¹ This work was made possible by courtesy of J. Papadimitriou, Ephor of Attica. The article was written in cooperation: the field-survey was carried out (spring 1955) by J. E. Jones and L. H. Sackett, who are responsible mainly for the bibliography, description, and interpretation, while C. W. J. Eliot has contributed the section on historical evaluation. Thanks are due for help and advice to M. S. F. Hood (Director of the British School), to whom we owe a number of improvements in the text, to J. Boardman, to J. M. Cook for permission to reproduce one of his photographs (Plate 32 a), and to members of the Agora Excavation Staff. Professors A. W. Gomme and H. T. Wade-Gery usefully discussed certain points, and A. J. Graham, T. Hodge, and R. Tomlinson, students of the British School, helped in the preliminary stages of surveying. The Royal Hellenic Air Force were approached for an air-photograph, and we are extremely grateful for the magnificent series which they have presented, and of which we publish a selection.

² The survey was carried out with tape and compass. The full survey (scale 1:1000) with a separate sheet of all the sallyports (scale 1:100) is in the library of the British School, Plate 29 is from an enlargement of Curtius and Kaupert's Karten von Attika no. vi, on which was superimposed the surveyed plan of the Dema, corrected for slope but modified in drawing so as to show details of plan.

³ The isolated knoll of Magouleza lying 700 metres east of the central saddle would not affect the distant view from the hill-tops or the command of the ground immediately in front of such a line.

⁴ The cutting made for the railway (PlateS 29, 30) would have removed another opening; that is, there would have been originally either fifty-one sallyports and two gates, or fifty sallyports and three gates.

⁵ It would not therefore be impossible for troops to infiltrate around the upper side of the wall and it is perhaps surprising that the wall was not carried up the slope for another 100 metres. We owe to Professor W. G. Hardy of the University of Alberta the suggestion that the wall ended up against a forest (there are still a few trees on the higher slopes); this would be a natural obstacle to any large-scale movements, though on the other hand perhaps cover for infiltration. Statius, , Theb. xii. 620Google Scholar, certainly describes Aigaleos as well wooded. See also the next note.

⁶ This ledge and channel is the ‘road with incised rut’ of Karten von Attika (Milchhoefer, , Text ii. 45Google Scholar); that only one rock-cut channel can be traced, and that moreover leading to a watercourse, seems to disprove this interpretation of the feature. The existence of a 700- or 800-metre-long channel leading from a watercourse now dry even in spring raises the question whether such a work would ever have been undertaken if the source had always been as poor, and whether a more permanent flow must not be presumed, indicative of a greater depth of soil and more forest on the slopes of Aigaleos.

⁷ Slighting would have left more blocks lying in front of the wall; it seems that many have been removed, perhaps as material for railway embankments close by.

⁸ Disturbances again perhaps to be connected with the building operations on the railway.

⁹ As preserved, the Rear Wall seems too slight and too short to be an effective barrier. But stone-robbing for the sheepfolds built later against its west side would account for a reduction in height and perhaps in length, particularly at the northern end where some general clearing of the ground appears to have taken place. Possibly the wall was never completed. Even so, as seen from the west and from some distance, it still gives the impression of being a well-sited breastwork.

¹⁰ In connexion with the uncertain relation of the Rear Wall to the Dema, it should be noted that the Rear Wall appears not to have had a properly built passageway through it to correspond with that through the Dema at Gate A. This inconsistency could imply a difference of function and of date; but it might also suggest that there was here, over the central saddle, no formal roadway. The present-day path passes through one of several narrow breaches in the Rear Wall.

¹¹ Possibly related to it is a short length of similar shallow rock-cut channelling noted far down the same central valley, but on the south side a little way up the slope. Correlation of the two would imply the assumption that the channel crossed the natural watercourse at some point, presumably by means of an aqueduct.

¹² The distance over the saddle and the long descent involved seem to invalidate Milchhoefer's suggestions that levering machines were used to connect them. The breadth and careful terracing of the road could certainly imply a provision for more than foot traffic, and stripping of the surface might reveal confirmatory wheel-ruts. In this case the links between the roads would require an equally adequate surface for wheeled traffic, perhaps in the form of a timber causeway or even a stone-built, zigzagging incline, but no traces of such are visible.

¹³ The return, in its present state of rough repair, abuts against the north end of 48–49, although in its original state it undoubtedly stopped short, leaving a sallyport. The lower courses seem certainly to indicate that the return was an original feature, though now the position of the jamb and the width of the sallyport can be judged only with approximate accuracy.

¹⁴ There are various possible solutions for the deviation from the normal overlap: (a) The existence of this feature in the original design, due to some overriding tactical need. But the nature of the terrain does not call for the provision of extraordinary safeguards. As a logical termination of the normal overlap and sallyport pattern, a final reverse overlap (= a sallyport-in-reverse) would be understandable, fulfilling the defensive scheme of sallies by facilitating the re-entry of the last of a series of sortie-parties (see below, p. 168); but sallyport 49 is neither the last sallyport nor the terminal point of terrain practical for sallying tactics. (b) An intentional change of plan during construction, designed by the military engineer to improve his original line, which was seen to incline too far eastwards, (c) A fortuitous change in plan, necessitated by the failure of working parties to follow the surveyed line, here or farther south.

For the constructional change in mid-length there are two plausible explanations: (a) Two working parties cooperating on one length, each to its own technical specifications. The two styles, however, represent neither different ideals of fortification nor the work of different contractors, but adaptations of one simple model to the needs of terrain —which cannot be taken to account for the change here in so short a distance, (b) A pressing need for economy or haste during the later stages of construction. An immediate change to the narrow style would save both time (in that only one worked face need be raised) and building material. That this is a plausible interpretation is shown by the subsequent cessation of work before a ramp could be added to the modified length or the wall farther north be raised up— two indications that the measure taken was insufficiently effective. A further implication is that the programme of construction had been to work from south to north.

¹⁵ It appears from the state of the ground that there has been no significant destruction or stone-robbing due to cultivation, or concealment due to silting.

¹⁶ At its south end on Aigaleos, the Dema employs the overlapping trace on ground equally rough and difficult of approach, and well beyond the probable limits of attacks directed against the south saddle. At this northern end, however, the laying-out line, which is all that the remains in situ appear to represent, gives no indication of providing for indentations, even so close as this to the saddle.

¹⁷ Even so its erection to a height of even 1 metre is surprising, since it seems that work was discontinued farther south, at the north saddle, where the labour expended on Parnes could have been more profitably employed. The evidence for a south to north progression in building the more important southern sector is not incompatible with the dispatch of a detachment forward to begin work independently on Parnes.

¹⁸ This last stretch from the second watercourse northwards was not noticed and marked in Karten von Attika vi. It accounts for the discrepancy between Milchhoefer's estimate of the length of the Dema (4,260 m.) and our own (4,360 m.).

¹⁹ To illustrate this point, it may be observed that the south end of the Dema, the main crests of the two central hills, the three saddles, and the south shoulder of Parnes are almost in direct line; little divergence from this straight line would be required of a ‘stone-wall’ defence designed to halt a hostile advance at its own foot, by sheer obstruction. That wall would be simple and also shorter (c. 2,500 m. rather than c. 2,950 m.); it would follow the highest ground, and still possess magnificent long-distance views from its two summits. But it would not everywhere command the ground in front.

²⁰ Note the immediate reversion to the narrow style for three lengths (36–37, 37–38, 38–39) in the dip between the flat summits of the north hill.

²¹ A typical 30-m. length of broad wall, built on flat ground to a height of 2·0 m. and a breadth of 2·75 m., together with its two ramps, each 5·0 m. long and 2·75 m. wide, would theoretically require 192 cubic metres of build ing material. It is more difficult to calculate the volume of the narrow style owing to its variation with the terrain. But an average section of the narrow wall, 30 m. long, 1·50–1·75 m. high, and 1·75 m. broad, with a ramp 3–4 m. wide (cf. Fig. 5) would have a volume of only 140–50 cubic metres.

²² The 192 cubic metres of the broad style (see previous note) would compare favourably with the 255 cubic metres of an equivalent length of this (hypothetical) narrow walling, 2·0 m. high and 1·75 m. thick, together with its heavier buttressing ramp, 5·0 m. broad.

²³ Good examples exist in narrow lengths 13–14, 15–16, 16–17, 18–19, 19–20, 20–21, 23–24, 24–25, 28–B, B–29, 29–30, 30–31, 31–32, 37–38, and a trace in the broad length 32–33.

²⁴ It is tempting to suppose that the builders had in mind some specifications (συγγραφαί—for the term and the method of exposition in public contracts, cf. IG ii.² 1665. 2; 1668. 2; AE 1900, p. 94, l. 46; and particularly for military building IG ii.² 463, with Hesperia ix (1940) 66–72 and AJA 1950, 337–50). The following units appear frequently in ancient wall construction: ἡΜιπόδιον (½ foot), πούς (1 foot), and πῆχυς (1½ feet = 1 cubit). It is likely that the Dema was designed in multiples of one of these, and probably in multiples of the foot or cubit, since accuracy to within a ἡΜιπόδιον might have been both pointless and unobtainable in rubble construction. Converted into Greek units (cf. Dinsmoor, W. B., The Architecture of Ancient Greece (1950) 54 n. 4, 161 n. 1, 241 n. 1, 247Google Scholar, on the length of the Greek foot), the rampart dimensions might read as follows: broad style—rampart (τεὶχος) g feet (= 6 cubits) broad and 6 feet (= 4 cubits) high, wall walk (πάροδος) 7 feet broad, parapet (ἔπαλξις) 2 feet thick; narrow style—rampart 6 feet (= 4 cubits) broad and 6 feet (= 4 cubits) high, wall walk 4 feet broad, and parapet 2 feet thick; the broad sections would be broader than the narrow sections in the simple proportions of 3:2. Such might have been the general dimensions agreed upon where terrain made it possible to follow them.

²⁵ The lengths in metres of individual sections are as follows (measurements are read to the nearest metre): narrow sections: 73, 70, 70, 73, 79, 71, 60, 47, 47, 51, ^**, 105, 97, 49, 45, 51, 47, 63, 70, 105, 73, 71, 78, 70, 73, 73, 70, 62, 71, 97, 50, 51, 73, 50, ^****, 47, 47, 51, ^******, 45, 47 (in south-to-north order, with asterisks marking the occurrence of broad sections); broad sections: 34 +, 58 +; 37, 30, 25, 28; 25, 28, 27, 24, 30, 36; 24, 26, ?35.

²⁶ An attempt to convert the measurements of wall sections into Greek would be complicated and not convincing. Any one of three units—the foot, the cubit, or the pace (βῆΜα = 2½ feet)—might have been the basic unit of length, and it would be impossible to identify the one in use while the method of measuring and its exactness remain uncertain.

²⁷ The one exception is at sallyport 49, due to the departure from the normal system of overlapping; even so, the opening there would have been oblique and partly screened: see above p. 163.

²⁸ Some broad sections have very shallow overlaps; at one sallyport the butt-ends do not overlap at all and are hardly even in line with each other.

²⁹ The variation appears fortuitous, and may depend on the ground, the material at hand, or the methods of working parties. For similar variation, cf. Philo of Byzantium, (for full references see below p. 168, n. 33).

³⁰ Rough walling closes several of the gaps but definitely postdates the original construction; other openings have been blocked by falls of rubble.

³¹ Cf. Vitruvius i. 5. 42 for an exposition of the principle. It is seen in operation, at widely different periods, at Tiryns (Schliemann, , Tiryns (1886)Google Scholar pl. ii), Stymphalus (Orlandos, , PAE 1926, 132, fig. 1Google Scholar), and Mantinea (Fougères, , Mantinée et L'Arcadie orientale (1898) 151–7, pl. viiiGoogle Scholar: hereafter called Mantinée).

³² For an apt description of these tactics cf. Onosander (mid-first century A.D.), Strategikos xl. 3 (Loeb ed., 1948):

³³ This use of sallyports is described by Philo of Byzantium in his treatise on city fortifications (ed. Diels, H. and Schramm, E. (Abh. Preuβ. Akad,, Phil.-Hist. Kl, (1919) no 12, 1920) vii, viii. 1. 33–35Google Scholar; Graux, C. (Textes Grecs, 1886) vi. 1–3Google Scholar; Thevenot, (Vet. Math. 1693) p. 82, ll. 15–26)Google Scholar: Cf. Fougères, , BCH xiv (1890) 89Google Scholar; Mantinée 159.

³⁴ Based on observation only; but excavation would probably produce little supplementary information here.

³⁵ See above, p. 159 n. 4; the railway cutting farther south is wide enough to have destroyed either a sallyport or a third gate.

³⁶ See above, p. 162. Any westward extension of the road to the plain, inevitable if it were to be a normal traffic route, would necessitate steep inclines quite at variance with the careful contouring of the known course.

³⁷ See above, pp. 155 f. At the present day a railway and a cart-road cross the south saddle, and mule-tracks the central and northern saddles, but the terraced road shows little sign of use.

³⁸ ‘Quarry-faced’ would be the terra most nearly applicable, cf. Scranton, R. L., Greek Walls 21.Google Scholar

³⁹ Professor A. J. B. Wace informed us that he and G. L. Cheesman had also noticed one drafted edge during the course of an earlier investigation of the Dema; cf. Hesperia xi (1942) 195, fig. 2 for another view of the same butt-end.

⁴⁰ R. L. Scranton, loc. cit.

⁴¹ Rather rougher work seems to occur in safer areas, and better work at more vulnerable points: Plate 33 c represents one piece of walling high up on Aigaleos, and Plate 33 e part of one of the mutilated sections across the south saddle; but these are extreme examples and the contrast is not usually as exaggerated.

⁴² Cf. Scranton, op. cit. 12.

⁴³ Aristotle's suggestion that fortifications should decorate as well as defend (Pol, 1331^a10) applied to city walls; as it is, the Dema is impressive and suited to its site.

⁴⁴ Wrede, Attische Mauern (1933) 11Google Scholar recognizes rusticity as a characteristic of a number of fourth century B.C. Attic walls.

⁴⁵ Inhabitants of nearby Ano Liosia recall that Italian troops patrolled the line of the Dema during the last war. The rock-cut trench to the east of the south hill may perhaps also be connected with these troops.

⁴⁶ Curtius and Kiepert, Karten von Attika vi (marked ‘Lager’); Milchhoefer, , Text ii. 45Google Scholar; Dow, , Hesperia xi (1942) 207.Google Scholar The remains were visited by the Director of the British School and J. E. Jones.

⁴⁷ Gardikas, , PAE 1920, 68Google Scholar; cf. Dow, , Hesperia xi (1942) 206–7, fig. 13Google Scholar for an impression of the size of the round base of the tower.

⁴⁸ Some disturbance has been caused by the hollowing out of the fill, as if for a rifle pit (see p. 171 n. 45). For the rest, the fill appears to be original and homogeneous with the face. It is not due to the collapse of a hollow structure, for the packing inside is solid to the full height of the facing, while outside the fall is scanty; furthermore, there is no trace of a doorway at ground level.

⁴⁹ Noted earlier by Gardikas and Dow, see n. 47 above.

⁵⁰ Leake, , Demi of Athens 143Google Scholar; Milchhoefer, , Text ii. 44Google Scholar; Gardikas, op. cit. 70; Wrede, op. cit. 43.

⁵¹ The top of the structure has been surfaced with concrete, and a trigonometrical point has been erected upon it; but (as at the Wall Tower) the height of the base, the scantiness of the fall outside, and the absence of a groundlevel doorway indicate that the packing of the base is original. Rough stone steps abutting on die eastern side are probably modern, erected for convenient access to the trigonometrical point.

⁵² If the figures and contours of Karten von Attika vi may be trusted, it can be calculated that the towers would be intervisible if they had been originally 15–20 metres high. This would give a height: diameter ratio of 2½ or 3 to 1, which accords moderately well with such figures as are available for towers founded on free-standing walls (Hesperia xxv (1956) 135). Towers founded on solid-stone platforms would very likely be as lofty as those founded on ring-walls (BSA xlix (1954) pl. 5 b illustrates a well-preserved round tower, 7 m. in diameter and c. 17 m. high, built on a ring-wall of a single thickness of ashlar blocks). This calculation is in itself no proof, but tentatively suggests that direct intercommunication between the two towers was practicable.

⁵³ Cf. Polybius x. 44–47 for some methods of signalling in antiquity.

⁵⁴ Published examples of towers, particularly round ones, associated with enclosures, seem to differ in siting, construction, and purpose; many occupy sites with little command of terrain, are hollow and entered at ground level, have other buildings associated with them, and may have had domestic, manorial, or industrial functions (Young, J. H., AJA lx (1956) 51–55CrossRefGoogle Scholar; Hesperia xxv (1956) 122–46, with full references to other publications). Round towers with solid bases and enclosure walls have not yet been observed elsewhere or not distinctively recorded. Other towers in Attica, whether farm towers or military towers, are not of this type, but are either rectangular, or round but hollow, based on ring-walls. The closest parallels available are Plakoto in Attica (Chandler, , JHS xlvi (1926) 14–15Google Scholar; Wrede, op. cit. 34, 35, fig. 8, 57, pls. 92, 93) and Nemea Station tower in the Peloponnese, (AJA xliii (1939) 80–81Google Scholar, fig. 1). Plakoto is indeed a round watch tower, set within a (rectangular) enclosure and upon a peak, built with squared stone similar to but better than the masonry of the Aigaleos Tower, and entered at first-floor level; but it is ring-walled and hollow inside down to the ground. The Nemea Station tower is circular and has a solid platform, but this appears to be a terrace with a battered slope rather than an upstanding solid drum.

⁵⁵ Curtius and Kiepert, Karten von Attika vi; Milchhoefer, op. cit. 44; Lowry, H. and Smith, E., ‘A Survey of Mountaintop Structures in Attica’ (Unpublished Papers of the American School of Classical Studies at Athens, 1954) 39–40 and App. iii.Google Scholar

⁵⁶ Marked as a tower on Karten von Attika vi.

⁵⁷ Leake, op. cit. 125; Chandler, op. cit. 5.

⁵⁸ Leake, op. cit. 143; Milchhoefer, op. cit. 14; JHS ii (1881) 122; Gardikas, op. cit. 70.

⁵⁹ Differences of quality in the Dema need only indicate divergences in tactical needs, and a hasty relapse during the building from good work to poorer; cf. pp. 163 f. above.

⁶⁰ This consideration is perhaps not absent from interpretations of the wall as a frontier-work between a territorially limited Athenian state and an independent Eleusis (R. L. Scranton, op. cit. 41–42; so also Beloch and Chandler, p. 153 above; Milchhoefer, , Text ii. 45Google Scholar, attributes the wall to the same early period but precludes its use as a permanent boundary). The wall should be regarded as primarily a tactical defence and as not essentially a territorial boundary; consequently its position between Athens and Eleusis need neither imply a limitation of the former's territory and enmity towards the latter, nor compel its attribution to a period when these two conditions were fulfilled.

⁶¹ Two occasions have been suggested, both based on the assumption of hostility between those in control of Athens and secessionists at Eleusis. Skias (p. 152 above) suggested the period of the Thirty Tyrants, and Kirsten (see Phillipson, p. 153 above) any one of the recurring occasions during the third century B.C. when Eleusis was separately governed. Both suggestions are valid to the extent that emphasis is laid on a conflict existing or impending between forces east and west of the Dema, rather than on their mere separation.

⁶² Hdt. vii. 176.

⁶³ Hdt. vi. 36; the wall would have been as long as the isthmus is wide, that is, 36 stades or about four miles long.

⁶⁴ Hdt. vii. 176.

⁶⁵ Hdt. viii. 40, 71.

⁶⁶ Thuc. iv. 3, 4, 13.

⁶⁷ Thuc. iv. 45. 2.

⁶⁸ Many walls were erected as part of siege works, as at Syracuse, but such walls of circumvallation are better not considered here in connexion with linear walls of a defensive kind.

⁶⁹ Xenophon, , Hell. v. 4, 9–10, 38–42, 49.Google Scholar

⁷⁰ Diod. xviii. 13. 3.

⁷¹ Polyb. x. 41. 5.

⁷² Among linear walls in Greece are structures which do not belong to the classical or Hellenistic periods: both Roman and Byzantine periods are represented in the East Wall at Thermopylae (Grundy, , The Great Persian War 291Google Scholar; Béquignon, Y., La Vallée du Spercheios 48Google Scholar; AJA lxii (1958) 210 n. 64, noting several walls at Thermopylae); the medieval in the walls near Cenchreae (Fowler in Fowler, H. N. and Stillwell, R., Corinth i. 1. 104–5Google Scholar); and the Turkish period in a high rubble-and-mortar wall with rounded bastions, parapet, and loop-holes, which crosses the long peninsula running eastwards from the port of Megara towards the north-west corner of Salamis.

⁷³ Fowler, op. cit. 55; BSA xxxii (1931–2) 68–90.

⁷⁴ The wall at Thermopylae must be on or close to the line of the Phocian wall, and has been identified as that fortification (Grundy, op. cit. 288–9; Béquignon, op. cit. 46–48, 240–1; AJA xliii (1939) 699–70; Marinatos, in VI Internationaler Kongress für Archäologie, Berlin 1939 (1940) 333–41Google Scholar). It ran down a spur of high ground to the sea so as to bar advance from the south. Starting from a square tower on the spur, the wall followed a zigzag course along the south edge of the ridge, was some 3 metres thick, and had (at least) two rather narrow gates. These remains represent more than one period of construction. The wall near Stylida likewise ran down from high ground towards the sea, having indentations formed by alternate 12–18-m. long curtains and 3-m. long offsets or flanks; the wall was 3 m. thick and strongly built with inner and outer faces of regular, hammer-dressed trapezoidal masonry (Béquignon, op. cit. 294–5, 298–9, pl. xiv; where the wall is identified as Leosthenes' wall, built to blockade Lamia. The distance from Lamia and the careful finish of the masonry suggest that the structure may not be this field-work, but a more permanent fortification. But its style does indicate a date in the late fifth or fourth century B.C. (Scranton, op. cit. 85–98)). In the present state of knowledge, neither wall throws much light on the tactical use of the Dema; their plans indicate that importance was attached to enfilading fire but not particularly to sallying.

⁷⁵ Texier, , Description de l'Asie Mineure iii (1849) 142–3, 147–8, and pls. 147–9Google Scholar; Judeich, , AM xv (1890) 137, 144–53, pl. iiiGoogle Scholar; Krischen, F., quoted in AA xxviii (1913) 476Google Scholar; Guidi, G., Ann, iv–v (1921–1922) 346–50.Google Scholar

⁷⁶ Martin, R., BCH lxxi–lxxii (1947–8) 85–92.Google Scholar

⁷⁷ von Gerkan, A., Miletos ii. 3Google Scholar, Die Stadtmauern (1935) 53–80, 124–6, pls. i, xiv, xvi.

⁷⁸ Fougères, G., BCH xiv (1890) 65–90CrossRefGoogle Scholar; Mantinée 130–61.

⁷⁹ Sallyports or posterns occur in many fortifications, but rarely so numerously and systematically; cf. Samiko, (Hesperia vi (1937) 526, fig. 1)Google Scholar.

⁸⁰ For these dangers see Fougères, , BCH xiv (1890) 88–90Google Scholar; Mantinée 160.

⁸¹ Philo of Byzantium discusses the proper protection of sallyports against artillery fire in the passage quoted above, p. 168 n. 33, and immediately before advocating multiple sallyports he describes the siting of defensive batteries. We are grateful to E. W. Marsden for discussing with us by correspondence the connexion between sorties and the development of artillery.

⁸² Collingwood, , Antiquity vi (1932) 272–3.Google Scholar

⁸³ Lawrence, A. W., JHS lxvi (1946) 99–107.CrossRefGoogle Scholar Thanks are due to Professor Lawrence for access to his notes on Greek fortified sites.

⁸⁴ After the expulsion of the Persians, Themistocles planned to fortify the Piraeus with walls so thick and so high as to repulse all attacks. They could then be lightly garrisoned by a few inferior troops, while the rest of the manpower served at sea (Thuc. i. 93. 6).

⁸⁵ During the Pe oponnesian War, before artillery was known in Greece, some sieges were marked by a ruthless exploitation of all the then known resources of war and the improvization of new devices; cf. the siege of Plataea, 429 B.C. (Thuc. ii. 75–78), and the flame-thrower used at Delium, 424 B.C. (Thuc. iv. 100). Dionysius of Syracuse, the first to use catapults, when besieging Motya in 398 B.C., made use also of other innovations, notably movable towers (Diod. xiv. 50–51). Alexander the Great, when attacking Tyre in 332 B.C., employed the first recorded stone-throwing ballistas and other improved siege-machinery (Arrian, , Anab. ii. 16 et seq., especially 21. 7Google Scholar).

⁸⁶ At the siege of Plataea (see previous note) the first precaution taken by the besieging forces was to erect a stockade to prevent sorties.

⁸⁷ Fougères, , Mantinée 139.Google Scholar The first use of artillery was in Sicily at Motya in 398 B.C. It became known in Greece, at Athens, by 371/370 B.C. (IG ii.² 1422. 9; cf. IG ii.² 120. 37). Mantinea was refounded by Epaminondas in 371 B.C., when the circuit now preserved was constructed (Fougères, op. cit. 132). A. W. Lawrence, op. cit. 102, postulates that a besieged garrison would soon learn to counter enemy artillery with artillery of its own, mounted on the city walls, but that there would be a time-lag between the first use of the new weapon and the changes brought about by it in the actual design of fortifications.

⁸⁸ E. W. Marsden sees in the meandering course and the re-entrants at the saddles a similarity to the enfilading devices of artillery-period warfare. He tentatively suggests that catapults were placed on or near the wall in a defensive role, and that the individual sloping ramps of the broad sections (so different from the normal means of access to ramparts, staircases built parallel to the walls) were for mounting these on top of the rampart. Ramps were rare in Greek fortifications; for a ramp at Miletus, see A. von Gerkan, op. cit. 63 and fig. 35.

⁸⁹ The peltast was originally, and especially as developed by the Athenian military innovator Iphicrates, a soldier armed with missile weapons and trained to skirmish. Later he was given a spear far longer than the hoplite's, perhaps to compensate for the disadvantage in close combat of insufficient armour, and used in the phalanx. The longer spear, which would be unsuitable for skirmishing, was sometimes anachronistically attributed to Iphicrates (Diod. xv. 44. 3).

⁹⁰ Hell, v. 4. 9–10, 38–42, 49.

⁹¹ Some indication of this is seen in Diod. xv. 32. 2–3.

⁹² Xenophon, op. cit. v. 4. 39:

⁹³ According to Scranton's chronology, the prevalence of coursing and quadrilateral forms, noticeable even in so roughly constructed a wall as the Dema (see p. 170 above), would seem first to exclude a date in archaic times suggested by a certain similarity between much of the rubble work and Lesbian techniques; secondly, to point to a chronological advance on the normal polygonal style of the fifth century B.C., and finally to recall the trapezoidal and ashlar techniques of the later fifth and fourth centuries B.C. Cf. Scranton, op. cit., chronological notes, and summary.

⁹⁴ Young, John H., Hesperia xxv (1956) 122–4 and pl. 34 b.CrossRefGoogle Scholar

⁹⁵ Pouilloux, J., La Forteresse de Rhamnoute (1954) 23–48 and pls. xv. 2–3, xxi. 2–3, and xxiv. 1–2.Google Scholar

⁹⁶ Op. cit. 31, 44, 50, and pl. xiii 1.

⁹⁷ Ibid. 43–66. Cf. also C. W. J. Eliot, review of J. Pouilloux. op. cit., in AJA lx (1956) 200.

⁹⁸ Op. cit. 80.

⁹⁹ Kourouniotes, K., ADelt xi (1928–1928) 49–51Google Scholar; Wrede, W., Attische Mauern 59Google Scholar; and Young, Rodney S., Hesperia xx (1951) 256–61.Google Scholar

¹⁰⁰ John H. Young, op. cit. 126 (‘The Golden Pig Tower’) and pl. 35c.

¹⁰¹ Krischen, , quoted in AA 1913, 476Google Scholar; Guidi, G., Ann, iv–v (1921–1922) 346–50.Google Scholar Iasus was a fortified city in the fifth century B.C.; it was destroyed and abandoned in 405 B.C., and was not refounded and rebuilt until 394 B.C. (Thuc. viii. 28. 2–4, 29. 1; Diod. xiii. 104). The walls of the island city probably date to this reconstruction and so possibly does the inland wall. But its style differs from that of the walls on the island, and may indicate its later construction as an additional defence. The coursed masonry is consistent with a fourth-century date. Martin, , BCH lxxi–lxxii (1947–1948) 92 n. 3Google Scholar, includes the inland wall in fortifications of the fourth or third century B.C.

¹⁰² Op. cit. 120–47.

¹⁰³ Martin, op. cit. 92 and n. 4.

¹⁰⁴ Noack, F., AM xix (1894) 428Google Scholar, discerns Mycenean origins; Scranton, op. cit. App. ii, considers it an early feature; Martin, op. cit. 135–9, classifies the relevant sites and draws subtler distinctions; the Dema belongs to his second group of later indented walls, dated from the midfourth century B.C. onwards, and including also Priene, Samothrace, Samiko, and the wall near Stylida.

¹⁰⁵ The siting of the wall along the south edge of the ridge and the provision of staircases along its north side suggest that it was built by forces occupying the north end of the pass against the advance of an enemy coming from the south—a situation not in keeping with the needs of either the Phocians or the Spartans. The staircases must surely therefore be later than 480 B.C., and possibly the main fabric of the wall, an admittedly much renovated one, represents a new layout of the fortification some time after that date. See Béquignon, op. cit. 47, for possible later occasions. The occupation of the pass by Philip in 346 B.C. would provide a plausible occasion for the building of a rampart facing south; Cf. AJA 1958, 211–13.

¹⁰⁶ The similarity in function and perhaps in general plan between the Dema and the defensive stockade erected in 378 B.C. by Athenian troops in Thebes is possibly to be regarded as confirming a fourth-century date. Likewise, if any weight may be attached to the conjecture that the Dema was planned by and for a professional soldiery, consisting of light-armed men, this general date finds support in the rise of professionalism in that century and a consequent increase in tactical skill, and also in the creation and independent use of the peltast arm, notably by Athenian and mercenary commanders, from the beginning till towards the end of the fourth century B.C.

¹⁰⁷ Cf. P 1830 (Hesperia iii (1934) 318, fig. 4: A 20).

¹⁰⁸ i.e. was it left by farmers living in the vicinity, by those who pastured their flocks in the valley, by those who built the wall, by those who manned the wall (if indeed it was ever guarded), or was it left by a combination of these and possibly other groups?

¹⁰⁹ There were also two unglazed sherds, one being from the rim of a lekanis.

¹¹⁰ Both type and date are based on the study of Greek lamps by R. H. Howland to be published in 1958 as a monograph subtitled Greek Lamps and their Survivals in the series The Athenian Agora, Results of Excavations conducted by the American School of Classical Studies at Athens.

¹¹¹ Although the last invasion of Attica during the Archidamian War took place in 424 B.C., the threat of further invasions would have persisted until the Peace of Nicias.

¹¹² Thuc. ii. 19. 2.

¹¹³ Thuc. ii. 13. 2 and 14.

¹¹⁴ Thucydides (v. 25) describes the period of peace as a time of unrest and suspicion. This could be a retrospective view on the part of the author. At all events, the average person was probably quite unaware of the unstable political condition.

¹¹⁵ The Thebans who joined the Spartans were particularly thorough in their plundering and Hellenica Oxyrhynchia xii. 4 tells us that they removed even wood and roof tiles (see Hardy, W. G., CP xxi (1926) 346–55Google Scholar for a full commentary on this passage). It is tempting to consider that we have a concrete example of this ruthless type of raiding in that we have found only a few scraps of roof tile in the area of the house.

¹¹⁶ This point was fully appreciated in antiquity and provision was made accordingly. Philo of Byzantium recommended that a space of 60 cubits be left in front of city walls. At Nisyrus an inscription of the fourth century B.C. set into the outer face of the wall of the city declared the 5 feet of ground adjacent to the circuit to be public property (see p. 118 above); this strip of open ground would be too little for tactical security, but it would prevent damage or encumbrance to the wall from private buildings and facilitate a general clearance of the area if needed.

¹¹⁷ Had the wall been obsolete at the time when the house was built, then a house might have been built in front of it. For economy, however, the house would probably have been placed so as to utilize the Dema as its rear wall as is the case with more recent structures along the Dema Wall. Also, a wall stylistically dated as not earlier than the fifth century B.C. would not be militarily obsolete by the end of that century; strategically it could have become obsolete due to Pericles' policy, but its potential value would still remain and so the above-mentioned factors of military tactics and private security would still operate.

¹¹⁸ It might be expected that the complete destruction visited upon the house would have affected to the same degree an existing (even if ‘obsolete’) rampart; if the wall did not exist at the time, the contrast is less surprising. Granted that the house could well have been razed down to its very last course by enemy action, natural decay, or modern cultivation, the sparse remains now preserved suggest nevertheless that the builders of the Dema may have levelled the already derelict site to remove all obstacles from in front of their wall. But such levelling (if any took place) could not account for a primary destruction of the house and the termination of its occupation, for this event occurred at a date when the construction of the Dema would have been strategically pointless (with the enemies of Athens holding Decelea) and historically improbable (since so bold an enterprise and a departure from the accepted (Periclean) strategy would hardly have been left unrecorded). It would follow that there was a lapse of time between the abandonment of the house and the erection of the rampart.

¹¹⁹ This fort, noted by Milchhoefer, A. (Curtius, E. and Kiepert, J. A., Karten vort Attika, Text vii. 7)Google Scholar and Chandler, L.JHS xlvi (1926) 15)Google Scholar, was examined and surveyed by E. Vanderpool and C. W. J. Eliot in 1953. The latter is preparing a study of the fort.

¹²⁰ Mrs. Carl Roebuck in the spring of 1954 kindly looked at some similar tiles from Attica and said she believed them to be of fourth century B.C. date. One could also mention parallels in the Athenian Agora from fourth century B.C. contexts, particularly A 432 (Hesperia, Suppl. iv (1940) 78–79 and 81, fig. 61), which has been placed in the third quarter of the fourth century B.C.

¹²¹ None of these fragments can be dated; the eaves tile, however, is not later than the fourth century B.C. Kalathoi with corrugations on the inside surface are normally thought of as Hellenistic, but they first appear in the latter half of the fifth century B.C. (cf. Athenian Agora P 11017, Corbett, P. E., Hesperia xviii (1949) 335Google Scholar and pl. 97, no. 95) and there are examples from the fourth century B.C. (cf. Athenian Agora P 4062, Thompson, Homer A., Hesperia iii (1934) 326, A 63).Google Scholar

¹²² According to G. Roger Edwards, to whom we are grateful, c. 310 B.C.

¹²³ Even if the wall and towers were built for a specific occasion, there is no reason to think that once that occasion had passed all elements of the system were entirely abandoned; the towers at least would have always remained valuable observation posts.

¹²⁴ Translated by J. O. Burtt, Minor Attic Orators ii (Loeb ed., 1954).

¹²⁵ Aeschines, In Ctes. 27:

¹²⁶ Op. cit. 27.

¹²⁷ This procedure is inferred from remarks in Aeschines' In Ctes. and Demosthenes' De Cor. relating to the latter's experience as a commissioner.

¹²⁸ Demosthenes, De Cor. 300 and Plutarch, , Mor, 851a.Google Scholar

¹²⁹ Aeschines, op. cit. 23.

¹³⁰ Ibid. 31.

¹³¹ Ibid. 17 and Plutarch, op. cit. 845 ƒ, (Demosthenes op. cit. 118 and Plutarch op. cit. 851 a record his gift as three talents, but neither decree can be wholly trusted for details.)

¹³² Demosthenes, , De Fals. Leg. 125.Google Scholar

¹³³ Had the Athenians done any work on the walls around Piraeus—presumably, their first consideration—it is hardly likely that Demosthenes only a decade later would have had to make such extensive repairs.

¹³⁴ J. Pouilloux, op. cit. 62–66.

¹³⁵ Cf. also Aristotle, Pol. 1321^b26.

¹³⁶ Demosthenes, , De Cor. 118Google Scholar, described his own position as This statement is bedded in an untrustworthy decree, and is itself suspicious since the official title for the appointment was quite different. We suggest that this title was designed post eventum and describes the nature of Demosthenes' work on the circuit of Piraeus, not the work of the board as whole.

¹³⁷ Translated by C. A. Vince and J. H. Vince (Demosthenes, De Cor. and De Fals. Leg. (Loeb ed.), 1926).

¹³⁸ Conon, for example, may well have had the machinery for such a task, but certainly lacked money. And would his friends the Thebans have understood his motive?