You are on page 1of 24

bs_bs_banner

The International Journal of Nautical Archaeology (2017) 46.1: 4871


doi: 10.1111/1095-9270.12212

The Harbour at Ptolemais: Hellenistic City of the Libyan


Pentapolis
Robert A. Yorke
Silver Birches, Bashurst Hill, Horsham, West Sussex, UK, RH13 0NY

David P. Davidson
Corner Cottage, Breedon Lane, Tonge, Derbyshire, UK, DE73 8BA

This article describes eldwork undertaken in 1972 at the harbour of the Hellenistic city of Ptolemais at Tolmeita in Cyrenaica,
Libya. This survey has shown that the city possessed a sophisticated Hellenistic harbour installation, now submerged by some
2 m. It consisted of a small shing harbour and a larger commercial harbour protected by extensive stone breakwaters built
around two offshore islands. Between the harbours, remains were found under water of an extensive block-built quay connected
to one island, where there was a system of submerged rock-cut tanks. These are interpreted as the foundations of a sh-processing
factory. Changes in sea-level are also discussed.
2017 The Authors

Key words: Ptolemais, harbour, underwater survey, block walls, sh tanks, sea-level.

I
n the summer of 1972, under the aegis of the in 2009, to publish the site in full using all the 1972
Society for Libyan Studies, the authors with John eldwork data. Moreover, recent sand movements
H. Little undertook a marine survey of the remains have exposed further shoreward sections of submerged
of the harbour at the site of Hellenistic Ptolemais, blocks and these have been added to the plans using
in Libyan Cyrenaica. The ancient site of Ptolemais is information from aerial photographs taken by Miron
still largely unspoilt. It lies between modern Benghazi Bogacki of the Polish Mission of 20062011 under the
and Derna at a point where the mountains of the late Professor Mikocki (Mikocki et al., 2010; Bogacki,
Djebel Akhdar come down to the sea and pinch off 2012). We thank Dr Monika Rekowska-Ruszkowska
the coastal plain, leaving a rocky steep coast as far as for providing this information.
Apollonia (modern Susa) to the east. In the hills behind The result is a two-basin harbour based on the
Apollonia lie the famous ruins of ancient Cyrene sandstone reef elements with inshore islets so common
(Fig. 1). all along the coast of North Africa. The scale of the
The interim report of the site was presented in installations is very comparable to the better-known
1972 (Yorke et al., 1972) and preliminary publication harbour at Apollonia nearby, if not quite of the same
appeared in Libyan Studies (Yorke, 1972), the IJNA level of complexity. The site is characterized by a
(Yorke et al., 1973), and the Proceedings of the headland, on which sits the modern lighthouse, with
Underwater Association Minisymposium (Yorke and two small islands to its east, and a long open sandy bay
Davidson, 1973). to its west (Fig. 2).
The eldwork in 1972 involved a comprehensive There is a small natural harbour in a bay to the
underwater visual survey of the whole harbour area east of the headland, where local shing boats moor
combined with some echo-sounder surveys. The up. In antiquity the westerly of the two islands would
resulting overall outline plan of the site was published always have been a promontory, affording even better
at the time. As part of the survey, detailed plans were protection, and to the east was the larger harbour basin
also made of areas of submerged blocks, rock-cut formed by the addition of articial rock breakwaters
tanks, and quarries, which formed an important part to the small island, known from the Stadiasmus Maris
of the harbour installations. The opportunity has Magni #55 (Muller, 1855) as Ilus. The present-day
therefore been taken, in the light of the authors westerly island has no modern local name, and is
subsequent researches and an informal visit to site referred to herein as the Islet.

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
Published by John Wiley & Sons Ltd. 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Figure 1. Location map. (R. A. Yorke and D. P. Davidson)

It was initially known as the Harbour at Barca for


the city of Barca on the Djebel Akhdar behind, which
sat on a fertile plain but had needed an outlet via the
sea. Ptolemais became a named city in its own right in
252 BC under the Ptolemaic dynasty of neighbouring
Egypt (Mueller, 2004: 1ff). It achieved great renown in
the Roman world for its trade in the medicinal herb
silphium.
The region was subject to massive earthquakes in AD
262, and especially in AD 365 (Ambraseys et al., 2005),
which is spoken about in contemporary literature by
one of its most famous inhabitants, Bishop Synesius.
His later letters (Synesius, 413) tell us so much about
life in the late Roman province.
In any event, the incursions of the tribes from the
Figure 2. View eastwards from the headland over the
south took their toll, so that by the time of Procopius
harbour at Ptolemais showing the two islands, the Islet and in the mid 6th century, the city was but a shadow
Ilus, with the Djebel Akhdar in the background. (R. A. Yorke) of its former self, mainly because the majority of the
populace had departed when the water supply from
the Djebel Akhdar was destroyed. There was a brief
The modern authority on the site has been Kraeling, recovery in the renewal under the Emperor Justinian
who dug for four seasons in the city (Kraeling, 1962). (Ananeosis), using a much smaller perimeter plan, as
He apparently did not explore the harbour in any detail also at Apollonia.
and followed early travellers who believed there had By the time of the Arab invasion in the 7th century
been an extensive articial harbour to the west of the AD the city was already in a reduced state as a
headland. The 1972 survey proved conclusively that consequence of constant incursions by the Garamantes
this was not the case. The site was visited both above tribes from the south. Ptolemais was so out on a
and below water by Carlo Beltrame (2012, 315326) limb, away from the main east-west routes along the
who conrmed the existence of underwater remains. A Djebel Akhdar, that the city was bypassed for conquest
Polish team excavated extensively here on the landward until the Arab eet arrived several years after the
side of the site from 2001 to 2010 (Zelazowski, 2011; fall of Cyrene, Barca and even Tauchira. Although it
http://ptolemais.pl/publications/). never recovered any signicant commercial importance,
maritime traffic must have continued at a modest level
Historical background through to the turn of the millennium and beyond
First settled, according to Herodotus (iv. 151160) (Kennet, 1994).
(Beloe, 1825: 419423), by Greeks in the 6th century The site was visited by several early Arab authors
BC (Boardman, 1966: 149156; James, 2005: 2), (see below). But there is then a great paucity of
Cyrenaica became a Roman province in 70 BC, and detailed information until the early European travellers
shortly afterwards became a joint province with Crete. ventured there from the end of the 18th century
Ptolemais appears in most of the ancient itineraries (Rekowska-Ruszkowska, 2012), all of whom found a
(Kraeling, 1962: 33ff, Jones and Little, 1971: 7173). deserted but virtually intact site with the exception of

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 49
NAUTICAL ARCHAEOLOGY, 46.1

Figure 3. General plan of the visible structures at Ptolemais Harbour. (R. A. Yorke and D. P. Davidson)

the harbour, which was by then completely submerged. for this movement was gathered in various parts of the
We rely almost exclusively on the survey undertaken by survey and is summarized below. In order to clarify
the brothers Frederick and Henry Beechey who came a description of the site, features described are given
this way in 1822 (Beechey and Beechey, 1828: 376385) letters and numbers on the plans.
and left a series of plans of extraordinary veracity of the
sites along the coast. Not until the survey of Kraeling
in 19561958 (Kraeling, 1962) do we learn more of
substance about the harbour. Ptolemais headland
The headland (H), which lies in front of the scatter
of houses that constitute the present-day town of
The survey and sea-level change Tolmeita, is the most salient point on the coast for
It was clear from the 1972 survey that there had been a many miles (Fig. 3). Eastwards the generally bare, rocky
rise in relative sea-level of around 2 m since antiquity. coast with occasional sandy bays is punctuated also by
The description here of the survey of the harbour small inshore islands such as these at Ptolemais, formed
(Fig. 3) should be considered and interpreted on the also from a Pleistocene reef. To the west we are told
basis that when the structures were built and the by the Stadiasmus that there is nothing much, apart
harbour was initially in use the sea-level relative to the from Benghazi, in the way of natural shelter all the way
land was 22.5 m lower than in 1972. The evidence through to the Gulf of Sirte (Muller, 1855, 447450).

50 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Figure 4. Plan of Ptolemais Headland. (R. A. Yorke and D. P. Davidson)

The dry part of the headland is bounded by a low remains of quarrying. Some of these blocks are still
cliff (H2), some 45 m in height with a reef (H1) awash in situ and incomplete, while others may possibly have
to seaward in front of it at, or just above, present sea- been extracted before being re-laid.
level, dropping into the sea at a depth of a metre or two In the north-west corner of the headland some of
(Fig. 4). The reef itself has been extensively quarried the quarrying forms sloping beds close to sea-level,
and there are a number of rocky outcrops (H4), most which has prompted various interpretations elsewhere
of which show signs of having been quarried. as slipways or shipsheds (Kraeling, 1962: 60, plate
The signicance of this promontory for shipping in IIIA; Laronde, 1981: 62; Kenrick, 2013: 98; Baika,
antiquity can have been no less than it is today, forming 2014: 497). The area is cut into the bedrock and slopes
as it does, sheltered bays to east or west according to the marginally down to the waters edge. In one of these
weather. It also may be adjudged to have been provided there are two rock-cut ridges that divide the rock into
with some sort of navigation warning, judging by the two compartments (H5, H6) having slightly differing
ancient foundations that appear in the footings of the widths. H5 is the narrower of the two at 4 m, and, while
modern lighthouse (H3). A lighthouse is mentioned its form is clear from the landward end, the ridge on the
by Kraeling (1962: 50, n41) and he joins the Beecheys north side is only partially complete (Fig. 5a). However,
(1828: 378) in viewing this as the logical place for the at the seaward end of H5 the rock ridge is very apparent,
ancient pharos, as did Pacho (Pacho, 1827: 178). effectively barring the potential for entry and exit of
At the highest level of the ground of the headland a boats.
number of mainly circular silos were found cut down In the walls of one of these dividers in H6 there are
into the bedrock, similar to many others found at horizontal ledges running at about one block height,
neighbouring sites (Hesein, 2014). There are no other 0.5 m, from the bedrock (Fig. 5b). These have been
obvious signs of building on the headland dating from proposed as mountings for the roof of a shipshed
antiquity. The other cuttings in the rock and a few (Baika, 2014), but appear to be simply the line of the
associated blocks of the native stone appear to be the cutting of the blocks extracted at the higher level. In the

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 51
NAUTICAL ARCHAEOLOGY, 46.1

Figure 5. a) Rock cutting H5 is not considered by the authors to have been a slipway; b) rock cutting H6 showing the horizontal
ledge on the right and the edge of the partially quarried block on the left; c) close up of the partially quarried block still engaged
in the quarry oor. (R. A. Yorke)

north-east corner of H6 one single partially quarried sea with a ledge at the waters edge and some unnished
block remains engaged in the bed of the quarry, adding blocks remaining engaged in a water-eroded form. In
further doubt to the interpretation as slipways (Fig. 5c). the north-west corner of H7 is a distinctive array of
Closer inspection in the shallow water 15 m beyond blocks, 3 m 8 m (H8), which have been cut on all sides
the seaward ends of these slipways shows extensive except the base, and remain engaged in the bedrock.
evidence of quarrying and partly quarried blocks on A few metres away to the south-east, in contrast, is a
the seabed (H14). There is further quarrying nearby group of cut blocks (H9) differing in size and shape that
at H15, now under water. This quarrying would have appear to have been deliberately laid for some unknown
been on dry land in antiquity and would have blocked purpose.
any access to the sea from H5 and H6. Furthermore To the south of the quarry H7 were found signs of
the fact that an upwards change in sea-level would also silos cut vertically into the rock at the higher level of the
have put the exit from any supposed slipways at perhaps headland. One of these, H12, had been partially eroded
1.5 m above the sea-level in antiquity adds considerable so that it was completely visible in section. It proved to
further doubt. Another important point to be taken be 34 m deep and 0.4 m in diameter at its entrance at
into account is that these features on the west side of the top, widening to 1.75 m diameter at the bottom. The
the headland face breaking seas driven by the prevailing oor, which was well above present sea-level, was lined
north-westerly wind that would have made launching with opus signinum and, so far as could be told, without
boats very dangerous. a drain. The entrance to a second similar bell-shaped
The conclusion is that these are simply rock quarries silo was found adjacent to H10, but was almost full
and could never have been slipways or shipsheds. of sand, and again with signs of opus signinum. Some
Beltrame (2012: 317) also came to the same conclusion. 1.5 m away was a further deep, and similarly belled,
Adjacent, to the south, H7 is a single quarry of 50 m cutting that was rectangular in form, also with signs of
width, but otherwise similar nature, sloping down to the opus signinum and communicating via a channel (H11).

52 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Signs nearby of yet a third circular aperture (H13) cross-walls run to the shore. (Lehmann-Hartleben, 1923:
indicate that there may be others below the ground 278, quoting Barth, 1849, Vol I: 399, 406)
surface. Similar silos are found in the rock at the waters
edge at sites in the province, including Apollonia and Barth took these to be shipsheds and Kraeling (1962:
Phycus (Arena, 2011: 167; Hesein, 2014: 135) but their 50) seems to buy wholeheartedly and directly into this
use remains uncertain. The silos on the headland may interpretation, as did Laronde (1986: 171). This is an
simply have been water storage tanks for buildings that unlikely location for the main harbour, however, and
have been effaced since antiquity. Alternatively, there a trial excavation of the wall conrmed this through
remains the possibility that they could have been used the nature of the structure and its levels relative to
for some wet product, perhaps salted sh or the sh- the sea. The full extent of the wall was also surveyed
sauce, garum. (Fig. 6). The trial excavation was undertaken under
the supervision of the local Antiquities Department in
order to answer two specic questions: what was the
The bay to the west nature of the structure; and could it have formed a part
To the west of the headland, bay J is open with a long of the harbour?
shallow sandy beach that is sheltered from an easterly This wall is built in ne ashlar masonry and is
wind, but the presence in 2009 of a small modern over 300 m in length and 5 m in width, having two
coaster lying wrecked just offshore attests to its general parallel single lines of blocks with numerous cross-walls
exposure to the prevailing westerly and north-westerly every 2 m to 3 m, forming a conventional casement
wind and seas (Fig. 3). conguration (Figs 6 and 7). It lies buried with its top
There is a general assumption that ancient meteoro- courses of blocks level with the sand, so that it is not
logical and wind patterns have changed little since always apparent. It runs roughly parallel with the edge
antiquity, discussed at length by Davis (2009: 33). The of the sea and with the land behind. At its western end,
Mediterranean Pilot shows winds at Derna station, just short of the Wadi Khambish, it turns a curve of
with the wind roses for January showing dominant approximately 45 towards the land and is then lost
winds from SW to NNW, and NW to NNW for July beneath the sand. There is no sign of it continuing to
(Mediterranean Pilot: v. 5, 2005: 29ff). This has been the west of the wadi.
taken into consideration in reviewing the suitability of At the eastern end, towards the headland and the
the sea defences in the different sectors of the harbour. line of the city wall, the terminus is indistinct and
At the same time the predominant sea current runs was not located. In the time available neither end of
anti-clockwise in the eastern Mediterranean, giving a the wall could be located, and a single section at the
west to east ow, and a dominant swell from the north- commencement of the curved section was chosen at
west sector (Mediterranean Pilot: v. 5, 2005, 21ff.). In which to excavate the structure down to its foundations
antiquity a similar pattern can be imputed from the (Fig. 6).
dominant directions of maritime trade (Aubet, 2001: The front wall, nearest to the sea proved to have
215). only two courses at the point investigated, with the
Four echo-sounder runs carried out during eldwork foundation level some 2.5 m above present sea-level.
showed no trace of any protective breakwater having The foundation layer was of ashlar header blocks
been constructed westwards from the tip of the approximately 1 0.5 m in plan, with the long axis
headland that could have formed a sea defence enabling towards the sea and a layer of stretchers of similar size
this bay to have been used as a viable harbour. laid on top. The base layer was laid directly on to clean
The most that can be said about the west bay from sterile clay, and surrounded by clean sand.
a maritime viewpoint is that it would always have been The second, landward element of the wall had three
available as a sheltered anchorage against a north-east further courses, taking the foundation down to 1.5 m
storm, and at any time small boats could have been above present sea-level. It was something of a surprise
drawn up on the beach. that the foundations of the rear wall were deeper than
those of the front wall, which was down the slope of the
Beecheys wall beach and closer to the sea.
In 1820 the Beechey brothers (1828: 355, 378) One of the cross-walls was also revealed and was
commented on a small internal harbour to the east of found to be of a single line of headers and stretchers
the headland, but noted the presence of a signicant and not engaged with either the up-slope or down-slope
wall on the beach to the west that has, from time to element. This is not unusual in such casement walls,
time, been linked with harbour works. Hermann Barth as seen at Ptolemais itself and at Tauchira (modern
passed this way in 18451847, and Lehmann-Hartleben Tocra) nearby (Bogacki, 2012: 83). However, all of these
reported (in translation) his conclusions thus: examples are only 23 m in overall width, whereas the
Beecheys wall is 5 m across.
. . . in a wide open bay to the west of the ancient city and In the bank behind the beach visible remains of
outside it, Barth also saw a long and strong wall running walls to the north of the Tauchira road were formerly
from the city walls parallel with the shore, from which reported, but there are no current signs of these, nor

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 53
NAUTICAL ARCHAEOLOGY, 46.1

Figure 6. a) General plan of Beecheys wall; b) detailed plan of the excavated area of Beecheys wall; c) section of Beecheys
wall. (R. A. Yorke and D. P. Davidson)

any obvious connection. There is no clear sign of Beecheys wall contains nothing but the ashlar blocks,
any further continuation of the structure at beach and these are on a different line and are at a different
level to the west of the wadi that is just to the west height.
of the curved end of the wall. However, there are The investigations showed that the lowest foundation
some jumbled remains of a wall, ashlar masonry and of the wall is 1.5 m above present sea-level, which
fragments of cisterns with opus signinum in the bank would have been at least 3.5 m above the sea-level
behind the beach to the west of the wadi mouth. But associated with rest of the harbour site in antiquity, and

54 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Little, 1971: 72, g. 6), and is apparent on aerial


photographs. This feature (H16, Fig. 4) is a long trench,
or foundation, cut into the rock of the reef about 60 m
long and now about 0.7 m under water. At its northern
end it is about 2 m wide emerging from the outer rock
boundary wall of the reef, and it widens at its southern
end before becoming less distinct and disappearing
under modern rubble below the modern lighthouse. On
the same alignment, some 50 m to the south, there is a
15 m-long construction of blocks (H17) at least three
blocks wide and 11 blocks long on the seabed in c.0.7 m
of water. Between H17 and the modern rubble to its
north the line of the rock-cut foundation is visible, but
any blocks that may once have been there appear to
have been robbed out to clear an entrance into the small
Figure 7. Excavated section of Beecheys wall looking north-
sandy bay to the west where shing boats are moored.
east towards the headland, scale 1 m. (R. A. Yorke) The line of H16 and H17 appears to align closely
with the street grid of Hellenistic Ptolemais (Fig. 3),
indicating that these features could have formed the
foundation of a continuous sea wall that ran along the
eastern side of the headland out to the edge of the reef.

Ilus and harbour entrance


The eastern of the two islands, Ilus (D) (Fig. 9), is
slightly higher than its twin, with its top level (D1) at
23 m above the 1972 water level. It is surrounded by
a reef (D2 and D6) at 12 m depth. It has obviously
been used for extracting stone, and shows three small
quarries D3, D4 and D5 (Fig. 10).
Quarry D3 faces the sea to the east at the eastern end
of Ilus. It slopes consistently down towards the sea, and
Figure 8. Modern shing boats moored in the bay to the east it terminates at a depth of 0.7 m below present sea-level.
of the headland. (R. A. Yorke) It is clear from its exposed location and its rough plan
that it has not been constructed as a boat slipway, and it
demonstrate conclusively that the wall could never have is likely therefore that it is indicative of at least a 1.5 m
formed a part of the seaward defences or installations. upward sea-level movement in order for quarrying to
Because the lower wall is on the seaward side, it also have been carried out on dry land in antiquity. The
seems unlikely that it was built as a defence against bases of quarries D4 and D5 are 0.51.0 m below sea-
further sea-level rises after the earthquakes of AD 365 level, and a similar argument therefore applies to rise in
and later. sea-level since antiquity.
While there is still much to learn about the detail and Ilus is devoid of other man-made structures, except
true nature of the Beecheys wall, it was clearly nothing for two small cisterns/tanks D8 and D9 that contain
to do with the ancient harbour. The most plausible opus signinum mortar indicating that they had been
interpretation at the present time is that it was some sort waterproofed for storing liquids. It is most probable
of esplanade along the beach between the town and the that their bases would have been above sea-level in
wadi. antiquity.
The only other visible features are two iron gun
barrels, presumed medieval (D10 and D11), that have
The bay to the east been let vertically into the surface of the rock in more
The bay to the east of the headland is sheltered from the recent times to act as mooring posts. They serve as
westerly prevailing wind and swell by the headland itself a reminder that long after the site of Ptolemais was
and from any easterly wind by the two islands, Ilus and abandoned as a classical city, boats continued to use
the Islet (Fig. 3). Small shing boats are moored here or the shelter for anchorage (Kraeling, 1962: 49; Laronde,
drawn up on the beach to the present day (Fig. 8), just 1981: 65; Kennet, 1991: passim), and, following the
as they must have been in antiquity in what is normally Italian invasion of Libya in 1911, the Italians are
a sheltered bay (G). reputed to have effected some modications in the
Where the beach joins the headland below the area to make the harbour more useful (Kraeling, 1962:
lighthouse there are tumbled stones and a linear feature 31). The removal of much ancient masonry, visible at
that has been marked on previous plans (Jones and locations on the seabed around the site, could therefore

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 55
NAUTICAL ARCHAEOLOGY, 46.1

Figure 9. Plan of Ilus and the eastern entrance to the harbour. (R. A. Yorke and D. P. Davidson)

have taken place at many points in the past two a building on the end of the mole that has subsided
thousand years. into the channel. A prole across the mid section of
The rock shelf D6 continues on the south side of the Mole C is shown in an echo-sounder trace (Fig. 11a).
island where it leads down to an area (D7) of stones, of From a depth of 5 m on its outer eastern edge the water
size approximately 0.20.3 m, among which are some depth reduces to 2 m over the top of the mole and then
squared stone blocks of c.0.8 0.4 0.3 m and pot increases back to 3 m in the lee of Ilus.
sherds in about 3 m of water. After 20 m or so these At a point halfway to the shore from the southern
disappear into a sandbank rising to 2 m depth. There end of Mole C the scattered blocks disappear under
is no indication of how far they extend under the sand. the sand, to be succeeded by a rising bank of rubble
Various small nds (D12) with scatters of small stones forming a southern breakwater, Mole A, emerging from
probably indicate the discharge of ballast in antiquity, the shore (Fig. 3). On the top of the mole there is an
and support the proposition that the shelf (D67) on the abundance of blocks, monumental masonry, columns
south of the island facing unobstructed, deeper water and column drums and there is a clear line of quarried
would have constituted a mooring point for vessels in stone on its outer (eastern) edge, conrming that it is a
the lee of the island. man-made structure.
From the eastern end of Ilus there is a man-made Mole C, which protects the main basin of the harbour
structure of rubble and masonry, topped with small- (K) from easterly and north-easterly seas, the deep
sized rubble and blocks, that has been articially channel B, and Mole A rising to the land indicate
deposited in the water. It runs in a south-easterly quite clearly that this was the eastern entrance to the
direction for 130 m from the island ending in a steep harbour. The presence of a large amount of masonry
edge down to channel B, some 6 m in depth, forming a on the seabed at the southern tip of C might suggest
substantial breakwater (Mole C) on the eastern side of that there had been a navigation light or lighthouse at its
the harbour. In the bottom of the channel there are a end in antiquity, a not unreasonable supposition since
substantial number of large cut blocks (C1) scattered this would have been the logical place to indicate the
on the seabed indicating that there might have been harbour entrance to incoming seafarers.

56 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Figure 10. Remains of quarrying on Ilus looking north-west across quarry D5. (R. A. Yorke)

Figure 11. Echo-sounder traces: a) Mole C, mid section; b) Ledge E, between Ilus and the Islet east; and c) west. (R. A. Yorke
and D. P. Davidson)

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 57
NAUTICAL ARCHAEOLOGY, 46.1

The ledge between Ilus and the Islet On its west side there is a wall (F2) 49 m long running
Between the two islands Ilus and the Islet there is a north-west under water consisting of a single visible line
rock ledge (E) covered with sand at a depth of c.2 m of header blocks, average size c.1.3 m 0.65 m, that
below the surface (Fig. 3). Echo-sounder traces (Fig. sit one course above the seabed. At its landward end it
11 b, c) show two examples of the water depth reducing disappears into the sand of the beach at F2.1 and at its
rapidly from 9 m in the open sea to about 2 m across seaward end it merges into an area of rubble and blocks
the ledge and into the main harbour basin. Even now, at F2.2 and becomes indistinct.
although submerged, the ledge breaks the force of the Running parallel with this wall towards the Islet,
north and north-westerly seas entering the harbour and 5 m to the east, there is similar line of underwater
basin K. On the basis of a rise in sea-level of at least blocks, 75 m long (F3), although the blocks do become
2 m since antiquity, the ledge would have been at or indistinct in three places. After 30 m the wall is
near the surface and in all probability is the remains constructed of two lines of headers (F3.2) for about
of the base of a breakwater that connected the two 10 m, which could indicate that F3 was originally two
islands. blocks, or 3 m, wide. At F3.3 the wall becomes two
The current depth of ledge E shows that there would courses high for 6 m and the second line of blocks re-
have been insufficient water depth between the two appears for a short distance at F3.4. The wall then turns
islands for this to form an entrance to the harbour through 10 to the north before terminating at right
in antiquity. It seems almost certain that the natural angles against the side of F7.
reef between Ilus and the Islet would have been built F7 comprises two parallel lines of blocks (1.5 m x
up to form a rubble breakwater topped by a roadway 0.65 m) separated by four spaced-out blocks 0.6 m
of blocks, which would also have provided access square (Fig. 15). Some 17.5 m of the northern line
for the stone quarried on Ilus to be brought to the and 16 m of the southern line remain in place. The
shore. This breakwater and the two islands would have eastern end of this structure (F7.1) is 0.8 m below the
formed a large and totally secure harbour basin of 7 surface and 13.2 m to the west (F7.2) the block surface
hectares. With the force of the waves over hundreds of is 1.3 m below sea-level, indicating a downward slope
years the breakwater has been attened to its current of 4% into the bay. There is also evidence from the
prole. aerial photographs (Figs 14 and 16) that the structure
probably continued further out into the bay. On the
The Islet and block structures assumption of a rise in sea-level of 22.5 m, the
This island is a rocky sandstone outcrop some western end would have been at sea-level and the whole
200 m offshore on which a considerable amount of structure can be interpreted as a 2 m-wide slipway for
construction has taken place in the form of walls made bringing small shing boats up on to the quay F1.
of blocks, rock-cut chambers and foundation trenches In the area to the east of F7, Beltrame (2012: 317)
(Fig. 12). The seabed around the island is now covered found within the area of F1 the remains of an ancient
with blocks and rubble that have fallen or been washed boat which he describes as mortise-and-tenon and
away from previous constructions. Most of the area treenail fastened, but with no secure date.
from the shore to the tip of the island is now shallowly A further 25 m to the east of F3 there is another
submerged by 11.5 m, and on the assumption of a parallel single line of blocks (F4). The shoreward end
sea-level rise of 22.5 m, the whole area would have been (F4.1) is marked by a line of some nine blocks that, with
dry land when the harbour was built. another four lines of blocks up to F4.2, form a small
Between the shore and the Islet there are four seemingly paved area between it and F5. Although there
parallel lines of submerged blocks and at their southern, is no knowing if this is the original surface of the quay
landward, end there is a cross-wall (Fig. 13). About 60% it would seem to be of similar type of construction. The
of these blocks were visible in 1972 and were accurately line of F4 north-westwards is visible mainly as a robbed-
surveyed in from the shore nearby. A short visit by out trench leaving no blocks visible in the seabed for
Yorke in 2009, and aerial photographs taken by the 41 m until F4.3, where it becomes a single course of
Polish mission from a balloon in 2008 (Bogacki, 2012: blocks, 0.7 m wide and c.1.2 m below sea-level. These
81, g. 7; Rekowska, 2016: 84, g. 39) showed that blocks are smaller, varying from 0.7 m 0.6 m to
the sand had receded from the area at that time and 0.7 1.0 m in plan, and the line becomes indistinct after
a further two lines of blocks that were under the sand 17 m, re-emerges at 22 m and nishes at F4.4.
in 1972 had become visible to the west (Fig. 14). It Joining F3 and F4 at their southern ends by the beach
also showed that there was a south-easterly extension there is a cross-wall (F6) of which at least 22 blocks
to the structures previously surveyed and a cross-wall (c.1 m 0.5 m) are visible on the occasions when the
of blocks (F6). The combination of both sets of data sand has been washed away. At its western end (F3.1)
gives a very clear layout of the overall block site with there is a semi-circular array of blocks, which might
its wide, level rubble core (F1) and will be described have been the foundation of a round structure. Either
as a whole. No excavation was undertaken, so all block side of F6 there is a large amount of masonry and
structures are those visible above the rather uid sand blocks lying in the water indicating that this may once
of the seabed. have been a more substantial structure.

58 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Figure 12. General plan of the Islet with submerged blocks and tanks. (R. A. Yorke and D. P. Davidson)

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 59
NAUTICAL ARCHAEOLOGY, 46.1

Figure 13. Detailed plan of the submerged blocks south-east of the Islet. (R. A. Yorke and D. P. Davidson)

60 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Figure 14. Aerial view looking north-east showing submerged tanks on the Islet and submerged block walls between the Islet
and the shore. Ilus is in the background. (Polish Mission to Ptolemais, Institute of Archaeology, Warsaw University, author:
Miron Bogacki)

From the eastern end of the cross-wall (F6), wall F5


runs for a distance of 73 m, parallel with, and about
5 m to the east of, wall F4, towards the Islet (Fig. 17a).
The wall consists of one course of blocks (average 1
0.7 m), 2.1 m wide, consisting of stretchers and headers
as far as F5.1, where there is a second level of blocks
and the headers and stretchers reverse. At F5.2 there is
a second course of blocks for 12 m with headers and
stretchers reversed from the course below (Fig. 17b).
The tops of the base course of blocks vary from 1.2 m
to 1.5 m below sea-level. At F5.3 wall F5 is abutted
by F9 and becomes indistinct. Wall F8 is of an entirely
different form of construction and cuts through the line
of F5 at an acute angle.
Figure 15. Two parallel lines of blocks with spacer blocks In between the parallel lines of blocks F4 and F5
between at F7, possibly part of a slipway. (R. A. Yorke) there is stone rubble and random large blocks indicating
that these two line of blocks may have formed a
retaining wall about 8 m wide.
The four walls F2F5 separate the two harbour At the northern end of F5 a wide structure of blocks
basins and it was sometime the custom of harbour- (F9), one course high, in a water depth of 1 m, leads
builders to incorporate a sluice channel between basins off east-north-east from F5.3. It is four blocks wide
to provide a ow of seawater to prevent silting up of of varying widths for 9 m to F9.1, where it reverts to
the harbours, much as found at el-Haniyeh nearby (De a pavement, 2.1 m wide, of alternating headers and
Graauw, 2012). If there had been such a feature in this stretchers. At F9.2 a second course is still in place for
harbour the only feasible place would have been to the 6 m (Fig. 18a). This course is constructed with headers
south of, and adjacent to, wall F6. There is no evidence and stretchers reversed from the course below, but is
of this as the seabed is packed with jumbled blocks missing its stretchers on the northern side.
(although some coherently ordered) but, based on the The lower course of F9 extends for 34 m and the tops
similar features at el-Haniyeh, the possibility of there of the blocks are 1.4 m below sea-level at its eastern end,
having been a sluice here cannot be excluded. an increase in depth of 0.4 m from its root at F5.3. The

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 61
NAUTICAL ARCHAEOLOGY, 46.1

Figure 16. Aerial view of the Islet showing submerged rock-cut tanks, submerged lines of block walls F3, 4, 5 and 9 and the
slipway F7. (Polish Mission to Ptolemais, Institute of Archaeology, Warsaw University, author: Miron Bogacki)

structure consists of about 49 blocks (counting headers To the south of F9/F10 the seabed (F13) is again
measuring 1.3 0.7 0.5 m) except for some smaller covered in concretion, small blocks and rubble for a
ones 0.5 0.5 0.5 m at its base. The construction of further 15 m in a water depth of about 1 m.
structure F9 is identical to F5. Continuing north-westward from F5.3 there is
Another single course of blocks (F10) (1.3 m 0.7 m another block structure (F8), 2 m wide, comprising
in plan) runs for 22 m parallel to F9 and 4 m to its south. four lines of blocks (single course), that runs straight
While the visible blocks of F10 are headers, those of F9 in shallow water at a depth of 0.2 m for 14 m
are stretchers. Nonetheless, both F10 and F9 appear to and then curves northwards for 2 m, and ends in
form the retaining walls of a structure such as a quay or sand at F8.1 just south of the rock-cut tanks on the
a roadway similar in construction to F4/F5. island. However, both the orientation of F8 and its
On the northern side of F9 by its root, the seabed is construction technique are quite different from the
covered with small blocks and rubble but from about adjacent structures F5 and F9, and the water depth is
12 m along its length the water depth increases to 0.2 m compared to 1 m where F5 and F9 join. The
2 m into an enclosed sandy lagoon (F12) (Fig. 13), details of the join between F8, F9 and F5 are not clear
on the bottom of which are random larger blocks, under water, but because the orientation of F8 is quite
column drums and a spiral-uted column at F12.1 different from F5, it may be a later addition.
(Fig. 18b). From the end of F9 the seabed is covered At the northern end of F8 a 2 m-wide rock-cut
with concretion, small blocks and rubble at a depth of channel (F14) runs initially westward at right angles to
2 m (F12.2) until it drops off the rock ledge to sand at the end of F8 at a depth 1 m. This channel then curves
3 m depth on which large blocks and a stone anchor northward round the rocky western edge of the island
were found at F12.3 (Fig. 18c). for c.60 m until it peters out on the wave-swept shelf at
To the north of the submerged lagoon at F12 there F14.1 (Fig. 12). Halfway along, at F15, a column drum
is a deep channel (F12.4), 3 m deep, that runs along and rectangular blocks in 3 m of water were found off
the southern edge of the Islet by wall F16.1 (see below) the edge of the island.
into the deeper water of the eastern harbour basin.
The bottom of the channel is lled with masonry and Discussion of block structures
rubble so the original depth is unknown. The existence The block structures described are currently in water
of deeper water surrounding F9 and F10 could mean depths varying from 0.2 m to 1.2 m and so would have
that together they form a quay 5 m wide, and boats been above sea-level in antiquity, probably topped by
could have been loaded and unloaded on both sides. further construction. It must be assumed that all blocks

62 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

above water were robbed out over the years, leaving only perhaps paved with blocks to form a smooth roadway.
those below the surface that were difficult to reach. The existence of many cut blocks lying on the seabed
The two double lines of blocks appear to form a pair between the walls adds substance to this interpretation.
of double retaining walls either side of a wide platform, Similar paved harbour structures are seen at Apollonia,
F1, some 40 m wide. The interior of the platform some 100 km to the east (De Graauw, 2014).
would have been lled with rubble and the surface The Islet would have been connected to the shore at
the time of the use of the harbour by this elaborately
constructed structure in ashlar masonry, in the manner
of a roadway and quay. The quay would have provided
access to the Islet and been at the heart of the harbour.
Ships could have been moored on either side and off-
loaded their cargoes on to its broad surface before
transportation into the city close by or on to the island
itself.
The structure formed by F9 and F10 to the east of the
platform could have formed a further quay some 35 m
long, and 5 m wide, where boats could be moored along
both sides. Being close to the Islet they could have been
used for loading or unloading sh or other products for
processing on the island.
The sloping parallel line of blocks at F7 may be
interpreted as a 2 m-wide slipway for pulling small
shing boats up on to the quay from the bay (G)
between the Islet and the headland.

Rock cuttings on the Islet


The rock surface on the southern half of the Islet
is rough but approximately level at a water depth of
generally around 0.2 m in calm water. It would be
difficult to determine whether this had been levelled
articially, or as a result of sea action behind the rock
outcrops to the seaward side of the island. There are
certainly outcrops in this atter section that are visible
up to 0.5 m above the water, but the most remarkable
feature is an elaborate layout below current water level
of tanks and slots cut in the bedrock (Fig. 19). While
Figure 17. a) Line of blocks comprising F5; b) header and they vary in the ways in which they relate one to
stretcher blocks on wall F5 looking south from point F5.2. another, they have been created in a clearly deliberate,
(R. A. Yorke) functional manner.

Figure 18. a) Second course of blocks on wall F9; b) spiral-uted column found at point F12.1; c) stone anchor found at point
F12.3. Scale intervals 0.10 m. (R. A. Yorke)

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 63
NAUTICAL ARCHAEOLOGY, 46.1

Figure 19. Detailed plan of the submerged tanks on the Islet. (R. A. Yorke and D. P. Davidson)

64 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

The most striking features are two long channels of 0.3 m. This tank connects into tank F66 at a depth
(F51 and F52) cut into the rock that run at an acute of 0.6 m via a long narrow channel (F82), showing
angle to each other towards the sea to the north, that water could ow from F83 to a pair of tanks
converging from 25 m apart in the south to 15 m apart F66 and then into F60 via a rock-cut sluice with eight
in the north, without appearing ever to have intersected. slots (F65).
The more western of the two (F51) runs almost due Along the east side of the main axis of the cruciform
north from the edge of the rock shelf in the south. It array (F60F63) is a further line of tanks (F68-F77) all
is 1.8 m wide with a slight kink at halfway and slopes at levels between 0.8 m and 0.6 m. None of them show
from a depth of 0.7 m at the south end to 0.9 m where it communication with each other at this level. Opposite
becomes indistinct in the north after 75 m. The second the sluice F65 in tank F60, and adjacent to tank F68 a
of these two angled channels (F52) terminates at a sill and a slot have been carved into a stone ag (F67)
different point in the edge of the rock shelf F50 in the let into the bedrock, of unknown purpose.
south, is the same width, and appears to be cut at a more At the base of the cruciform array is a further
constant depth of 0.9 m. separate system of tanks (F88F92), which also show
Between these two channels there are no tanks cut, blocks attached in situ (F94 and F93), perhaps part of
but several linear rock ridges (F55) have been left some superstructure.
protruding from the rock surface, which therefore must The foundation levels of the rock cutting all lie within
have been articially levelled in that area. A channel a range of up to 1 m below present sea-level. They are all
0.2 m wide (F54) leads at an angle from the eastern awash at all tidal conditions, but must have been 12 m
channel (F52) to the edge of the rock shelf to the south. above sea-level in antiquity. This conclusion, of course,
A further channel (F53), 1 m wide, cuts across the two raises the question of when in antiquity the change in
long channels (F51 and F52), but at a slightly shallower relative sea-level took place, and when the complex was
depth. This peters out at its eastern end, but, where constructed and went into, and out of, use.
it crosses the larger channel (F51) to the west, it is
barred by a rock-cut system of slots in the rock before Surrounding wall
continuing into a system of tanks (F80 and F81). F81 On the south edge of the rock-cut channel F14 there is
is notable in this location in having an array of part- a line of 15 masonry blocks (F11) that runs westward
cut blocks still engaged in the rock, although a similar from the edge of F8 for 12 m following the curved line
feature is seen on the west side of the headland (Fig. 4). of F14 (Figs 12 and 13). This line of blocks is clearly
Further along the length of the western channel associated with structure F14. Given the position of
(F51) is one small tank (F56) with which it F14 following the edge of the island, it is likely that
communicates at a constant level, and a second this channel was cut to house the footings of a seawall
one with one curved end (F57), from which it is clearly that ran round the western side of the island to protect
separated. A channel leads off F51 around the curved the interior from incoming seas driven by the prevailing
edge of tank F57 leading to a cruciform tank (F58) with NW wind. In its position right on the edge of the rock-
side tanks (F59). The western end of the cruciform tank cut foundation, F11 could have formed part of the
terminates crudely and next to the rocky outcrop (F93) ashlar-block facing of that wall.
that is adjacent to an area of rectangular quarrying. To the north-east of the rock-cut channel F14, there
Nearby the single small tank (F61) is not connected. are intermittent signs of the channel continuing (F16),
Nor are two further tanks to the south-east, (F78 and and along the line there is a considerable amount of
F79). blocks and masonry. At F16.1 the channel re-appears
Along the main axis of the cruciform array extends and within it there is a course of blocks comprising
a collinear set of four tanks (F60F63) at decreasing headers and stretchers c.2 m wide that continues for
depths, so that the southernmost (F63) is at the highest 12 m. This would appear to conrm that the rock-cut
level of 0.3 m depth. This tank, and the only one channel, and the sea wall within it, extended all the
apparent in the entire system, shows extant signs of opus way along the southern side of the Islet from F14.1
signinum waterproof cement lining (F64) conrming its on the west to F16.2 on the east. As well as providing
use as a holding tank for liquids. This run of tanks ends protection from the sea, a wall along the south of the
in a rock outcrop (F86) that has rectilinear cuttings island would have provided security for activities on the
in it, which appear to be un-detached blocks (F85), Islet itself.
and a further eroded tank adjacent (F87). There is
no indication how many others of the tanks might Fish processing on the Islet
have had opus signinum lining, and this clearly would The remains of rock-cut trenches from F14.1 to F16.2
inuence the interpretation of the levels and what (Fig. 12), and associated external blockwork, could
structure there might have been above the existing rock be interpreted as the foundation trench for a wide
foundations. protecting wall that extended around the whole island,
Aligned with this, but unconnected, are two further although all signs on the seaward side presumably have
tanks (F84 and F83). Tank F84 is separate and at a been lost to erosion. Given the extensive nature of the
depth of 0.7 m, while the other, F83, is at a higher level tank constructions on the island there would have been

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 65
NAUTICAL ARCHAEOLOGY, 46.1

a need to protect this industrial activity from the waves waters edge (Felici, 2012). The parallel with the tunny
breaking on to the island, and also provide security shing at Pachino with its lookout tower is particularly
from unauthorized access. interesting in view of the descriptions we have of life in
The features found on the Islet are very remarkable Cyrenaica and Ptolemais, in particular in the writings
and appear to constitute the remains of a sh- of Synesius. One of the letters of Synesius (Ep. 57) talks
processing factory with the series of rock-cut tanks about a tunny lookout tower (thynnoskopeion) for the
accurately cut in carefully sequenced levels. These prospection for incoming schools of sh at Ptolemais.
workings have been reported elsewhere as quarries The facilities at Pachino were found with what seem
(Beltrame, 2012: 316), but this detailed survey reveals to be the foundation sockets for a traditional lookout
something a great deal more purpose built. tower (Felici, 2012: 116, g 20). Such towers indeed
There are in excess of 30 tanks in the array, plus survive to this day in Croatia (Anon, nd).
associated channels and sluices. The current differences There seems to be little doubt that in some way
in levels between the bases of the tanks could indicate the facilities at Ptolemais were, at least in the time of
that there was a sequence of ow between sets of tanks Synesius in the late 4thearly 5th century AD (Ep. 57),
as at Kenchreai, one of the ports of Corinth (Scranton used in the harvesting of pelagic sh such as tunny
et al., 1978). The overall arrangement is highly (Wilson, 2004: 148). Quite how the tanks were used
suggestive of what the Roman author Varro (de Re and what role each one played in the processing
Rustica III 17.4) calls a paint-box of tanks (piscinae presumably for saltingis not easy to visualize at
loculatae). Unfortunately we have no graphical present, but it represents an important addition to
representations of anything remotely comparable, even our growing knowledge of the shing economy in the
in Higginbothams comprehensive study of the sh classical world.
tanks (piscinae) of Italy (Higginbotham, 1997).
Tank F83 owing into F66 and then through a sluice
into F60 is an example of sequence of ow. However,
at their base, or foundation level, there are few other Sea-level changesthe evidence
interconnecting channels between tanks in the spine There are a number of areas throughout the site that
of the cruciform array F60F63 and F68F77. On the indicate there has been a relative sea-level rise since
contrary, there are clear ows into the long channel F51 the harbour was built in the Hellenistic period. Mostly
at F53 and F57 and it is possible to conclude from the these are construction features that would have been
falling levels that the channel F51 owed northward built on dry land and are now below the water. There
into the sea beyond the harbour. is only a small tidal movement of c.0.20.3 m in the
At least one of the tanks on the Islet was lined Mediterranean, but in order to remain dry against tide
with waterproof opus signinum type mortar, and it and wave action it can be postulated that the surfaces
is reasonable to assume that they had signicant of quays or quarries would therefore have needed to
walls built up on their rock-cut foundations to give be between 0.5 m and 1 m above the ancient sea-level
them reasonable holding capacity. However, the precise (Table 1). However, the depths for examples 2, 4 and
nature of how the process worked is not clear to the 5 in Table 1 were taken to the tops of the blocks, and
authors from the remains that exist. No precise parallels so it is necessary to add the thickness of these blocks
have been found for such a complex elsewhere in the (c.0.3 m) to reach the foundation level that would have
classical world. had to be dry.
The nearest parallel might be seen at a 2nd4th On the assumption that the deepest place would have
century AD site at Pachino in Sicily, where there are been dry in antiquity it is reasonable to assume that
signs of extensive tunny shing and the vats for salting the sea-level has risen by between 2 and 2.5 m since the
the processed sh are cut in the rock close to the harbour was rst in use.

Table 1. Evidence for sea-level rise at Ptolemais

Estimated Sea-level Rise (m)

No. Feature Current Submersion (m) Minimum Maximum

1 Quarry oor on Ilus 0.7 1.2 1.7


2 Top of slip F7.1 0.8 1.3 1.8
2 Top of slip F7.1 adjusted 1.1 1.6 2.1
3 Channel F51 on the Islet 0.9 1.4 1.9
4 Top of blocks F5.2 1.5 2.0 2.5
4 Top of blocks F5.2 adjusted 1.8 2.3 2.8
5 Top of blocks F2 1.2 1.7 2.2
5 Top of blocks F2 adjusted 1.5 2.0 2.5

66 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

Figure 20. Schematic reconstruction of Ptolemais harbour. (R. A. Yorke and D. P. Davidson)

Discussion and conclusions headland would have been used by the local shing
and coasting eet (Fulford, 1989), for vessels the size
The harbour of the contemporary Kyrenia ship (Casson, 1994). The
The survey of 1972 conclusively demonstrated the main harbour basin of 7 hectares would have provided
layout of the ancient harbour and recent research has shelter and anchorage space for a large number of ships
added further insight into how the original installations after arrival and awaiting off-loading, although Rouge
would have operated. Strong evidence has been found concluded that it was never part of the Roman Annona
to support a proposition that sea-level at Ptolemais has system (Rouge, 1966). The size of this basin is the same
risen by 22.5 m since antiquity. This is comparable as that of Apollonia just along the coast, where the
with other sites along the coast of Cyrenaica as far inner, military harbour was 5 hectares, slightly larger
as Apollonia and beyond (Tusa, 2009: 191216; De than the smaller basin at Ptolemais. The entrance to
Graauw, 2012, 2014). the harbour was between the two easterly moles, one
The research has shown that the harbour consisted of running south from the small island called Ilus, spoken
two basins located to the east of the headland (Fig. 20). of in the 2nd century AD in the Stadiasmus Maris
The smaller basin of about 2.5 hectares under the Magni, and a short mole facing it running north from

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 67
NAUTICAL ARCHAEOLOGY, 46.1

the shore. On the southern tip of the mole from Ilus water as an adjunct to the processing (Higginbotham,
there are signs that there could have been a lighthouse 1997: passim).
or beacon marking the entrance. The headland would have had no less signicance to
The main harbour basin would have been enclosed shipping in antiquity than it does now, providing shelter
from the north by a breakwater between Ilus and the in bays to the east or the west according to the weather.
Islet, which would have protected it from the prevailing Undoubtedly, there would have been a light tower at its
wind and sea. There would have been a road along northern end, judging by the ancient foundations that
the breakwater to provide access to the stone quarries appear in the footings of the modern lighthouse. It is
on Ilus, the lighthouse and to ships moored in the lee possible that it also served as a shing lookout tower
of the island. Kraeling noted in the 1960s that larger for the eet moored in the bay below.
vesselsshing eet leaderssheltered under the lee of The main use of the headland, however, appears
this easterly island. to have been as a large quarry for stone to build
The westerly island, the Islet, would have been the town and the harbour. The conclusion has also
connected to the mainland by a wide quay built in been reached that what some commentators have
ashlar masonry, quite analogous to the road pavements described as slipways, or shipsheds, on the western side
on shore at Ptolemais, and at neighbouring Tauchira of the headland are simply rock-cut quarries, which
(Bogacki, 2012: 83). Ships in the main easterly basin continued out seaward, and could never have been
would have docked against this quay for loading and used to launch vessels because of their height above
unloading, as well as at a separate quay by the island. ancient sea-level and their exposure to the prevailing
The smaller second basin lay between the Islet and sea.
the headland with direct access northwards to the
sea. This basin was probably home to the local shing Beecheys wall
eet, as it is today, and it is reasonably sheltered. In The open bay to the west of the headland, while useful
extremis, boats could have been drawn up on to the as a shelter when the weather is from the east, was
beach. The shing eet could have landed their catch certainly not the location of the main harbour as some
on to the main quay for further sale or processing. commentators have suggested, because it is too exposed
There are the remains of what appears to be a small to the prevailing wind and sea. Beecheys wall at the
slipway at the north-western end of the quay that could back of the beach was found to be set at a height that
have been used to pull smaller boats on to the quay for would always have been signicantly above sea-level at
maintenance. any historic period, and was not part of the harbour
The Islet itself would have been encircled by a works. The most plausible interpretation is that it was
substantial wall, at least 2 m wide, to protect it both some sort of esplanade along the beach between the
from the sea and from unauthorized access. Within town and the wadi.
these walls there appears to have been a sizeable sh- The style of constructionoutside faces widely
processing factory. Fish landed by the local eet on separated, interior divided into squarish rubble-lled
the west of the quay could have been processed on the voidsis very similar to the wall foundation either
island, and then what was not for local consumption side of the Tocra Gate, and would therefore appear
would have been loaded for export on to larger to be contemporary with the Hellenistic defensive
ships in the main basin. This interpretation of the circuit. The nish of the inner face is also neat work,
complex of rock-cut tanks on the Islet as a sh factory quite unlike the Byzantine works at Ptolemais that use
demonstrates the importance of these remains and second-hand materials that do not t together very well.
provides a reason why there was a need for such a The conclusion may be drawn that Beecheys wall is
substantial quay connecting the Islet to the mainland. more likely to be Hellenistic than later (Philip Kenrick
It also potentially shows that Ptolemais was a more and David Blackman, pers. comm.).
important harbour than had previously been supposed,
and that, from an early date, it possessed a sophisticated Relative dating and phasing
factory for the processing of tunny, along the lines of Surface survey and recording did not produce
that seen at Pachino (Felici, 2012). any denitive dating evidence for the phasing and
It may not be entirely co-incidental that the quay development of the harbour. However consideration
connecting the Islet to the shore appears to align of historical evidence and the practical necessities of
with the position where the aqueduct running down operating a harbour at Ptolemais can help to provide a
through the city could have arrived at the shore. This possible chronology.
is clearly shown in the Beecheys plan of the site, but In the 6th century BC Ptolemais was reported as
not by Kraeling. If this is an intentional adjunct to being the harbour for Barca and the city itself developed
the resources provided on this island, either for the extensively in the Hellenistic period from 322 BC. The
provisioning of boats or as a part of the processes taking Stadiasmus Maris Magni of c.1st century BC mentions
place there, it would have fundamental implications for the harbour, but with just the one island that it calls Ilus.
the place of the island in the laying-out of the Hellenistic It would seem reasonable to assume that the harbour
city. A number of major sh facilities in Italy used fresh developed from the 6th to the 1st century BC to make

68 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

it a safe anchorage and loading point for ships in all It is, however, the position of the tanks on the island
weathers. that is important. They have been cut into the rock
At that time, when relative sea-level was 22.5 m at the point nearest to land whereas all the extensive
lower, there may originally have been shallow sand bars general quarrying activity is less accessible and beyond
from the shore to the Islet, and between the Islet and them. It would seem likely therefore that this area was
Ilus. It would still, however, have been necessary to reserved for some sort of commercial activity from
build a stone causeway to the Islet to provide quays for the earliest stage of the development of the Islet. It
mooring and unloading ships, and to provide access to remains a possibilty that there could also have been sh-
the quarries on the Islet. It is also likely that walls F4 processing facilities here in the Hellenistic period, which
and F5 were part of the early or Hellenistic construction were then further developed by the Romans.
as they gave access directly to the Islet and would have Despite the destruction recorded in Cyrenaica
provided quay facilities. especially in Cyrenefrom the earthquakes of AD
In addition it would have been necessary to build a 264 and of AD 365 there is actually no evidence that
stone breakwater between the Islet and Ilus to protect the harbour at Ptolemais underwent any fundamental
the anchorage from the prevailing north-westerly seas change in use or deterioration from the 1st to the 6th
and northerly storms, without which the harbour would century AD. Diocletian had made Ptolemais the capital
have been untenable. It is therefore likely that stone of Cyrenaica under the reforms of AD 296 and there
quarried from both islands would have been used to is circumstantial evidence that Ptolemais was made
make the initial sea defences. The harbour would still capital of the province of Creta and Cyrenaica after the
have remained exposed to storms and seas from the east earthquake of AD 365 critically damaged the previous
and the north-east and so it is likely that the eastern capital, Gortys, in Crete. In the early 5th century AD
breakwater, Mole C, would have been added during this Synesius reports on-going shing activity and trade at
period using stone quarried from Ilus. Ptolemais.
In this development Ptolemais would not have been None of this would have happened if the harbour at
alone. During the Hellenistic period the harbour at Ptolemais had not possessed the facilities required of a
Apollonia was built in a similar fashion. The two coastal city of some importance.
offshore islands there were connected with a stone
breakwater. Quays were built out from the land to When did submergence occur?
the islands, with buildings and slipways on them and The precise period in which this submersion took place
an eastern breakwater with a lighthouse at its end is not immediately clear. Strong earthquakes causing
protected the entrance to the harbour (Flemming, severe damage as far as Cyrene are known to have been
1971: 99, g. 14; Taylor, 1965: 173). The conguration experienced in the locality in AD 264 and AD 365, but
of the harbour at Ptolemais is thus very similar to there is no visible evidence at Ptolemais of destruction
Apollonia and so it is not unreasonable to assume that under water, neither is there at Apollonia (Laronde,
similar designs and technologies could have been used 1996: 349). It is believed that commercial activity
contemporaneously at Ptolemais, on a smaller scale, but continued through the time of Synesius and the reign
with harbour basins of similar areas. of Justinian, but by the time the rst Europeans visited
Cyrenaica and the cities of the Pentapolis passed into the area in the 18th century, the harbour works were
Roman hands on the death of Ptolemy Apion in 96 BC, entirely below water. Laronde was of the opinion that
and were in a state of neglect from their Hellenistic both Ptolemais and Apollonia continued as working
heyday. It was probably not until the 1st century AD administrative centres beyond the 7th century (1996:
that much new public building took place. It seems 47). It is a plausible possibility, for which, however,
logical that expansion of the Hellenistic harbour would there are no witnesses, that the submergence of the
have commenced in this period to provide the facilities harbour could have occurred in the earthquake in AD
that the growing city required. If not already built, the 796, during the long silent period following the Muslim
causeway to the Islet (F) could have been expanded incursion in AD 642. This huge earthquake in the
to include F2 and F3, providing a large working quay Hellenic Arc and centred on Crete (Ambraseys et al.,
for loading and unloading ships, and the additional 2005: 26) did much damage at Alexandria and was
quay F9 would presumably have been added at that probably also responsible for the famous submergence
time. of Thonis-Heraclion and Alexandria Portus Magnus
Causeway F also provides access to the important in the Nile delta. It is also the case that there was
sh-processing facilities on the Islet. The fact that a multitude of smaller seismic events in unrecorded
these are rock-cut tanks would seem to place them medieval history that may well have caused a gradual
technologically with those at Matala in southern Crete sinking of the coastal area (Synolakis et al., 2014).
(Mourtzas, 2012: 2392ff), which belong to the 1st Ptolemais appears in the Cosmographia of Ravenna
century AD and the period when Crete and Cyrenaica of around AD 700 and the later re-work by Guido
were conjoined as a Roman province. The presence of of Pisa in AD 1119 (Campbell, 1987); but not
opus signinum in one of the tanks seems to signify a in Abd-el Hakm (Leuven, nd). We nd references
Roman date. in el-Bekri (Slane, 1859) in the 12th century, and in

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 69
NAUTICAL ARCHAEOLOGY, 46.1

Aboulfeda (Reinaud and Guyard, 1848) in the 13th (portolani) (Nyssen, 2010), but none of the authors seem
century, but without clear evidence that the place was to have visited the site, any more than apparently did the
more than a sheltered anchorage for light shipping. originators of the Peutinger Itinerary or the author of
Kennet bases his evidence for Islamic presence on a the Cosmographia. Otherwise there is no evidence that
few (27) sherds, mainly of the 12th century (Kennet, Ptolemais had any signicant commercial role after the
1994: 275). According to Kennet (1994: 278279), al- Arab invasion in the 7th century AD.
Yaqubi mentions three ports in 9th-century Cyrenaica;
Tolmeita, Benghazi and Ajdabiya, 6 miles south of Conclusion
Benghazi, of which only Benghazi seems to have been This underwater survey has shown that the important
able to offer ships good protection, while Tolmeita was Cyrenaican city of ancient Ptolemais possessed
usable only at certain seasons of the year. a sophisticated commercial harbour installation
The 15th-century map and text of Benincasa comparable even with that of Apollonia further along
referring to medieval Ptolemais (Brentjes, 2012) cited by the coast. This is attested by the extensive letters of
Rekowska-Ruszkowska (Rekowska-Ruszkowska, 2012: Bishop Synesius in which he portrayed the busy life
15) supposedly following the 11th-century Al-Idrisi and commerce at Ptolemais in the 5th century AD.
(Dozy and Goeje, 1866) may be incorrect. It seems, However, the historical sources appear to show that the
through some possible medieval transposition, to be harbour became of less signicance by the 9th century
referring to Ptolemais in Palestine, to which this and so it is not unreasonable to suppose that the
description bears a remarkable resemblance (see plate present-day submergence that brought its commercial
in Calmet, 1720, 470 Vol. II. E). Ptolemais appears life to an end could have been caused by the earthquake
on many of the lists of ports of the Middle Ages of 796 AD.

Acknowledgements
Our thanks are due to the Society for Libyan Studies who sponsored the initial project, Faculty of Classics, University of
Cambridge, the Russell Trust who provided nancial support, the Honor Frost Foundation who funded the digitisation of the
maps and plans, Andrew Yorke who prepared the plans, and to Awad Sadawya, Abdul Hameed and Budzama Abdusalem and the
contemporary Libyan Department of Antiquities. Thanks are also due to two anonymous referees who made valuable comments
that have resulted in improvements in the text.

References
Ambraseys, N. N., Melville, C. P. and Adams, R. D., 2005, The Seismicity of Egypt, Arabia and the Red Sea: A Historical Review.
Cambridge.
Anon, nd, Big game shing in CroatiaTuna hunting. http://www.nautilus.hr/tuna_hunting.htm, (accessed 2/7/2014).
Arena, G., 2011, Siti costieri ed attivita produttive nella Cirenaica tardoantica. Babesch 86, 167184.
Aubet, M. E., 2001, The Phoenicians and the West: Politics Colonies and Trade. Cambridge.
Baika, K., 2014, Ptolemais in D. J. Blackman and B. Rankov (eds), Shipsheds of the Ancient Mediterranean, 494500. Cambridge.
Barth, H., 1849, Wanderungen durch die Kustenlander des Mittelmeeres, ausgefuhrt in den Jahren 1845, 1846 und 1847. Wilhelm
Hertz.
Beechey, F. W. and Beechey, H. W., 1828, in Proceedings of the expedition to explore the northern coast of Africa, from Tripoly
Eastwards, in 1821 and 1822, 355, 376385. London.
Beloe, Rev. W., 1825. Herodotus, translated from the Greek, with Notes. London.
Beltrame, C., 2012, New Evidence for the Submerged Ancient Harbour Structures at Tolmetha and Leptis Magna, Libya. IJNA
41.2, 315326.
Boardman, J., 1966. Evidence for the Dating of Greek Settlements in Cyrenaica. The Annual of the British School at Athens 61,
151152.
Bogacki, M., 2012, Archaeological kite aerial photography in Ptolemais in years 20062008, in J. Zelazowski (ed.), Ptolemais in
Cyrenaica. Studies in memory of Tomasz Mikocki, Ptolemais 1, 7791. Warsaw: Institute of Archaeology University of Warsaw.
Brentjes, J., 2012, Medieval portolan charts as documents of shared cultural spaces, in R. Addellatif, Y. Benhima, D. Konig and
E. Ruchaud (eds), Acteurs de transferts culturels en Mediterranee medievale, 135147. Munich.
Calmet, A., 1720, Dictionnaire historique, critique, chronologique, geographique et litteral de la Bible. London.
Campbell, T., 1987, Portolan Charts from the Late Thirteenth Century to 1500, in J. B. Harley and David Woodward (eds), The
History of Cartography. Volume 1, 371463. Chicago.
Casson, L., 1994, Ships and Seafaring in Ancient Times. London.
Cretu, D., 2011, The Silphium Plant and Cyrene Perfume. Transylvanian Review XX (2:1), 5566.
Davis, D. L., 2009, Commercial Navigation in the Greek and Roman World. Austin, TX.
De Graauw, A., 2012, El Hanieh ancient anchorage in Ancient Ports online database. http://www.ancientportsantiques.
com/a-few-ports/el-hanieh-ancient-anchorage/ (accessed 6/10/2010).

70 2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society.
R. A. YORKE & D. P. DAVIDSON: SURVEY OF THE HARBOUR AT PTOLEMAIS

De Graauw, A., 2014, Apollonia in Ancient Ports online database http://www.ancientportsantiques.com/a-few-ports/apollonia/#1


(accessed 6/10/2016).
Dozy, R., and Goeje, M. J. de, 1866, Geographie del Idrisi. Description de lAfrique et de lEspagne. Leiden.
Felici, E., 2012, Un impianto con thynnoskopeion per la pesca e la salagione sulla costa meridionale della Sicilia (Pachino, SR).
Eliano, Oppiano e la tonnara antica, in E. Felici, P. Gianfrotta, L. Quilici, S. Gigli, and E. Tortorici (eds), Topografia Antica
2: Tradizione, Tecnologia e Territorio, 107143. Rome.
Flemming, N. C., 1971, Cities in the sea. Garden City, NY.
Fulford, M. G., 1989, To East and West: the Mediterranean Trade of Cyrenaica and Tripolitania in Antiquity. Libyan Studies
20, 169191.
Hesein, M. A. H., 2014, Ancient marine resource exploitation in the coastal strip of Cyrenaica: Some evidence of sh-related
industry in Cyrenaica, in V. Leitch and E. Botte (eds), Fish and Ships: Production and commerce of salsamento during Antiquity,
129141. Paris, Aix en Provence.
Higginbotham, J., 1997, Piscinae: Artificial Fishponds in Roman Italy, Studies in the History of Greece and Rome. Chapel Hill:
NC.
James, P., 2005, Archaic Greek colonies in Libya: historical vs. archaeological chronologies? Libyan Studies 36, 120.
Jones, G. D. B. and Little, J. H., 1971, Coastal Settlement in Cyrenaica. Journal of Roman Studies 61, 7179.
Kennet, D., 1991, Some Notes on Islamic Tolmeita. Libyan Studies 22, 8389.
Kennet, D., 1994, Pottery as Evidence for Trade in Medieval Cyrenaica. Libyan Studies 25, 275285.
Kenrick, P. M., 2013, Cyrenaica (Libya Archaeological Guides). London.
Kraeling, C., 1962, Ptolemais, City of the Libyan Pentapolis. Chicago.
Laronde, A., 1981, Variations du niveau de la mer sur les cotes de Cyrenaque a lepoque historique. Les Dossiers de lArcheologie
50, 6065.
Laronde, A., 1986, Les ports de la Cyrenaque : Ptolemas et Apollonia, in LAfrica Romana Atti del III Convegno di studio,
Sassari 1985, 167177. Sassari.
Laronde, A., 1996, Apollonia de Cyrenaque: Archeologie et Histoire. Journal des savants, 1996.1, 349.
Lehmann-Hartleben, K., 1923, Die Antiken Hafenanlagen des Mittelmeeres. Klio XIV, 6774.
Leuven, K. U., nd, Trismegistos interdisciplinary portal of papyrological and epigraphical resources. http://www.trismegistos.
org/index.html (accessed 27/10/2016).
Mediterranean Pilot: v. 5 (Admiralty Sailing Directions), 2005, London: UK Hydrographic Office.
Mikocki, T., Yacoub, G. and Lewartowski, K., 2010, Archaeological research in Ptolemais (Cyrenaica) in years 20012007. Libya
Antiqua 5, 99107.
Mourtzas, N. D., 2012, Fish tanks of eastern Crete (Greece) as indicators of the Roman sea level. Journal of Archaeological
Science 39.7, 23922408.
Mueller, K., 2004, Dating the Ptolemaic city-foundations in Cyrenaica. A brief note. Libyan Studies 35, 110.
Muller, C. F. W., 1855, Stadiasmus Maris Magni (Africa section), in Geographi Graeci Minores, Vol 1, 427514.
Nyssen, M. W., 2010, From Varro to Peutinger. www.semafoor.net/actueelSEMsiteTP-RavDOCeng.pdf (accessed 28/10/2016).
Pacho, J.-R., 1827, Relation dun voyage dans la Marmarique., 2 Vols., Paris.
Reinaud, J. T. and Guyard, S., 1848, Geographie dAboulfeda. Paris.
Rekowska-Ruszkowska, M., Kijak, A., 2016, In pursuit of ancient Cyrenaica: two hundred years of exploration set against the
history of archaeology in Europe (17061911). Oxford: Archaeopress.
Rekowska-Ruszkowska, M., 2012. Ptolemais in early European research and topography of the city in travellers accounts, in J.
Zelazowski (ed.), Ptolemais in Cyrenaica. Studies in Memory of Tomasz Mikocki. Ptolemais I, 134. Warsaw.
Rouge, J., 1966, Recherches sur lorganisation du commerce maritime en Mediterranee sous lEmpire romain. Paris.
Scranton, R. L., Shaw, J. W. and Ibrahim, L., 1978, Kenchreai, Eastern Port of Corinth: Vol 1: topography and architecture. Leiden.
Slane, M. G. de (transl), 1859, Description de lAfrique septentrionale par El-Bekri. Paris.
Synesius. (d. AD 413). Letters. Retrieved from http://www.livius.org/susz/synesius/synesius_letters.html (accessed 6/10/2016).
Synolakis, C. E., Findikakis, A. N., Valle, B. L., Kalligeris, N., Okal, E. A. and Melilla, L., 2014, Plausible megathrust tsunamis
in the eastern Mediterranean Sea. Proceedings of the ICE - Engineering and Computational Mechanics 167.3, 99105.
Taylor, J. du P., 1965, Marine Archaeology: Developments During Sixty Years in the Mediterranean. London.
Tusa, S., 2009, Rapporto preliminare della missione italiana per lo studio dellarcheologia costiera. https://www.academia.
edu/10004545/ (accessed 20/8/2015).
Varro, Marcus Terentius, de Re Rustica, Liber III., in W. D. Hooper and H. B. Ash, (trans.), 1934, Cato and Varro on Agriculture.
Loeb Classics, London.
Wilson, A., 2004, Cyrenaica and the Late Antique Economy. Ancient West and East 3, 143154.
Yorke, R. A., 1972, A Survey of Ancient Harbours in Cyrenaica. Libyan Studies 3, 34
Yorke, R. A. and Davidson, D. P., 1973, Tolmeita and TocraRoman Harbours of Cyrenaica, in Davis, M. (ed.), Underwater
Association Minisymposium, 2023. http://www.academia.edu/5180894/Tolmeita_and_Tocra_-_Roman_Harbours_of_
Cyrenaica_Yorke_and_Davidson._UA_Minisymposium_1973 (accessed 27/10/2016).
Yorke, R. A., Davidson, D. P. and Little, J. H., 1972, Pentapolis Project 1972A Survey of Ancient Harbours in Cyrenaica
interim report. Unpublished report.
Yorke, R. A., Davidson, D. P. and Little, J. H., 1973, A survey of ancient harbours in Cyrenaica. IJNA 2, 200201.
Zelazowski, J., 2011, Polish Archaeological Research in Ptolemais (Libya) in 2007-2009. Preliminary Report, in Annual of the
Institute of Archaeology of University of Warsaw VIII, 928. Warsaw.

2017 The Authors. International Journal of Nautical Archaeology 2017 The Nautical Archaeology Society. 71

You might also like