1The Rhodian fleet became an important force in the early Hellenistic period. Explicit figures are only available for the late third century or later. For earlier times we can only guess. Clearly the three cities of Rhodes had ships before the synoikismos of 408 BC (with ports of varied quality), but it is likely that development was limited until the fleet had its main base in the great natural harbour of Rhodes itself (fig. 1). Dio Chrysostom in his Rhodian Oration (31.103), delivered under Titus, refers back to a Golden Age, when Rhodes ‘used to send out a fleet of 100 ships or more, and again of 70 and then of 30, and sometimes not disband them for three or four years,… whereas now they appear with merely one or two undecked ships every year at Corinth’. This must refer back to the fourth/third centuries, I think (and the figure of 100 is a maximum); so too must the graphic description of the ports and the dockyards of Rhodes in the Rhodian Oration ascribed to Aelius Aristides (25.3-4): ὅτε τῆς θαλάττης ἐκρατεῖτε... He describes how some of the harbours received those sailing from Ionia, others those sailing from Karia, and still others those coming from Egypt, Cyprus and Phoenicia, as if each harbour was constructed to receive visitors from a specific place; and lining (epikeimena) these harbours were the neoria, the dockyards. (I omit discussion of the problems raised by his description of the neoria, except to emphasize that the speaker specifically links the neoria to the Rhodian thalassocracy.)

2These rhetorical passages are not precisely datable. The tradition of the Rhodian thalassocracy is clear: Rhodian trade interests required them to maintain a fleet and combat piracy. The fleet size is usually estimated to have been 40 (Casson) or 50 (van Gelder) major units in the Hellenistic period; I previously suggested 40-50 major units and a lower number of smaller ships, but I would now, like Gabrielsen, put the figure somewhat higher. After 227 BC, when gifts from Ptolemy and Seleukos included the replacement of 10 triremes and 20 ‘fives’ lost in the earthquake, we start to get precise figures, with squadrons of between 9 and 36 ships (the basic unit of three must reflect the triple character of the Rhodian state, as Gabrielsen has suggested)1.

3Berthold has added up the various figures for 190 BC, and concluded that in that year Rhodes put 75 ships to sea, but never more than 55 at a time; this figure (55) may well indicate the maximum number of ships that Rhodes could man with citizen crews2. Berthold argues for a long, slow decline of the fleet after 167. What is clear, ever since their resistance to the (much larger) fleet of Demetrios Poliorketes in 305, is the great seamanship skills of the Rhodians, who (as has been said) relied on citizen crews.

4The military harbour was Mandraki, whose shoreline might just have managed to accommodate 100 ships, but it is not clear that such a large number of ships was ever based at Rhodes itself. I will return to this point.

5Some traces of shipsheds have been found on the west side of Mandraki harbour, but the main surviving evidence comes from the south side. Under the monumental Roman tetrapylon at the north end of a main street (P31) remains of shipsheds were excavated by an Italian team during the Second World War (1940-41), but the records were lost; the remains were re-studied and published by Paul Knoblauch, Angeliki Yiannikouri and myself3. We were able to identify the dividing wall between sets of shipsheds of different clear widths: 6-6.30 m housing the larger units – kataphrakta (triremes to pentereis); 4.20-4.40 m housing the smaller units – the aphrakta (which would include trihemioliai and phylakides, the latter not a distinct ship type but a generic term).

6We only have a narrow strip (some 10 m) of the back part of six shipsheds, with two phases clearly defined: before and after the great earthquake of 227 BC (figs. 2, 3). A clear fourth-century phase was not found. The second-century phase did not change the dimensions of the third-century phase of shipsheds. What we have here is evidence for the standard ‘battleships’ and the phylakides – the ships used for policing the seas, notably against pirates.

7With this experience I was invited by the 22nd Ephorate of the Dodecanese, at the suggestion of Dr Sampson, to look at remains which he had found in 1980 when excavating a Neolithic site on the island of Alimnia (ancient ‘Eulimna’) which lies just north of Chalke off the west coast of Rhodes (fig. 4)4. Alimnia must have belonged to Chalke, which was a ktoina of the deme of Kamiros. It has the best natural harbour in the area (and was an important Italian naval base until 1943).

8Two groups of rock-cut slipways were found, ten on the south-east side of the main harbour of the Agios Georgios and eleven on the south side of the Bay of Emporeio, to the east (figs. 5, 6). I investigated these in 1991-92, and in 1997 collaborated with the Ephorate of Underwater Antiquities in underwater investigation of Emporeio Bay; limited excavation of one of the shipsheds had to be suspended after storms, and unfortunately we have not been able yet to finish this project. Agios Georgios Bay is still to be studied. The slipways seem very short (length 16 to 20 m), and we wonder whether their lower ends have been eroded. Unlike Rhodes we have at Alimnia no clear evidence of roofing: the slipways may have had only light roofs – some sort of covering would have been desirable – but strong enough to resist the wind. The slips are cut in the rock at irregular intervals; most have a width ofc. 8-10 m (Emporeio) or 9-11 m (Agios Georgios); one in each is much wider (13 m in Agios Georgios, 18.5 m in Emporeio). They could each have housed two small ships (in one case perhaps three). The slips do have a gradient, though this has become difficult to define because of weathering.

9Dating is uncertain: we have not yet done enough excavation. There is clear evidence on the shore and in the water of activity in the Early Christian Period. But there are clearly Hellenistic walls in the vicinity of Emporeio Bay, and above lies the Hellenistic (and later) fort of Kastro. I have suggested that as the fort has a panoramic view of the west coast of Rhodes, it must have served as a lookout station, from which the ships below could be alerted, and also signals could be sent to towers on the mainland coast and up to the city of Rhodes.

10We therefore have here an outpost of the Rhodian fleet, guarding the south-western approaches from Crete to Rhodes and also the western approaches from the Aegean (the bay later saw non-military use in the early Christian and Byzantine periods). The ships would have been phylakides. In suggesting some sort of early warning network of naval bases for the Rhodian fleet, I was hoping that we could find further evidence to support the hypothesis – and I noted with interest that Gabrielsen was soon emphasizing the importance for the Rhodian fleet of a ‘number of strategically situated anchorages and ports, at home and abroad, as permanent bases from which detachments of their fleet operated’, in addition to the fleet headquarters in Rhodes5. In the mid-1990s he could only quote Sampson and myself on Alimnia, but now we have more evidence.

11The first piece of new evidence, from the mainland of Karia, is an inscription found in 1992 in Yeşilyurt (formerly Pisiköy), north of the Keramic Gulf, at the site of ancient Pisye. It was first published by Ender Varinlioğlu6 (the Turkish member of the Franco-Turkish research team carrying out the survey of south-east Karia), and fully published in Les Hautes Terres de Carie edited by P. Debord and E. Varinlioğlu7. This is an excellent publication; I am in full agreement, and want mainly to follow up some of the arguments. The proposed dating seems to me plausible: second or third quarter of the third century; our French colleagues are greater experts than I on the Hellenistic inscriptions of the region (and on the letter forms). The preamble runs as follows:

[ο]ἵòδòεò [---------------------------------------]
ἐπαγγειλάμενοιò τ[ῶ]ι δòήòμòωιò [------------------- εἰς τὴν]
[τῶ]νò νεωρίων κατασκευὴν Δ[------καὶ βουλόμενοι φανερὰν]
ποιεῖν ἣν ἔχοòυòσιν αἵρεσιν εἰς..[---------------καὶ τὸ πλῆ]-
θòος τὸ Πισυητῶν καὶ Πλαδασέ[ων τῶν μετὰ Πισυητῶν ἔδωκαν]
χòρήματα δωρεάν vacat
[Φ] ίλαγρος Στρατοκλέους Πιστια(νός) ρ’Ι [----------------------]

12‘The following [….] who have promised to the demos [….] (for the) construction of neoria [….] and who [want to make clear] their (favourable) attitude towards [….] and towards the population of the Pisyetai and the Pladaseis [united with the Pisyetai, have made] the following gifts of money: [Ph]ilagros, son of Stratokles, Pistia(nos) 100 drachmas’ (etc.)…. Even if the opening lines are not easy to read or restore, the main point is clear: this is the preamble of a public subscription (gifts, not interest-free loans) for the construction of neoria; the list of names continues on at least two columns on Face A and on Face B, and on the left side. The stone is broken on the right and below; the editors note correctly that the text begins with the largest contributions. By the last lines of column (d), the second column on Face B, the sums are smaller, and we may be nearing the end of the text, in my view.

13The editors suggest that the largest contributions will have come from the communities most interested in the construction of the neoria; they discuss at length the communities involved, which seem to lie between Pisye and the coast, near which lived the Pladaseis. The editors calculate that the first column (Aa) preserved contributions of 2,000 dr., in total, and assume that the second column (Ab) – where the figures are lost – contained contributions at a similar level: say + 2,000 dr. total. Face B preserves the numerals in column (c) and all but the numerals in column (d). Here the numerals are all much smaller (total 1,200-1,350 dr.), and the editors calculate that the parts preserved contained a total of just under a talent. The preserved height of the stele is 0.60 m, and the editors conclude that if we assume an original height of 1 m-1.50 m, the total epidosis may have been c. 2 talents.

14The editors find that this text is in the normal range of Hellenistic public subscriptions, with individual contributions of 100 dr., appropriate for the well-to-do. We are not here dealing with anything like the massive grants made to Rhodes by the great Hellenistic powers after the earthquake there in c. 227 BC, which badly damaged the Rhodian dockyard, inter alia8. Could we at Pisye be dealing with the after-effects of that earthquake? I think not, for two reasons: first, the inscription says κατασκευήν and not ἐπισκευήν: ‘construction’ and not ‘repair’, although that argument is probably not decisive; secondly, the lettering seems to favour an earlier date.

15The decision to (appeal for or) collect money must have been made by the Pisyetai and Pladaseis. Were they acting as a demos or a koinon? The koinon of the two ethnics is well attested in later inscriptions, though not all the Pladaseis joined it. What are the words missing in line 2, and in line 4? The editors plausibly assume that the preamble stretched across two columns only, which means that the missing space will take 23 letters (omegas) to 27 letters (iotas). It is argued that there is not space for the full title of the Pisyetai and Pladaseis in line 2 (after δòήòμòωιò), and they are not referred to in the missing part of line 4, which clearly refers to another authority for which the neoria were to be built.

16Which is the authority to which the contributors also wished to show their αἵρεσιν (favourable attitude) – in addition to the Pisyetai and Pladaseis? The answer to this question involves also the question of the date of the inscription: the alternative candidates are Rhodes (already in control by the mid-third century?) or the Lagids (or possibly the Seleukids who briefly had interests in the region c. 260)9. The Rhodians do not appear in the surviving text of the preamble nor as donors. Line 4 could, the editors admit, be easily restored as εἰς τòὸò[ν δῆμον τὸν τῶν ῾Ροδίων καὶ τὸ πλῆ]|θòος (a line of 26 letters with 3 iotas and 2 omegas), and Riet van Bremen had suggested to restore a reference to the Rhodians already in line 2 τ[ῶ]ι δòήòμòωιò [τῶν ῾Ροδίων] or [τῶι τῶν Ῥοδίων]. This seemed to the editors logical, though it involves accepting that the promises of contributions were made to a body other than the koinon of Pisyetai and Pladaseis; and this involves accepting that at the time of the subscription (somewhere in the mid-third century), the Pisyetai and Pladaseis had come under Rhodian control and had lost their autonomy (perhaps at this stage becoming a simple koinon).

17The alternative is to see a ‘royal’ presence in the area, in the mid-third century, exerting pressure on their Karian subjects to help their naval operations from a small base in the Keramic Gulf. In this case, the editors suggest restorations of line 2 to refer to βασιλέα Πτολεμαῖον... or a Seleukid monarch10. In this case, the ‘État’ of the Pisyetai and Pladaseis did not come under Rhodian control till the last part of the third century.

18The editors come down in favour of Rhodian control by the mid-third century. They quote Livy (33.18.1) describing the Rhodians in c. 200 as having occupied the Peraia ‘maioribus suis’ (which means for some decades at least) and note that Rhodian possession of the nearer part of the Peraia – the Chersonese – goes back to the last quarter of the fourth century at least, and control of the Chersonese in reality to the end of the fifth century. I shall come back to a site in the Chersonese, but first let me say that I am convinced by the argument for Rhodian control of these sites in the Keramic Gulf in the mid-third century BC, with a naval base whose location has been plausibly proposed by the editors. We are not here in inland Karia.

19The neoria must have been built on the bay of Akbük at the one breach in the coastal cliffs of the north side of the Keramic Gulf, below the fortified site of Sarnıç which seems to have been the main centre of the Pladaseis. The peninsula, 2 km long and 1 km wide, offered two anchorages, the better on the north-east side (which probably then extended farther inland into what is now marshland). There are ancient remains on the peninsula, which is heavily overgrown: a rectangular tower and some submerged walls. The suggested location for the neoria seems plausible11. The site is small, but the bay is large, and the important point was a suitably gentle shoreline. The neoria were clearly modest and so were the contributions – we are not to look for parallels at Piraeus, or even Rhodes, but rather Alimnia. So we can ignore the figures for the construction of the Piraeus dockyard (Isokrates claimed that the shipsheds cost 1000 talents)12, which was for c. 300 ships in the fifth century and 372 in the later fourth century.

20In any case, the contributions listed in the inscription may have covered only part of the cost: the editors quote the evidence from Migeotte and suggest that the neoria at Akbük may have cost 6-10 talents. They take the high-cost figures for Piraeus (c. three talents per slipway) and think of a dockyard with 2-4 slipways, but in view of the number of slipways at Alimnia they wonder if there may have been more. I think that this may well be so, and that the slipways may have cost less per unit. We do, however, also have evidence for small naval stations, for example the Athenian bases at Atalante, Boudoron and Sounion itself13; also, at Naxos in Sicily we are defining a dockyard with four trireme shipsheds, for a Greek polis (admittedly fairly small)14. We know nothing about the later history of the port in the Rhodian period. The citadel at Sarnıç had a Rhodian garrison, so the bay continued to have military (as well as, no doubt, commercial) importance, with the neoria continuing to serve as a base for the Rhodian fleet.

21The second piece of new evidence from the mainland of Karia is not from the recesses of the Keramic Gulf, but from the nearest point to the island of Rhodes: the small city of Loryma at the southern tip of the Karian Chersonese, the heart of the Rhodian Peraia; now Bozukkale. Its strategic importance for Rhodes is clear: it lies across the strait from Rhodes, and on a main shipping route.

22We hear of an attack on the city by an Athenian squadron in 413/12; a large Persian fleet (90 ships) was based there in 395; Demetrios Poliorketes gathered all his fleet there (370 ships) before his attack on Rhodes in 305; and Cassius did the same in 43 BC15. It is not certain when Rhodes established its control over Loryma: Blümel argues for the mid-fourth century; Held for the end of the fourth century16. The fortress denoted Loryma III was probably built by the Rhodians to defend their naval base, and the epistatas, now epigraphically attested, presumably commanded the fort and the naval base17.

23It seems to me most likely that the fortress was built by the Rhodians as an outer defence after the siege by Demetrios Poliorketes, as part of the Rhodian reaction to that bitter experience – just as recent excavations in the city of Rhodes have shown that a wall was added on the ‘Mole of the Windmills’ to ‘plug a gap’ which had been exploited by Demetrios who landed 400 troops on ‘the unfortified breakwater’ 18. I. Pimouguet-Pédarros argues that the Loryma III fort was ‘certainly built by the Rhodians to defend Bozuk Bay’ between the late fourth century and the early first century (!); and the acquisition of the subject Peraia gave them the funds and the motivation to defend their naval base. She emphasizes that the Rhodian fleet became important in the third century, but does not exclude the second century, when Rhodes had many overseas naval bases19.

24Certainly in the second century Rhodes tried to impose its maritime supremacy in the south-eastern Aegean, and had by then a large number of bases. Pimouguet-Pédarros cites Samos, Halikarnassos, Myndos and Kaunos, well beyond Rhodian territory; and the revived Nesiotic League under Rhodian hegemony had a federal fleet base at Tenos. But that is another era. I think that the acquisition and fortification of a key naval base so close to their city and main harbours would have been a priority for the Rhodians from the end of the fourth century, after they had successfully repelled the siege of Demetrios in 305. This is the date plausibly argued for by Held in publishing the new epistatas inscriptions from Loryma: the first has letter forms of the first half of the third century20.

25Remarkable evidence for the naval base at Loryma was discovered on the beach immediately west of the walled city area, near the sanctuary of Artemis Soteira, by W. Held in 1995 and 1998-200121. An inclined row of 20 socket-holes was observed in 1995 in the vertical cliff side that delimits the bay on the west (figs. 7-8). They are spaced c. 70 cm apart, lie 3-3.30 m above ground level and have a gradient of 1: 7 (8.1°). A vertical slot in the rock indicates the line of the back wall, and from there what are clearly slots for roof-beams can be traced for 15 m, as far as the present shore line; but the buildings may have continued farther out (there seems to have been a rise in sea level since antiquity). Held then carried out an electromagnetic survey on the beach, which identified six parallel lines, 12-13 m apart, which he has plausibly related to the cuttings in the rock as the walls of shipsheds (fig. 9). Weaker readings between these lines may indicate a subdivision of each shipshed into two slipways. Their original height may have been up to 5 m, if we allow for a relative rise in sea level. (The area is partly covered by the remains of a large structure known as ‘the Byzantine Arsenal’ – Loryma served again as a naval base in the fifth to seventh centuries AD.)

26Even without excavation the identification of the site is clear: we have a set of six shipsheds, for twelve warships; perhaps warships smaller than triremes. Rhodes made great use of patrol vessels (νῆες φυλακῖδες), and phylake was a key word in Rhodian naval hegemony – well discussed by Gabrielsen22; these may have been trihemioliai, but the dimensions seem rather wider than the narrow category of shipsheds identified at Rhodes, so maybe they were hemioliai (fig. 10). Without excavation we can only date from context, but as we have argued, a naval base here is likely to have been a Rhodian construction not later than of the early third century BC.

27It was not only Rhodes that saw the importance of its peraia as part of its naval defences: the Samians built shipsheds for guardships below a watch-tower on the mainland opposite the south-east coast of the island, on the north shore of Cape Mykale, commanding the channel between Samos and the mainland (much narrower than that between Rhodes and Loryma) (fig. 11). Remains were noted by Wiegand in the 1890s, and slight traces of a shipshed have just been rediscovered by H. Lohmann (fig. 12)23.

28I was particularly interested to learn of Dr Held’s discovery at Loryma, since our work at Alimnia had led me to argue that the Rhodian fleet had a series of out-stations at strategic points in its territory: Alimnia covered the approaches from the south-west, and it was logical to think of similar out-stations at key points in Rhodian territory on the mainland, and also on off-shore islands as far as Megiste.

29I was reassured to find out that Gabrielsen had independently come to the same conclusion in 1997; he also emphasized that Rhodes was a serious naval power by the midthird century, ensuring phylake of the south-east Aegean islands from then on24. We can now add growing evidence that the phylake extended to Rhodian possessions on the Karian coast from that time, not only reflecting Rhodian interests there, but also serving as part of a ‘naval early-warning system’ for the island and cities of Rhodes herself.

30This is still an interim account, which I hope can be supplemented by future finds on the coast of Karia and Lykia to fill in the picture of Rhodian naval strategy. I close with some speculations about possible sites on the coast of Karia and Lykia. The likeliest site for a further Rhodian naval out-station is clearly Megiste, an early Rhodian possession off the Lykian coast, but none of the remains found along the shore can be related to such an installation – and I do think that slipways will be the diagnostic feature (maybe that is natural for me to say!). Megiste clearly had a fortress and a garrison and an epistatas25.

31Another candidate is Daidala, a further Rhodian exclave, in the Gulf of Fethiye (Telmessos), where a dedication by a Rhodian epistatas was found; the inscription was actually found on the island of Tersane in the bay. This is an evocative name, and the island (which I have not visited) would seem to be a more suitable site for a naval station than Daidala, but the evidence from Tersane looks rather late (not before the second century BC) and Tietz has shown that the inscription came originally from Daidala, and it is not certain that Tersane Island formed part of the Rhodian exclave – it may have been rather a port of Kalynda, Telandros, mentioned in the inscription from Xanthos published by Balland26. Held mentions a third site where a Rhodian epistatas is epigraphically attested – Idyma in the subject Peraia27. But there is no archaeological evidence for naval installations there.

32Clearly Rhodes was very concerned for its security from attack, not only from a distance but also from the nearby coast of Karia and Lykia.