Greek fortifications of the fourth century bc

1In late Classical and early Hellenistic times the construction of fortifications intensified all over the Greek world and larger and more sophisticated systems were constantly developed. In terms of sheer size and cost this must count as one of the greatest achievements of Greek architecture. There are several reasons for this development. One is the changing political situation: as new states or empires emerged, new borders and new frontiers were created, for whose protection new defences were needed. The frequent wars rapidly led to new developments in siege techniques which again necessitated an equally rapid development in defence strategies and new or improved fortifications2.

2In very many cases the precise dating of these walls has to be based on an estimation of what stage they represent in the general developments in siege techniques. Although the fortifications of a city were in many cases repaired and modernized during their life-time, we may hope to get some idea of the first construction date by analyzing the most characteristic and original traits. This I aim to do for Halikarnassos.

3Sicily is often regarded as the place where some of the most decisive new advances in fourth-century BC siege warfare were made3. The catapult is thought to have been invented on the initiative of Dionysios I of Syracuse in 399 BC, and although this was probably a non-torsion oxybeles for shooting arrows, it initiated a rapid development of artillery4. By the invention of stone-throwing torsion catapults, lithoboloi, a new type of siege-machine was created, which could in time be employed directly against fortification walls. The development of very large lithoboloi, however, is generally believed to have been the work of the Macedonians after the middle of the fourth century; and as both the oxybeles and the first lithoboloi were ‘antipersonnel’ artillery, they did not constitute a serious threat to the fortification architecture for a very long time. But Dionysios of Syracuse also had fresh success with the employment of effective rams and large siege towers with gangways for his siege of Motya in 3975, and perhaps this had a more immediate effect on the lay-out of Greek fortifications in the early fourth century. Among the new strategies and systems that were developed for fortifications in this period, the ‘great circuit’ or the ‘Geländemauer’ was particularly popular, no doubt because it was effective in preventing the besieger from getting his battering rams and siege towers close to the walls.

4The Greeks on the mainland and in western Asia Minor may have had much to learn from Dionysios about siege warfare and fortifications, but to what extent the developments in mainland Greece and western Asia Minor were actually dependent on the new developments in Sicily is unclear. Knowledge could have spread very fast with travelling artisans and mercenaries, and Dionysios and his experts may themselves have been instrumental in bringing the new techniques to the Aegean, when he interfered on the side of Sparta in the north-eastern Aegean shortly before the King’s Peace.

5Syracuse itself had its great circuit of about 27 km in the late fifth century. This must be the largest of all Geländemauern and many new city walls followed in the fourth century. The fifth-century walls of Athens were reconstructed by Konon. Although they were very long, they can hardly be described as Geländemauern: they had to be reinforced with outworks such as ditches and proteichismata. But many new city walls of Geländemauer-type were erected on the Greek mainland during the fourth century. Among these, some of the most impressive were a result of the new political situation after the defeat of Sparta by the Thebans. The well-preserved walls of Messene constitute the spectacular culmination of the Geländemauer-type on the Greek mainland in the fourth century.

6In western Asia Minor the historical situation in the early fourth century was very different from that of Sicily and mainland Greece. Decades of fighting between Athenians, Spartans and Persians finally came to an end in 387/6 BC, when the Persians with the Spartans forced the Athenians to accept the conditions of the so-called King’s Peace or Peace of Antalkidas. The peace conditions dictated that from this year on western Asia Minor should be Persian territory and politically the situation was back to what it had been before the great Persian wars almost a century earlier. The peace, however, initiated a new period of prosperity and stability under Persian rule and western Asia Minor experienced one of the most flourishing periods in the history of Greek monumental architecture: the Ionian Renaissance.

7The Peace also marked the beginning of a new and intense era of fortification building in Asia Minor, which was hardly inferior to that of Sicily and the Greek mainland; but unlike these areas western Asia Minor had been through a severe economic recession in the fifth century, which had stopped almost all new monumental building for a century. These conditions probably also affected fortification building. Although the construction of defence walls must of necessity sometimes have had priority over religious and representative architecture, it seems unlikely that any of the great, ambitious, Geländemauern, which are so characteristic of western Asia Minor in the fourth century, were new creations of the difficult times of the fifth century. This impression is corroborated by the written sources, if one takes Thucydides literally when he states that (in 427) Ionia was ‘unfortified’6. Winter thinks that fortresses must have been built and cities fortified or refortified during the last decade of the Peloponnesian War, and it seems indisputable also from Xenophon’s account of the events of the early fourth century that many towns of western Asia Minor had functioning fortifications. But they were hardly large Geländemauern, and in general I find it difficult to follow Winter’s opinion that the period from the late fifth century until c. 375 was a time in which many new fortifications were built ‘incorporating new developments, such as thicker curtains, higher and more numerous towers and more numerous sallyports’7. I would rather see the large fortifications of Geländemauer-type as the most important innovation of fortification building in Asia Minor after 375 BC.

8In Asia Minor the Geländemauer type had been known already in the Archaic period in some of the large city states like Samos and Miletos. Its strategic qualities were so important that it was reintroduced and considered superior to other kinds of city fortifications during the entire fourth century. The main strategic advantage was, of course, that it included the heights around the city and so forced any attackers to make their way uphill on difficult ground under constant fire from the defenders. Stones and bolts discharged from the higher position would have a larger range than those of the attacking forces downhill. For anyone who has studied these city walls in the field it is easy to imagine in what state any attacking force would be if it finally reached the foot of the wall unharmed by arrows and stones before even attempting to climb the wall and start fighting man to man. But, most important of all, the steep and broken terrain in front of most Geländemauern made it extremely difficult for the attackers to bring their siege engines up to the wall. The advantages of the Geländemauer-type of fortification are evident.

9The disadvantages or limitations of the Geländemauer were, however, considerable too. First, it required a very large population to man such long walls, while the majority of cities mentioned in the written sources were more like towns and villages than large and populous cities. Karia, for instance, is known for its numerous hilltop towns mostly surrounded by a dry-stone wall of only slightly worked stones quarried on site. Individually these towns would have had great difficulties in withstanding an Agesilaos or even an Athenian patrol coming for the collection of tribute, which were some of the threats with which there was relatively fresh experience in western Asia Minor. The most efficient solution was to synoikize smaller towns into larger ones. Synoikisms and refoundations of cities to new and better sites are indeed frequent in the first half of the fourth century. Secondly, a Geländemauer was extremely expensive and therefore would often be the prerogative of large and wealthy cities, or was erected on the initiative of powerful states and leaders like Epaminondas. As McNicoll has pointed out, the long circuits in the Hellenistic period were generally built by kings8. Kingship, rich financial resources and synoikisms are all characteristic features of the time after the ‘King’s Peace’9.

10The long Geländemauer-type became outdated in the mid-Hellenistic period as new techniques and strategies in poliorcetics developed, especially as a result of the developments in artillery. The heavy artillery from an attacking army had to be met with just as dense a network of large catapults from defenders, from numerous high and strong towers at any point in the circuit that the enemy would choose to attack. To meet such needs circuits were made shorter and stronger.

11The Geländemauern of the fourth and third centuries in Asia Minor constitute a very impressive and distinctive phase, and there is reason to believe that the city wall of Halikarnassos stands at the very beginning. Among its followers were Myndos, Alinda, Latmos, Kaunos, Alabanda, Knidos, Priene (probably), Herakleia under Latmos, Ephesos, Samos (fourth-century fortifications) and Pergamon (fourth-or third-century fortification), and probably many others, as well as a great number of watch-towers and fortresses. No doubt the knowledge of new developments and inventions in Greek warfare and poliorcetics were quickly interchanged between different parts of the Greek world, leading to a certain homogeneity. But the cities just mentioned have at least one particular feature of masonry in common, the ‘double corner bond’, which seems to be restricted to Asia Minor and some of the adjoining islands.

Asia Minor after the ‘King’s Peace’ of 387/6 bc

12With the conclusion of the ‘King’s Peace’ the situation in western Asia Minor changed completely. The difficult economic conditions of the fifth century improved remarkably fast. The new political stability and security combined with access to the unlimited markets of the Persian empire probably constituted a milieu for production and commerce not paralleled since the Archaic period10. The financial means were now present for large-scale building. The new political situation had also completely changed the military situation and the need for fortifications. Western Asia Minor was now part of the Persian empire, and the western border of the empire was constituted by the Aegean coast. The formerly free Greek cities on the coast from now on formed part of the defence of Persia’s western frontier. The planning of the Persian defence and the fortifications of the land and the cities was probably put in the hands of individual satraps, among whom the satrap family of Karia, the Hekatomnids, appears to have been particularly enterprising in almost all respects: politics, economy, defence, art and architecture11. They, and perhaps Maussollos in particular, were most likely responsible for the defence of Karia.

13In recent years there have been several new studies on Karian fortifications, of which some treat a specific site, while others take a broader perspective12. Among the latter may be mentioned that of A. W. McNicoll and the detailed and valuable study by I. Pimouguet-Pédarros on Classical and Hellenistic fortifications in Karia, which is now essential for all work on the subject13. From slightly different angles, Pimouguet-Pédarros and I reach approximately similar conclusions about a large, coherent, Hekatomnid fortification system for the whole of Karia based on a chain of strong fortified cities supplemented by watch-towers and smaller mountain fortresses. In a wider perspective, McNicoll wrote about Maussollos, that ‘…in the matter of fortifications he anticipated developments which in Asia Minor only became common a generation or more after his death. For this reason the defence works of Maussollos and his successors cannot be omitted from an examination of Hellenistic fortifications in Asia Minor’14.

Halikarnassos and the defence of the Halikarnassos Peninsula

14As the western frontier of Persia was moved to the Aegean coast, the fleet gained a much more important role in the defence against potential aggressors from the Greek mainland. This may well have been the decisive reason for moving the satrapal capital from inland Mylasa to Halikarnassos. Halikarnassos had a very good natural harbour opening up to a larger bay which is relatively well protected from the open sea by the large island now known as Kara Ada. The natural conditions with an isthmus connecting the Zephyrion peninsula to the mainland had fine potential for the establishment of a naval base that could harbour a substantial part of the fleet of the Karian satrapy. In addition to the naval harbour of Halikarnassos, the Hekatomnid fleet may have had several other good harbours at its disposal such as Kaunos, Knidos, Myndos, Iasos and Miletos. It is not known precisely in which year Halikarnassos was made capital of the Karian satrapy, though it must have been after 377, when Maussollos became satrap and before 362, when the palace in Halikarnassos is mentioned by Diodoros in connection with the Satrap’s Revolt15, and it is generally thought that it happened before the synoikism of Kos, traditionally dated to 366. Strategically, however, the need for strengthening the coastal area became a reality immediately after the ‘King’s Peace’ and it became even more urgent when the second Athenian Confederacy was created in 377. Accordingly, it seems most likely that the creation of a new fortified capital at the old site of Halikarnassos took place early in the satrapy of Maussollos. A similar point of view was expressed by Hornblower and Pimouguet-Pédarros16.

15Archaeologically, the history of Halikarnassos goes back to Mycenean times, but in the early fourth century it was probably rather small and had neither a size nor fortifications that could in any way answer the needs of a new capital for the ambitious Hekatomnids. The town was entirely rebuilt by Maussollos with a new, orthogonal town plan, temples and sanctuaries, a satrapal palace, a unique dynastic tomb and possibly a theatre and a gymnasion, too. In addition to this, comprehensive naval installations were built and, last but not least, the huge new city wall of the Geländemauer-type.

16On the Halikarnassos peninsula the new principles for land defence seem to have been applied in a very systematic way. The existing small hilltop settlements in the area were impossible to defend against the kind of intruders Asia Minor had been frequently subjected to in the period before the ‘King’s Peace’. Therefore a comprehensive synoikism was systematically carried out by Maussollos17. Though the details are unclear it seems that six Lelegian towns were given up and their inhabitants transferred to three large and well-fortified cities: Halikarnassos in the centre and – at approximately equal distance – Theangela to the east and Myndos on the west coast of the peninsula. It seems likely that on this occasion Myndos was transferred from an inland hilltop settlement (at Bozdağ?) to the present location at Gümüşlük, with its fine, natural harbour. All three sites were supplied with large Geländemauern, which at Myndos and Halikarnassos included the harbours and secured access and provisions from the sea. Theangela was situated further inland, but its access to the sea at the bay south of the city was apparently defended by a small fortress on the east side of the valley south from Theangela to the coast18.

17Watch-towers constructed of strong ashlar masonry in regular courses were distributed over the peninsula, and even today it is possible from many of them to see other towers or parts of the city fortifications. I agree with Pimouguet-Pédarros that these towers must have had primarily a defensive purpose, probably for the kind of communication prescribed by Aeneas Tacticus19. This system of strong city fortifications supplemented by watch-towers and smaller fortresses was probably established for the whole of Karia20.

The fortifications of Halikarnassos

18For a discussion of the fortifications of Halikarnassos the detailed description of Halikarnassos by Vitruvius is essential for understanding the general topography, while the accounts of Alexander’s siege in 334 BC by Diodorus Siculus and Arrian are extremely valuable for the understanding of the city wall and its effectiveness in war21. In modern times the wall has been described in some detail22. The description in the present paper is to a very large extent based on observation made as part of the great Turkish restoration project in 1998-2000, the ‘Ericsson-Turkcell city wall Restoration Project’23.

The trace of the wall

19The city wall of Hekatomnid Halikarnassos is a true Geländemauer (fig. 1). Its length can be estimated at c. 7.5 km, if it is presumed that it started at one end of the harbour entrance and ended at the other24. The trace of the wall was evidently determined to a very large degree by the natural configuration of the site. A great effort was made to include all the high points that are within reasonable distance from the city centre. This means in particular the Salmakis promontory west of the harbour entrance, the Göktepe hill to the north-west, the north-east salient and finally the Zephyrion peninsula east of the harbour entrance. The wall runs along the outside of these high points and where possible along the outside of the ridges connecting them so as to give the defenders of the city the advantage of height.

20In very many places the wall is well protected by the steep terrain leading up to it, making it very hard for attackers to get close. This is particularly true of the part that encircles the north-east salient and runs down along the east side of the city towards the coast, since this is protected by a steep ravine along most of its course: a formidable natural fortification ditch. But from our recent investigations it has become clear that, in places where the natural defence was considered insufficient, a ditch was often excavated at some distance in front of the wall. One feature in relation to ditch-construction should be mentioned. Ditches could be 2 m deep and 5 m wide, so for instance on the north-west slope of Göktepe hill. The sides of this ditch are vertical and its floor is approximately horizontal. But if the ditch was dug on sloping ground, only the side nearest the wall would be about 2 m in height, the outer side lower. If the ground sloped more than 2 m over a distance of 5 m, this ditch would have no outer scarp (fig. 2) but would be reduced to a level, flat area (the bottom of the ditch) ending in a vertical scarp (the inner side of the ditch). This is how it appears in several places at Halikarnassos: a vertical, sometimes slightly irregular rock-cut escarpment running 5-10 m in front of the wall and with a level area in front that in parts look almost like a road. So what is, in principle, one strategic feature may in some places appear as a precise and carefully cut trench and in other places – mainly in steep and rocky terrain – as an irregular, more or less vertical scarp hewn in the rocky hillside at some distance in front of the fortification wall25. When these facts are taken into account it is evident that the trace of the city wall of Halikarnassos was very carefully laid out according to the configuration of the landscape so as to make the best of its potentials for natural defence and – when necessary – supplemented by ditches or escarpments.

21Because of these strategical considerations the city wall of Halikarnassos comprised a much larger area than was needed for habitation. Already in the Hellenistic period some areas inside the city wall to the north-west were used for tombs, especially on the western and southern slopes of the Göktepe hill above the theatre. The extent of the built-up area is not known in detail, but it centred around the harbour and from here spread out westward and northward almost all the way up to the Mars terrace near to the city wall. It appears that the districts closest to the city wall to the east and west were used for polluting industrial processes such as the making of pottery. Remains from pottery production have been observed on the surface inside the city wall south of the Myndos gate, while pits filled with broken and misfired pottery have been observed in a modern excavation for an underground parking garage south of the Mylasa gate. The area north of the Mylasa gate had the large stadion, probably because this area too was on the outskirts of the habitation area with space for such a large construction26. In the area of the north-east salient there is comparatively much pottery on the surface (especially just outside the north wall, almost as if the slope had been used as a dump); as this distant quarter of Halikarnassos apparently had its own necropolis along the road just outside the gate located here (the ‘Pedasa gate’), this district was perhaps not quite as desolate and remote as was previously thought.

A description of the city wall

22The description starts west of the harbour entrance and ends with the Zephyrion peninsula to the east (fig. 3)27.

23Until some time in the twentieth century, underwater remains of old harbour moles could be seen at both sides of the entrance to the harbour, leaving only a narrow entrance open (fig. 3)28. All of this is presently covered by the modern moles, but it seems likely that the older remains here formed part of the ancient fortification system, and so probably carried a wall, perhaps ending with bastions or towers flanking the harbour entrance in such a manner that it could be blocked by a chain. Of such a harbour wall no trace is known and no remains of its continuation westwards along the water front are preserved in situ. The walls of the Ottoman arsenal, however, have a large number of re-used, hard, greyish-blue limestone blocks of irregular shape, which strongly suggest that the fortification wall in this area was constructed in the same technique as most of the western wall all the way to the Myndos gate. On Newton’s map there are indications of traces of wall running west from the Arsenal promontory across the beach.

24The wall no doubt continued to the top of the ‘Kaplan Kalessi’ and there joined a fortress, the Salmakis fortress. The Kaplan Kalessi rock, which reaches a height of c. 60 m, is one of the most conspicuous topographical features in this part of the city (fig. 4). It forms a parallel to the rocky height on the Zephyrion peninsula, which presently carries the castle. Both consist mainly of hard greyish-blue limestone. It is one of two heights (akra) to which the Persians retreated the night after Alexander broke through the city wall in 334 BC, the other being the citadel on the Zephyrion peninsula29. Although almost no traces of walls are found in situ, it was – in 1978 – possible to follow the line of the wall in the rock-cut beddings running from the north-west towards the summit of the rock and at one point dividing into one branch along the north side and another running along the south side, thus together entirely encircling the summit. So, not surprisingly, it looks as if there was a fortified stronghold at the top of the Salmakis hill. A number of other beddings were visible close to the Ottoman tower on the summit, but not enough to decide if they had anything to do with the fortification, or with the sanctuary of Aphrodite. Jewellery of gilded clay is said to have been found at the summit of the Salmakis hill. On the water front to the east of the hill is the large Salmakis fountain complex where the famous inscription was found in 1995 by the Turkish military authorities30.

25From the north-western flank of the Salmakis hill the wall can be followed either as blocks or as a low ridge consisting of decomposed remains of the wall, covered by bushes and small trees. It passes across a level area with a small Greek chapel and is then cut through by the road to Gümbet. The state of preservation of this stretch is so bad that it is not possible to tell how many towers there were nor if there was a gate as one would expect. The materials here are three of the most typical: hard reddish to brown and violet andesite, an almost white and very soft andesite, and a hard blue-grey limestone. While the andesite is mostly cut into regular ashlars, the hard limestone is mostly applied in an irregular polygonal technique making best use of the natural size and form in which the stone is found on or near the construction site. The ashlars of soft white andesite are often in a state of complete decomposition.

26At 1 (on fig. 1) the outline of a tower about 10 m wide could just be discerned on the surface. It is of hard, reddish andesite, while the curtain-wall here is of soft white andesite. The wall continues towards the Myndos gate across the hilly landscape, which offers possibilities for two projecting bastions respectively at k1, where almost nothing is preserved and at k2, which has more clear indications of constructions of some importance. Small clandestine excavations (probably during the 1980s) revealed that there are structures of soft white ashlars inside the bastion. On the outside of the bastion at a distance of some meters is a remarkable level area running like a road in front of the wall – most clearly observed to the north and west. This is probably a level area resulting from the excavation of an escarpment in front of the wall (see above). The curtain-wall in between these two bastions constitutes a good and relatively well-preserved stretch of polygonal masonry of hard blue limestone. There is at least one tower (2) here: exceptional in being of reddish hard andesite ashlars, but nevertheless bonded into the polygonal curtain-wall. A look at the map of the part of the circuit described so far illustrates very clearly how the Arsenal promontory, the Salmakis fortress and the two strong advanced bastions offered exceptional possibilities for enfilading the stretches between the strong-points where the wall recedes in a long curve. A comparable strategic situation can be observed at the north-east salient.

27This situation changes dramatically north of the bastion (k2). The wall runs in straight lines and the only possibility for enfilading fire to meet attacks on the curtain-wall is by means of towers. In fact, remains of at least 15 towers in the stretch of wall from bastion k2 to the fortress on top of the Göktepe hill are visible. The size of the towers varies from less than 50 to over 100 m². The distances between the towers vary according to the character of the landscape and to the risk of attack at each particular place. The ground around the Myndos gate is relatively flat and particularly exposed to the threats of siege machines and therefore the towers are placed with an interval of only 50-55 m.

28The curtain-wall continues for some unknown distance, perhaps about 100 m, north of bastion k2 as a double wall whose inner and outer face are constructed of irregular polygonal masonry of hard limestone (fig. 5). It then continues all the way to the Myndos gate as an ashlar wall built of large regular ashlars of soft white andesite. As this material very quickly disintegrates when exposed to the weather it is poorly preserved except for parts that were covered up by earth or by disintegrated material from the upper part of the wall. This is, in fact, the material and the construction technique that was used for the curtain-wall up to the top of the Göktepe hill. Most of the towers are built of more durable materials, mostly of hard reddish or brownish andesite or hard, grey and blue limestone. All towers in the western part of the city circuit were built in ashlar technique no matter what techniques were used for the curtain-wall.

29The last two towers before the Myndos gate were measured and studied in some detail. The first, in the garden of the hotel ‘L’ Ambiance’, is beautifully built of strong, grey limestone ashlars. It is preserved up to 4 m (south-west corner) above the present ground level (fig. 6). The length of the south wall could be measured to c. 7 m and the west wall to 6.8 m. At this point, the main wall has totally disappeared and this offers a rare possibility for studying the rear wall of the tower. Like the next tower to the north, this one stands on its own without being bonded to the curtain-wall. It can be clearly seen that the rear wall of the tower is less well constructed–it contains blocks that were reused or rejected from another structure – because it was originally built against the city wall and not meant to be visible. This tower illustrates well what would happen if the curtain-wall collapsed during a siege: the tower would remain standing and for some time at least remain a threat for the attacking forces. The tower was massively filled up with earth and rubble in its lower section. No excavation was carried out, but some blocks in the centre of the tower suggest that this tower like most others of the Halikarnassos fortification had interior cross-walls in the first storey. Just north of this tower a small road runs through the city wall. It passes close to a late-antique necropolis with large two-storied tombs. The tombs, as well as the foundations for a small sanctuary of unknown date close to the south side of this road, were excavated by a Turkish team. This road may go back to late Antiquity too, a time when the large Myndos gate complex had been given up.

30The next tower along to the north was constructed of soft white andesite at least as far as its foundation, which is the only part preserved (fig. 7). It measures c. 7.2 m east-west and 6.1 m north-south. The north, west and south walls were well built, but the east wall is poorly constructed and not carried down to bedrock like the rest. The existence of the east wall of the tower indicates, however, that this tower was not bonded to the main wall either.

The Myndos gate

31After the recent excavations and restorations the Myndos gate stands out as a very impressive architectural structure, carefully designed and of good craftsmanship (fig. 8). The most prominent feature of the complex today is the two large projecting towers situated at about 26.6 m in front of the face of the curtain-wall. They are joined to it with connecting walls 3 m wide (fig. 9). In this way a large gateyard of 26.6 x 23.6 m is formed in front of the main gate, which however has completely disappeared together with most of the curtain-wall at precisely this spot. Two large stones of hard andesite were found ex situ however, with cuttings on the upper side probably for the pivot fastenings that held the door-panels, as suggested by the excavators of this area, Kudret Ata and Habib Karaçet. The gate opens directly to the large main street of Halikarnassos, running along the north side of the Maussolleion terrace and possibly leading directly to the other main gate, the Mylasa gate, in the east wall. There are no signs of arrangements for a gate between the two large towers. No doubt there was none. The gateyard only allowed a limited number of attackers to approach the city gate at a time, and they would be exposed to fire from both of the towers and from the walls all-round the yard. The wall walks were reached by a stair of which remains can be seen along the main circuit just south of the gate complex (fig. 10).

32Two symmetrically placed gate openings of 2.43 m each in the northern and the southern connection wall lead to the areas in front of the city wall (fig. 11). The purpose of these may have been to give the defenders easy access to the area behind the proteichisma and the ditch. Together with the main gate they may constitute the three gates that gave the whole gate complex its ancient name, the ‘Tripylon’31. The material and the masonry of the large connecting walls (in so-called ‘maniera romana’) is reminiscent of that of the foundations of the peribolos wall of the Maussolleion. The gateyard was never properly levelled and no trace was found of any ancient paving or road surface. Two carefully made drainage openings are placed in the northern connecting wall (fig. 12). Another drainage opening of exactly the same design was observed on the curtain-wall about 200 m further up the slope. This and other similarities suggest strongly that the whole western course of the city wall from here to the top of the Göktepe hill formed part of one single building phase, either the original Maussollan, or a later restoration phase.

33The two projecting towers are about 10.8 m wide in north-south direction and 9.7 m east-west and thus cover an area of c. 105 m² (figs. 13 and 14). They are solid up to the first floor. A narrow trench was excavated on the west side of the south tower revealing that the wall of the tower rested on a foundation slab at the bottom of a c. 1 m-deep foundation trench, which had been excavated in the living rock of soft white andesite. Unfortunately, no pottery was found in the foundation trench to give an indication of the date of this structure32. The outer walls of the towers are constructed of strong rusticated ashlars of hard brownish andesite. A cleaning of the surface of the towers showed that the tower wall is a double wall, and that the wall constituting the inner face is made of the same soft white andesite that was used for other parts of the Myndos gate complex except the facing of the two large towers.

34In the normal courses no clamps or dowels were used, but at the beginning of the second storey of the tower there is a special ‘string’ course of headers going deeper into the masonry than the ordinary ashlars of the tower. This course – the tenth course visible above the surrounding rock surface – is lower than the normal courses, and each block is fastened to its neighbour by two dove-tail clamps. It thus constitutes a strong ring of masonry on which the upper walls of the tower rest. At the same point the interior of the tower changes. Up to this level it is solid, with a fill of earth and rubble. A very strong ashlar wall constructed of large slabs of white andesite runs through the tower from east to west and forms a continuation of the great connecting wall with a width reduced from 3 to about 1.5 m. Its purpose is not clear. Inner walls are extant in almost all towers in Halikarnassos, but this one is much stronger than normal and it could be thought that it was constructed to support a large catapult. This can hardly be the case though, as on top of this inner wall a few remains are preserved of two crossing walls that form four rooms of approximately equal size in the next storey of the tower (fig. 15). As these cross-walls were apparently built of a slightly different, darker kind of soft andesite it cannot be ruled out that the cross-wall in the second storey belongs to a second building phase.

35The top of the string course, and so also the approximate floor level of the second storey of the south tower, is about 5.6 m above the rock surface west of the tower (the wall, however, continues with two more regular courses and a shallow foundation course c. 1 m further down in the foundation trench). The north-west corner of this tower stood much higher even as late as 196033, when the corner had still 16 courses above the string course. Though the height of each course diminishes towards the top, it can be estimated that the north-west corner, at least until about 1960, reached a height of c. 13 m above the surrounding area and had altogether 28 courses of masonry above the foundation course at the bottom of the foundation trench. During the recent restorations the structure was rebuilt above the first storey for the sake of stability.

36Although the Myndos gate area may seem very flat and level on modern plans, it was not quite so in antiquity. The ground slopes considerably west of the gate area and the towers must have appeared very impressive for anyone approaching from the west on the road from Myndos through the necropolis. In spite of this, the configuration the landscape offers no natural defence comparable to that of the rugged heights on the other sides of the city wall circuit. Therefore the place was strengthened by a large fortification ditch (fig. 16). The ditch begins about 20-30 m south of the gate in front of the first regular tower of the circuit. It then turns west to encompass the large projecting towers of the Myndos gate-complex, which therefore must have existed when the ditch was constructed. In front of the gate the course of the ditch is interrupted by the road leading up to the gate. The road runs on a rise of the natural rock, which had been left for this purpose when the ditch was excavated, so there can have been no bridge here as one might have expected from Arrian’s description of the dramatic events taking place here during Alexander’s siege34. According to Arrian, the Persians suffered great losses by the Tripylon gate when Ptolemaios drove back the Persian sallying party across a narrow bridge that collapsed into the ditch in front of the gate. The losses in front of the actual gate were apparently even greater, because the defenders had closed the gates in panic before the sallying troops were able to get back inside; they were left outside the gate to be killed by the Macedonians. If this story is true, there must have been a provisional wooden bridge somewhere near. It is difficult to get a full understanding of the place because of the modern cemetery immediately north of the gate complex. Did the ditch continue further north? Immediately north of the ancient road it was apparently never finished. Work had commenced, however, and it is tempting to ask whether the reason for its unfinished state might be that it was made in great haste by Memnon, after he learned that Alexander was on his way from Miletos to Halikarnassos35.

37The ditch in front of the Myndos gate is the largest found in Halikarnassos yet, but much smaller than that described by Arrian, which was 30 cubits wide and 15 deep. The dimensions are not so easy to tell, partly because they vary quite a lot and partly because the edges are completely eroded away due to the softness of the rock (fig. 17). West of the southern tower the ditch starts about 14 m out from the tower; it is about 5 m wide at the bottom and 7 m halfway up. The bottom of the ditch is at a very constant level about 4 m below the area in front of the tower but little more than 2 m below the surface west of the ditch.

38It may be concluded that the Myndos gate complex was an extremely well-planned and well-built architectural structure, reminding one of the tetrapyrgion in Theangela in its strict regularity and symmetry. And like the Theangela tetrapyrgion fortress it is an entity of its own, positioned without much consideration for the direction of the city wall where it is attached to the main circuit. Instead, it seems to be oriented according to the main street of Halikarnassos and thus at this place connects the city to its fortifications. It may not seem totally in harmony with other parts of the city wall but the technique and material of the connection walls and the drainage openings are very similar to what is found on the adjoining parts of the city wall south and especially north of the Myndos gate.

39The large ditch is a clear example of the great efforts made in those parts where the natural defences seemed insufficient.

The circuit from the Myndos gate to the Göktepe fortress

40The curtain-wall running up to the top of the Göktepe hill is mostly built of ashlars of soft white andesite and it is therefore in a poor state of preservation. It was furnished with a number of towers of several different kinds of stone. The first tower (8) is inside the modern graveyard. Of this tower no details are visible. The second tower (9) is situated in the very busy square, where the old main road of Bodrum, Turgut Reis Caddesi, meets the great highway running north of the city (fig. 18). The road-builders made a clean cut through the curtainwall immediately south of this tower and so produced an instructive section showing both the foundation cutting as well as the emplekton technique of the wall. The tower here was also built of soft white andesite.

41From here 6 further towers follow up to the top of the Göktepe hill – some of hard grey and blue limestone (10, 11, 12 and 13) and some of soft white andesite (14 and 15). Some are bonded to the city wall, while others are not, e. g. (10) (figs. 19 and 20). The tower (13) in the north-west corner is c. 7 x 8.2 m and commands a large area below to the north and west (fig. 21). The greatest preserved height is c. 2.76 m. It is built of hard limestone, except for the part of the south wall that was covered by the curtain-wall. This part is of more irregular stones of hard, reddish andesite. On the inside of the curtain-wall there are remains of a flight of stairs leading up to the wall walk and the upper storey of the tower. Though these staircases are not always preserved, it seems that they existed in connection with most, or even all, towers. Often the staircases have disappeared entirely but they can sometimes be identified by a line of blocks that constituted the lowermost courses for the inner wall of the staircase. In contrast to the other known towers of the city wall, the two final towers before the Göktepe fortification were solidly built of large ashlars of white andesite of the kind that can be quarried on the spot (fig. 22). The towers were severely damaged by illegal excavations and a number of large ashlars lie dispersed on the ground around them. These destructions make it possible to observe that the first storeys were solid and constructed entirely of large ashlars (fig. 23). During further destructions in early 2000 the building-blocks were removed all the way down to bedrock in the centre of the tower, and it could be observed that the tower was founded on a large, square, bedding in the rock surrounded by an irregular border of unworked rock. Five metres in front of the tower a line in the rock surface could be seen running parallel to the wall of the tower. We excavated a narrow test trench across this line and discovered that a small but well-cut fortification ditch runs in front of the wall; it is 5 m wide and 2 m deep. The course of the ditch can be followed further on as a shallow depression on the ground. The ditch runs in front of the tower and then turns at a right angle towards the curtain-wall and continues in front of this at a regular distance of c. 5 m (fig. 24). It continues towards the summit of the hill and then turns north around the Göktepe fortification, but the mountain side is so steep here that the ditch has no outer side (for reasons explained earlier), but is reduced to a level shelf bordered by a vertical escarpment in front of and parallel to the city wall.

42A little before reaching the summit of the Göktepe hill (c. 160 m) a branch of wall splits off from the curtain-wall and forms the beginning of the south side of what must have been a small fortress. It is not possible to study this because of the number of television and telephone antennae with their small sheds which completely cover the structure; however, a few courses of rusticated masonry with corner drafting (hard brown andesite) can be seen (fig. 25). Newton noticed the foundations for a building inside this fortress, measuring 35 ft east-west and 27 ft north-south (c. 10.7 x 8.3 m) and a large cistern close by. I am convinced that he is right in assuming that this is a ‘small citadel’ with a tower and not the famous Sanctuary of Ares as suggested by both L. Ross and G. E. Bean and J. M. Cook36.

The north east salient and the eastern line of the wall

43From the top of the Göktepe hill the city wall runs down the steep east side of the hill, across a valley, where there must have been at least one small gate, judging from the composition of the foundation courses and from some remains of graves and a tomb inscription found just outside it. The wall is not well-preserved here and the area in front of it has not been carefully investigated; accordingly, it is not known if there was any fortification ditch here. From the bottom of the valley the character and materials of the wall change and from here the city wall is constructed of different kinds of hard limestone for the rest of its entire circuit. The masonry technique changes according to the nature of the stone, which must therefore have been quarried close to the spot where the wall was erected37. The first very long stretch of wall is constructed of irregular polygonal blocks, but further north in the north-east extension of the circuit the rock appears in layers and breaks naturally into blocks with parallel upper and undersides. These are sometimes used in the wall having been dressed only very slightly (fig. 26).

44This part of the wall runs across very rugged and hilly ground, and when following its trace it becomes very evident that it is placed with great skill according to the natural defensive potentials. The largest towers are clearly placed in the highest positions from where they command great areas of the mountain slope below. Towers 29, 25 and 21 are very good examples of this. Tower 16 is constructed in an irregular polygonal masonry style like the curtain-wall at this place, but it is an independent construction, not bonded to the curtain-wall. The interior of the tower is divided by cross-walls into four compartments. The interior walls are very coarse and they must have been hidden in the earth and rubble fill, which no doubt filled up the first storey. There are remains of a staircase on the inside of the curtain-wall. At this particular point, the homogeneity of the masonry and the staircase on the interior support the view that wall and tower are parts of the same building phase. It may therefore be argued that the lack of bond between wall and tower had a well-considered strategic aim as prescribed by Philon38.

45Tower 16 has an unusually commanding position in relation to the valley below and catapults or arrow-shooting machines could cover a large area from here. In spite of the high position of the tower, the rock has been worked to make a steep escarpment at some distance outside the wall. Below the escarpment there is a level area which looks almost like a road running parallel to the wall, no doubt for reasons mentioned earlier. In this area as well as in some places on the eastern part of the circuit, the wall follows an indented course, a feature that has sometimes been seen as an early substitution for towers because it likewise gives possibilities for opening enfilading fire on enemy forces attacking the curtain-wall. This has been seen as especially characteristic for Priene.

46About 170 m further north there are remains of a gate flanked by two towers (fig. 27). Newton had a water-colour made of the plan of this gate39. From the gate a mountain road leads down the slope past a number of looted tombs across the river-bed below and up the next mountain to the north-west in the direction of the ancient Lelegian site of Pedasa. The gate, which we may call the Pedasa gate, was protected by two flanking towers of moderate size, and although the complex is only preserved to the height of a few courses, a doorstep clearly shows that these small towers had open rooms in the first storey, unlike all other towers of the Halikarnassos circuit. The masonry has many surprisingly small blocks and the gate complex thus appears as very different from the monumental Myndos gate complex. In this part of the wall a surprising observation was made. In some places remains of mortar could be seen deep in the joints between the individual stones of the wall (fig. 28). There is no particular reason to think that this part of the wall was repaired and reused in Byzantine or more recent times, and indeed the very small stones applied rather suggest some means for strengthening the construction. However, it would be most extraordinary if the wall had been erected with regular use of mortar in late Classical times, and I suppose that we will have to postpone attempts at explanation until further investigations can be done.

47A little north of the Pedasa gate we excavated what seemed on the surface to be one block of the step of a staircase. It proved to be so, and the staircase was found to be well-preserved below the surface (fig. 29). At its base were found different kinds of domestic pottery and loom weights, suggesting that the area here at the northernmost extension of the city wall circuit was probably not as deserted as we had previously thought40. The small necropolis and dumps of ancient pottery on the slope north of this area suggest more permanent habitation at some point in the Hellenistic period (perhaps a garrison?). Less than 100 m north of the Pedasa gate the city wall ends in a protruding bastion (c. 145 m above sea-level). It is c. 5 m wide and projects c. 5 m from the curtain-wall adjoining the bastion to the east. This bastion must have functioned as a small tower, and it looks as if a staircase led up to the level of the wall walk along the inner side of the curtain-wall from the south. This bastion commands the deep gorge below to the north and west (fig. 30). At the bastion the wall makes a sharp turn eastwards and after about another 70-80 m it makes a 90 degrees turn right and begins the long way southwards. At the actual corner there is a large tower of almost 90 m² on the verge of the ravine, totally inaccessible to any kind of siege machines. The tower is constructed of large ashlars of limestone (fig. 31), different from the bastion just mentioned and from the curtain-wall, which in this entire area is built of somewhat irregular ashlars that were applied almost as they broke free from the layers of the natural rock. But only the outer face of the tower is so strongly and carefully built. The wall is double and the inner face is built of small irregular stones (fig. 32). A stone-robbers’trench in the centre makes it possible to see that two inner walls cross each other and divide the massive fill of the first storey of the tower into four compartments. Here too it is not clear whether the inner walls were made for stabilizing the earth fill or for supporting walls or catapults on a higher level. The size of the tower as well as its magnificent location would probably make it suitable for the positioning of some kind of artillery.

48About 50-60 m further south there are traces of another large tower with interior crosswalls. A closer look at the area west of this tower revealed substantial, but quite unexpected, remains of a wall, a diateichisma, running from this tower westwards to the north tower of the Pedasa gate. Approximately in the middle of this diateichisma there is a small tower built of large ashlars, and it looks as if there was a tangential gate next to the tower. The discovery of the diateichisma shows that this northernmost part of the north-east sector constituted a fortress by itself (fig. 33). A bastion and three towers guard the four corners of the fortress which forms, in fact, a real tetrapyrgon. The extremely strong natural defence of the site and the towers, of which two at least may have been large enough for artillery, makes the whole complex a real stronghold with great strategic importance for the entire city wall. This part of the Halikarnassos fortifications, though larger, has strong similarities with the tetrapyrgon at Theangela, and may have had a tangential gate like it. It also has strong points of resemblance with the fortresses in the Latmos fortification. In particular, the north-easternmost bastion of the Halikarnassos fort may be compared to the projecting bastions in the north fort at Latmos41.

49The east wall of the north-east salient continues southwards in a less well-preserved condition. No towers have been recorded for quite a long stretch, but this does not have to mean that there were none. There is, however, at least one example of indented trace, possibly with a staircase on the inside. About 200-300 m south of the northernmost tower of the northeast sector just described the landscape changes character. The river-bed and the steep ravine leading down to it, runs farther eastwards and away from the course of the wall. The area in front of the wall turns into a gentle slope for some distance, and this is clearly an endangered stretch of wall. When looking down from the large, north-eastern tower towards this area a distinct depression can be seen on the ground running at some distance in front of and parallel to the wall (fig. 34). This surely indicates the position of a fortification ditch. This could be the place where Alexander broke through the wall in 334 BC. It may even be possible to discern the place where the Macedonians filled up the ditch for the siege towers where the depression of the ground is levelled out.

50Another large and strong tower of 7.7 x 9 m is located on a height at 23. It is built in good, regular ashlar masonry of hard limestone. The curtain-wall coming from the north-west is 1.85-1.95 m wide, while the curtain-wall continuing to the south-west could be measured to c. 2.1 m. The curtain-wall runs up against the sides of the tower, but is not in bond with the tower. The surface of the tower is covered with bushes and could not be inspected. The wall then turns right and after little more than 100 m there are some faint traces of a tower (24), on the high point inside the wall, which is unusual. The wall then continues almost entirely in irregular polygonal masonry of quite large stones. After a stretch with one clear example of indented trace, the wall reaches the highway, which cuts right through it (fig. 35). This offers another possibility for studying a section of the wall. But there is very little to be seen here. The foundations are quite superficial and it is not possible to discern any remains of a ditch. There probably was no ditch here but to the south of the highway there clearly is an escarpment parallel to the wall, as can easily be seen from the large freestanding tower on the height to the north-east (fig. 36). On the height just south of the highway there are remains of a remarkable but badly preserved tower (25) – at present the last one to be seen on the eastern part of the circuit (fig. 37). The wall meeting the wall from the north is 2.25 m wide and the width of the foundations close to the tower indicates that there must have been a staircase c. 1.65 m wide along the final part of the curtain-wall against the tower. The dimensions of the tower are c. 7.7 x 7.8 m. Immediately against the south side of the tower there are remains of a postern, only c. 1.7 m wide, flanked by doorposts of green andesite of the Koyunbaba type, a stone not otherwise seen in the fortifications of Halikarnassos. This pattern–tower, staircase, sallyport – is typical of Hellenistic fortifications and has so far only been found this once, so it may be suspected that this complex has been altered in the Hellenistic period. A bit further on there is a very good example of irregular polygonal masonry of large stones, which ends in a little road where one has a fine section of the wall (fig. 38).

51Together with Ali Uçarer from Bodrum Museum I once traced a few remains of the city wall in the private gardens south of here, but it disappears entirely for the rest of its eastern trace. It clearly followed the west slope of the torrent which created such a formidable ravine further north. As it nears the coast the ground becomes more level, but it is not possible to see today if the diminished effect of the river-bed was improved by any kind of artificial outworks. T. Spratt noticed some large blocks in a fig garden where one would expect the wall to have run, and shortly before the wall reached the isthmus it was probably observable by some foundations for a massive ‘Cyclopean’ wall on the shore42. Some houses and the church of Agios Nikolaos were built on part of this wall, which Spratt thought continued from the coast up the rocky hill that presently holds the school. These remains are not visible today. From here the city wall must have continued towards the Zephyrion fortress to which it must have been connected.

The Zephyrion peninsula and the harbour

52The Zephyrion peninsula is a rocky peninsula about 30 m high, connected to the mainland by a sandy isthmus (fig. 39). It is sometimes called an island in the ancient sources, but whether this actually means that it was naturally separated from the mainland at the time of Maussollos, or perhaps that a channel had been dug through the isthmus for strategic purposes, is not clear. The Zephyrion peninsula must have constituted an ideal place for the first settlers at Halikarnassos, easy to defend and with a large and fertile plain on the adjoining mainland. In the Turkish-Danish excavations carried out on the peninsula under the direction of Bodrum Museum in 2002-200443, an almost uninterrupted sequence of pottery was found from the Mycenean period through Late Geometric, Archaic and all subsequent periods until today44. It looks as if in some periods Halikarnassos was reduced to a small settlement on the Zephyrion peninsula.

53At the time of the Hekatomnids the most important sites on the Zephyrion peninsula were probably the old and venerable sanctuary of Apollo on the summit and a newly constructed satrapal palace further down, no doubt with buildings both for residential and representative purposes as well as quarters for military garrisons. In my opinion, there was also a shipshed in the small bay on the south shore of the peninsula with room for one or two royal ships. Extensive traces of beddings for an ancient fortification wall can be followed on the rock surface near to the water-front on most of the south and east sides of the peninsula, except where the rock is covered by earth or the surface has been eroded away (fig. 40). A few traces of bedding for a wall can also be discerned on the west and the north-west part of the rock. The peninsula must have been completely surrounded by a fortification wall, and the complex beddings on the projecting extremities of the rock to the south-east and the south-west probably indicate that the southside was finished at either end by towers. The interior of the peninsula still has very substantial remains of ancient foundations and walls, which probably belong to the palace of Maussollos as some characteristics of technique and materials are strikingly similar to those of the Maussolleion. The crusaders no doubt demolished everything that was to be found of the ancient fortifications and made use of the material for building the castle. Of the huge walls that once rested on the beddings of the rock surrounding the peninsula, I have only seen one small fragment of an ashlar still in situ in its rock-cut bedding. It was of the red-brown, hard andesite found in some parts of the city wall.

54While the siting of the wall surrounding the ‘island’ seems logical in itself, there has to be some special explanation for a stretch of fortification wall running from the south side of the wall-circuit northwards into the interior of the ‘island’. Remains of this wall were first discovered in 1994 when Bodrum Museum and INA carried out a small excavation inside the Crusader chapel, revealing tombs of the Crusader period among the remains of an ancient fortification wall running approximately north-south45. In 2003 we opened a trench on the south side of the chapel and discovered a fine piece of this same wall, still preserved in three courses of which the uppermost had a height of c. 0.56 m and the next about 0.61 m (fig. 41). A drainage channel was cut into the surface of a line of smaller blocks along the east side of the wall. The width of the wall is 1.8-1.9 m. The material is mainly a hard brownish andesite and the masonry is of good ashlar technique. As so little has yet been excavated we cannot be sure about the purpose of this wall, but in my opinion it should most likely be seen in relation to the underwater remains of a slip-way or perhaps a shipshed (the ‘secret harbour of Vitruvius’?) in the little bay between the rocks on the south coast of the Zephyrion peninsula (fig. 42). If there was a shipshed on the waterline here it would leave a dangerous opening in the fortification unless a line of wall turned north, around the shipshed and back to continue the course along the coastline.

55There can be little doubt that the Zephyrion peninsula with the important sanctuary of Apollo and the palace of Maussollos, all surrounded by very strong fortifications, must be the fortress ‘on the island’, where some of the Persians took refuge when Alexander had succeeded in breaking through the city wall far out to the north-east. The other Persians had fled to the Salmakis promontory to the west of the harbour entrance46. Both of these fortresses were very well-defended and they were situated directly on the coast. At this time the Persian fleet was still in control at sea, and the Zephyrion fortress probably had a tiny, well-protected landing for one or two ships, which facilitated contact between the Persian fleet and the defending garrison in the fortress.

56The entrance to the great, natural harbour was probably guarded by moles with walls (see above), but it is not clear if there was an entire land wall along the coast. A very remarkable feature is the large underwater mole running for about 150 m in a north-south direction in the eastern part of the harbour, starting from the modern quay. Its southernmost part is widened eastwards to form a platform. The remains of both this underwater mole and of the moles at the harbour entrance are carefully rendered on the valuable map accompanying T. Spratt’s article of 1856. The large underwater mole is also clearly visible on the aerial photographs of Halikarnassos of 1967 kept in Bodrum Museum (fig. 43)47. The surface of the mole is about 1 m. below present sea level, but as this is estimated to have risen 1-1.5 m. since antiquity, the surface of the structure as it is preserved today would have reached just above sea level in antiquity. The structure has generally been taken to have formed part of the ‘secret harbour’ mentioned by Vitruvius48, but this is a complicated problem because it does not fit Vitruvius’description on essential points, for instance that the secret harbour was situated ‘…sub montibus latens…’. But it seems reasonable to assume that the submerged mole formed part of the military installations of ancient Halikarnassos. I agree with Jeppesen that the mole probably in some way constitutes the west side of the naval harbour and that the naval harbour with its shipsheds and other naval installations should be looked for on the western part of the present isthmus. It may therefore be assumed that the large mole carried a fortification wall that may have ended in a large tower on the platform on its southern extremity.

57When the authorities of Bodrum removed part of the ancient mole to facilitate the manoeuvering of the tourist-boats about two decades ago, an ancient fortification tower of rusticated ashlars was found and demolished, where the mole meets the quay49. In it was found a large number of ballista-balls for catapults. This supports the view that the naval harbour was protected by a fortification wall. Harbour fortifications like this have not to my knowledge been investigated systematically and I wonder if any parallels have been recorded for a remarkable detail in relation to the submerged mole. As can be seen on the aerial photograph, and as can be easily ascertained on the site, a ridge runs in front of and parallel to the sunken mole at a distance of some metres. It reaches approximately the same level as the surface of the mole itself, meaning that the top of this ridge would have reached just up to sea-level in Antiquity. After observing what great troubles the builders of the city wall took in excavating ditches and cutting escarpments all along the wall to prevent enemies from getting their siege machines close to the walls, it struck me that there might be similar considerations behind the construction of this ridge. Is it, so to speak, a ‘reversed ditch’ constructed to prevent enemy ships from coming close to the wall and to lower gangways down on top of the wall as from a siege tower?

The exterior wall

58In 1994 I argued that the outer line of wall as indicated on the maps of Spratt and Newton probably never existed50. I have not had reason to change this view, and I am still inclined to think that only the very large freestanding tower (29), finishing the hypothetical line of the outer fortification to the south, has anything to do with the fortifications of Halikarnassos. Some years ago the Marmara Hotel was built on top of this hill, and the tower can now be seen in its garden (fig. 44). The tower measures c. 12.2 x 15 m and is divided by cross-walls into four compartments of unequal size. Into one of these compartments a deep mortar-lined cistern was built. It seems that the tower was massive up to some level – perhaps to the present surface (the result of a minor surface-excavation that must have taken place in recent times). A gargoyle rests on the surface of the east wall of the tower but is perhaps not in its original place. The tower occupies a very commanding position above the modern Mylasa road. Judging from the location of the ‘tomb of a Karian Princess’ and other graves, it is obvious that the ancient road too must have passed closely below the tower, although it is not quite clear on which side. The size of the tower as well as the cistern and the interior rooms suggest that it may have functioned as a permanently manned watch-tower in critical periods. The rough masonry of large limestone blocks may be characterised as ‘coursed polygonal’. It is not very different from what is found among the great variation of masonry used in the eastern part of the main wall. There is in my opinion no reason to doubt that this tower belongs to the Maussollan fortification. It has a close parallel in the ‘Nördlicher Aussenturm’ (11.7 x 7.3 m) in Latmos, but is not quite as well built. In particular the interior walls of the Halikarnassos tower are constructed of smaller and more irregular stones compared to the large ashlars of the Latmos tower51.

Some conclusions and some questions

59After this survey of the fortifications of Halikarnassos I shall try to summarize the general character of its main constituent parts and to define some significant characteristics that may help to determine the place of this particular fortification in the development of Greek fortifications.

The trace of the circuit: the curtain-wall

60The fortifications of Halikarnassos constitute a very carefully planned city wall of the Geländemauer-type52. The layout of the circuit forms a coherent logical entity, designed in one single process. For historical reasons this can only have been in relation to the refoundation of Halikarnassos by Maussollos. No remains of an earlier line of walls have been observed, nor any indications that the course of the wall was ever altered in later times. The foundations of the wall are only known in a few cases. They are carried down to bedrock on the Zephyrion peninsula, at the Myndos gate and probably elsewhere whenever possible, but the foundation of the curtain-wall sometimes seems surprisingly weak. The great towers of the Myndos gate are founded in a c. 1 m-deep foundation trench cut in the rock. Both masonry techniques and materials vary all along the circuit. The technique of the masonry changes between polygonal, coursed polygonal, irregular ashlar, and regular ashlar or isodomic53. As building materials are found soft white andesite; hard reddish-brown andesite (tufa); and blue and grey limestone of varying texture. As is to be expected, there is a correspondence between masonry technique and building material. Blue and grey limestone was used for all kinds of techniques, but soft and hard andesite only for isodomic ashlar masonry. The style and the material applied over the 7.5 km course of wall does not appear to signify successive building phases or repairs, but is rather a result of what was at hand near the construction site54. The only exception to this principle of convenience seems to be that the outer face of most of the towers is constructed of isodomic ashlars of hard stone. The curtain-wall is mostly only c. 1.8-2.2 m wide except for the part of the western wall that is built of large ashlars of soft white andesite. It runs from a little south of the Myndos gate to at least the summit of the Göktepe and has been measured at several places to 2.5-2.6 m in width.

61In general, the curtain-wall of the fortifications of Halikarnassos does not appear to be very strong and in some places it is constructed of surprisingly small stones (in some places with remains of mortar of unknown date). The relative weakness and the lack of uniformity could be seen as indicators of its belonging early in the evolution of fourth-century fortifications. The curtain-wall is not yet constructed to resist the stone-throwing catapults of the later fourth century. The possible shortcomings of the curtain-wall were remedied, however, by several other very efficient means:

• the approach was very difficult, either due to natural configuration of the site or to artificially made ditches and escarpments;
• a number of strong interior fortresses were integrated into the circuit, placed at strategically important heights;
• a number of strategically placed towers were included in the circuit.

Ditches and escarpments

62Ditches and escarpments have been noticed in many places, except where the natural slope is very difficult to climb and impossible for siege machines to pass. They are found both on level ground as in front of the Myndos gate, and on the highest parts of the circuit on the Göktepe hill and in the north-east salient, at places where the ground in front of the wall was accessible to siege machines. They vary in size and appearance. The date of these precautionary outworks is questionable. One could see them as a deliberate result of the quarrying for building stones for the fortification wall. This would mean that they are contemporary with the original Maussollan building phase. On the other hand, both Arrian and Diodoros relate that Memnon, who was in charge of the Persian forces on Alexander’s attack in 334 BC, made preparations to strengthen the defence of the city in places where the strong natural defence was not enough55, and Arrian expressly mentions that the Macedonians filled up a ditch c. 30 cubits wide and 15 cubits deep, which the defenders had excavated in front of the wall56. The apparently unfinished state of the ditch north of the road leading to the Myndos gate would support the view that the ditch was made under time pressure as Alexander approached from Miletos. But the question remains open.

Interior fortresses and strongholds

63The existence of a number of strong fortresses in the circuit was essential for the entire concept of the defence of Halikarnassos and it is interesting that these are expressly mentioned by Diodorus57. In addition to bastions in the south-western part of the circuit of which we know next to nothing, the fortification contained four complete fortresses: the Salmakis and Göktepe fortresses, the fortress in the north-east salient and the Zephyrion fortress with the satrapal palace. Two of these are situated on the coast and thus could get supplies and men from the sea as long as the Persian fleet was in control at sea. The strategic importance of the interior fortresses is evident and the system worked brilliantly during Alexander’s siege. When the defenders had to give up the defence of the large circuit, some sailed away on the Persian fleet, while others withdrew to the fortresses with provisions to sustain a prolonged siege. Before withdrawing to the safety of the fortresses the defenders tried to burn down the city during the night so as to leave neither weaponry nor provisions to the Macedonians. Their strategy seems very well-considered, and if the enemy had been a less powerful adversary, the Persians would probably have been able to hold the interior fortresses until an army of neighbouring satraps or of the Great King could come to the rescue. These, however, had very serious problems of their own and Alexander therefore could move on while the fortresses were still in the hands of the Persians. But according to Arrian, Alexander had to leave 3000 mercenaries and 200 horsemen in Halikarnassos under the command of Ptolemaios to continue the siege58.

Towers

64The number of towers in the fortification can only be estimated very approximately. McNicoll saw 18 towers in 1968. It is now possible to identify 23 safely and add at least three59. The original number must have been closer to 35 or 40. The size of the towers varies between c. 44 m² in the tower south of the Myndos gate to c. 105 m² for each of the two large towers of the gate itself. Large towers of c. 60, 70 and 90 m² were placed on commanding high positions of strategic importance. Although the towers vary they have general characteristics in common. Only the two small flanking towers of the Pedasa gate appear to have had accessible rooms at ground level. In all other cases the ground floor of the towers is solid with a fill of earth and field stones. Often, or perhaps always, there are walls in the fill forming two or four compartments. In some cases the interior of the towers has been partly emptied by irregular excavators, and it may be seen that the interior walls are of rather poor quality. The towers have an inner and outer wall with a fill of earth and small stones. The inner face of these double walls is sometimes of the same simple dry stone technique as the interior cross-walls and generally of a poorer quality than the outer wall-face. The purpose of the interior cross-walls in the fill of the tower is not clear. As they are not strong, they were hardly made to keep the fill of the tower in place. At this early time it also seems unlikely that they were constructed to support catapults, as these were still only very small arrow-firing machines60. Most likely they were built as foundations for the walls on the next level of the tower, which may have had open rooms. The two towers of the Myndos gate are basically similar to the other towers, but much stronger and built entirely of ashlars. But even here the outer walls were constructed of a stone of much better quality than that of the inner walls. As stone-throwing artillery was still mainly ‘antipersonnel’ at this early stage, I am inclined to think that the solid and strong first floors of the towers were mainly constructed to withstand the blows of large rams. In the Hellenistic period the first floor of the towers was normally hollow and could be used for storing weapons or to give shelter to the guards. This suggests, perhaps, that it took quite some time before the torsion catapults were developed to such a stage that they constituted a real danger to the fortifications.

65At Halikarnassos some towers are bonded to the wall, others are not. For a relatively narrow curtain-wall as that at Halikarnassos the towers could have had a positive function as buttresses, when the two structures were bonded together. But, as pointed out by Philon, a firm bond between the curtain-wall and the towers also has the disadvantage that if the besieger managed to overturn a tower this was likely to take the curtain-wall down with it when it collapsed and vice versa. Therefore towers and curtain-wall should not be bonded together. There is no obvious reason why both systems are found in Halikarnassos. There may be a chronological explanation: the towers that are not bonded to the curtain-wall are perhaps later additions. However, as these towers too seem to be necessary for the entire fortification, they most probably all belong to the original, Hekatomnid city wall. A very particular technical feature of the Halikarnassos towers is the peculiar ‘double bond’ at the corners (fig. 45). It appears at almost all corners made of regular ashlar masonry, and it is sometimes very systematic and at other times more irregular. Basically, the system is as follows: at one side of the tower every second course of ashlars is finished at each end with two narrow headers side by side. The outermost header of course continues as a stretcher on the other side of the corner, but the one next to it is often a very narrow header not going very deep into the masonry. When one looks at the adjoining side of the corner, it will be observed that the system is also applied here but in alternating rhythm. As the walls of the towers are two-faced, it might be thought that the second header in the corner bond belongs to the inner face of the adjoining wall. But this is not the case.

66At a number of places fragments of roof-tiles were observed on the surface next to the towers. Though these might in some cases be of more recent date, it seems likely that most of them belong to an original ancient roofing of the towers. This was necessary to protect whatever kind of catapults or arrow-shooting machines were stationed in the towers as well as to protect the defenders against the arrows of the attacking troops.

67In addition to the bastions and towers the wall had a few indentations or ‘jogs’, which likewise offered the defenders a possibility to enfilade along the curtain-wall.

68The walls, towers, bastions, ditches etc. of the Halikarnassos fortifications show many variations. Are these due to alterations and repairs over the centuries? Or are they a sign of an early date, when fortification-building on such a large scale had not yet reached a very high level of uniformity and systematization? In the case of the Pedasa gate it may be argued that the small towers with their open ground floors represent a later stage and should be dated to the Hellenistic period – not least, perhaps, because the masonry seems to contain some mortar.

69The monumental towers of the Myndos gate differ from the rest of the fortification – but only in quality and size; in plan and strategical concept they do not seem to differ from the rest. So perhaps the monumental appearance of the Myndos gate is due partly to its dangerously exposed location and partly to its role as representational architecture opening directly on to the monumental main street of Halikarnassos. The ditch in front of the Myndos gate was clearly made at a time when the gate complex was already there, and it also seems quite certain that the ditch was there when Alexander attacked the city. But whereas the ditch may be conjectured to have been hastily established by Memnon, the towers and the gate complex certainly do not look like something quickly built up to meet an immediate danger. Therefore I think that at present the evidence points to a Hekatomnid, and presumably Maussollan date for the Myndos gate and its huge towers as well as for the great majority of the rest of the towers of the Halikarnassos fortification.

The gates

70Five gates are presently known to me, and except perhaps for the small gate next to tower 25, they are all placed with a view to serve the lines of communication between Halikarnassos and the neighbouring towns. The Mylasa gate was probably the main gate of Halikarnassos. Outside it was a large necropolis, which was partly excavated by Newton. This gate has completely disappeared, but I did notice a strong ashlar foundation in the bottom of the dry river-bed in the 1970s, which must belong to an ancient bridge leading across the stream directly to the Mylasa gate. The river-bed itself has now been converted into an asphalted street for a very long stretch, covering the ancient foundation. But the old Turkish road leading diagonally from the north-east down past the east side of the bus station and on to the harbour still makes a turn across the old river-bed 30-40 m south of the foundation.

71The second main gate was the well-preserved Myndos gate or Tripylon with the three openings leading north, south and east from the gateyard. This gate follows the orientation of the Hecatomnid town plan and can thus be said to link the city architecturally to its fortifications. There was also a large necropolis in front of the Myndos gate and a well-constructed tomb chamber can be seen on a little height (‘Esentepe’) west of the gate. From the very top of the north-east salient the small Pedasa gate with its flanking towers gave access to the mountain road leading north towards the important Lelegian town of Pedasa a few kilometres to the north. At this gate too there was a small necropolis of which the looted tombs and their big coverstones remain.

72In the bottom of the valley between the Göktepe hill and the north-east salient there are remains of the fortification wall in which the ashlars of the masonry show that a small gate must have opened here, where a path still leads through. The road leading through here may have joined the Pedasa road a little further up on the hill. At least one tomb inscription found here suggests that there were tombs outside this gate too. West of it there are many tombs also inside the city wall on the slope of the Göktepe. There are likely to have been at least two more small gates – one facing west in the saddle just north of the Salmakis promontory towards the bay of modern Gümbet. Another faced east towards the large fertile coastal plain and the bay which may have constituted the ‘Emporion’ of Halikarnassos as suggested by Jeppesen61. The road leading east from this conjectural gate is likely to have passed across the bay and continued eastwards along the coast towards the plain below Theangela62.

73Although Halikarnassos may have had seven gates or a few more, it seems in our present state of knowledge that these were all placed to serve the roads leading to the most important neighbouring towns. There may have been a few sallyports or posterns in addition to the one (secondary?) noted at h by tower 25, but generally the concept is definitely that of a ‘passive’ fortification. In this respect the wall was already outdated when Halikarnassos was besieged by Alexander. It was clearly a serious problem for the defenders that whenever they had made a sally to set Alexander’s siege machines on fire or to make a surprise attack, they had great difficulties in getting back inside the gates before the Macedonians cut off their retreat. The lack of strategically well-placed posterns in some cases had disastrous consequences for the defenders.

74The study of the gates of the fortifications of Halikarnassos supports the impression that, in spite of its many well-considered elements, the city wall was not designed to meet the new siege techniques and strategies of the time of Alexander.

Double corner bond – a hallmark of Asia Minor fortifications of the fourth and third centuries BC?

75I. Pimouguet-Pédarros has tried to define a particular Hekatomnid masonry style in regular ashlar masonry63. Among the criteria are that the walls are of emplecton type, they are of header-and-stretcher technique, the block face is either quarry-faced or hammer-dressed, the horizontal joints are precise and the vertical are either right-angled or slightly oblique, and finally the corners have a vertically drafted margin. These criteria alone are hardly sufficient to distinguish Hekatomnid masonry from other fourth-century fortification masonry, but if we add to these criteria that of the particular double corner bond mentioned already in the description of the towers, it seems that we have good possibilities for identifying the masonry made by the Hekatomnid architects and their successors in Asia Minor.

76The system can be safely dated to the Hekatomnid period, as it was applied at the corners of two terraces at Labraunda and at least one in Amyzon64. I am not aware of any older examples and I have never observed any examples on the Greek mainland or in the west. The double corner bond appears to be another typical creation of the Ionian Renaissance, invented by the architects of Maussollos and continuing in use in Western Asia Minor until well into the third century.65 The system can be observed in a great number of towers and fortresses in addition to the city walls of Halikarnassos, Theangela, Myndos, Latmos, Alinda, Alabanda, Kaunos, Knidos, Herakleia, Ephesos, Samos and in a few places at Priene and Pergamon66.

77L. Karlsson has suggested that the header-and-stretcher masonry of the Hekatomnids was introduced to Maussollos by masons coming from the Peloponnese or Sicily67. It is certainly possible that there was a direct influence through masons from the Greek mainland as there was no doubt a lack of experienced local stonemasons in Western Asia Minor when the new intense building activity began at the time of Maussollos, and the influence of mainland workmen and architects can be seen in other fields of the Ionian Renaissance. But I agree with Pimouguet-Pédarros that the origin of the Hekatomnid fortification masonry was probably more complicated68. In addition to a possible mainland influence, there were Archaic fortifications still standing and in use in many places. For instance, the Archaic city wall of Miletos – a large Geländemauer built of ashlars with hammer-dressed face – may still have played a role when Alexander laid siege to Miletos in 334 BC. And in addition to the old Ionian fortifications there was a rich Karian or Lelegian tradition of building fine walls without mortar. Parts of the city wall of Halikarnassos, especially in the north-east salient, have good parallels at Alazeytin and Pedasa.

78So the fortification masonry of the Hekatomnids could be a creation based on several sources, and perhaps it began quite early in the century. It is remarkable how highly developed and varied the masonry of the Maussolleion was. In the peribolos alone, five different kinds of stone were applied for different purposes according to their structural and aesthetic value: header-and-stretcher technique was applied in the terrace wall; masonry of the ‘maniera romana’ was used in the foundations for the peribolos wall; double-faced ashlar masonry was used for the peribolos wall itself, with diatonoi (headers going through the entire wall and appearing on both the inner and the outer wall face); rusticated ashlars were used in the euthynteria of the peribolos wall, in the southern terrace wall and in the low terrace surrounding the actual tomb building. This and the sophisticated architecture of the Maussolleion itself points to a high technical level, which would hardly be dependent on influence from the west for the creation of simple, rusticated fortification masonry.

Other Hekatomnid city walls

79There are many fortifications in Asia Minor that have the characteristic masonry of the Hekatomnids with double corner bond. But as this – like other special features of the Ionian Renaissance – seems to have been in use from the time of Maussollos until well into the third century, other supplementary criteria are needed to distinguish which of them may belong to the Hekatomnid period. Very long circuits, modest thickness of curtain-wall, interior fortresses, few gates, towers with solid first floor, interior cross-walls and double corner bond are all features that are found particularly well represented in Theangela and Latmos, which should therefore be counted as Hekatomnid fortifications. For historical and strategic reasons Myndos must belong together with Halikarnassos and Theangela in the first group. The extremely varied masonry of the Halikarnassos circuit should probably be taken as an indication that this is the first of the group, although the variety is no doubt also due to the changing character of the stone quarried on the construction site. The more regular character of the masonry at Myndos can thus be explained by a slightly later date and by the availability of green andesite from the Koyunbaba quarries, just as the gneiss and granite found in abundance at Latmos can explain the regularity of the ashlar masonry here. Alinda will have been fortified already when Ada lived there in the 340s, but the upper and the huge lower circuit are different: only the small enceinte on top of the hill has double corner bond, and therefore I would suggest that only this is of Hekatomnid date, while the great circuit with many well-preserved towers is more likely to be of high Hellenistic date. A number of other city fortifications with double corner bond and very large circuit – like Alabanda, Knidos and Kaunos – may be of Hekatomnid date, but this seems unclear at present.

80Priene constitutes an interesting case of its own. The close architectural connections to Halikarnassos and the Ionian Renaissance are witnessed both in the written sources and in architectural remains. Priene must have been in the sphere of Hekatomnid influence and the overlife-sized head of a Hekatomnid woman (Ada?) found in the temple of Athena may be compared to those Hekatomnid ruler statues that were set up by many other cities such as Latmos, Kaunos, Miletos and Erythrai, all of which had reason to pay homage to the Halikarnassian satraps. Nevertheless, some scholars have claimed that the refoundation of Priene did not happen until after the advent of Alexander69. The city wall of Priene has not been published in detail yet, but a comprehensive study has been made by U. Ruppe, with whom I have discussed the relation between the city walls at Halikarnassos and Priene70. According to Ruppe, there are several cases of general similarity in the less well-built part of the wall and I believe that the double corner bond can be seen in a few places at Priene, although less frequently than at Halikarnassos. The relative weakness of the city wall of Priene and the lack of posterns are some of the indicators that the wall can hardly be much later than the Halikarnassos wall. Although it is less typically Hekatomnid than Latmos, Alinda, Myndos and Theangela I would still see it as belonging to the periphery of the Hekatomnid fortifications rather than as Athenian or mainland Greek. But a more precise decision will have to await the publication by Ruppe.

81Among the followers of the late Classical fortifications in the Asia Minor tradition, Samos constitutes an interesting example. The fine ashlar wall of Period II has good examples of double corner bond, clearly indicating that it was built in the Hekatomnid tradition, even though it should be dated to 310-290 BC according to H. Kienast71. There are other similarities between the city wall of Samos and the fortifications of Halikarnassos72. For instance, it is remarkable that the ditch in front of the western part of the city wall of Samos seems to be very similar to that found and excavated west of the summit of the Göktepe hill at Halikarnassos–but at Samos it is assigned to the time of Polykrates73.

A Hekatomnid fortification system for the Karian satrapy

82A number of cities in the Karian satrapy were fortified at the time of the Hekatomnids. It seems that Halikarnassos was among the very first, at a time when some of the strategical principles and fortification techniques were still being developed. At the same time, or very soon after, city fortifications were probably erected in Myndos, Theangela, Alinda and Latmos. Circumstantial evidence suggests strongly that Kaunos, Iasos and Alabanda were part of the same Hekatomnid defence system for Karia, although the architectural evidence is perhaps not so conclusive at present or needs further study. As pointed out by McNicoll, Pimouguet-Pédarros and others including myself, the system of city fortifications in Karia seems to have been supplemented by a number of fortresses and watch-towers with intervisibility, which together constituted a strong fortification system for the entire satrapy74. In some respects the fortification techniques and strategies developed under the Hekatomnids continued after Alexander and it is not clear at present if Knidos, Kaunos and Priene are to be dated to the time of the Hekatomnids. But siege techniques changed quickly and the development of artillery in both siege and defence of cities necessitated great changes. Although the particular double corner bond in the towers seems to continue as a special hallmark of the Asia Minor fortification masonry, the circuits changed in other respects during the Hellenistic period.