood-dykes along the banks of the stream and this area is drained across the higher ground to the west.

¹⁴⁴ Fig. 19, based on drawings by Dr. Corbett. Note that only the upper courses of the andesite socle are shown. See also Plate 17 a and c, where the mortar shows clearly.

¹⁴⁵ Plate 17 c.

¹⁴⁶ Plate 17 a; JHS lxx. 13, fig. 9.

¹⁴⁷ Plates 16 b and 18 d (from the south). On the fitting of this ashlar masonry see also p. 97 and Fig. 32.

¹⁴⁸ Plates 16 b and 18 d (from the south) and 17 b (from the north-east).

¹⁴⁹ See the section, Plate 8.

¹⁵⁰ Plate 17 b; see also p. 95 and Fig. 30.

¹⁵¹ This second tower has yielded virtually no pottery and is not closely datable in terms of associated finds. The close sequence of building suggests that it should be of approximately Wall 2 date; the brick dimensions (pp. 103, 124) favour its having been a very early part of Wall 2.

¹⁵² It would seem somewhat unlikely that the drastic breach effected in the east face of the Wall 1 tower in the construction of its successor (p. 72) would in fact have been made if there were already in existence an approach from the east. But it is noteworthy that, on the admittedly limited indications as yet available to us (pp. 52 f.; resources were not available for a systematic excavation of the gate), the approach to the Wall 2 south-east gate does not seem to have aimed at exposing the unprotected right flank of an attacker. However, here the nearness of the estuary and the resultant general subjection to fire from the whole length of the wall probably made the situation very different.

¹⁵³ It probably continued down at least as low as modern sea-level.

¹⁵⁴ Its own backing was actually preserved to just over 2 m. above sea-level and the brick ‘core’ behind it (n. 156 below) stood even higher.

¹⁵⁵ The thickness of the fill varies because of the difference between the orientation of the brick ‘core’ and that of the face (see below).

¹⁵⁶ i.e. the stone socle of the outer face should have continued up to this level at least.

¹⁵⁷ On the plan, Fig. 18, it is largely obscured by the Wall 3 brick laid above it, being only shown where the latter has been cut through by some unrecorded previous excavation (Fig. 18, z) and in a small ‘window’ showing its junction with the face-backing. However, it continues under the later brick to the northern and southern limits stated and to others not yet established to the east and west.

¹⁵⁸ Fig. 19, Plate 16 b.

¹⁵⁹ At least, it seems more likely that the south face surviving at a higher level (Fig. 18, g) is of Wall 3 date.

¹⁶⁰ p. 103.

¹⁶¹ Cf. p. 66 above.

¹⁶² Section κ–κ′, Plate 8.

¹⁶³ This interpretation, which I find completely convincing, I owe to Dr. A. E. Ozanne.

¹⁶⁴ If, as seems probable but not completely certain, the smaller 0·3-metre-wide brick was used, this somewhat favours a date in the first half of the seventh century (i.e. as part of the Wall 2/3 construction phase—pp. 125 f.). The pottery from this structure, however, will hardly allow of a date before the middle of the century.

¹⁶⁵ It probably also constituted a somewhat late part of the Wall 2/3 line on the north side of the tell (see n. 164 above), but, in view of this lateness and the uncertainty in any case obtaining as to its dating, the designation ‘Wall 3’ has been retained for it.

₁₆₆ The quantities of stones in the dissolved mudbrick below this level had presumably fallen from Wall 2, since Wall 3 seems to have lost very little masonry. The mudbrick debris above, however, seemed to be from Wall 3. This seems easily the most probable interpretation. If correct, we may infer a stepped crowning to the socle as found elsewhere on the site and assume that the lower ‘steps–, being simply a facing and not structural members, were easily plundered later. However, although unlikely on architectural and excavational grounds, we cannot yet completely exclude the possibility that the ‘steps’ may simply have been set at the top of a clay glacis, consisting of seventh-century mudbrick debris, that quite concealed the Wall 2 face.

¹⁶⁷ Plates 8 and 17 d. At the other points cleared only a single trench had been dug (for the upper ‘step’).

¹⁶⁸ See n. 164 above, p. 103 below.

¹⁶⁹ Plate 16 b foreground; note that in this photograph the fill behind the face has been removed.

¹⁷⁰ At 5·37 m. above sea-level.

¹⁷¹ pp. 66 ff.

¹⁷² See n. 174 below and p. 74.

¹⁷³ e.g. to the north-west of the fourth-century ‘round-tower’ at Fig. 18, b.

¹⁷⁴ Here, however, there might be intrusion from the fourth-century debris (Fig. 18, r) to the north.

¹⁷⁵ See also Plate 16 c, the κ–κ′ section at an early stage in its excavation viewed from the north, showing this face and those of Walls 3 and 1 behind.

¹⁷⁶ Precisely of this kind is much of the material re-used in the backing of the modern terraces round the tell.

¹⁷⁷ It is probably to be assumed that such an extensive stone backing originally lay behind the Wall 4 face in the now badly plundered stretches to the east and south of the city, e.g. Sections λ–λ′ (p. 50) and υ–υ′ (p. 56).

¹⁷⁸ pp. 50, 53, 68.

¹⁷⁹ Fig. 18 o, Fig. 20, Plate 17 e. By describing them as ‘decorative’ I merely wish to stress that they can hardly have fulfilled their ostensible practical role of providing an alternative path up on to the cobbled way—a role stressed by the continuation of the line of the top step in the form of a flagged path (Fig. 18, n; see below). They doubtless served some practical function, but it seems unlikely to have been this. The width of the top step is too slight for one to be able to walk along it against the face of the city wall in any comfort, whilst the small detour involved seems somewhat pointless unless further steps some way away to the east led on down to the estuary. They would, however, have been excellent for displaying articles for sale. Perhaps they constituted one of the amenities of a small open-air market here by the gate. The wall bounding them to the south-east (Fig. 18, p) may then have been simply an enclosure wall, but could conceivably have helped to support an awning.

¹⁸⁰ Sample dimensions: in the photograph, the block showing complete in the top step, 1·55 × 0·69 × 0·32 m., and that showing complete in the bottom one, 1·16 × 0·48 × 0·32 m. Normally the masonry associated with the fortifications has been fitted empirically. These blocks are the rare exceptions pre-dressed to size. However, in view of the probability that they are re-used material dressed down again for their second role, I should prefer not to be too definite about the foot-standard in terms of which they were originally marked out (see pp. 104 f.). Possibly it was the 0·327-metre foot.

¹⁸¹ The latter type of dressing is attested as early as the seventh century B.C. at Smyrna, being used, for example, on one of the pilaster bases, perhaps itself re-used material, in the late seventh-century B.C. temple pylon.

¹⁸² The full account of the seventh-century B.C. temples will be published separately. The names used for the different building phases, ‘Temple Platforms I’, ‘II’, and ‘III’ (dating to the early seventh century, early third quarter of the century, and late seventh century respectively), are those that will be employed there, as also the distinction between the ‘inner’ and ‘outer platforms’. Roughly speaking, the ‘inner platform’ is the smaller rectangular area to the north-east occupied by solid stone fill, whilst the ‘outer platform’ consists in the extension of this to the west and south by stone walls backing an earth fill.

¹⁸³ The approximate limits of these investigations by J. Boardman and Miss D. H. Gray are shown on Fig. 21. They have been omitted on the detailed temple plans to be published later in the interests of avoiding confusion.

¹⁸⁴ Section θ–θ′, Fig. 23. This is not, however, conclusive, since the thicker brick was still used in the Wall 2 tower at the north-east gate (p. 71; see also p. 103 below).

¹⁸⁵ p. 103.

¹⁸⁶ See Section ι–ι′, Fig. 22. Much damage may have been done in the earthquake postulated around 700 B.C. (see p. 124), but this particular fault-line appears to be the product of a later calamity when Wall 3 was already in existence. On its closer dating see p. 79.

¹⁸⁷ Its construction is well shown on one of Dr. Corbett's elevation drawings to be published with the account of the seventh-century temples.

¹⁸⁸ With the important reservation that the south face of this mudbrick was apparently not cleared down in the excavation. The dimensions of the bricks themselves seem to have tallied with those of Wall 2 to the north (see above), save that they were somewhat thicker. See p. 103 below.

¹⁸⁹ Against the latter interpretation speaks the rarity of specifically votive material of the eighth century, only one such piece seeming certainly assignable to it on stratigraphical grounds, part of a faience figurine (S.F. 1350) apparently found under the foundations of Platform I. For this reason this eighth-century construction has not been included in the publication of the temples proper. In any case, the building of Temple Platform I shortly after 700 B.C. marked a radical change and the emergence of a public cult of the first magnitude.

¹⁹⁰ On the chronological aspect see pp. 125 f.

¹⁹¹ p. 103.

¹⁹² p. 81.

¹⁹³ The wall above it to the south-west in E3.xvi7 seemed to be of the fourth century B.C.

¹⁹⁴ p. 103.

¹⁹⁵ The dating material from its fill was a consistent lot spanning the second half of the sixth century B.C.

¹⁹⁶ Plate 18 a, from a photograph by J. M. Cook.

¹⁹⁷ However, towards the north this upper stratum represents chippings washed down by rain from the higher ground to the south at a much later date after the abandonment of Wall 4.

¹⁹⁸ The date is based on a few black-glazed sherds found in the earth fill behind it. It has not been verified whether this terrace linked up with the unfinished Wall 3/4 line to the west and its relation to the undated terrace to the east (p. 73 above, Fig. 18, a) is also unknown.

¹⁹⁹ The adze-dressed Lesbian and ashlar blocks incorporated in it had presumably been plundered. See Plate 18 b and c (from photographs by J. M. Cook and B. B. Shefton respectively).

²⁰⁰ The stone fill backing the upper part of the face has not survived the destruction of the latter, save for a thin layer of stones from the bottom of it securely bedded in the fallen mudbrick beneath. But even these were preserved up to about 8 metres above sea-level to the south at Fig. 21, m.

²⁰¹ Fig. 21, o, Plate 18 b and c.

²⁰² The column-drums themselves will be dealt with in the forthcoming temple publication.

²⁰³ It would seem most reasonable to assume that the terrace wall passed out of use when Wall 4 was built. If so, a late dating for Wall 4 becomes probable. See further p. 135.

²⁰⁴ But at w, though not at x, also associated with occupation debris of the sixth to fourth centuries B.C.

²⁰⁵ The full account of the extremely complex sequence of building in this area will be found in the detailed treatment of the domestic architecture of the city to be published separately. A few advance judgements are given here simply to enable the fortifications to be correlated with the occupation sequence. They doubtless involve many errors that the thorough consideration of the house-architecture may be expected to set at rights and certainly are very far from being final. As here presented they are simply based on an examination of the sections and have drawn very heavily on the preliminary assessments of the pottery evidence. Because this is so essentially a preliminary statement and because I do not wish to prejudice the final judgement, in the plan (Fig. 24) I have incorporated all the principal building remains from Middle Geometric times until the late seventh century B.C. without differentiation, though not normally those of earlier or later date. Their significance will emerge more clearly with the publication of the individual plans of the different building periods drawn up by our Turkish colleagues, of which Fig. 24 is a simplified conflation.

²⁰⁶ Extending, if we may judge from the probe in Jxviii–xix (pp. 9 and 39), far down beneath modern sea-level.

²⁰⁷ p. 68 and Plate 8.

²⁰⁸ Plate 9, Fig. 24 in ‘window’ b. The position of another probe through the Wall 1 fill to this Monochrome debris is indicated by the ‘window’, Fig. 24, c, but neither of these cuts is shown to its full extent in the interests of not obscuring the later walls above.

²⁰⁹ See the section, Fig. 25. I regret that I have not a detailed north–south section of this cut to hand and so have to proffer an east–west one. This, too, has its idiosyncrasies, the topmost seventh-century level being shown on the axis of the stepped cellar, but all the levels beneath it as they appear in the south face of the cut.

²¹⁰ p. 102.

²¹¹ In the small areas cleared no traces of stake holes for a palisade or similar defence work were observed. There may, however, have been localized terracing. To the north-east of this area (see the section, Fig. 25) there is a considerable depth of Protogeometric occupation on the edge of the tell and the Protogeometric and Early Geometric burnt strata at this point do not slope down towards the north but, if anything, slightly upwards. Whatever the cause, this area seems to have blossomed into the highest part of the tell in the early Iron Age.

²¹² The terms ‘Early’, ‘Middle’, and ‘Late Geometric’ as used in this account are those of the preliminary pottery classification. It is to be observed that the Early Geometric phase as then defined may represent a much shorter period than the middle and late ones.

²¹³ pp. 40 and 102.

²¹⁴ pp. 41, 122.

²¹⁵ On Fig. 24 the area occupied by this fill is shown in black.

²¹⁶ n. 211 above, Fig. 25.

²¹⁷ But there seems, quite apart from this, to have been something of a fashion in sunken floors in the early Greek levels.

²¹⁸ See also p. 122.

²¹⁹ Saddle querns occur throughout. A similar cupped door-socket stone was found in the tenth monochrome level in Jxviii–xix.

²²⁰ This jar may belong with a great depot of sunken pithoi further to the west, apparently representing a major store of pre-Protocorinthian Late Geometric date. For simplicity, these have been omitted from the plan, Fig. 24.

²²¹ This seems the most likely interpretation. See, however, p. 126.

²²² Entirely of pre-Protocorinthian date.

²²³ It seems to have been abandoned before the end of the century, perhaps around the end of the third quarter, yielding not only local vases and early Ionic graffiti but also imported Attic Geometric and Protocorinthian wares on its floor. The walls of this house may be seen at the bottom of Plate 19 b. For Protocorinthian pottery from its floor see p. 139 (Group D—the building had been assumed to be apsidal until the remainder of it was cleared in 1952).

²²⁴ pp. 79. 125 f.

²²⁵ Cf. p. 103.

²²⁶ As against the lower part's being taken for Wall 2 and the upper part for Wall 2/3 speaks the apparent unity of the pack.

²²⁷ Plate 19 b and c, taken during the excavation by our Turkish colleagues before the deep sectional cut was made to the north-west.

²²⁸ p. 47.

²²⁹ pp. 79, 134.

²³⁰ e.g. outside and under the later wall face at Fig. 24, x and z.

²³¹ Fig. 24, r and s, Plate 9; on the siege-mound see pp. 88 ff.

²³² See also Plate 19 d.

²³³ But these are isolated pieces and seem insufficient in themselves to establish that this is more of the Wall 3/4 construction (p. 81) as opposed to being of Wall 4 date.

²³⁴ The whole of the glacis was fully exposed up to the time of the abandonment of the city in about the third quarter of the fourth century B.C., as was shown by the pottery found against it.

²³⁵ p. 81.

²³⁶ I am indebted to Professor M. S. Şenyürek for making a preliminary pronouncement on these bones as being apparently those of a man in his forties and as showing a degree of fossilization, as compared with the early Hellenistic sarcophagus-burial in N3.xiv4–5, that suggested that they were of a respectable date in antiquity.

²³⁷ It seems to represent a somewhat summary disposal of remains (perhaps those of an executed criminal?).

²³⁸ Formidable enough when one remembers that the top of the occupation platform inside the city even at the close of the seventh century B.C. when this mound was built only occasionally exceeded 10 metres above sea-level. On the mound see further pp. 24 f. and 128 ff.

²³⁹ e.g. the classical stepped glacis, p. 87.

²⁴⁰ See Section α–α′, p. 2, Fig. I.

²⁴¹ Extending about as far as the 12-metre contour on the plan of the tell, Plate 74.

²⁴² Op. cit. 167–8; Miltner, H., Eski İzmir 16–17Google Scholar; see p. 36 above. To the traces of beams of round section found by Dr. Baki Öǧün and described below should probably be added the square ones reported by the Miltners.

²⁴³ The full sectioning of the mound and the unravelling of what was probably a crowded and exciting sequence of events must await the advent of subsequent investigators with resources many times greater than those at our disposal.

²⁴⁴ pp. 86 ff.

²⁴⁵ It was so located to avoid the Miltners' probes (n. 242 above) as well as embarrassment to any subsequent sectional work on the axis itself. Actually, however, there had been previous excavation at this point, too, represented by a trench 2 metres or more wide and reaching in part to a depth of 15·7 m. above sea-level. It was dated by a modern Greek orthodox medallion. This probably extends to the ε–ε′ section line (Fig. 27) just south of the north-west corner of the probe where the strata of mudbrick debris have vanished, but the limits of the disturbance could not be clearly distinguished when the section was drawn (see n. 246 below).

²⁴⁶ These are by no means as informative as I should like them to be, having had to draw them at the end of a long rope dangling over a crumbling abyss. Probably the masses of compact soil shown in the sections cloak many gradations in the fill that it was not possible, under the circumstances, adequately to differentiate.

²⁴⁷ Plate 20 a, from a photograph by D. T. D. Clarke, shows the impression left by one of them in the south-west part of the cut.

²⁴⁸ As with the nexus of beams and branches, Plate 20 b, photographed by D. T. D. Clarke.

²⁴⁹ The slope from east to west shown by the burnt stratum, Fig. 27, j may well be fortuitous, although it is not impossible that there may have been more than one original crown to the mound.

²⁵⁰ p. 131 below.

²⁵¹ Except that no weapons were recovered from the narrow cut extending down from 11 metres above sea-level, although this circumstance may be entirely due to the smallness of the area tested.

²⁵² Cf. p. 87.

²⁵³ p. 87, Fig. 24, r and s.

²⁵⁴ pp. 91 ff.

²⁵⁵ A preliminary analysis of the distribution of the different arrowhead types over the surface of the tell and outside it is given on pp. 129 ff.

²⁵⁶ Many of the branches and beams that have toppled down to the south in the ε–ε′ and ζ–ζ′ sections may have served such a purpose before they were flung holus-bolus into the fill. This seems even more likely in the case of the coherent mass of timber reported by the Miltners (see n. 242 above).

²⁵⁷ This fire, too, seems evidenced by the arrowheads found inside the city to the south in the probe in Kxi. See p. 132.

²⁵⁸ The incidence of loomweights, wash-basins, oven-dishes, &c, seems to preclude the possibility that it might have come from graves, the only tenable alternative.

²⁵⁹ pp. 15, 125, 128.

²⁶⁰ Plate 1, where the area apparently involved is shown stippled.

²⁶¹ p. 92.

²⁶² Like the seventh-century houses to the north of the larger probe in Kxi, it was very well preserved as a consequence of having been buried under the spill from the mound. Cf. pp. 88 ff.

²⁶³ Indeed, if the whole area was as densely built up as at that point, then probably to the west of the block partially cleared in Kxi there may have run a narrow lane followed by another block two houses deep, separated by a further lane from the house in K9.x–x1. It would seem unlikely that there was room for further building on the platform beyond again to the west.

²⁶⁴ In K7.viii7–K9.ix2 the mudbrick debris, possibly of Wall 2 date, continued on down below 3 metres above sea-level.

²⁶⁵ As compared, e.g., with the buildings to the north-east in Kxi.

²⁶⁶ Op. cit. 167; p. 36 above.

²⁶⁷ See p. 36 and the references there given.

²⁶⁸ Plate 1. On the question of the wall lines possibly linking this area with the tell see also pp. 128 and 135 f.

²⁶⁹ pp. 15, 91, 125, 128.

²⁷⁰ From K.vii6 to K4.vii9.

²⁷¹ Thus the possibility that the face preserved may have been an inner one cannot be ruled out.

²⁷² Plate 20 c, from a photograph by A. G. Hunter.

²⁷³ Probably that already observed farther to the north (p. 92) as being produced by the decay of the wall bounding the west side of the tell, perhaps itself a construction of Wall 2 date.

²⁷⁴ The maximum probed at Fig. 29, ƒ. Quite probably there may have been an inner face to this wall here on the lower, flatter ground.

²⁷⁵ Plate 14 C.

²⁷⁶ JHS lxxi. 248, fig. 7, to left behind ranging pole. The reader is also referred to the forthcoming detailed publication of the Smyrna temples.

²⁷⁷ With so little exploration done no sure interpretation is possible. I can only advance the very tentative suggestion that these walls may perhaps have fronted on an apron-like enclosure behind a main gate. The south-east ‘extension’ (if such it be) might then have served to revet the side of a road from it up on to the tell.

¹ An exception, where it occurred several times, was the mudbrick pack at Fig. 10, d(p. 56). It is an ingredient in the modern baked bricks from the area.

² A heavy stratum of similar hard, dark, clayey soil commonly occurred immediately above the floors of the houses (cf., e.g., Plate 9). The mudbrick debris from the walls seemed usually to have come down rather later on top of this earth stratum. Perhaps it was used because it did not crack as readily as clay, but I doubt if it could have been rendered impervious without a heavy plastering of clay or some similar material.

³ p. 70 above; Figs. 18, f, 19; Plates 8 and 17 b.

⁴ Plate 17 b.

⁵ Plate 8.

⁶ The closest parallel for this type of construction seems to be that afforded by a rather ill-preserved seventh-century circuit wall at Corinth, (Corinth xv. i, 14, pls. 4.A (top right), 51Google Scholar; xv. ii, pl. 60). Here the faces of the socle seem to have been of stone but its ‘core’ to have been composed of earth fill packed into long chambers (about 5 m. long and 1·8 m. wide) separated by boulder partitions. One may also compare the use of earth fill in later Greek walls, e.g. at Gortys in Arcadia, (BCH lxxi–ii. 95Google Scholar, fig. 7) and again at Corinth, (Corinth iii. ii, 93).Google Scholar I would, however, draw attention to the purity of the earth fill in the Smyrna tower and the way it has been packed in hard. Its resistance to collapse (see Plate 8) is remarkable. In the excavated parts of the Smyrna fortifications it is unique. On the changed construction at this point see p. 97 n. 17.

⁷ pp. 93 f.

⁸ Dry-stone fills, like those of the modern terrace-walls around the tell, seem never to have been used in the ancient fortifications.

⁹ p. 106 below.

¹⁰ However, the Wall 4 fill outside the Wall 3/4 face to the north-west of the temple (p. 81) constituted a very even pack of somewhat larger river stones.

¹¹ As to the geological identification of these stones, it should be pointed out that we have had no reports on our specimens. Philippson, A. (Petemanns Mitteilungen, Heft 172Google Scholar, 18) describes the whole of the south slope of the Yamanlar Daği, embracing the areas of both these quarries, as of andesite. The grey-stone quarries he describes at length; the red-stone ones he does not mention. Other writers have alluded to the local stones as trachyte.

¹² p. 68; Fig. 19; Plates 16 b, 17 a, b, 18 d.

¹³ p. 43.

¹⁴ p. 68; Plate 16 a; Figs. 16–17.

¹⁵ e.g. by the unfinished column drums from the temple, p. 81, Plate 18 b and c.

¹⁶ On the large threshold block, p. 50, Fig. 6.

¹⁷ Basically the andesite socle may be considered as fulfilling precisely the same role here as elsewhere. The change is in the superstructure, where here white stone ashlar with an earth core has been substituted for mudbrick. Such a change in two stages has been argued in the fortifications of the sixth city of Troy and was clearly shown at Tower Vl.g (Dörpfeld, Troja und Ilion i. 143 ff.; Blegen, Caskey, Rawson, Troy iii. 82 ff.), where traces of an earlier mudbrick construction were found preserved. However, in the Smyrna Middle Geometric circuit the all-stone construction at the gate seems to have been an original part of the fortification project, enhancing the beauty and strength of the gate as the most commonly frequented and most vulnerable part of the defences. Presumably considerations of economy prevented its more extensive use as in the all-stone superstructures of Troy VI (see above) and the Phrygian circuit at Gordion (AJA lix. 12 ff., figs. 26–27, pls. 6–9; lx. 257 ff., pls. 87–91).

¹⁸ Except that the somewhat irrelevant small, thin spacer-blocks sometimes used in the fitting of the larger ones are not here shown. In arrangement this masonry bears comparison with apparently contemporary but distinctly more regular work in Palestine, e.g. Crowfoot, , Kenyon, , Sukenik, , Samaria-Sebaste i, pls. xii. 2, xiii, xiv. 1.Google Scholar But the Wall 1 circuit appears otherwise essentially Greek in conception and perhaps the masonry of the superstructure of its tower is simply a direct local development from Mycenaean ashlar work. The use, however, of an ashlar superstructure may possibly be of Anatolian inspiration (see n. 17 above), but its finish and fitting on this more modest Smyrna construction are far superior in quality to the more ambitious work at Troy and Gordion.

¹⁹ p. 68.

²⁰ The outer face of the Wall 3 socle on the east side of the city tended to be better finished towards its top, the highest courses being at some points worked with the adze.

²¹ Here a consistent run of technical changes enables a measure of independent stylistic dating. This material will be published in the section on the Iron Age domestic architecture.

²² The reader is referred to the forthcoming publication of the seventh-century temples at Smyrna. These seemed to suggest that a purely angular polygonal was more current in the earlier part of the century than the developed Lesbian. On the significance of these technical terms see the next note.

²³ The terms ‘rubble’, ‘polygonal’, ‘Lesbian’, &c, are used as far as possible with the connotations given them by R. L. Scranton (Greek Walls,passim), but without the chronological implications there attached to them.

²⁴ P. 93, Plate 20 C.

²⁵ p. 86, Plate 19 b, C.

²⁶ pp. 40, 68, Fig. 19, Plates 10 a—includes re-used material probably from it—17 a, c.

²⁷ pp. 40, 43, 53, 56, 64, 71, Plate 10 a.

²⁸ pp. 48, 52, 55, 58, 65, 66 ff., 72 f., Plates 12 c and d, 16 a, c, 15, c background (note that this face had been partially plundered between excavation seasons; cf. Fig. 13).

²⁹ p. 48 n. 60, Plates 7, 12 d; p. 67, Fig. 16, Plate 16 a.

³⁰ p. 58, Fig. II.

³¹ p. 72, Plates 8, 16 c—marked ‘3’.

³² c. 640 B.C. This structure will be amply illustrated in the publication of the seventh-century temples.

³³ Where it presumably formed part of a temple platform anterior to that whose cella had the internal Aeolic columns—Koldewey, Neandria (51st Berliner Winckelmannsprogramm) 26–27, fig. 52.

³⁴ AJA xxxvii. 280, fig. 11; Noack, , Eleusis pl. 21e.Google Scholar

³⁵ AJA lx. 259–60, pls. 87; 89, fig. 32; 91, fig. 34.

³⁶ pp. 87, 89, 114, 135, Plates 9, 19 d. Of the oriental variety, good examples in stone lie round parts of the Hittite circuit at Boğazköy; but the closest in space and time to the Smyrna structure is the stepped glacis round the Persian defences at Gordion, (AJA lx. 253–4, pls. 83–84, figs. 11–15).Google Scholar However, in view of the several uncertainties remaining so far as the Smyrna structure is concerned, it would perhaps be rash at this stage to infer from it a Persian defence system there also.

³⁷ One possible interpretation is that these ledges may have carried vertical posts supporting projecting battle ments (much like the inner face projection of the chemin de ronde suggested on p. 114); this lends some point to Iliad xii. 258 and disposes of the objection that a pronounced stepping-back of the crowning of the socle otherwise makes the base of the wall dead ground from the point of view of the defenders mounted above; against it speak later Greek practice and the absence of timber traces where they might well be expected to survive in the debris outside the outer face. Projecting battlements are known on a model from Urartu (Iraq xii. 5 ff., pl. 1), but there they are simply carried on horizontal beams protruding from the top of the wall.

³⁸ Iliad xii. 258, 444; Lorimer, H., Homer and the Monuments 433.Google Scholar In view of the poor preservation of the Wall 2 outer face, however, it is probably wiser at this stage not to exclude the possibility that the first appearance of this form of construction may have fallen rather earlier.

The rendering of ‘battlements’ is also known for this word from the lexicographers but, although the stepped triangular crenellations so widespread under the Persian Empire (p. 109) might perhaps aptly illustrate the stepped pyramidal form of construction which Herodotus (ii. 125) may have envisaged, this interpretation is more inadequate in the Homeric contexts. If the structural suggestions advanced in n. 37 above can be substantiated, a confusion between the battlements and the associated stepping of the socle face below might seem not unnatural, but here we embark on pure speculation.

³⁵ pp. 41 ff., 52, 57 f., 65, 66, 72 f., 86, Plates 10 b, c, 13 b, 19, b, c.

⁴⁰ pp. 50, 53, 56, 64, 65, 68, 73 f., 81, 87, Plate 16 c foreground.

⁴¹ p. 93.

⁴² pp. 40 f.; cf. pp. 107, 117.

⁴³ Such a mudbrick ‘core’ is attested even as late as the close of the classical period, Corinth iii. 2, 282 ff.

⁴⁴ The way in which it has resisted collapse in Wall 2 in Section λ–λ′ is particularly noteworthy. See Plate 7, p. 43.

⁴⁵ p. 106.

⁴⁶ Cf., however, p. 95 n. 1.

⁴⁷ Sea-shells occurred as well, e.g. in the Wall 3 brick in Section λ–λ′. Pottery and other refuse were to be found in quantities that varied considerably and presumably depended on the sources of the clay, sand, &c., employed.

⁴⁸ Occasionally its carbonized flecks were visible, but usually not. However, as all mudbricks from the site that have accidentally been baked by fire while in an intact state seem to show clear straw impressions, it was probably used throughout. The significance, however, of a very thin carbonized coating occurring on the surface of some mudbricks immediately under the mortar remains unknown.

⁴⁹ See the diagram, Fig. 33.

⁵⁰ Cf. Aristophanes, Birds 1143 ff.

⁵¹ It must, however, be stressed that unbaked bricks in their moist condition underground become quite pliant and suffer all manner of distortions to their original shape and size. For this reason isolated measurements are quite useless and only averages struck over a very great number of the more perfect examples occurring in a pack have any validity. The dimensions given here have been arrived at in this way. On their relation to the original sun-dried size of the bricks see p. 104 below.

⁵² ii. 3, 3–2.

⁵³ e.g. at Assur (Andreae, , Assur ii. 14)Google Scholar and Babylon, (Koldewey-Wetzel, , Die Königsburgen von Babylon ii. 71Google Scholar, s.v. Ziegelmasse—a useful table of dimensions, but in this instance essentially for baked bricks).

⁵⁴ e.g. Schmidt, , Persepolis i. 159Google Scholar (treasury), 260 (harem of Xerxes).

⁵⁵ Naumann, R., Architektur Kleinasiens 46.Google Scholar

⁵⁶ P of M ii. 519.

⁵⁷ Pernier, , Festos i. 228, 299, 354Google Scholar; Durm, , Baukunst der Griechen 37Google Scholar, fig. 19 (from Gournia); P of M iv. 69, BSA xxx. 59 (these last perhaps only fragments).

⁵⁸ e.g. at Lerna, (Hesperia xxiii. 24, xxiv, pl. 23, f)Google Scholar, Tiryns (Tiryns iii. 83) and Korakou, (Blegen, , Korakou 75).Google Scholar Possible rectangular variants are less well attested (Blegen, , Zygouries 12Google Scholar, Goldman, , Eutresis 31)Google Scholar and may in part be the product of poor preservation or the necessary manufacture of fractional bricks. But rectangular bricks had been used early in the north (Tsountas,Προϊστορικαὶ Ἀκροπόλεις Διμηνίου και Σέσκλου 79, 110, 127).

⁵⁹ Goldman, , Eutresis 61, 63Google Scholar (0·3 × 0·15 × 0·08 m. A solitary square brick also occurred in a Middle Helladic pit.)

⁶⁰ History of Technology i. 475.

⁶¹ Petrie, , Naukratis i. 89 ff.Google Scholar

⁶² e.g., cf. the bricks of Troy VI whose dimensions have been most recently listed by Naumann, op. cit. 46; sixthcentury B.C. bricks at Gordion measured 0·48 × 0·28 × 0·1 m. (Archaeology vi. 159 ff.).

⁶³ Nauman, loc. cit. Rectangular bricks also occurred amongst the Etruscans (e.g. Rivoira, , Roman Architecture 3)Google Scholar and the Romans (Vitruvius, loc. cit.).

⁶⁴ Schliemann, Tiryns (English ed.) 260: Mycenae: ? × 0·35 × 0·08–0·09 m.; Tiryns: 0·47–8 × 0·36 × 0·1, 0·52–3 × ? × 0·09, 0·43 × 0·25–0·26 × 0·09 m.

⁶⁵ Noack, , Eleusis 70Google Scholar (Mycenae: ? × 0·285 × 0·08, 0·59 × ? × 0·085 m.). The identification of the buildings involved is not very clear and the slight possibility remains that some of Noack's (and Dörpfeld's) bricks may be from post-Mycenaean structures); Blegen, , Zygouries 37Google Scholar (0·35 × 0·22 × 0·085 m.); Heurtley, , Prehistoric Macedonia 24Google Scholar, 102 (c. 0·35 × 0·28 × 0·07–0·08 m.).

⁶⁶ SCE ii. 476 (at Idalion: 0·42 × 0·225 × 0·12, 0·42 × 0·31 × 0·12, 0·525 × 0·42 × 0·12 m.).

⁶⁷ Liverpool Bulletin ii. 1–2. 41 (at Paphos: 0·55 × 0·32 × 0·13 m.).

⁶⁸ Kübler, , Kerameikos v. i. 10Google Scholar: 0·49 × 0·44 (0·41) × 0·09 (0·1) m. Similar bricks from Athens, possibly of Geometric date, reported in BCH lxxxi. 507: 0·50 × 0·45 m. × thickness not stated.

⁶⁹ Noack, , Eleusis 69Google Scholar: 0·44–5 × 0·24–6 × 0·095–0·1 m. The date of the brickwork measured by Noack seems most uncertain. Perhaps it was, as he took it to be, original to the ‘Peisistratan’ circuit. But most of the superstructure of this wall seems likely to have been renewed in early classical times. Perhaps the variant dimensions quoted by Philios (Ἐλευοίς 102: 0·45 × 0·45 × 0·08 m.) relate to this fifth-century work. Cf. also n. 70 below.

⁷⁰ ADelt xiv, παράρτ. 14: 0·44 × 0·44 · 0·09 m. (apparently Cimonian work).

⁷¹ Arvanitopoulos, Θεσσαλικὰ Μνημεῑα i. 77: 0·5 × 0·33 × 0·08–0·088 m. Noack, (Eleusis 70 n. 2)Google Scholar quotes a variant thickness of 0·095 m.

⁷² AA 1954, 644: 0·4 × 0·4 × 0·07–0·08 m. Cf. also p. 107 n. 132.

⁷³ Olynthus viii. 49, 225; xii. 225 n. 6: 0·265 × 0·25 × 0·084 n.; variants: Olynthus viii. 40, xii. 225.

⁷⁴ Corinth iii. 2, 282 ff.: 0·45 × 0·45 × c. 0·09 m.; also halfbricks; late fourth-century work.

⁷⁵ IG ii². 1672, ll. 55–57. This interpretation, originally put forward in terms of the Corinthian bricks (n. 74 above), seems the most attractive, even though it involves the difficulty of the continued use of the short foot in Attica for some brickwork at least at a time when only the long foot seems to have been used in monumental architecture. On the metrological aspects see pp. 104 ff.. The alternative explanation of treating τριημιπόδιοι as a synonym for Λύδιοι (e.g. Noack, , Eleusis 70 n. 2)Google Scholar and of applying it to such bricks as those at Demetrias (n. 71 above; interpreted in terms of the long foot of 0·327 m.) seems much less likely.

⁷⁶ e.g. Urschweiz xvi. 61 ff.; Bittel, and Rieth, , Die Heuneburg an der oberen Donau 22Google Scholar: 0·4 × 0·4 × 0·07 m., i.e. πεντάδωροι of the 0·327-metre foot like those at Gela.

⁷⁷ pp. 40, 82.

⁷⁸ The greater length of 0·57 m. (instead of 0·51) was also attested in brick possibly of this date (p. 56, Fig. 10, d).

⁷⁹ In Gxii. They were laid as stretchers throughout, save that the bricks themselves had to be broken to fit round the curves.

⁸⁰ In Jxix. This and the oval house just mentioned will be dealt with in detail in the publication of the domestic architecture of the city.

⁸¹ e.g. in the debris from a possible Late Bronze Age circuit (p. 82) and in the ninth Monochrome level in Jxviii–xix. The other dimensions have not yet been established, but must have been over 0·27 by over 0·19 m.

⁸² See the section, Fig. 25.

⁸³ Cf. p. 121.

⁸⁴ p. 43.

⁸⁵ pp. 53, 56. But a variant length of 0·57 m. also occurs (p. 96 n. 100). See also p. 91.

⁸⁶ p. 71.

⁸⁷ p. 72, Fig. 18, u.

⁸⁸ p. 77, Fig. 21, l and m.

⁸⁹ p. 77, n. 188, Fig. 21, i.

⁹⁰ p. 53.

⁹¹ Πλαίσια (Aristophanes, , Frogs 799 ff.)Google Scholar rather than ταρσοί (FD iii. v, 131 ff., no. 26, 1. 13; BCH lxxx. 514 ff.).

⁹² c. 0·5 × 0·3 × 0·08 m. to the north-west of the temple (p. 79) and 0·51 × 0·3 × 0·07–8 m. in the main occupation probe (p. 86). Also, if, as seemed likely, the bricks of the superstructure of the ‘Wall 3’ tower at the north-east gate measured c. 0·5 × 0·3 × 0·08 (p. 72), it becomes most attractive to assign its building to this stage as well.

⁹³ pp. 47, 53, 58, 79.

⁹⁴ To be published in detail in the forthcoming account of the Iron Age domestic architecture.

⁹⁵ The mudbrick wall, perhaps of a shed, in Section λ–λ′ (p. 49, Plate 7, Fig. 2, y).

⁹⁶ For this also the reader is referred to the coming publication of the Iron Age domestic architecture.

⁹⁷ p. 72, Fig. 18, u.

⁹⁸ Cf. below.

⁹⁹ p. 65.

¹⁰⁰ That the difference arising in this way was very slight is also suggested by a comparison with similar bricks accidentally baked by fire in the occupation levels.

¹⁰¹ See immediately above.

¹⁰¹ The figure apparently suggested by the fifth-century East Greek metrological relief in Oxford (Chandler, Marmora Oxoniensia no. 166; Michaelis, , Ancient Marbles in Great Britain 559 no. 83Google Scholar; JHS iv. 335 ff., pl. 35; Langlotz, , Bildhauerschulen 133, 136Google Scholar; Hesperia Supplement viii. 338 n. 10; Berriman, A. E., Historical Metrology 131 ff.Google Scholar Regrettably the two most recent attempts to touch on this relief from a metrological point of view seem to disregard the small but real progress made over the past two or three generations in the understanding of Greek linear standards.) I suspect that with the earlier material with which we are here dealing the unit may have been slightly longer, possibly approximating to the later Roman foot of 0·296 m. or perhaps being even longer still. But we cannot expect to arrive at a precise figure from mudbrick standards alone. Later the unit became distinctly shorter, at least to judge from the figures established with considerable precision at Priene (Wiegand, and Schrader, , Priene 86)Google Scholar and Didyma, (ÖJh xxxii. 132).Google Scholar It is simply in the hope of avoiding confusion that the length of 0·294 m. fixed there is elsewhere employed throughout in in this account.

¹⁰³ p. 101 n. 52.

¹⁰⁴ ii. 3, 3. Probably quite to be distinguished from the πλίνθοι τριημιπόδιοι (cf. p. 102 n. 75).

¹⁰⁵ See above.

¹⁰⁶ AJA xxi. 147 ff.

¹⁰⁷ e.g. Olynthus viii. 45 ff.

¹⁰⁸ In the table that follows foot indicates the unit of 0·294 m. (but cf. n. 102 above) and FOOT that of 0·327 m.: Protogeometric houses, &c., and Wall 1:

Greco-Anatolian rectangular: 1¾ feet (variant 1¾ Feet) × 1 foot × ⅜ Foot.

Wall 2 (except north part) and Wall 2/3:

Greco-Anatolian rectangular: 1¾ feet (variant 1¾ Feet) × 1 foot × ¼ foot.

Wall 3, later seventh-century houses, &c.:

Greco-Anatolian rectangular: 1¾ Feet × 1½ feet (= c. 1⅜ Feet) × ¼ Foot.

Sixth-century houses:

Λύδιοι: 1½ feet × 1 foot ×¼ foot.

Fourth-century furnace:

Τριημιπόδιοι half-bricks ?: ? × ¾ Foot × ¼ Foot.

This table is an essentially provisional attempt to interpret the dimensions of the bricks in terms of the two established Greek feet. If the interpretation placed on the fourth-century bricks is justified, it will necessitate some modification of the statement above (p. 101) about the absence of fractional bricks, although the position in relation to the earlier material and all the packs in the fortifications remains unchanged. At this stage such a table cannot be regarded as definitely proving the coexistence of the 0·327-metre and the 0·294-metre feet in Ionia as early as Protogeometric times. However, such features as the early fluctuation in length between 0·57 and 0·51 m. render such a hypothesis extremely attractive. Should it become an established fact, the inference would be that the conflicting linear standards of classical Greece, oriental in their remotest origins, must first have appeared in Greek lands prior to the Iron Age amongst the conflicting cultures of the Bronze Age. Already a foot of c. 0·3 m. has been detected in the Early Helladic palace at Lerna (a longer version of the 0·294-metre foot?) and I am indebted to Miss E. B. Wace for the observation that measurements in the houses at Mycenae seem to be in terms of a unit of c. 49 cm. (the 0·49-metre ell of the 0·327-metre foot?). The brick dimensions quoted on p. 101 are also of great interest in this regard, notwithstanding that it is not always possible to evaluate the different degrees of reliability with which they have been taken. But much more than this is needed and above all else a systematic study of the major Bronze Age buildings as architecture. One of the lesser achievements of such work would probably be to solve the riddle now before us.

In the above account I have taken it that both length and width of the earlier rectangular bricks are fixed directly in terms of the linear standard rather than, as commonly today, through the medium of an unalterable length-width ratio dictated by the bond. For a bond requiring a c. 5:3 ratio of Wall 1 and 2 brick and a c. 5:4 ratio of Wall 3 brick argues such a monumental type of construction as we have little reason to expect outside military work; and in the superstructure of the fortifications the battered faces must have prevented the systematic use of any single bond; rather must broken fractional bricks have been of necessity employed, as they are in the preserved mudbrick ‘cores’.

¹⁰⁹ The only measurements to which this unit has any possible application at all in the Smyrna bricks are the thicknesses of the Wall 1 and 3 brick and of that of the fourth century B.C. But since it has no relevance to the greater measurements it becomes most unlikely that it was ever used at all in this connexion. The very existence of the unit itself has been categorically denied by Dinsmoor, (Architecture of Ancient Greece 124, 137, 257).Google Scholar So far as classical times are concerned, the chief source for its employment is probably Herodotus ii. 168. However, it is noteworthy that Herodotus does not designate the Egyptian unit that he equates with the Samian ell as ‘royal’. Hence from this passage alone it cannot definitely be said whether the ‘Samian ell’ alluded to is that of 0·525 m. (equivalent to the Egyptian royal ell) or the 0·44-metre ell of the 0·294-metre foot (themselves probably ultimately of Egyptian origin and approximately equivalent to the ordinary Egyptian units). In one other passage Herodotus (i. 178) specifies the royal ell. However, here he would seem to be dealing with a Babylonian or Perso-Babylonian unit. It is 3 dactyls longer than the ‘ordinary ell’, presumably a Greek unit. If the Ordinary ell’ is the 0·44-metre ell of the 0·294-metre foot, this would give the ‘royal ell’ a value in the vicinity of the 0·49-metre ell (actually a little over) of the 0·327-metre foot, which is not unreasonable since both of these measures probably originally hail from this periphery. In that case Herodotus, in using the short foot, would be speaking rather as an Ionian, although, as we have seen, no very hard and fast distinction is possible. But it would be remarkable in that case that no comparison is drawn with the longer Greek unit. If, however, he were addressing himself rather to his Athenian audience, then the ‘Ordinary ell’ would be the 0·49-metre ell of the 0·327-metre foot and the ‘royal ell’ would have a value of c. 0·55 m. Either way, these passages do not justify the equation of the Perso-Babylonian and the Egyptian royal ells as made by Pliny (N.H. vi. 26) and followed by Berriman (op. cit. 29 and passim). The further solution of this problem is the domain of the orientalist, not the classicist. We are left with the issue of whether the Samian ell is a measure of 0·44 m. or of 0·525. In the case of the Assos standard (Clarke, , Bacon, , and Koldewey, , Investigations at Assos 71 ff.)Google Scholar, the actual linear standards were marked on metal bars, now plundered, that were set in the stone; but we may take it as certain that the units themselves were somewhat shorter than the surviving cuttings made to receive the bars. Not very much more can be said for sure, but it is certain that the longer of the fathoms shown must have been shorter than that of the 0·525-metre ell. In the case of the Oxford metrological relief, for which a Samian provenance has been suggested (n. 102 above), the left hand is a modern restoration and the total length of the fathom shown can only be guessed at. However, if any symmetry is to be allowed to the carving, the unit shown must approximate to the fathom of the 0·525-metre ell and must surely be distinctly longer than that of the 0·49-metre ell. Here let the matter rest. The Oxford relief, without recourse to the less secure evidence of inductive metrology, suggests that Dinsmoor's rejection of the long Samian ell may have been somewhat hasty, although it is possible that that unit may not have been used very extensively. See also p. 137.

¹¹⁰ On the metrological issues involved see further Lehmann-Haupt, RE s.v. ‘Stadion’. On the determination of the 0·294-metre foot see p. 104 n. 102 above; on that of the 0·327-metre foot (the actual figure quoted is Dinsmoor's) see AM xv. 167 ff., Paton, and Stevens, , Erechtheum 277 ff.Google Scholar

¹¹¹ pp. 96, 100.

¹¹² p. 43.

¹¹³ Such an explanation for the curious construction at E3.xix3 (p. 66) seems hardly tenable.

¹¹⁴ ÖJh xxvii Beiblatt 162.

¹¹⁵ The arrangements in this regard shown in Fig. 34 are purely conjectural.

¹¹⁶ e.g. Eleusis, , ‘Peisistratan circuit’: four such drains shown on the plan, Kourouniotis, Eleusis pl. 1Google Scholar, two of them shown in greater detail by Noack, (Eleusis 40, figs. 17–2, 71, fig. 33, pl. 24.b)Google Scholar; Miletus, , Kalabaktepe, : Milet i. 8, 28 fig. 20Google Scholar; Buruncuk-Larisa, : Larisa am Hermos i, pl. 5.b.Google Scholar

¹¹⁷ The reader is referred to the forthcoming publication of the domestic architecture of the site. Two such sixthcentury drain outlets appear on the section, Plate 9.

¹¹⁸ pp. 52 f.

¹¹⁹ Wall 4 seems to have had no distinct superstructure, being apparently an all-stone construction perhaps continuing no higher than the platform of the tell, and the same may well have been true of the small part of the Wall 3/4 line (p. 81) actually completed.

¹²⁰ That these were of considerable height and bulk is usually clear from the extent of the debris from their collapse where this has been left relatively intact.

¹²¹ Cf. p. 97 n. 17.

¹²² pp. 68, 97.

¹²³ p. 72, Plate 8.

¹²⁴ pp. 88 ff., 128 ff.

¹²⁵ p. 100.

¹²⁶ Of course these generalizations only apply to mudbrick-and-stone construction. It would seem most reasonable to assume also an unbroken tradition in all-stone Greek fortifications (cf. p. 818).

¹²⁷ But with some exceptions. Cf. p. 100.

¹²⁸ Cf. Pausanias viii. 8, 8; Olynthus viii. 228 ff.; I am also indebted to Mr. J. K. Anderson for his observations on the resistance of solid mudbrick packs to modern artillery.

¹²⁹ pp. 68 ff., 97, Figs. 19, 32, Plates 8, 16 b, 17 a, b, 18 d.

¹³⁰ p. 72, Plate 8.

¹³¹ Kourouniotis, , Eleusis, a Guide to the Excavations and the Museum 11, fig. 3.Google Scholar

¹³² AA 1950–2, 263, fig. 56; 1954, 643 ff., figs. 95–96; AJA liii. 386–7, pl. 53 C–E; lix. 311–12; JHS Archaeological Reports 1955, 53, fig. 4; Archaeologia Classica ix. 44 ff., pls. 39–40; inaccessible to me in writing this have been: Atti Acc. Agrigento 1953–4, 18 ff.; Griffo, Scavi delle fortificazioni greche in località Capo Soprano; idem, Agrigento—Guida ai monumenti ed agli scavi.

Less appositely, one may also draw attention to the wellpreserved, all-mudbrick face of the fortifications of the Lydian period round the Küçük Hüyük at Gordion, (AJA lxi. 324, pl. 89, fig. 14).Google Scholar

¹³³ Cf. Thucydides iii. 20,3, where unplastered patches on the face of the siege-wall at Plataea revealed the brickwork beneath. Such plaster was found preserved on the wall at Gela, (AA 1954, 644).Google Scholar In the case of the Smyrna fortifications, if anything may be inferred from the preserved mudbrick house-walls on the site, a clay plaster would seem more likely than a lime one.

¹³⁴ University of Pennsylvania Mus. Bull. xvii. 4, 17 ff., figs. 12–14; ILN Jan. 3rd 1953, 20 ff., fig. 1; AJA lix. 1 ff., pl. 1, fig. 2. One may compare the somewhat different Hittite timber revetting, e.g. Koldewey, , Ausgrabungen in Sendschirli ii. 107 ff.Google Scholar

¹³⁵ Larisa am Hermos i, pls. 3, 4c, 28, 29, 33.

¹³⁶ pp. 68, 97.

¹³⁷ p. 72.

¹³⁸ AJA lix. 12 ff., figs. 26–27, pls. 6–9, lx. 257 ff., pls. 87–91; ILN Sept. 17th 1955, 478 ff.

¹³⁹ Iliad xii. 28–30, 36–37 (unless the genitive denotes the object aimed at), 259–60 (unless στήλας προβλῆτας designate a stone outer face foundation). Many of our difficulties of interpretation arise from our ignorance as to how far we have to deal with a simple stockade such as mere mortals might raise in a day and how far, in the glory of heroic achievement, we have to visualize a far more permanent form of construction. However, a great many details in the narrative favour the latter interpretation. See further p. 99. n. 37.

¹⁴⁰ Timber revetting or the use of a timber frame is, of course, normal in the palatial house-architecture of the period, but it seems generally absent from contemporary military building. See also p. 137.

¹⁴¹ All-stone ones would seem to be ruled out on account of their weight, which would have been excessive at the edge of a mudbrick structure. Wooden ones would also seem unlikely in view of subsequent practice and the ill accord between their flimsiness and the massiveness of the structure below. They have been claimed on the evidence of Homer, and it is conceivable (cf., however, n. 139 above) that on the Achaean wall at Troy a mixed construction is envisaged employing timber, perhaps as on the Delphi bowl (p. non. 161). It is doubtful, however, whether Iliad xii. 444 and Odyssey vii. 45 entitle us to view the battlements of Troy and Scheria as no more than a mere wooden palisade (a more probable explanation is perhaps offered on p. 119); still more open to doubt would seem any attempt (e.g. Life Apr. 29th 1957, 34 ff.) to invoke such a palisade construction to form the battlements of all-stone Mycenaean fortifications, particularly as no cuttings or sockets in the stone to receive the vertical timbers have been reported. On the perplexing issue of Mycenaean battlements see also pp. 116 f. If the variant Iron Age construction with projecting battlements suggested on p. 99 can be verified, the probable material for them would presumably still be mudbrick, especially with supports from below to carry their weight, although conceivably they might be laced with timber against outward collapse as perhaps on the Toprakkale model (p. 99, n. 37).

¹⁴² Cf. p. 111, n. 177 and p. 114, n. 187 on IG ii². 463.

¹⁴³ AA 1954, 644, figs. 95–96.

¹⁴⁴ Rodenwaldt, Korkyra ii, Die Bildwerke des Artemistempels figs. 68, 69, 73, pls. 25, 26; Altdorische Bildwerke in Korfu pls. 29, 30.

¹⁴⁵ AM lx–lxi. 274, 277.

¹⁴⁶ Iliad xxiv. 734.

¹⁴⁷ Homeri Opera, ed. Allen, v, Ilias Parva fr. xix. Cf. also Antiquité classique vi. 5 ff., AJA lviii. 293 ff.

¹⁴⁸ These observations are based on an examination of the casts in the Museum of Classical Archaeology in Cambridge.

¹⁴⁹ Cf. p. 107 n. 133.

¹⁵⁰ A useful recent analysis of some of these representations is afforded by Naumann, , Architektur Kleinasiens 290 ff.Google Scholar, and to this the reader is referred. An exhaustive bibliography, as also any consideration of the ethnic questions involved, lies beyond the scope of this study which must confine itself to observations on possible western connexions.

¹⁵¹ Whitaker, , Motya 180–1, fig. 20.Google Scholar Such battlements also appear on the towers in the siege scene on the Phoenician silver bowl from Amathus (n. 160 below).

¹⁵² The authority for the restoration of such battlements on the palace at Pylos in the drawing, ILN Aug. 27th 1955, 346–7, Apr. 7th 1956, 256, fig. 1, is unknown to me. Earlier restorations of such battlements on Mycenaean fortifications are without foundation.

¹⁵³ e.g. on some of the buildings on the Musasir relief slab showing the looting of a temple or similar building.

¹⁵⁴ Naumann, loc. cit. Cf. n. 150 above.

¹⁵⁵ Cf. Demangel, R., La Frise ionique 60 ff.Google Scholar

¹⁵⁶ e.g. Koldewey, Neandria (51st Berl. Winckelmannsprog.) 48, fig. 68; Mon. Piot xxxviii. 55 ff. fig. 1, pl. v; Studia antiqua A. Salač oblata pl. viii, 1; Larisa am Hermas ii, pls. 2, 5, 6, 8–12, 15–17, 36, 40, 47; Åkerström, Architektonische Terrakottaplatten in Stockholm, pls. 1, 2, colour pl. 2.

¹⁵⁷ Perhaps not uninfluenced by the apparent ‘shorthand version’ of the Assyrian monuments mentioned above.

¹⁵⁸ Most striking is the way they are continued around the antefixes, e.g. Koldewey, , Neandria 46, fig. 66Google Scholar; 48, fig. 67; BCH xlv. 141, fig. 24; M. Launey, Études thasiennes i. fig. 17, pl. x. The dog's-teeth on disc-acroteria are probably in some way related, although the precise form of this relation is much more obscure.

¹⁵⁹ MemAmAc iii, pls. 20–21.

¹⁶⁰ JHS liii. 25 ff., pls. i–iii, Op Arch iv. io, pl. 6; for the crenellations on the towers see JHS liii, pl. iii.

¹⁶¹ Perdrizet, FD v, pls. xviii–xx.

¹⁶² Fellows, , Ionic Trophy Monument at Xanthos pl. 3Google Scholar; Smith, , BMC Sculpture ii. 19, slabs 869, 870, 871b, 876b, 877, 878Google Scholar; Mon. Inst. x, pls. 15–16; Brunn, , Denkmäler 216Google Scholar; Winter, , Kunstgeschichte in Bildern 264 ff.Google Scholar; AM lii Beil. xiii. 5–6, xiv. 1–2; Picard, , Manuel d'Archéologie grecque, la Sculpture ii. 869, fig. 352Google Scholar; Eichler, Heröon fig. 9.

¹⁶³ Benndorf, Heröon von Gjölbaschi-Trysa pls. xii–xiii; Eichler, , Die Reliefs des Heröon von Gjölbaschi-Trysa 22 ff., pls. 18–21.Google Scholar

¹⁶⁴ Benndorf-Niemann, , Reisen i. 144, fig. 86dGoogle Scholar; AM lii. 143, fig. 6; note, however, that, although the similarity in the type of the representation is clear, the battlements cannot be adequately distinguished on the published drawing.

¹⁶⁵ Benndorf-Niemann, Reisen i, figs. 36–37; AM lii. 141, fig. 5; Eichler, Heröon fig. 10.

¹⁶⁶ Cf. p. 111 n. 177.

¹⁶⁷ See below, p. 116. It is possible, further, that Iliad xii. 262–4 may imply the presence of crenellations on the Achaean wall and conceivable that the popularity of the plain key-pattern (i.e. devoid of meander elements) in Geometric art may reflect an interest in square crenellations (cf. the Munich hydria, n. 171 below). But such arguments involve too much speculation to be fully admitted as archaeological evidence.

¹⁶⁸ FR pls. 11–12; ABV 76. 1.

¹⁶⁹ e.g. in the jointing of the heavy masonry over the light and presumably wooden lintel.

¹⁷⁰ Caskey, and Beazley, , Attic Vase Paintings in Boston 31Google Scholar, pl. xiv and further bibliography there given.

¹⁷¹ Mon. Inst. i, pl. 34; Benndorf, , Heröon 153, fig. 141Google Scholar; Greece and Rome i, pl. 9; Antiquité classique vi, pl. 4; AJA lviii, pl. 59, figs. 19 a–b; ABV 362. 27.

¹⁷² Inghirami, Vasi Fittili iv, pl. ccciv; Thiersch, , Tyrrhenische Amphoren 64Google Scholar; ABV 95. 8.

¹⁷³ Cf. Thucydides iii. 22, 4.

¹⁷⁴ Milliet-Giraudon, , Recueil i, pl. 18Google Scholar; Babelon, , Cabinet des Antiques 129, pl. xiGoogle Scholar; Benndorf, , Heröon 152, fig. 140Google Scholar; De Ridder, Cat. no. 179; CVA Bibliothèque Nationale i, pls. 11. 9–11, 12. However, it is nevertheless far from clear why people's heads are shown in front of the crenellations, if such they be.

¹⁷⁵ Wrede in AM xlix. 173 ff. and fig. 6. The crenellations had angled returns for better protection.

¹⁷⁶ Wrede, op. cit. 176 fig. 7; Blouet, , Expédition de la Morée i, pl. 39.Google Scholar fig. iv, pl. 41, fig. iii.

¹⁷⁷ e.g. at Herakleia on the Latmian Gulf (Krischen in Milet iii. ii passim) and as has been claimed (Krischen, op. cit. 51; Wrede, op. cit. 178) for Cnidus, Iasos, Acarnanian Chalcis, Samos, and Aegosthena, not to mention Miletus itself (A. von Gerkan, Milet ii. iii as restored on figs. 10, 11, 31, 44–45). In mudbrick, as opposed to stone, and with a tile covering it was employed on the late fourth-century B.C. walls of Athens (IG ii².463; seen. 187 below). However, the date of the first introduction of this type of battlement, whether still well back in classical times or not until the beginning of the Hellenistic period, seems to me much more uncertain. The Aegosthena circuit, for example, is tentatively dated by Scranton (Greek Walls 81, 176) in the early fourth century B.C.

¹⁷⁸ Cf. p. 64 above. The reader is also referred, for example, to material from the temple pylon to be published in the full account of the seventh-century temples. Further examples from other parts of the site will probably be included in the publication of the domestic architecture.

¹⁷⁹ Those found outside the fountain-house must surely belong to that building and not to the fortifications behind.

¹⁸⁰ Such copings, presumably of stone, seem to be restored in the reconstructions, Larisa am Hermos i, pls. 28–29. The excavational evidence for them at Buruncuk is not recorded.

¹⁸¹ To anticipate briefly the publication of the domestic and temple architecture, it has been inferred that two types of roofing were current in pre-classical Smyrna: a pitched, thatched one and a flat planked one covered with loam. These are not matters capable of absolute proof on present evidence, but the inferences are strong ones.

¹⁸² It would seem that such details were not of interest to the artist who carved the fortification wall on the Corfu pediment.

¹⁸³ Cf. AJA liv. 353.

¹⁸⁴ p. 41, Plates 7, 10 c.

¹⁸⁵ p. 65 (at grid O7.xxii5).

¹⁸⁶ In both areas there may have been a gate in the vicinity. Alternative explanations to that advanced seem, however, less tenable. The breadth of the projection is, for example, too slight for it to have served as the foundation for wooden staircases giving access to the chemin de ronde. An analogous construction occurring on the Wall 3 outer face (e.g. at E3.xix3) has already been discussed above (pp. 98 f.).

¹⁸⁷ AJA liv. 337 ff., where reference to previous studies will be found. Holland's interpretation of this ἴκρια-construction is, from an architectural point of view, by far the most attractive yet advanced; however, it does involve a very awkward transition in the text of the inscription at ll. 69–70, unless the original phrasing of the document be convicted of confused wording.

¹⁸⁸ pp. 87, 97, Fig. 24 μ Plates 9, 19d.

¹⁸⁹ p. 5 and pp. 134 f.

¹⁹⁰ Olynthus viii. 39 ff.

¹⁹¹ Wrede, Atisiche Mauern fig. 2; marked on plan: Archiv für Religionswissenschaft xxxii. 54, ADelt xiv. παράρτ., Kourouniotis, Eleusis.

¹⁹² von Gerkan, A., Milet i. 8, pl. xvi. 2.Google Scholar At Smyrna such masonry would be more likely in the seventh century B.C. than in the eighth, but such criteria are not yet ones in which one can place too much confidence (cf. p. 98).

¹⁹³ Corinth xv. i. 14–15, pls. 2D, 3A, 4A (top right), 51, xv. ii, pl. 60. Poorly preserved, it has a width of c. 2·4 m. The socle apparently consisted of stone faces with a ‘core’ of earth fill subdivided by boulder partitions (cf. p. 95 n. 6); the superstructure is presumed to have been of mudbrick. The main body of the wall may be as early as the first half of the seventh century B.C., but a trench cut to receive an excessively deeply founded part of its inner face was dated to the third quarter of that century.

¹⁹⁴ Kinch, , Vroulia 90 ff.Google Scholar, pls. 2, 19a and plan. Stone socle and mudbrick superstructure. The relative lightness of the wall commands comparison with the classical circuit at Olynthus (n. 190 above). Note the quarry-fosse outside the wall. If, as seems likely, the ‘tower’ is later than the circuit wall, then a seventh-century or earlier date for the latter seems assured. Was the original form of the gate not perhaps a simple ‘overlap’ type of construction?

¹⁹⁵ i.e. the earliest construction phase, IA, which I take only to be certainly evidenced to the north-east. It is dated by its excavator around 600 B.C. (Hesperia ix. 390 ff., 430). Is not the second stage at this gate (IB), with which I would suggest taking the earliest extant circuit wall with the round towers, perhaps rather later than stated?

¹⁹⁶ JHS Archaeological Reports 1954, 20; ILN Jan. 30th 1954, 159 ff; Apr. 23rd 1955, 740ff.; Dec. 31st 1955, 1144 ff. Apparently not later than the early seventh century and possibly distinctly earlier.

¹⁹⁷ p. 4 and references there given; cf. p. 125; Plate 5 a.

¹⁹⁸ e.g. the ‘Cyclopean’ circuit at Asine (Persson and others, Asine 33) has been omitted because of the implied late date for part of it (Scranton, Greek Walls 53 ff., 67 ff.).

¹⁹⁹ Scranton, op. cit. 25 ff., 137, 159 ff.

²⁰⁰ Scranton, op. cit. 145 ff., 184 ff.

²⁰¹ Scranton, op. cit. 39 ff., 155, 186; Hesperia xi. 193 ff.; Wrede, Attische Mauern figs. 28–29.

²⁰² See Jones, , Sackett, , and Eliot, , BSA lii (1957), 152 ff.Google Scholar

²⁰³ Strategic considerations may have dictated the inclusion of a more considerable area in the case of Gla, the defences following the edge of the ‘island’. Whether the first Milesian circuit (Milet i. 8. 73 ff.; Istanbuler Mitteilungen vii. 106 ff., 114 ff.) may be regarded as a true city wall is not yet clear. Similar doubts obtain at Phylakopi.

²⁰⁴ Wace, , Mycenae 49 ff.Google Scholar; Matz, in Hdb. d. Klass. Alt. vi. 2. 285.Google Scholar

²⁰⁵ Müller, and Sulze, , Tiryns iii. 1 ff.Google Scholar

²⁰⁶ Persson, , New Tombs at Dendra 3 ff.Google Scholar

²⁰⁷ Kavvadias and Kawerau, ἡ Ἀνασκαφὴ τῆς Ἀκροπόλεως passim; Judeich, , Topographie von Athen ²113 ff., pl. 3aGoogle Scholar; Wrede, op. cit. fig. 1; Hill, , Ancient Athens 8 ff.Google Scholar; Hesperia xv. 73 ff.

²⁰⁸ BCH xviii. 271 ff.; AM xix. 405 ff.

²⁰⁹ JHS Supplement iv, pl. ii.

²¹⁰ Cf. n. 216 below and p. 97 n. 17.

²¹¹ Karo, , Die Schachtgräber von Mykenai 106–8, 173 ff., pl. cxxii.Google Scholar I take it that the architecture is Aegean in conception. This is by no means universally conceded. Cf. Persson, , New Tombs at Dendra 182 ff.Google Scholar

²¹² Uncertainty exists as to the significance and even the relevance of the further material such as Bulle, , Orchomenos pl. xviii. 2–6Google Scholar, Rodenwaldt, , Der Fries des Megarons von Mykenai 37, fig. 19.Google Scholar

²¹³ However, curious V-shaped openings appear on the fragment, AM xl, pl. viiia, although these may conceivably represent part of a decorative crowning set, not on the fortification wall, but at the top of a non-military building behind it. On the coping see further n. 233 below.

²¹⁴ e.g. Wace, , Mycenae 97.Google Scholar

²¹⁵ Though apparently attested in possibly earlier work in the Cyclades, , AE 1899, 115 ff.Google Scholar

²¹⁶ Dörpfeld, , Troja und Ilion 107 ff.Google Scholar; Blegen, , Caskey, , and Rawson, , Troy iii. 81 ff.Google Scholar

²¹⁷ As further evidence for this west Asia Minor tradition one may perhaps instance the Prehellenic wall at Buruncuk-Larisa, , Larha am Hermos i. 15, fig. 3, 44, pls. 2a, 4, 32, 33.Google Scholar The Bronze Age fortifications of Smyrna itself have yet to be explored.

²¹⁸ As at Phylakopi and Tiryns.

²¹⁹ e.g. at Gla.

²²⁰ Scranton, op. cit. 149 ff.

²²¹ Cf. Tiryns iii, pl. 4.

²²² AM xix. 425 ff., Fimmen, Kret.-Myk. Kultur fig. 26.

²²³ e.g. at Tiryns and Phylakopi. Cf. also AE 1899, 115 ff.

²²⁴ Cf. Scranton, op. cit. 151 ff.

²²⁵ AM xlix. 173 ff., fig. 6; p. 111 above.

²²⁶ AA 1916, 238, fig. 17; here apparently used with ‘window battlements’, for which see p. 111 n. 177.

²²⁷ Blouet, , Expédition de la Morée i, pl. 39, fig. ivGoogle Scholar; AM xlix. 173.

²²⁸ Cf. p. 118.

²²⁹ The indented trace is not represented at Smyrna until classical times and then only at the upper classical fort where it may occur once or possibly twice. See p. 134 and Fig. 36 (south side) and ÖJh xxvii Beiblatt 131, fig. 64 (west side).

²³⁰ More intimate details of their construction can probably only adequately be obtained by area excavation on a scale beyond the resources of the recent Anglo-Turkish investigations.

²³¹ Cf. pp. 15, 125.

²³² Cf. Wycherly, , How the Greeks Built Cities 40.Google Scholar See also p. 115 n. 203 above. Late Cypriot fortifications (e.g. see below) suggest the possibility that the Greeks may have taken cognizance of such changes quite early.

²³³ This is not to be taken to mean impossible. Mudbrick was used for the upper stories of houses and palaces above the high timber-laced stonework. However, the surviving fortifications seem often to attain a full and sufficient height without there being any call to invoke an addition of brick. However, strange though the resultant construction might seem, it is possible that we may have to concede the employment of brick to form the battlement parapet of these walls. Indeed it is not altogether easy otherwise to account for the coping set above it on the Mycenae rhyton. It is noteworthy that brick battlements on a stone wall are perhaps also suggested much later by the François vase (p. 111). However, such a construction is completely different from that under consideration. See also p. 137.

²³⁴ That such may await discovery is far from being ruled out. In any case, in view of the Cypriot material (see below), it seems somewhat unlikely that the influence of local Anatolian practice over the period following on the Aeolian and Ionian migrations was a major determining factor in this regard. Further, at Troy itself the main brick superstructure had long since been replaced by one of stone (p. 97, n. 17). The construction of the upper part of the Buruncuk wall is admittedly unknown. But only at Smyrna does it seem possible that we may have a mudbrick fortification wall as late as the Late Bronze Age (pp. 68, 82, 121), although no definite pronouncement about even this can be made until such time as it has been possible to excavate it.

²³⁵ Liverpool Bull. ii. 1–2. 33, 37–39, 53, figs. 2–3, 5.

²³⁶ SCE ii. 476, 516, 518, 522.

²³⁷ JHS Archaeological Reports 1954, 28, fig. 1.

²³⁸ However, the gap after the Mycenaean period is not as yet bridged by datable Greek examples. John Boardman kindly informs me that he considers the citadel wall at Emborio, Chios (p. 115 n. 196), to have been entirely of stone and it may thus be, for the present, the earliest confidently dated instance of this construction in the Greek Iron Age (apart, of course, from the Wall 1 tower at Smyrna).

²³⁹ However, in the environs of Smyrna itself, it is noteworthy that in the earlier circuit at the Bel Kahve hill-fort (p. 4, Plate 5 a) part of the top of the socle is preserved with its level surface destined to carry a brick superstructure. The classical forts above the city (pp. 5, 134) were perhaps of all-stone construction like Wall 4 round the tell.

²⁴⁰ Cf. Herodotus i. 178–9; Thucydides ii. 78, iii. 22–24.

²⁴¹ Liverpool Bull. ii. 1–2, 34–35, 54.

²⁴² p. 115 n. 194 above.

²⁴³ Iliad vii. 436–41, viii. 343, ix. 349–50, xii. 28–30, 36–37. 258–60, 444, &c, xv. 1, 344–5.

²⁴⁴ p. 108 n. 139; cf. also p. 99 n. 37.

²⁴⁵ Iliad xviii. 177.

²⁴⁶ Odyssey vii. 45.

²⁴⁷ p. 110.

²⁴⁸ pp. 98 f.. As has been seen there this may, however, be earlier in its beginnings and, in an extreme form, farreaching in its structural implications for Ionian fortifications. But on both counts more evidence is needed.

²⁴⁹ Noack, , Eleusis pls. 24b, 26, 27a–cGoogle Scholar; Wrede, Attische Mauern figs. 10–13; ADelt xv. 104, fig. 24; plans Archiv fr Religionswissenschaft xxxii. 67, ADelt xiv. παράρτ. (facing 30), Kourouniotis, Eleusis, Guide to Excavations and Museum (end).

²⁵⁰ Larisa am Hermes i. 27 ff., 44 ff., 116 ff.

²⁵¹ p. 115 n. 194. Presumably there was also a longmaintained tradition in lighter defences capable of holdingmen at bay but not machines.

²⁵² The archaic examples cited by Scranton (op. cit. 154 ff., 186) rest purely on stylistic criteria or historical inference, but some of these seem attractive. On its isolated employment at Smyrna in classical times see p. 117 n. 229

²⁵³ Cf., however, p. 115 n. 195. At Smyrna one may instance the construction by the Wall 4 north-east gate (p. 73, Fig, 18, Plate 16 c) and, more appropriately, that at the north-east of the upper classical fort (pp. 5, 134, Fig. 36).

²⁵⁴ p. 111.

²⁵⁵ Mantinea is a good example: Fougères, , Mantinée et l'Arcadie orientale 137 ff.Google Scholar

²⁵⁶ Commonly, however, covered from a tower or salient.

²⁵⁷ At Smyrna this construction is attested by the gate to the upper classical fort—Weber, Le Sipyle et ses Monuments, pl. 1; ÖJh xxi–xxii. 225, xxvii. 140 ff., fig. 68; Miltner, H., Eski İzmir 4 ff., fig. 5Google Scholar; Cadoux, , Ancient Smyrna, pl. facing 41 top. Here pp. 5, 134 ff.Google Scholar

²⁵⁸ Corinth iii. 2. 107 ff., 121 ff.

²⁵⁹ Koldewey, , Wetzel, , Die Königsburgen von Babylon i, Die Südburg 38 ff., pls. 5–8.Google Scholar

²⁶⁰ Not that the north Greek material is any better dated than that anywhere else in Greece (apart from Corinth) or Asia Minor; however, some of the Macedonian vaulted tombs have been dated, perhaps rather sanguinely, in the fourth century B.C. (e.g. BSA xxiii, pl. xii; BCH lxxix. 276, figs. 10–11). See now further Lawrence, A. W., Greek Architecture 229.Google Scholar

¹ Cadoux, C. J., Ancient Smyrna 23–104.Google Scholar

² pp. 9 ff.

³ Thus the analysis by Dr. Baki Öğün of the Monochrome and bucchero pottery must be awaited with interest, in particular if we wish to learn whether or not there is any break in the antecedent Monochrome sequence. Further excavation to provide Mycenaean evidence for the dating of the latest purely Monochrome levels would be a great help, since the handful of Mycenaean sherds so far recovered (JHS lxxii. 104, fig. 10) were from a Protogeometric context and seem to have been thrown up by pits of that date.

⁴ p. 10.

⁵ See the forthcoming accounts of the Iron Age domestic architecture and the Bronze Age levels.

⁶ p. 102.

⁷ Cf. p. 39.

⁸ Apart, of course, from the subsequent accumulation of the siege-mound. For possible debris from its last circuit wall see pp. 68, 82, Plates 8–9, Fig. 25. The Monochrome occupation levels towards the north edge of the main occupation probe show that here the earlier settlement spread out as far as its immediate Iron Age successors (cf. pp. 9, 82, Figs. 24, b–c (and under parts of ƒ, h, n), 25).

⁹ Cf. p. 66. The deep-level probe in Jxviii–xix (pp. 9, 39) affords a fixed point not far from the east edge of the Iron Age cities that is safely inside the settlements of the Middle and Late Bronze Age.

¹⁰ The presence of Early Bronze Age habitation in MNxiv suggests that in this vicinity the westerly limit of the settlement may not have changed greatly from that period until the collapse of Iron Age Wall 2.

¹¹ p. 39.

¹² Plate 74.

¹³ Despite such phenomenal undertakings as the Wall 1 platform fill (p. 41) the southern part of the later city suffered severely from erosion (cf. pp. 55, 66 n. 136; also on the erosion of the southerly edge of the Bronze Age tell cf. the exposure of the rock outcrop in Nxiv, p. 1).

¹⁴ p. 68, Plate 8; p. 82, Plate 9; p. 118 n. 234.

¹⁵ p. 82, Plate 9, Fig. 25.

¹⁶ p. 82, Plate 9, Fig. 24, b, Fig. 25, on the Monochrome habitation overlying it.

¹⁷ The dating of these pre-Protogeometric Monochrome levels is most uncertain. An absolute chronology certainly cannot yet be advanced with the pottery unstudied and the investigations confined to small soundings lacking correlation.

¹⁸ p. 68, Plate 8; p. 82, Plate 9. At both these points the crown of this pile of debris where one might expect such a stockade to be set seems to have survived relatively intact. Cf., however, p. 82 n. 211.

¹⁹ pp. 40, 82, cf. also p. 56; on the implications of this construction, pp. 100 ff., 107 ff., 117 ff.; see further Figs. 2, 7, 10, 24, Plates 7, 9, 10 a.

²⁰ pp. 68 ff., 95, 97, 106, Figs 18, ƒ, 19, 30, 32, Plates 8, 161, 17 a–c, 18 d.

²¹ pp. 102, 104 f.

²² At this early date perhaps more likely to be found attested in palaces or princely dwellings than in public temple buildings?

²³ pp. 68 ff., 75, 82 ff.

²⁴ pp. 40 f.; cf. Fig. 7.

²⁵ Cf. pp. 56, 123.

²⁶ pp. 65, 123.

²⁷ p. 41

²⁸ pp. 68 ff.

²⁹ P. 53.

³⁰ pp. 82 ff.

³¹ p. 68.

³² pp. 40 f. The dating material was, however, extremely limited and undue stress cannot be placed on the presumed chronology of isolated sherds. But a securer dating for this part of the Wall 1 circuit can probably only readily be obtained by digging the inner edge of the ‘platform fill’ where it abuts on the original tell and where occupation pottery may be expected to be plentiful.

³³ i. 150.

³⁴ Cf. p. 122 above. An alternative, and less probable, theory is that a temporary wall-line or stockade did in fact enclose the tell but that it did not follow the edge of it but was located in the low ground outside it. Such theories, too, we were not equipped to test.

³⁵ P. 13.

³⁶ pp. 40 f. Cf. n. 32 above.

³⁷ pp. 103, 104 f.

³⁸ p. 77.

³⁹ pp. 15, 124. Cf. p. 77 n. 189.

⁴⁰ pp. 116, 119.

⁴¹ p. 41.

⁴² e.g. p. 53. On the general structural questions see: stone fill, p. 96; stone faces, pp. 98 f.; brick ‘core’, pp. 100 ff., 103, 104f.; brick superstructure, pp. 106 ff.

⁴³ pp. 71 f.; Fig. 18, Plates 8, 17 d bottom.

⁴⁴ pp. 66, 41 ff., 52; Figs. 2, 7, 18, Plates 7, 10. On the projecting step at the base of its inner face in Section λ-λ′ see also p. 114 and Fig. 35.

⁴⁵ P. 53, Figs. 2, 8.

⁴⁶ pp. 56 ff., Figs. 10, 13.

⁴⁷ p. 64.

⁴⁸ pp. 65, 94.

⁴⁹ pp. 91 f., 84, 77 ff.; Figs. 21–22.

⁵⁰ pp. 43, 53, 91.

⁵¹ p. 103.

⁵² The new standard may thus, on the present chronology, have been first adumbrated and finally established in the second quarter of the eighth century B.C. Whether this circumstance has any wider implications (cf. p. 123 above) it is impossible to say.

⁵³ p. 72, Plates 8, 17 d—well down near the water. The Wall 3 brick above was subsequent to this fault, which consequently must be the product of a considerably earlier tremor than that which occurred in the sixth century B.c. (pp. 79, 87).

⁵⁴ Most notably p. 53, Fig. 8, and perhaps also p. 91.

⁵⁵ e.g. pp. 43 f., 53 ff.

⁵⁶ pp. 14, 139 ff. (Group E).

⁵⁷ pp. 14. 55, 86.

⁵⁸ p. 103.

⁵⁹ pp. 15, 77.

⁶⁰ pp. 15, 91 f., Plate 1 (the area shown stippled).

⁶¹ See below, pp. 125 f.

⁶² p. 126.

⁶³ pp. 4, 115 n. 197, 118 n. 239.

⁶⁴ For the source material see Cadoux, op. cit. 78 ff., p. 14.

⁶⁵ pp. 25, 129 ff.

⁶⁶ The faint possibility remains, and it is a very faint one indeed, that, despite the physical difficulties and the absence of surviving traces on the hill-side (cf. p. 128), the vast extramural habitation area was fortified and that it was its circuit that came into question on this occasion; this seems unlikely, but the possibility cannot be denied without far more extensive exploration.

⁶⁷ It seems at least desirable to point out that the chronology usually accepted (and here used) for Gyges rests on the Assyrian identifications. If, for example, the Herodotean chronology be adopted, then Gyges' attack will take place in the late eighth century B.C. shortly before the earthquake that destroyed the city's defences. Such a hypothesis would allow of a siege; but, although damage arising from the assault would be almost indistinguishable in some cases from that caused by the earthquake, yet the rarity of weapons in the earthquake stratum would probably still preclude the possibility of a sack. Or is our pottery dating too high? On Lydian Chronology see Historia vii. 1 ff.

⁶⁸ p. 79. The date is based on a small amount of pottery actually found in the mudbrick. See also p. 103 and Figs. 21, 23.

⁶⁹ pp. 93, 98, Fig. 29, Plate 20 c.

⁷⁰ pp. 86 ff., 98, 103, Fig. 24, Plates 9, 19 b and c.

⁷¹ pp. 72 f., 98, 103 n. 92, Fig. 18, Plates 8, 16 c.

⁷² p. 103.

⁷³ p. 86 n. 226.

⁷⁴ p. 79.

⁷⁵ e.g. such a burial was made at E8.xii9 under a floor occurring at about 6·8 m. above sea-level (cf. the section, Fig. 25).

⁷⁶ By the burial in the temple area, Fig. 21, y, p. 77 above.

⁷⁷ At G4. xii by the child burial in an amphora appearing in Section η–η′ (Plate 9).

⁷⁸ pp. 44 ff., 55, Figs. 3–4, Plate 11.

⁷⁹ pp. 48, 49, 66, 93, Figs. 5, 28.

⁸⁰ For further views on the subject see CR xvii. 83 ff.; Eitrem, , Hermes und die Toten 4 ff.Google Scholar; Nilsson, , Geschichte der griechischen Religion i ². 175.Google Scholar

⁸¹ The seventh-century B.C. adult cemeteries have still to be discovered at Smyrna; they probably lay quite a distance from the city because of the extensive extramural settlement. However, the burial grounds in use from the sixth century B.C. onwards are extensively known (see Plate 1).

⁸² pp. 58 ff., 124, Figs. 10, 12–15, Plates 14–15.

⁸³ pp. 46 f., Plates 7, 12 a, b.

⁸⁴ PP. 52 f., 123.

⁸⁵ See immediately above on Wall 2/3.

⁸⁶ p. 103.

⁸⁷ pp. 72 f.; see above, p. 125 n. 71.

⁸⁸ PP. 73, 65 ff.

⁸⁹ pp. 66 ff, 98 ff., Figs. 16–17, Plate 16 a.

⁹⁰ pp. 65, 48, 52, cf. p. 98, Figs. 2, 7, Plates 7, 12 c, d.

⁹¹ pp. 52 f., Figs. 2, 8, Plate 13 a, b.

⁹² pp. 52, 57 f., 64 f., 98, Figs. 10, 11, 13, Plate 15c—rear.

⁹³ p. 58.

⁹⁴ p. 61.

⁹⁵ P. 98.

⁹⁶ p. 65; on the ‘step’ against the inner face see p. 114.

⁹⁷ P. 93; P. 125 n. 69.

⁹⁸ p. 94 n. 277.

⁹⁹ The seventh-century debris shown in Fig. 28 (p. 93) might be from it.

¹⁰⁰ p. 88.

¹⁰¹ pp. 86 f., 79, 125 nn. 68, 70.

¹⁰² p. 15 Fig. 3.

¹⁰³ pp. 15. 91 f., 125.

¹⁰⁴ Cf. p. 125 n. 66.

¹⁰⁵ p. 36 n. 2.

¹⁰⁶ pp. 135 f.

¹⁰⁷ pp. 92 f., 135.

¹⁰⁸ Tiles were unknown at Smyrna in the seventh century B.C.; indeed they do not even occur in the sixth century, although this circumstance is probably rather to be seen as testimony to the reduced importance of the town in the Lydian and Persian periods. They do not in fact make their appearance until the later fifth century B.C. In the fourth century they are extremely common.

¹⁰⁹ See Plate 1.

¹¹⁰ pp. 23 ff.; Cadoux, op. cit. 84–85; on the date see also pp. 143 ff. (Group H).

¹¹¹ pp. 24, 88 ff.

¹¹² Perhaps these timbers were intended to restrict the collapse of the mound in the event of its being undermined (cf. Thucydides ii. 75–76; such mines seem to have been excavated at Paphos, , Liverpool Bull. ii. 1–2. 33 ff., 53 ff.)Google Scholar, but they did not seem to have been laid in any order.

¹¹³ pp. 91, 125. If, as seems likely, the cylindrical beams are from the planking of house-roofs, their architectural implications are considerable.

¹¹⁴ Cf. Herodotus i. 162. The siege-mound excavated at Paphos in Cyprus has been attributed to the Persians (loc. cit.). It may further be asked whether the great mound (variously interpreted in the past as a temple site, a natural hill, and a tumulus) against the south-east part of the fortifications of the Lydian period round the Küçük Hüyük at Gordion may not also be in fact a siege-mound (see the air photographs: Archaeology iii. 196, fig. 1; vi. 159, fig. 1, and the plan: Archaeology iii. 197, fig. 2; AJA lx, pl. 95 fig. 52). The details of the recent excavations as they have been disclosed to date seem to fit such a hypothesis uncommonly well (Pennsylvania Mus. Bull. xvii. 4, 26 ff., Archaeology vi. 159 ff., AJA lxi. 324, lxii. 140–1). If this interpretation is correct, then this mound probably represents the point at which the forces of Cyrus carried the defences of Gordion.

¹¹⁵ Thucydides ii. 75 ff.

¹¹⁶ The sling-bolts would seem to have been confined to rounded stones. As even the identification of these is a some what subjective matter, it was not possible to obtain statistics for them. Metal bolts appear not to have been used.

¹¹⁷ This is simply based on the routine inventory de scriptions and excavators' notes and must contain many imperfections. It will be completely superseded later by the detailed publication of the weapons from the site save in one important respect: a considerable part of this material has not survived cleaning and storage and the statistics that follow, with all their inadequacies and probable occasional inaccuracies, will still constitute the fullest record we shall have of the distribution of the different types over the field of battle.

¹¹⁸ The iron ones are much rarer and they do not have the same clear typology.

¹¹⁹ pp. 131 ff. The name has been chosen for simplicity only and implies nothing as to the section of the head, which is definitely three-flanged.

¹²⁰ The Smyrna material has already been referred to by MissLorimer, , Homer and the Monuments 294 n. 7.Google Scholar An example is illustrated on Plate 6d. In view of the wide distribution of this variety it is not possible to give an exhaustive bibliography in this space. The following instances from the Greek and Anatolian regions may serve as a guide to the reader as to the appearance of the type:

Athens: Dodwell, , Tour in Greece 159Google Scholar, fig. 2 (right-hand example); Acropolis: ADelt i. παράρτ. 28, fig. 29 θ, ι, κ North slope: Hesperia ii. 341, fig. 13a–c; iv. 114, fig. 4 (middle row and left part of upper row).

Thermopylae: AA 1940, 200, fig. 47 (except examples in centre); AJA liii. 700, fig. 3.

‘Marathon’ (the British Museum and Karlsruhe material shows a strange diversity in date): British Museum Guide to Greek and Roman Life ² fig. 102 (top row, third to fifth from left); Schumacher, , Beschreibung der Sammlung antiker Bronzen, Karlsruhe 144 ff. no. 748, pl. xiv. 28Google Scholar; examples in Breslau referred to by Walter, , AA 1940, 200.Google Scholar

Olympia: Olympia iv. 178, pl. lxiv, especially nos. 1083–8; Olympische Forschungen i. 162 ff., pl. 69l–o.

Delphi: FD v. 97, fig. 338.

Asine: Persson, and others, Asine 334.Google Scholar

Olynthus: Olynthus x. 397 ff. ‘Type F’, pls. cxxv–cxxvi.

Chios: BSA xxxv. 151, pl. 32. 13–14.

Dodecanese: BMC Bronzes nos. 2803, 2807; Lindos i. 196 nos. 608–10, pl. 23.

Smyrna area: Behn, Vorhellenistische Altertümer der östlichen Mittelmeerländer no. 427.

Various: BMC Bronzes 346, ‘Type F’; Coll. Gréau 142.

Cyprus: Richter, , Metropolitan Mus. Cat. Bronzes 404 nos. 1487 ff.Google Scholar; Cesnola, , Atlas iii, pl. lxxiii. 4Google Scholar; Myres, , Cesnola Coll. 490 no. 4789Google Scholar; cf. also the classical varieties, SCE ii, pl. clxxiv (Type 8); iii, pl. xlv (Type 3) and pls. lxxxvii. 9, cxlix. 4–5.

Gordion: Archaeology vi. 166, fig. 10 (two right-hand ones).

Alisar: Chicago Oriental Inst. Publications (hereafter abbreviated OIP) xxix, fig. 496 (top row except second from right, second from right in second row), xxx, fig. 107 (apparently d 427, c 2522, d 57, d 37, c 431).

Boǧazköy: Bittel, and Güterbock, , Boǧazköy, Neue Untersuchungen i, pl. xi. 26.Google Scholar

Kerkenesdaǧ: Przeworski, , Die Metallindustrie Anatoliens pl. x. 8e–f.Google Scholar

Carchemish: Carchemish ii, pl. 22b (certain of the examples in lower row).

On the wider distribution of the type and its subsequent development and modification see, e.g., Schmidt, H. in Pumpelly, , Explorations in Turkestan i. 183 ff.Google Scholar; Bulanda, , Bogen und Pfeil bei den Völkern des Altertums 28, fig. 22Google Scholar; 44 fig. 29; Vulpe, , L'Âge du fer dans les régions thraces de la péninsule balcanique 65 ff.Google Scholar; Bonnet, , Waffen der Völker des alten Orients 165, fig. 73, leftGoogle Scholar; Przeworski, op. cit. 61 ff.; H. Lorimer, op. cit. 285 ff., 294–5; Petrie, , Tools and Weapons 34 ff.Google Scholar, especially pl. xli. 55, 60, 61, 73, pl. xlii. 220–4, 226, 239–42; Minns, , Scythians and Greeks 68, 190, fig. 82Google Scholar (especially second and third rows except 10, 12, 14, 17, 18, 201). Closer to hand, and so more specifically, one might also mention Petrie, , Naukratis i, pl. xi. 2Google Scholar; Gerar, pl. 29. 14–17, 19; Tanis ii, pl. xxxix. 9; Syria viii. 208, fig. 15i, j; 56, fig. 14c, pl. xxc.

That this variety was already by the sixth century in regular use amongst the vast complex of Scythian peoples there seems little doubt. Petrie, (Tools and Weapons 34)Google Scholar argues that it first appears in Egypt as a consequence of the great Scythian invasion of Syria in the later seventh century.

But its origins are less clear. In south Russia an example occurs in a seventh-century Taman grave which Rostovtzeff considers Cimmerian, (Iranians and Greeks in South Russia 40, pl. v. 2).Google Scholar Przeworski (loc. cit.) suggests that it may have arisen in Iran in the eighth century. Its distribution throughout the civilized world seems probably to be associated with the great Scythian migrations (and perhaps those of the Cimmerians also) and whether its appearance in Greece and Anatolia might in any way have anticipated these is possibly doubtful. But the identity (and preference in arrowheads) of the pre-Scythian archers seemingly equipped with composite bows that appear on eighth-century Greek vases remains unknown and, against this background, the certain identification of the early specimen from Smyrna, no. 1 below, becomes an issue of some importance. Further, it seems difficult to dissociate the socketed ‘leaf’ heads (see below) completely from the socketed ‘triangular’ ones, especially in the case of the ‘triangular’ variety equipped with a single side-barb (e.g. Carchemish ii, pl. 22b and the mould, pl. 23b). They are commonly found together and Smyrna (p. 133) offers a rare opportunity of establishing a distinction in their use.

In the absence to date of any other early ‘triangular’ variety in western Anatolia (although such have been claimed elsewhere, e.g. von Sacken, Das Grabfeld von Hallstatt, pl. viii, fig. 10), it seems most likely that the Homeric τριγλώχιυ alludes to arrowheads of our type and has the meaning of ‘three-flanged’ rather than of ‘three-barbed’. For although early three-barbed ‘triangular’ heads may perhaps occur earlier farther east (e.g. at Carchemish), there is nothing to suggest their presence in western Anatolia prior to the sixth century. The example from Megalopolis commonly cited to explain the Homeric term (Helbig, , Das homerische Epos aus den Denkmälern erlautet ²341, fig. 134Google Scholar; Lorimer, op. cit. 285–6, fig. 36) is surely actually of fourth-century B.C. date. It is perfectly paralleled by specimens from Olynthus bearing Philip's name (Olynthus x, pl. cxx). Even so, when the archaeological evidence is better documented, it may still be found that the references to three-flanged arrowheads are a factor indicating a low date for parts of the Iliad.

Over the sixth century other ‘triangular’ varieties seem to have become common as well, but that under consideration is still in regular use in contexts of Persian War date in Greece and it is frequent at Persepolis, at least in the earlier stages (Schmidt, , Persepolis ii, pl. 76).Google Scholar However, the Scythian tombs suggest that it may have suffered an eclipse thereafter. Thus the fifth-century grave, Prähistorisches Zeitschrift v. 11 ff., contained only one of our type as against 376 of other ‘triangular’ varieties.

¹²¹ Plate 6d.

¹²² The following comparative material may be cited:

Athens: ADelt i. παράρτ. 28, fig. 29–ε, Ʒ, η; North slope: Hesperia ii. 341, fig. 13 g (cf. larger variety, e and ƒ); iv. 114, fig. 4.

‘Marathon’ (cf. n. 120 above): British Museum Guide to Greek and Roman Life ² fig. 96 (left, second from top), fig. 102 (first and second from left, top row); Schumacher, , Antike Bronzen, Karlsruhe 144 no. 748, pl. xiv. 25, 38.Google ScholarThermopylae: AA 1940, 200, fig. 47.

Olympia: Olympia iv. 177, pl. lxiv. 1076, 1078; Olympische Forschungen i. 162, pl. 69e–g.

Delphi: Fouilles de Delphes v. 97, fig. 337.

Olynthus: Olynthus x. 381, pl. cxx no. 1898.

Lindos: Lindos i. 195, nos. 601–3, pl. 23.

Chios: BSA xxxv. 151, pl. 32. 11–12.

Magna Graecia: Atti e Mem. della Società Magna Grecia 1932, 110, fig. 63.

Naucratis and Tell Defenneh: Petrie, , Naukratis i, pl. xi. 3–4Google Scholar; Tanis ii, pl. xxxix. 16.

Various: Petrie, , Tools and Weapons pl. xli. 133, 135, 136–7Google Scholar; Coll. Gréau 143 (top centre and one to left of it); BMC Bronzes 2803–4, 2806–7, 2812–13.

Alisar: OIP xxix, fig. 496, centre row, e 1156; xxx, fig. 107 (two at bottom left).

Boǧazköy: Bittel, and Güterbock, , Boǧazköy, Neue Untersuchungen i, pl. xi. 18–20.Google Scholar

Kerkenesdaǧ: Przeworski, op. cit. pl. x. 8a–d.

Kumkale: Przeworski, op. cit. pl. xviii. 3.

Tarsus: AJA xli. 278, fig. 33.

Gordion: Archaeology vi. 166, fig. 10 (three left-hand ones).

South Russia: Minns, , Scythians and Greeks 68, 190, fig. 82. 1, 7, 10, 201.Google Scholar

The form of the barb or ‘spur’, shape of the flanges and socket, and general technique all seem to link this variety with the early ‘triangular’ one (see above). Its distribution and history seem to have been similar and Petrie, (Tools and Weapons 34)Google Scholar remarks that its appearance in Egypt seems to follow the same pattern. In Cyrus' attack on Gordion this, rather than the ‘triangular’ variety, seems to have been the principal weapon of the Persians, (Archaeology vi. 165–6, AJA lxi. 324).Google Scholar In general it is very common in Anatolia, where Przeworski (op. cit. 60 ff.) considers it to have been preceded by an earlier plain leaf type without a barb (cf. Schmidt, , Schliemanns Sammlung trojanischer Altertümer 260 no. 6538Google Scholar; Bulanda, op. cit. 105, fig. 73b; other examples will be found at certain of the references above; cf. also the Iranian specimens, e.g. Bulanda, op. cit. fig. 20, left and right). A leaf variety with a barb consisting in the prolongation of one of the flanges occurs in a seventh-century Taman tomb which Rostovtzeff (op. cit. 40, pl. v. 2) considers Cimmerian.

¹²³ Cf. BMC Bronzes 346 ff., ‘Type E’.

¹²⁴ e.g. at the later part of the Imperial Hittite period:

Bittel, and Naumann, , Boǧazköy-Η̮ attuša i. 100, fig. 24Google Scholar; Bittel, and Güterbock, , Boǧazköy, Neue Untersuchungen i, pl. xi. 24Google Scholar; Garstang, Prehistoric Mersin pl. xxxia (top left); H. Koșay, Alaca Hüyük Kazisi 1937–8 pl. lxxxv, fig. 1 (fourth from left), OIP xxix, figs. 290–1 (b 2151 and d 2791, classified, however, as spearheads); Goldman, , Tarsus ii. 291 nos. 79–85, pl. 427Google Scholar; a development parallel to that of our own seems to occur in Phrygian times; Belleten xi, fig. 41, OIP xxix, fig. 496 (e 1156, e 2271); Garstang, Prehistoric Mersin pl. xxxiia (top row, fourth from right). See further Przeworski, op. cit. 59 ff.

¹²⁵ Ann. vi–vii. 153, fig. 76; 220, fig. 142. Cf. Blegen, Prosymna fig. 461; Petrie, , Tools and Weapons pl. xli. 94.Google Scholar As well as the Anatolian variety referred to above, these arrowheads relate themselves to a more widespread Aegean type, e.g. Evans, Prehistoric Tombs at Knossos fig. 28, Bulanda, op. cit. fig. 48. However, the gap from Mycenaean times until the seventh century B.C. is a vast one (most imperfectly bridged by Kerameikos iv. 27, pl. 38) and the possibility of the interplay of further Anatolian influence in the interval cannot be excluded.

¹²⁶ Most fully discussed recently, Proc. Society of Antiquaries xxxii. 152 ff. This classical variety is distinguished from the archaic one, inter alia, by the different treatment of the curious ‘haft’ above the tang. It is abundantly attested at Olynthus, Olynthus x, pls. cxx–cxxii. But earlier examples resembling our c. seventh-century specimens from Smyrna have also been found, e.g. Olympia iv. 178, pl. lxiv no. 1095; Olympische Forschungen i. 160 ff., pl. 69b; FD v. 97, fig. 336a; Jantzen, U., Griechische Greifenkessel 58, pl. 64. 1–2 (moulds)Google Scholar; BSA xxxv. 151, pl. 32. 10 (fragmentary); OIP xxx, fig. 107 (c 1130); Schmidt, , Schliemanns Sammlung trojanischer Altertümer 260, no. 6534Google Scholar; Bulanda, op. cit. 105, fig. 73a.

¹²⁷ Petrie, , Tools and Weapons 33 ff.Google Scholar

¹²⁸ The inventory description (although it possibly suggests the leaf variety) is quite inadequate and judgement on the identification of this important piece must accordingly be deferred.

¹²⁹ p. 89.

¹³⁰ The possibility that a few of these might be of later, e.g. Persian War, date cannot be excluded on typological grounds but is not supported by the overall excavational evidence.

¹³¹ Again, one or two of those found in later levels could conceivably be later.

¹³² p. 91.

¹³³ A further triangular arrowhead was picked up from the surface of about the middle of the tell. It could well have washed down from the north. Neither it nor a further specimen, also triangular, found in the fill of a tumulus of the Lydian period near the north edge of the later cemetery is included in the overall statistics given below.

¹³⁴ Cf. p. 131.

¹³⁵ An undisturbed area of about 10 square metres in the temple pylon alone yielded some seventy-five examples (p. 24).

¹³⁶ It is conceivable that the roof of the pylon may also have been demolished and the cross-wall barring the access-way constructed at this stage (i.e. in that event antedating the rest of the boulder substructures). For it is otherwise (i.e. if this wall be taken to be sixth century with the remaining substructures) difficult to see how the spear-points found lying against its face in the undisturbed sack stratum escaped destruction when it was built. These matters will be debated further in the publication of the temple buildings.

¹³⁷ P. 131.

¹³⁸ The four triangular arrowheads found in the mound suggest the possibility that they may also have been using a very small number of triangular heads, but do not prove it, since these might equally be unexpended Lydian ammunition or ricochets.

¹³⁹ pp. 48. 93.

¹⁴⁰ pp. 79, 87. The occupation levels in Kxi and elsewhere suggest the possibility of a further earthquake in the early second quarter of the fourth century B.C. The cracks in the masonry of the fountain-house (p. 63) are probably the product of these and perhaps of other undetected tremors.

¹⁴¹ pp. 80 f.

¹⁴² p. 81, Fig. 23, Plate 18 a.

¹⁴³ PP. 5 ff. (to which the reader is referred for a fuller account and bibliography), 32, Plate 1.

¹⁴⁴ ÖJh xxvii Beiblatt 130–46.

¹⁴⁵ Marked ‘4th century fort’ on the plan, Plate 1.

¹⁴⁶ Fig. 36. Note also that the rock-cuttings, which are all that survive at this point, suggest the presence of a ‘jog’ of ‘indented-trace’ type on the south side of the fort.

¹⁴⁷ p. 5. Marked ‘fort’ on the plan, Plate 1.

¹⁴⁸ op. cit. 146–9.

¹⁴⁹ It is no longer possible to obtain permission to see it. An idea of the conflicting opinions already held may be gained by comparing the Miltners' account with ÖJh xi Beiblatt 161, xxi–xxii Beiblatt 225, AM xvi. 224.

¹⁵⁰ pp. 87, 89, 99, 114, Plates 9, 19 d.

¹⁵¹ pp. 92 f., 128, Fig. 28.

¹⁵² pp. 81, 87, Figs. 21, 23–24.

¹⁵³ pp. 73 ff., Figs. 18, 20, Plates 16 c, 17 e.

¹⁵⁴ pp. 50, 53, 56, 64, 65, 68, Figs. 2, 7, 8, Plate 7.

¹⁵⁵ p. 114.

¹⁵⁶ p. 87.

¹⁵⁷ P. 81.

¹⁵⁸ p. 56. Cf. also the burials in Section λ–λ′; (p. 49).

¹⁵⁹ P. 73.

¹⁶⁰ P. 93.

¹⁶¹ p. 81, Figs. 21–22, Plate 18 b, c.

¹⁶² p. 63.

¹⁶³ pp. 36, 92, 128.

¹⁶⁴ Plate 1. The solution suggested there involves the assumption that the fourth-century wall in Kv–viii ran straight up the hill-side to the point from which the ashlar blocks were plundered in 1948: it also involves the assumption that the surviving face of this wall is an outer one, although its weight and finish seemed better suited to an inner face. An alternative seems to be to regard tell and hillside as being linked by two parallel walls no great distance apart, of which that in Kv-viii is the more southerly. The single huge face-block in Hi–ii might then almost be in position and represent all that survives of the more northerly wall. Of the two theories, the distribution of surface pottery perhaps favours the first.

¹⁶⁵ P. 34.

¹⁶⁶ At Fig. 18, i.

¹⁶⁷ P. 34.

¹⁶⁸ p. 11 n.2

¹⁶⁹ The sarcophagus burial was at Nxiv (p. 31). It could not be determined whether the ‘burial’ at Fig. 24, γ (p. 87) was of this period or later.

¹⁷⁰ P. 93.

¹⁷¹ pp. 64, 65, 93.

¹⁷² PP. 36. 92. 128, 135.

¹⁷³ As at Fig. 18, j (p. 74).

¹⁷⁴ p. 36.

¹⁷⁵ pp. 53. 55 f., 65.

¹⁷⁶ pp. 79. 87. 134.

¹⁷⁷ p. 38 n. 18.

¹⁷⁸ A small part of Wall 1 also showed above ground at the north-east gate, but this was probably first uncovered by the modern terracing of the tell.

¹⁷⁹ Cf. p. 74 n. 176.

Article contents

Old Smyrna: the Iron Age fortifications and associated remains on the city perimeter

Extract

Access options

References

Save article to Kindle

Save article to Dropbox

Save article to Google Drive

Reply to: Submit a response

Your details

You have entered the maximum number of contributors

Conflicting interests