Maritime Archaeology and Ancient Trade in the Mediterranean
Edited by Damian Robinson and Andrew Wilson
Edited by Damian Robinson and Andrew Wilson
9 781905 90 5 1 7 1
Published by the Oxford Centre for Maritime Archaeology
at the School of Archaeology, University of Oxford
This monograph comprises of twelve papers that look at the shifting patterns of
maritime trade as seen through archaeological evidence across the economic cycle of
Classical Antiquity. Papers range from an initial study of Egyptian ship wrecks dating
from the sixth to fifth century BC from the submerged harbour of Heracleion-Thonis
through to studies of connectivity and trade in the eastern Mediterranean during the
Late Antique period. The majority of the papers, however, focus on the high point in
ancient maritime trade during the Roman period and examine developments in
shipping, port facilities and trading routes.
Oxford Centre for Maritime Archaeology Monographs
Maritime Archaeology and Ancient
Trade in the Mediterranean
List of Illustrations xi
List of Tables xvii
List of Abbreviations xix
List of Contributors xxi
Introduction: Maritime archaeology and the ancient economy 1
Andrew Wilson and Damian Robinson
1. e shipwrecks of Heracleion-onis: a preliminary study David Fabre 13
2. Developments in Mediterranean shipping and maritime trade from the Hellenistic 33
period to AD 1000 Andrew Wilson
3. Ancient sailing-routes and trade patterns: the impact of human factors Pascal Arnaud 61
4. Ceramic assemblages and ports Candace Rice 81
5. Constructing port hierarchies: harbours of the central Tyrrhenian coast Katia Schörle 93
6. Technology, innovation, and trade: research into the engineering characteristics 107
of Roman maritime concrete John Oleson, Christopher Brandon and Robert Hohlfelder
7. Heracleion-onis and Alexandria, two ancient Egyptian emporia Franck Goddio 121
8. Lapis transmarinus: stone-carrying ships and the maritime distribution of stone 139
in the Roman Empire Ben Russell
9. Dolia shipwrecks and the wine trade in the Roman Mediterranean Karen Heslin 157
10. Location, location, location: characterizing coastal and inland production and 169
distribution of Roman African cooking wares Victoria Leitch
11. A reconstruction of the maritime trade patterns originating from western Asia Minor 197
during Late Antiquity, on the basis of ceramic evidence eodore Papaioannou
12. Maritime connectivity in Late Antique Lycia: a tale of two cities, Aperlae and Andriake 211
2: Developments in Mediterranean
shipping and maritime trade from the
Hellenistic period to AD 1000
is paper attempts an overview of developments
in maritime trade between 200 BC and AD 1000. It
looks ﬁrst at what the totality of shipwreck evidence
may say about trade, then moves on to identify some
relationships between ship design and construction and
maritime trade, and compares these to parallel develop-
ments in harbour technologies.
Shipwrecks and maritime trade
Any survey of maritime trade over the longue durée
must be heavily indebted to A. J. Parker’s catalogue of
1,189 Mediterranean wrecks datable before AD 1500.1 His
graph of the number of wrecks per century (Figure 2.1)
was seized on by ancient historians from Hopkins onward
as a means of illustrating ﬂuctuations in the levels of
Mediterranean maritime trade, taking the number of
wrecks in any period as a more or less straightforward
reﬂection of the intensity of maritime shipping and
therefore trade, and has featured in numerous discussions
of the Roman economy, being made to bear a far greater
interpretative superstructure than Parker himself had
ever intended.2 Parker’s graph shows a progressive
increase in the number of known wrecks from about 600
BC to 200 BC, followed by a rapid rise to peak in the ﬁrst
century BC, dropping very slightly in the ﬁrst century
AD, rather more so in the second century AD, and then
sharply in the third century, with continued diminution
in the fourth to ﬁfth centuries. A slight recovery in the
sixth century3 is followed by further sharp drops and
thereafter wreck numbers before 1500 do not regain even
the levels attained in the sixth century AD. Overall, the
Classical, Hellenistic, Roman and Late Antique periods
stand out as having much larger numbers of wrecks
than the Bronze Age or the Medieval period, and the last
two centuries of the Roman Republic and the ﬁrst two
centuries of the empire have exceptionally high numbers
of known wrecks.4
Figure 2.1. Mediterranean shipwrecks by century (n=1189), graphed using the mid-points of date ranges. (After Parker 1992a: ﬁg. 3.)
1 Parker 1992a.
2 Hopkins 1980: 105–6 (using an early version); Gibbins
2001: ﬁg. 10 (but n.b. pp. 273–83 on the strengths and
weaknesses of wrecks as evidence for trade); de Callataÿ
2005; Davies 2006: 84–5; Jongman 2007: 188; Morley
2007a: 572–3; 2007b: 98 (with caveats about under-rep-
resentation of African trade). Whittaker 1989 pointed out
many of the problems in using these shipwreck data to
chart changes in trade over time. See Parker 1992a: 8–9
for a discussion of the wreck statistics; Parker has since
published an updated version of the graph (Parker 2008:
187 ﬁg. 12).
3 Parker 1992a: 8.
4 As Parker 1992a: ﬁg. 4 (presenting the chronological dist-
ribution of wrecks by periods in a pie chart) and discussion
on p. 8 showed very clearly.
simply ‘Roman period’ (150 BC–AD 400) is to count each
instance as 0.09 of a wreck in the second century BC and
0.18 of a wreck per century in each century from the ﬁrst
century BC to the fourth century AD. ese wrecks thus
account for a total of 6.8 wrecks in the second-century
BC column and 13.7 wrecks in each of the columns for
the ﬁrst century BC to the fourth century AD.
e overall shape of the graph has changed somewhat;
the gradual increase from 600 to 200 BC is still there, as
is the sharp increase after 200 BC, but the absolute peak
now occurs in the ﬁrst century AD, not BC. e second-
century AD column has now fallen to under half that of
the ﬁrst century AD, level with the third century; this is
the result of spreading those generically ‘Roman’ wrecks
across the six centuries in which they might have sunk,
rather than concentrating them in the second century.
Parker’s catalogue was published in 1992 and, obviously,
more wrecks have been found and published since
then. With the assistance of Dr Julia Strauss, the Oxford
Roman Economy Project updated the wreck database
with new material from the maritime archaeology
literature since 1990.7 is is still work in progress but
has (so far) increased the number of known and datable
Mediterranean wrecks before AD 1500 from 1189 to
1646 (Figure 2.3). e shape of the graph continues
to change, so diminishing returns have not yet set in,
and the accumulation of more data continues to be
worthwhile. e new graph further accentuates certain
features already visible in my regraphing of Parker’s data.
ere is now a steeper climb in the Late Republic to a
more pronounced ﬁrst-century AD peak and the fall
We shall see that there are several reasons why the
graph of known wrecks is not a simple reﬂection of ancient
maritime trade levels—and Parker himself did not assume
it was; indeed, he stressed that the wrecks needed to be
studied on their own terms and drew attention to biases
in reporting, and to archaeological factors (transport of
amphorae, roof tiles and marble cargoes) which might
over-represent Hellenistic and Roman wrecks.5 But ﬁrst
we need to consider a consequence of the method Parker
chose to deal with imprecisely dated wrecks. He graphed
all wrecks at the mid-point of the date range assigned
to them,6 thus a wreck dated 75 BC to AD 25 would
be graphed in the ﬁrst-century BC column because the
mid-point of the range is 25 BC. A considerable number of
the wrecks in Parker’s catalogue were reported simply as
‘Roman’, which he quite reasonably interpreted as between
150 BC and AD 400. e mid-point of this range is AD
125, and a large part of the allegedly second-century AD
column on the graph is accounted for by these generically
‘Roman’ wrecks, which is clearly misleading.
Figure 2.2 uses the data from Parker’s catalogue, but
assumes that there is an equal probability that a ship sank
in any particular year within the date range assigned to a
wreck. is probability is then accumulated for each time
period on the x-axis (in this case, a century). us a wreck
dated 75 BC to AD 25 has a 75 per cent chance of having
sunk in the ﬁrst century BC, and a 25 per cent chance of
having sunk in the ﬁrst century AD; it therefore counts
as ¾ of a wreck in the ﬁrst-century BC column and ¼ in
the ﬁrst-century AD column. e eect of this on the
76 Mediterranean wrecks in the catalogue reported as
Figure 2.2. Mediterranean shipwrecks by century (n=1,189), but graphed according to an equal probability of sinking in any year
during the date range for each wreck. (Data from Parker 1992a.)
5 Ibid.: largely ignored by many subsequent users of his
6 Ibid.: 8.
7 Wilson 2009. The Oxford Roman Economy Project is
directed by Alan Bowman and Andrew Wilson, and is
funded by a 5–year grant from the AHRC and a bene-
faction from Lorne Thyssen; see http://oxrep.classics.
what the 1,646 wrecks look like if distributed across
half-century brackets, a graph which suggests that
whatever the reason for the second-century drop, it was
not the Antonine Plague because the drop occurs already
in the ﬁrst half of the second century. e further drop
in the fourth century does not occur until after AD 350.
is view receives some support from the graph of
only those 596 wrecks which can be dated to within
half-century brackets, graphed by 50-year periods
(Figure 2.6). is is broadly similar, but intriguingly
suggests dips in the second half of the ﬁrst century BC
(during the civil wars but also the subsequent Augustan
peace), and an apparent slight recovery in numbers in
the early fourth century AD.
What relationship do these graphs bear to levels of
maritime trade over time? Figures 2.2–2.6 show a massive
drop in shipwreck numbers between the ﬁrst and second
centuries AD, just when there was major investment in
harbour works at Portus, and just when African Red Slip
exports from North Africa take o. e pan-Mediterranean
to second- and third-century levels looks even more
dramatic; under a third of the number of wrecks of
the ﬁrst-century column. Later centuries look similar to
before: the number of fourth-century AD wrecks is not
far short of the third century, but the number of ﬁfth-
century AD wrecks is less than half that of the fourth
century, and below that even of the third century BC.
ere is another major step-change between the seventh
and eighth centuries AD, with very few wrecks at all
known from the period AD 700–1000. e Roman peak
is therefore more pronounced, but the drop-o after AD
100 is very clear.
We can test the validity of this method of graphing of
all wrecks by stripping out the noisy data—the wrecks
with long date ranges—and comparing Figure 2.3 with
the graph of only those 1,062 wrecks which are datable
to within a century (Figure 2.4). e basic shape does
not change much, inspiring some conﬁdence in the
method of graphing all wrecks. Can we probe the data
further, by using smaller time periods? Figure 2.5 shows
Figure 2.3. Mediterranean shipwrecks by century (n=1,646), graphed according to an equal probability of sinking in any year during
the date range for each wreck. (Data collected by Julia Strauss.)
Figure 2.4. Mediterranean shipwrecks datable within 100-year ranges (n=1,062), graphed according to an equal probability of sinking
in any year during the date range for each wreck. (Data collected by Julia Strauss.)
2: Developments in Mediterranean shipping and maritime trade
to shorter coastal voyages, entailing a dierence in risk
at dierent periods. Russell’s analysis of stone cargoes8
suggests that patterns of wrecking are indeed correlated
with the relative risks of dierent sailing routes that may
in turn be determined by fashions for particular marbles
at dierent periods—in the case of stone cargoes, wreck
numbers peak in the second and third centuries AD when
transport of eastern Mediterranean coloured marbles
to Italy and the central Mediterranean becomes more
common, as these involved the traversing of dangerous
waters o southern Italy and Sicily in a way which the
transport of Luna marble down the coast of Italy from
Carrara to Rome and the Bay of Naples did not. Moreover,
if practices of winter sailing became more common at
dierent periods—as there is some evidence that they did
under the Roman empire9—then this will have increased
the risk of sailing and therefore the likelihood of wreck.
distribution of this tableware shows that maritime trade
links were not declining; yet most of this evidence comes
from terrestrial excavations. ere is therefore something
anomalous about the underwater evidence.
Attempts to equate the graphs of wrecks over time with
ﬂuctuations in maritime trade rest on two fundamental
(1) e probability that any voyage will end in wreck is
the same at all periods.
(2) Wrecks are equally visible archaeologically at all
e ﬁrst of these assumptions might be questioned, if
climatic change is felt to have increased the incidence of
storms at certain periods; or if changes in shipbuilding
technology over time aected (reduced?) the propensity
to wreck; or if there were changes in sailing practice
and the balance of long-distance open-water voyages
Figure 2.6. Mediterranean shipwrecks datable within 50-year ranges (n=596), graphed according to an equal probability of sinking in
any year during the date range for each wreck. (Data collected by Julia Strauss.)
8 Russell, this volume (Chapter Eight). 9 Beresford 2005.
Figure 2.5. Mediterranean shipwrecks by half-century (n=1,646), graphed according to an equal probability of sinking in any year
during the date range for each wreck. (Data collected by Julia Strauss.)
Barrels were an invention of northern Europe, probably
of Celtic origin, and they are referred to in literary sources
from the second half of the ﬁrst century BC, and are
known archaeologically, reused as well linings, from
Roman sites along the Rhine frontier from the late ﬁrst
century BC.13 From the ﬁrst to third centuries AD they
are not uncommonly shown in Roman reliefs, particularly
in Gaul, Germany and Britain, but also in Lusitania
(Portugal).14 By the late third or early fourth century
AD the taris for unloading and handling wine brought
down the Tiber to Rome’s river port called Ad ciconias
nixas specify charges per barrel; there is no mention of
amphorae in this riverine trade.15 Is part of the apparent
sharp drop in the shipwreck graph from the ﬁrst to the
second century AD due to an increasing use of barrels?
But if so, it can hardly account for all of it—amphora
usage in the Mediterranean remained common until the
sixth or seventh centuries.
Another explanation worth considering, at least as
a partial factor, for the apparent mismatch between
the underwater and terrestrial evidence is the lack of
modern underwater survey work along the North African
coastline. If trade with Africa was increasing between the
ﬁrst and second century AD, as the ceramic evidence of
North African exports on land suggests that it was, we
should expect sailings to and from Africa to be under-
represented in the wreck evidence, given that we are
likely to ﬁnd their wrecks only at the non-African end of
Examination of the relatively small number of 87
Roman and Byzantine wrecks carrying stone cargoes gives
a rather dierent picture chronologically and reinforces
the idea that the shape of the main wreck graph is
heavily inﬂuenced by chronological fashions in amphora
usage, rather than necessarily being representative of all
shipping. Russell’s paper in this volume (Chapter Eight)
deals with these wrecks in more detail; here I conﬁne
myself to noting that they start in the second century
BC and climb sharply through the late Republic and
early Empire to reach a peak in the third century AD
(Figure 2.7). e fourth-century column is accounted
for almost entirely by a proportion of the long-dated
generically ‘Roman’ wrecks, as is apparent when we
remove these from the dataset and concentrate only on
the 42 wrecks datable to within a century (Figure 2.8). A
single Byzantine wreck, carrying the stone ﬁttings for a
church, is found thereafter. e evidence for stone cargo
wrecks is much more congruent with the evidence from
e second assumption is demonstrably false.
Shipwrecks are found, usually by SCUBA divers in
relatively shallow coastal waters, because they show up
as a mound of cargo on the seabed, or, in the case of
early modern wrecks, the iron cannon are seen. e ship’s
timbers, where not protected by the cargo, have usually
rotted or been scattered. Ships are therefore unlikely
to be found if they were carrying largely non-durable
cargo, like slaves, grain, textiles, or other perishables in
sacks. e main durable cargoes likely to preserve ancient
shipwrecks are therefore cargoes of stone, or of goods
carried in amphorae, the main container until at least the
early Imperial period. Although a few of the wrecks in the
dataset are represented only by shipboard equipment, or
have been found as sunken hulls in terrestrial excavations
of harbours or river ports, the vast majority were
discovered because their cargoes were spotted on the
seabed. e ‘wreck’ graphs may therefore be considered
as graphs of known cargoes. Are they in fact primarily
graphs of amphora usage? 10
By the Medieval period, at least in the central and
western Mediterranean, the barrel had largely replaced the
amphora as the preferred transport container for liquids
and even some solid goods, such as salted ﬁsh, having
a better volume to weight ratio, more ecient stacking
capability, and greater manoeuvrability on land. e
relatively short-lived phenomenon of dolia shipwrecks,
discussed by Heslin in this volume (Chapter Nine),
may have been a Mediterranean eort to compete
with the new container technology. In the southern
and eastern Mediterranean, where timber resources
were less plentiful, the amphora lasted longer in the
early Islamic world; but it is safe to say that in the
western Mediterranean during the second half of the ﬁrst
millennium AD the switch from amphorae to barrels was
largely complete. Possible iron hoops from barrels have
been found in the Port-Vendres C wreck of the second
century AD,11 but the construction of many Roman
barrels with wooden tied hoops or withies further reduces
the chance of survival.12 e concomitant reduction in
visibility of wrecks may therefore account for some of the
reduction in the number of known wrecks of this period;
in other words, we cannot be entirely conﬁdent that
either the second-century or the early Medieval decline
in trade was as sudden or complete as the graphs would
seem to suggest.
e real question is when this shift in container
technology commenced, and how fast it occurred.
10 Cf. Horden and Purcell 2000: 371–5.
11 Parker 1992a: 332.
12 Barrel in Istres museum, ﬁrst century AD: http://www.
consulted 1 June 2011). Barrels found reused as well lin-
ings at Silchester were six feet tall and made of silver ﬁr,
probably from the Pyrenees, bound with wooden hoops:
St. John Hope and Fox 1898: 121–2 and pl. VIII; Reid 1901;
Boon 1957: 159.
13 Marlière 2002: 27: 174–5. On origins, ibid.: 170–3.
14 Ibid.: 117–57.
2: Developments in Mediterranean shipping and maritime trade
AD 90 African Red Slip ware began to be exported
in quantity from Africa Proconsularis (modern Tunisia)
and rapidly became the dominant Roman ﬁneware class
in the western and central Mediterranean, achieving
a pan-Mediterranean distribution from the third to
seventh centuries AD, surviving the split between eastern
and western empires.17 Signiﬁcantly, the distribution of
known shipwrecks is concentrated around the northern
shores of the Mediterranean, largely because of modern
political circumstances limiting underwater survey along
the North African coastline. But if, as the land evidence
suggests, there was actually an increase in exports from
North Africa in the second century AD, this will be under-
represented in the wreck evidence, further complicating
interpretation of the graphs.
e end of this intensive, connected trading system has
been a matter for prolonged debate. Pirenne considered
that Mediterranean long-distance trade continued
essentially unchanged until the Arab expansion of the
seventh century;18 Hodges and Whitehouse showed that
this view was based on an optimistic reading of the
documentary sources, and that archaeological evidence
suggested a steep decline in trade in the western
Mediterranean in the sixth century,19 a view largely
supported by more recent analyses.20 To the extent that
state incentives, especially tax and customs exemptions
for ship-owners providing services to the annona,
stimulated maritime trade by subsidising the costs of
particular voyages, changes in the annona system may be
expected to have had repercussions on the entire system.
Particularly signiﬁcant in this regard were the loss of Africa
to the Vandals in the ﬁfth century, and the Persian capture
of Alexandria in 617 which severed Constantinople from
its Egyptian grain sources and brought an end to the
annona trac to the Byzantine capital.21
e ceramic evidence from land sites shows that the
seventh century generally saw a further major downturn
in the already diminishing volume of maritime trade,
especially in the west and central Mediterranean.
Following the Arab invasions of North Africa in the 640s,
exports of African Red Slip ware dwindled and ultimately
ceased by the time of the ﬁnal conquest of North Africa at
the end of the seventh century.22 e majority of imports
even to the city of Rome stopped, though the exceptional
deposit from the Crypta Balbi shows that imported
ﬁnewares from North Africa and amphorae from North
Africa, southern Italy and the eastern Mediterranean
continued to reach at least ecclesiastical or monastic
consumers in the second half of the seventh century.23
In the eighth century Rome’s table pottery—Forum
sites on land, which shows high levels of imported marble
for architecture from the ﬁrst century BC to the early
third century AD, and for sarcophagi from the second
century AD through the third century AD.
e picture of Mediterranean trade suggested by
ﬁnds on land certainly does not suggest a downturn
of maritime trade in the early second century.16 From
the Augustan period through to the mid-ﬁrst century
AD, Italian Terra Sigillata pottery achieved a wide
distribution throughout the Mediterranean, and was
even exported to India. From the early ﬁrst century
AD onwards it was increasingly replaced by Gaulish
Samian ware, the distribution of which centred on
the central and western Mediterranean. From around
15 CIL VI.1785 = 31931; Rougé 1957.
16 Cf. Rice, this volume (Chapter Four).
17 Hayes 1972; Fentress et al. 2004.
18 Pirenne 1937; 1952.
19 Hodges and Whitehouse 1983.
20 Hayes 1998; Wickham 2005: 693–824; McCormick 2001:
83–119; Reynolds 1995.
21 McCormick 1998.
22 Fentress 1998.
23 Saguì 1998; Fentress 1998: 5 on its exceptionality.
Figure 2.8. Well-dated wrecks, datable within 100-year ranges,
carrying stone cargoes by century, 200 BC–AD 600 (n=42). (Data
supplied by Ben Russell.)
Figure 2.7. Wrecks carrying stone cargoes by century, 200 BC–
AD 600 (n=87). (Data supplied by Ben Russell.)
ships were, of course, common at all periods, and we
must keep in mind that ships of less than 75 tons were
common throughout the Roman period, as they were
before and afterwards.27 Houston points out that the
great majority of merchant shipping at any period before
the mid-twentieth century was made up of small ships—
of less than 100 tons each—and this will have been true
for the Roman period too.28 Nevertheless, during the
period 100 BC to AD 300 we ﬁnd wrecks of ships that
carried cargoes of well over 100 tons, even over 350 tons,
which we do not before about 100 BC or between AD
400 and 1000.29 e attestation in wrecks of shipping of
this size is important in the perspective of the longue
durée, suggesting that the intensity and volume of Roman
trade was such as to justify investment in larger merchant
ships (and, as we shall see, of the harbour infrastructure
to receive them) than was the case in the Classical or
early Hellenistic periods, or in the early Middle Ages.
e sample of wrecks for which tonnage can be
estimated is small, and there is textual evidence at least
before 100 BC for larger ships—Hieron of Syracuse’s
superfreighter, the Syracusia, has been estimated at over
4,000 tons displacement including nearly 2,000 tons of
cargo.30 is was of course exceptional, but an indication
of sizeable ships in regular use is given by the asos
harbour regulations of the later third century BC which
restrict use of one part of the harbour to ships of 80
tons or more, and of the other part to ships of 130 tons
or more.31 Archaeologically attested evidence for ships
over 100 tons dates from c. 100 BC onwards, and two
of the largest known Roman wrecks, the Albenga and
Madrague de Giens wrecks, date from the period of the
Late Republican wine trade between Italy and Gaul in the
ﬁrst half of the ﬁrst century BC. e Albenga wreck sank
o the coast of northern Italy c. 100–80 BC, with a cargo
estimated at between 11,500 and 13,000 Dressel 1B wine
amphorae, together with hazelnuts and grain (probably
in sacks above the amphorae)—the total burden was
Ware—was locally produced or came from no further
aﬁeld than 30 miles away; in the ninth century, Forum
Ware began to be exported in small quantities around
the central-west Mediterranean, from Sicily to Provence.24
McCormick has shown that the ninth century saw some
revival of shipping communications and trade, but that
overall the Roman pattern of often direct, open-water
sailing between principal ports had given way to habits
of predominantly coastal voyaging.25 Trade in the eastern
Mediterranean, however, remained more vibrant than in
the west, but overall the Mediterranean of the seventh
and eighth centuries was much less connected than that
of even the fourth to sixth centuries, let alone the ﬁrst to
third, when the levels of maritime trade were such that
Italian bricks were exported as return cargoes for African
exports of grain, olive oil and salted ﬁsh.26
Even though deriving levels of trade from graphs of
shipwreck numbers remains problematic, the evidence
from terrestrial distribution of amphorae and ﬁne wares
suggests that the Hellenistic and Early Roman periods
saw a massive increase in trade, followed by a reduction
in the seventh and eighth centuries. Despite the problems
of archaeological visibility that seem to be aecting the
shipwreck graphs from the ﬁrst century AD onwards, the
Hellenistic and Roman periods together with, to a lesser
extent, the fourth to sixth centuries AD, look exceptional
by comparison with any other pre-modern period. is
observation is further supported by an examination of
changes in the technology of ships, and developments in
The technology of merchant shipping in the
e wreck evidence does suggest that there were
important changes in the size of the largest shipping
between the Hellenistic and early Medieval periods. Small
24 Wickham 2005: 735–6. On Forum Ware, see Whitehouse
25 McCormick 2001.
26 Wilson 2001; Wilson, Schörle and Rice forthcoming.
27 Confusion is caused by the different types and spelling of
modern measurement units, and by different ways of mea-
suring ships’ tonnage, which may include total displace-
ment of the ship, displacement when empty, the weight of
the cargo, or various ways of converting the capacity of the
ship from a volume measure to tons; see Lane 1964 on all
this. Three different tons are in use: the tonne or ‘metric ton’
of 1,000 kg (2,205 lb), the UK ‘long ton’ of 1,016 kg (2,240
lb), and the US ‘short ton’ of 907.1,847 kg (2,000 lb). I use
‘ton’ here to mean the long ton, following UK English us-
age. Since even in the US, in writings on naval architecture
the unqualiﬁed use of ‘ton’ refers to the long ton rather
than the short ton which is the normal US usage in non-
naval contexts, I assume that Casson (1971) and Houston
(1988) both use the long ton. Because the long ton is very
close to the metric tonne (1 long ton = 101.605 per cent
metric tonne) used by non-Anglophone European authors,
and because estimates of ancient tonnage are imprecise
anyway, I have not considered the difference between the
long ton and the metric tonne signiﬁcant when reporting
estimates of ancient tonnage. Therefore, when Pomey and
Tchernia (1978: 234) calculate the burden of the Madrague
de Giens ship as 375–400 tonnes, I have simply converted
this as 375–400 tons rather than achieve a spurious precis-
ion by reporting it as 369–394 tons.
28 Houston 1988.
29 Parker 1990a: 340–1; 1992a: 26; 1992b: 89 and 90, ﬁg. 1;
Pomey and Tchernia 1978.
30 Athenaeus Deipnosophophistae 5.206d–209b; Casson
1971: 185–6; Turfa and Steinmayer 1999.
31 IG XII. Suppl.: 151, no. 348, with SEG XVII: 417; Casson
1971: 171, n. 23.
2: Developments in Mediterranean shipping and maritime trade
tons.35 is implies that such ships might be aordable by
some private ship-owners. Indeed, the ﬁnancial burdens
of munera were so heavy that elite landowners were thus
encouraged to invest in large shipping to escape them.
e larger merchant ships of the Roman Imperial period
thus compare well, in terms of capacity, with Venetian
shipping of the mid-ﬁfteenth century: the Venetian
merchant marine consisted of merchant galleys which
carried 150–240 tons of cargo, and some 300 round ships
of 100 tons or more, with perhaps 30–35 of these carrying
240 tons or more, up to c. 360 tons.36
Several prior technological advances had been necessary
prerequisites for the development of large shipping.
e development of the keel is already attested in the
Ma’agan Michael ship of c. 400 BC and the fourth-century
BC Kyrenia wreck.37 e keel and its associated wineglass
hull section improved stability and reduced the leeway
that a ship made when sailing into the wind. Although
many merchant ships had rounded hulls, some, like the
Grand Congloué, Dramont A and Chrétienne A, had
sharp hulls.38 e Madrague de Giens ship had a keel
1 m deep (Figure 2.9),39 and a prow with a concave proﬁle,
ending in a jutting cutwater like those shown on later
mosaic representations of merchant ships of the second
and third centuries AD, from the statio of the shippers of
Sullecthum in the Piazzale delle Corporazioni at Ostia,
and from the baths of emetra in Tunisia (Figures
2.10–2.11).40 is prow and cutwater design would also
substantially reduce leeway.
Maritime archaeology continues to produce
new evidence on the techniques employed by
ancient shipwrights, and while it is increasingly clear
that shipbuilding technology did not remain static
throughout the Roman period, the picture formerly held
probably in the order of 500–600 tons. At the Madrague
de Giens, a ship of 375–400 tons sank o southern Gaul
c. 75–60 BC while carrying a cargo of 6,000–7,000 Italian
wine amphorae inserted into a layer of pozzolana for
stability, with Campanian black gloss wares and cooking
and coarse pottery packed in wooden crates above the
amphorae.32 e biggest amphora-carriers, in fact, belong
to the ﬁrst century BC.
Ships carrying stone cargoes show a somewhat
dierent chronological pattern; as Russell discusses in his
contribution in Chapter Eight, the majority were small
ships: 68 per cent of stone cargoes for which a weight can
be estimated weigh under 50 tons. But among the larger
ships, of 200–350 tons, all the known wrecks except
the Mahdia ship of the early ﬁrst century BC belong to
the second or third centuries AD, and are found o the
southern coasts of Italy or Sicily, carrying eastern cargoes.
Russell suggests that ships used to carry marble were
getting larger—or, put another way, marble shipments
became larger and more frequent as the marble trade
developed over the Imperial period. e largest cargoes
represented by wrecks of the Late Republican period are
dominated by amphorae, while the largest cargoes of
the High Empire are marble shipments. e transport of
stone obelisks weighing between 200 and 460 or even 500
tons from Egypt to Rome in the reigns of Augustus and
Caligula, and again under Constantine in AD 337, must
give some indication of the size of the ships needed to
carry these exceptional cargoes.33
Some of the largest ships, especially in the Roman period,
may have been primarily grain transports which have not
been found as wrecks for the obvious reasons of cargo
perishability; the largest of these have been estimated
at some 1,000 to 1,200 tons.34 Most telling, perhaps,
is the late second-century AD regulation exempting
ship-owners from civic munera if they put at the state’s
disposal a ship of c. 340 tons, or several ships of c. 70
32 Tchernia et al. 1978; Pomey 1982; Liou and Pomey 1985:
559–67; Parker 1992a: 249–50. The pozzolana that
stabilised the amphorae may have been intended for sub-
sequent sale for use in harbour construction.
33 Estimates for the weights of the largest obelisks moved to
Rome vary. According to Wallis Budge (1926: 143, 181,
219 and 255, followed by Habachi and Vogel 2000: 103–
6), the two obelisks now in Piazza Montecitorio and Piazza
del Popolo, which were moved in 10 BC, weigh 214 and
235 tons; the Vatican obelisk, moved by Caligula in AD 37,
weighs 326 tons, and the Lateran obelisk, shipped in AD
357, weighs 460 tons. Engelbach (1923: 30) reports est-
imates of 331 tons for the Vatican obelisk and 455 tons for
the Lateran obelisk. Wirsching (2000: 274, Table 1) gives
higher ﬁgures: 230 tons for Piazza Montecitorio, 263 tons
for Piazza del Popolo, 330 tons for the Vatican obelisk, and
500 tons for the Lateran obelisk. LTUR s.v. Obeliscus Con-
stantii: Circus Maximus gives a ﬁgure of 522 tonnes for the
Lateran obelisk. With the exception of Engelbach, none of
these authors gives sources for their estimates.
Pliny says that the ship transporting the Vatican obelisk
also carried 120,000 modii of lentils as ballast (Naturalis
Historia 16.76.201); its total tonnage has been estimat-
ed at 1,100 tons. The other loads will also have needed
stabilising ballast and imply ships of substantially greater
tonnages than the weights of the obelisks themselves.
Wirsching’s reconstruction (2000; 2003) of the design of
obelisk-carriers is wholly unconvincing.
34 Casson 1971: 186–8, on the basis of Lucian’s description
of the Isis, which he purports to have seen in the Piraeus
(Lucian Navigium 5–6).
35 Scaevola apud Digestam 50.5.3; cf. Casson 1971: 171 n.
36 Lane 1966, especially 4–5.
37 Steffy 1994: 40–49.
38 Casson 1971: 175; Benoit 1961: 130 ﬁg. 75 and pl. 28; cf.
Throckmorton 1972: 68.
39 Parker 1992a: 250.
40 Pomey 1982: 140–51.
simple evolutionary picture is complicated by the recent
discovery of wrecks in Tantura lagoon near Tell Dor, of
the early sixth (Tantura A) and early ninth centuries AD
(Tantura B), which already show skeleton-ﬁrst construc-
tion, with planking nailed to the internal frames, and no
edge-joints.45 e earlier of these wrecks pre-dates the
seventh-century Yassı Ada wreck, which was still built
using the shell-ﬁrst technique. Meanwhile the Dor D
wreck, of the mid-seventh century, has hull planking
which was edge-joined by relatively close-set but loose-
ﬁtting unpegged mortise and tenon joints, whose
purpose was probably for alignment of the planking
which was nailed to frames by means of treenails. is
suggests skeleton-ﬁrst construction, but with continued
use of mortise and tenon joints.46 e St. Gervais II
wreck, also seventh-century, may have been built in
similar fashion.47 Evidently, the transition from shell-ﬁrst
to skeleton-ﬁrst construction was a lengthy process and
of a gradual evolution in methods of hull construction
is now becoming increasingly complicated and region-
ally diverse.41 e shell-ﬁrst method of construction,
using edge-joined planks held together with mortise and
tenon joints and reinforced by internal bracing timbers or
frames is attested already in the Uluburun wreck of the
fourteenth century BC and persisted into the Byzantine
period. By the mid-ﬁrst century BC some large ships, like
the Madrague de Giens wreck, had double-skinned hulls,
adding strength.42 Over time, the mortise and tenon
joints became smaller and more widely spaced (Figure
2.12), and more reliance was placed on the internal
structures of frames, keelsons, ceilings and decking.43 is
process was originally seen as adumbrating the trans-
ition to skeleton-ﬁrst construction clearly attested in
the eleventh-century Serçe Limanı wreck, in which the
frames were laid ﬁrst and the hull planking then nailed
to them, without the need for edge joints.44 However, this
Figure 2.9. The Madrague de Giens ship, c. 70–60 BC. (a) cross-sections of the hull; (b) detail of cross-section of the keel; (c)
longitudinal proﬁle of the ship, with reconstructed elements dashed. (Reprinted from Wooden Ship Building and the Interpretation of
Shipwrecks by J. R. Steffy 1994: 63, ﬁg. 3-49, by permission of Texas A&M University Press.)
41 For syntheses see Steffy 1994; Hocker and Ward 2004;
Pomey and Rieth 2005: 156–83. Rival 1991 discusses
Roman carpentry techniques in shipbuilding.
42 Steffy 1994: 62–5.
43 Ibid.: 84.
44 Ibid.: 84–5 is cautious about adopting a simple evolution-
ary perspective, noting the importance of regional differ-
45 Kahanov et al. 2004.
46 Kahanov and Royal 2001.
47 Cf. Pomey 2004: 33.
2: Developments in Mediterranean shipping and maritime trade
both methods continued in parallel for a while. Dierent
regional traditions of construction also persisted and
further complicate an attempt to construct a simple
evolutionary narrative;48 the very ancient tradition of
sewn-planked ships persisted in northern Italy, especially
the Po Valley, for river and estuarine craft throughout
antiquity and the Middle Ages.49 In the North Sea and
Channel regions, in the late Iron Age and the Roman
period, ships were clinker-built, skeleton-ﬁrst with
overlapping hull planking, and the relationship of this
tradition to the emergence of skeleton-ﬁrst construction
in the Mediterranean remains enigmatic.50 e reasons
behind the transition from shell-ﬁrst to skeleton-ﬁrst
construction were probably connected primarily with
the costs of ship construction. Shell-ﬁrst construction,
in which the hull planking determined the form of the
hull and the frames provided internal strength, in fact
conferred greater ﬂexibility on the hull, giving greater
resistance to stresses inﬂicted during rough weather.
However, it was very labour-intensive. Skeleton-ﬁrst
construction, by contrast, was cheaper in terms both of
labour and materials; and the limited evidence currently
at our disposal possibly suggests that the transition to
this new, cheaper technique may have taken place earlier
in small coasters than in larger merchant vessels.51
All wooden ships leak, and water must be bailed or
pumped if the ship is not ultimately to sink.52 e
wineglass hull section enabled the water that seeped
through weeping seams between the timbers, or was
otherwise taken on board unintentionally, to collect at
the lowest point in the hull, in the bilge, which could
be ﬂoored over. is minimised humidity in the hold
and kept the cargo drier. Larger ships would obviously
collect more bilge water, and the height of a large ship
would set a limit to bailing out the bilge water by hand.
Or rather, unless a more eective method could be
found, the problem of emptying the bilge would set a
limit on the size of ships. Signiﬁcantly, one of the features
of Athenaeus’ account of Hieron’s superfreighter the
Syracusia is the use of an Archimedes screw to empty
the deep bilge.53 However, the Archimedes screw was a
suboptimal device for bilge pumping in a pitching and
rolling ship, and although it had a high discharge rate it
had a relatively low lift.
48 Pomey and Rieth 2005: 166–7.
49 E.g., the Commacchio wreck, ﬁrst-century BC (Pomey and
Rieth 2005: 164–5). Cf. Virgil Aeneid 6.413–4, where
Charon’s ferry is clearly a sewn boat. It is not irrelevant in
this context that Virgil was from Mantua, in northern Italy.
50 Cf. Pomey and Rieth 2005: 173–5.
51 Kahanov et al. 2004: 126.
52 Oertling 1996: xv.
53 Athenaeus Deipnosophistae 5.208f.
Figure 2.10. Mosaic from the statio of the traders of Sullecthum
(Salakta, Tunisia) in the Piazzale delle Corporazioni at Ostia,
showing a three-masted and a two-masted ship. (From Pomey
Figure 2.11. Third-century AD mosaic from the frigidarium
of the baths at Themetra (Tunisia), showing a merchant ship
with inclined foremast (artemon), concave prow proﬁle and
sharp cutwater. (From Pomey 1997: 89.)
Figure 2.12. Development of edge-joining techniques for hull
planking, using mortise and tenon joints (a) Kyrenia ship, fourth
century BC; (b) Yassı Ada ship, fourth century AD; (c) Yassı
Ada ship, seventh century AD; (d) Serçe Limanı ship, eleventh
century AD. (Reprinted from Wooden Ship Building and the
Interpretation of Shipwrecks by J. R. Steffy 1994: 84, ﬁg. 4-8, by
permission of Texas A&M University Press.)
pump operated from below, since it would have been
dicult and inecient to attach any of the suggested
drive devices for a chain pump to the lower wheel or roller.
If it was a chain pump, it must have been discharging into
scuppers between decks, to save lifting the water all the
way up to the top deck, but still operated from the top of
the chain, below the top deck. Oleson may well be right,
however, to see this as a force pump.56
In Roman wrecks the wooden disks and lead tubing
from the discharge pipe of a bilge pump are sometimes
found; the Nemi ships, the Los Ullastres wreck, and
the wreck in Ponza harbour also preserved parts of the
upper wheel or the lower guide rollers.57 e earliest
evidence comes from the La Cavalière wreck of c. 100
BC.58 e bilge pump on the Madrague de Giens wreck
had been recovered in antiquity by salvage divers.59
Elements of chain pumps are found in wrecks between
100 BC and the seventh century AD; after this the
trail goes cold again until the ﬁfteenth century AD.
Mariano Taccola in 1451 regarded the chain pump as a
Tartar device, while Portuguese and Spanish navigators
of the sixteenth century found Chinese naval chain
pumps alien and superior to the force pumps used in
By 100 BC, though, a new type of pump had been
designed, not described by any ancient author but
reconstructable from evidence in wrecks and similar in
fact to known pumps from eighteenth-century ships.54
e chain pump consisted of a series of wooden disks
threaded on a loop of rope or chain, passing over a kind of
cog-wheel at the top and guided by a roller at the bottom
of the loop, which was in the bilge (Figures 2.13–14). On
their upward journey the disks entered a tube which they
ﬁtted tightly. e lower end of the tube was below the
level of water in the bilge and the disks thus drew water up
into the tube, discharging it into a trough at the top, from
where it was emptied through the gunwales via a pipe.
How the loop was rotated is unclear; various scholars have
suggested a windlass, crank or treadwheel attached to the
upper wheel. Despite earlier doubts, it has in recent years
been shown that the principle of the crank was known in
the Roman world, and a wooden disk with an eccentric
square socket from one of the Nemi ships has been
interpreted as the disk from a crank handle for a chain
pump. e one textual reference we have to the operation
of a bilge pump is a passage of Paulinus of Nola, in which
an old man was below decks pumping bilge-water.55 It
is dicult to believe, however, that this was a chain
Figure 2.13. Elements of Roman bilge pumps: (a) upper sprocket from the Nemi ships; (b) lower guide rollers from the Nemi ships;
(c) wooden disks from the Madrague de Giens wreck. (© Carre, M. B. and Jézégou, M. P. (1984). ‘Pompes à chapelet sur des navires de
l’antiquité et du début du moyen âge’, Archaeonautica 4: 198, ﬁg. 14. CNRS Éditions.)
54 Oertling 1982; Carre and Jézégou 1984; Foerster Laures
1984; 1989; Carre 2007.
55 Paulinus of Nola Epistulae 49 (discussed by Oleson 1984:
56 Oleson 1984: 65–7.
57 Ucelli 1950: 183–5 (with incorrect reconstruction); Foerster
Laures 1984; Galli 1996.
58 Foerster Laures 1984: 93.
59 Pomey 1982: 139; Parker 1992a: 250.
60 Oertling 1996: 56–8.
2: Developments in Mediterranean shipping and maritime trade
Judging the ecacy of the Roman chain pump is
problematic, not least because the sole evidence we have
is archaeological, and therefore comes from wrecks which
by deﬁnition were not prevented from sinking by the
operation of the pump. Eighteenth-century experience
suggests that the chain pump would certainly have been
a help in many storms, but that ships could still be
swamped by seas that overwhelmed the capacity of the
pump.62 Nevertheless, since the chain pump facilitated
the building of larger ships, which, other factors being
equal, could ride out storms more easily than smaller
ships, it may have reduced the overall propensity to
wreck. Under-representation of larger ships in the wreck
assemblage, for reasons both of seaworthiness and
because many were grain ships, may be suspected but
seems impossible to prove.
Large ships required more eective propulsion. e
addition of a foremast is seen already on some ships of the
sixth century BC,63 and becomes a regular feature of larger
Roman merchantmen. e Madrague de Giens wreck
of the ﬁrst century BC had a main mast and an inclined
foremast,64 and the same arrangement is common in
iconography, as on an early third-century mosaic from the
frigidarium of the baths at the small port of emetra near
Sousse—the foremast sharply raked forward and carrying
a square sail (Figure 2.11).65 e ﬁrst three-masted ship
of which we have a record is Hieron’s superfreighter, the
Syracusia, which also had three decks; but in the second
century AD Lucian and Philostratus mention three-masted
ships in the context of the grain ﬂeet; and one appears in
the mosaic from the statio of the shippers of Sullecthum in
the Piazza delle Corporazione at Ostia, c. AD 200 (Figure
2.10).66 e relatively small size of the third sail shown, at
the stern, suggests that it would have contributed more to
manoeuvrability and steering than to propulsion.
Large ships might also raise a topsail when sailing
across open water—a ﬂattened triangular sail on the
mainmast.67 Fore-and-aft sails were used in the ancient
world for small ships, and a lugsail with a short lu
is shown on the second-century AD tombstone of
Alexander of Miletus, found near Piraeus (Figure 2.15).68
e triangular lateen rig may also have been known,
The chain pump was a sufficiently effective means
of lifting water through several metres, from the bilge
to the gunwales of a large ship; like the Archimedes
screw it was to some extent self-cleansing so could
cope with sludgy dirty water, but it could also cope
far better with the pitching and rolling of a ship
at sea. Chain pumps were an important feature of
wooden ships from the eighteenth to the nineteenth
centuries,61 and may be considered a key enabling
technology for the construction of large merchant
vessels in antiquity. Their disappearance from the
archaeological record in the seventh century is
probably related to the smaller size of shipping of the
early Middle Ages, for which such pumps were less
necessary; while the pump itself may not entirely have
disappeared, the smaller number of wrecks known
after the seventh century reduces the statistical
likelihood of Medieval pumps being discovered.
Figure 2.14. Reconstruction of the chain pump from the Los
Ullastres wreck (© 1984 F. Foerster Laures, ‘New views on bilge
pumps from Roman wrecks’, International Journal of Nautical
Archaeology 13: 88, ﬁg. 4a. John Wiley & Sons Ltd.)
61 Oertling 1982; 1996.
62 Oertling 1982.
63 Casson 1971: 70, 240.
64 Pomey 1982: 141–2.
65 Foucher 1958: 21–3 and pl. VIIIb, XIb, XIIb; 1967.
66 Lucian Navigium 14; Pomey 1982: 151; Meiggs 1973:
pl. XXIVb; generally on the Piazzale delle Corporazioni,
Meiggs 1973: 283–8, with references.
67 Casson 1971: 241–3.
68 Casson 1956; 1971: pl. 181; Guilleux La Roërie 1956;
Bowen 1956; Basch 1989; Campbell 1995; Medas 2008.
and may have evolved from the square sail, brailed up
or furled to present a triangular surface area to the
wind (Figure 2.16).69 By the seventh century the lateen
had become the standard rig, and square sails largely
disappear, until their re-emergence in the fourteenth
century. e disappearance of the square sail has often
been attributed to the supposedly superior sailing
qualities of the lateen rig, which allowed sailors to point
closer into the wind, but these have been exaggerated,70
given the awkwardness of tacking in a lateen-rigged
ship, which involves ‘wearing ship’—turning downwind,
setting the yard upright and then re-setting the sail, a
procedure which involves a larger crew, and more free
deck space. In fact, the general disappearance of the
square sail from the seventh century onward is part and
parcel of the return to the use of smaller shipping.
e hulls of early Medieval ships became more rounded
and box-like, losing the wineglass section of some of the
earlier vessels to allow more space for more cargo as
overall ship sizes became smaller (Figure 2.17).71 But the
rounded hulls of early Medieval ships would have made
a lot of leeway, and the combination of a lateen rig and
rounded hull may have held a course not much closer
into the wind than a brailed square rig and the more
wineglass hull and straight cutwater of earlier, Roman,
ships.72 e change in rig is likely to be due instead to
the disappearance of the larger merchant ships which
required a square rig because large lateen sails would be
too awkward to handle.
Some merchant ships, besides their sails, also carried
oarsmen, both to enable them to make progress during
calms, and for manoeuvring in port.
By the sixth century BC the use of the sounding
weight, usually a lead weight of hemispherical, conical,
bell-shaped or similar form with a suspension lug for
attaching a rope, and a cup on the underside for holding
tallow to bring up a sample of the sea-bed, allowed sailors
to measure the depth of water and assess the nature of
the bottom, thus enabling them to judge when they were
approaching land. ere is some tendency over time for
the weights to become taller, and perhaps some increase
in the security with which tallow was retained in the
cup by means of nails, or internal divisions (septa) which
appear from the ﬁrst century BC (Figure 2.18); weights
with the suspension lug cast in one piece rather than
added separately become near-universal after 100 BC,
69 Campbell 1995: 8; Medas 2008: 88–102.
70 Campbell 1995.
71 Cf. Castro et al. 2008; Steffy 1994.
72 Whitewright 2008. Cf. Pomey 1982: 152–3 and n. 37 on
the likely sailing qualities of the Madrague de Giens ship.
2: Developments in Mediterranean shipping and maritime trade
Figure 2.15. Tombstone of Alexander of Miletus, from the
Piraeus, showing a sprit sail with a short luff. (Athens, National
Museum. From Pomey 1997: 41.)
Figure 2.16. Square sail brailed to create a triangular surface.
(From Medas 2004: 197, ﬁg. 85.)
Harbours and Ports
is peak in the size of the largest ships in the Hellenistic
and Roman worlds is matched by an increase in the
provision of port infrastructure; the story of Hieron’s
Syracusia, which was too large to dock at most
Mediterranean ports except Syracuse and Alexandria,77
illustrates the necessity for harbour facilities to keep
pace with developments in shipping. While the practice
of beaching small or even medium-sized ships or of
unloading through the shallows existed at all periods
before the twentieth century,78 the number and scale of
artiﬁcial harbour and port facilities built and maintained
around the Mediterranean between 200 BC and AD 300
stands out as unusual for any period before the Industrial
Phoenician and Punic cothons were artiﬁcial basins
excavated on the landward side of the coastline, such
as the well-preserved example at Motya in Sicily, or
the famous commercial harbour and circular military
harbour of Carthage. is design obviated the complex
perhaps because it was realised that the lug was thus less
likely to break o resulting in loss of the weight. However,
overall little development of the sounding weight is
visible over the Roman period, perhaps because already
by the Classical period the design of the weight was
well suited to its purpose.73 Oleson has suggested that
by the time of Aristotle the use of the sounding weight
had enabled the acquisition of depth measurements for
much of the waters of the Mediterranean and Black Sea.74
Iconography and literary sources thus combine to
suggest that really large ships, of several hundred tons and/
or with three masts carrying square sails, were built from
the Hellenistic period to perhaps the fourth century AD,
but not before, nor afterwards for some centuries; and
this is congruent with the wreck evidence. We lack direct
evidence for three-masted ships after the fourth century,
but they are mentioned again by Anna Comnena at the
end of the eleventh century (a three-masted pirate ship
in the Adriatic).75 Larger ships start reappearing in the
twelfth and thirteenth centuries, with the rise of Italian
dockyards at Genoa, Venice and Pisa building large round
ships with two or three decks,76 a single lateen sail on each
mast; twin steering oars, and no deep keel. e square rig
reappears on large merchant ships with the renewed rise
of trade in the fourteenth and ﬁfteenth centuries, and,
this time with multiple square sails on each mast, once
again became the standard rig for large sailing ships from
the Age of Discovery onwards.
73 Oleson 2000; 2008.
74 Oleson 2008: 127–30.
75 Alexiad 10.8 (trans.Sewter 1969: 315); Pryor 1988: 31.
76 Pryor 1988: 30.
77 Athenaeus Deipnosophistae 5.209b.
78 Houston 1988: 560–4.
Figure 2.17. Development of hull shapes in the Middle Ages.
Yassı Ada: 7th century AD. Bozburun: AD 874. Serçe Limanı:
c. AD 1025. Culip VI: c. 1300. Contarina I: c. AD 1450.
(© 2008 F. Castro, N. Fonseca, T. Vacas and F. Ciciliot,
‘A quantitative look at Mediterranean lateen- and square-rigged
ships (Part 1), International Journal of Nautical Archaeology
37 (2): 358, ﬁg. 6a. John Wiley & Sons Ltd.)
Figure 2.18. Roman lead sounding weights. (From Oleson 2000:
e artiﬁcial harbour built at Portus by Claudius to
provide a sheltered deep water harbour for the grain
ﬂeet and other merchant ships supplying Rome, which
previously had had to anchor o the river mouth at Ostia,
also relied on concrete construction for the breakwaters,
mole and lighthouse foundation. e hexagonal basin
which formed part of the Trajanic expansion of the
facilities at Portus was dug as a kind of Roman cothon, but
again with the use of concrete for massive quays, wharves
and warehouses around its sides.85
e harbour at Lepcis Magna, extensively developed
under Nero and remodelled on a more lavish scale by
Septimius Severus, used a wadi mouth as a natural shelter
further enhanced by concrete breakwaters and moles
(Figure 2.22). Silting was prevented by damming the wadi
further upstream and diverting it around the south-west
of the town. e harbour basin enclosed some 10 ha, with
c. 1,200 m of wharf space.86 Again, the breakwaters were
wide enough to accommodate temples and warehouses.
engineering problems of building breakwaters out into
the sea; the principal requirement was a large labour
force to shift the volume of earth that had to be removed
in digging the basin. e drawbacks were the relatively
limited size of the basins thus created: 0.18 ha or less
than 170 m of wharf space for the cothon at Motya;
0.78 ha or less than 370 m of wharf space for Mahdia in
Tunisia (Figure 2.19). ese ﬁgures are tiny by compar-
ison with the data presented by Schörle in this volume
(Chapter Five) for the sizes of Roman harbours along the
e Hellenistic period saw more ambitious harbour
works, notably of course the vast harbour at Alexandria,
with breakwaters and moles linking oshore islands
and reefs to create a massive sheltered basin two km
across, further subdivided into smaller port sections
by artiﬁcial works and breakwaters, and equipped with
the massive Pharos to signal the approach (Figure
2.20).80 Harbour construction became a science to
which Philo of Byzantium (ﬂ. c. 260–220 BC) devoted
an entire book.81
e Roman invention of hydraulic concrete, which
set underwater, opened up entirely new possibilities
for harbour development, enabling the construction
of breakwaters, jetties and moles on shores without
natural protection.82 Schörle shows how this technology
enabled the creation of artiﬁcial harbours along the
Tyrrhenian coast of Italy, allowing the development of
a façade maritime with a complex hierarchy of ports
of dierent sizes, all of which, from the largest down
to the smallest, relied on concrete construction.83 First
pioneered in Campania, the new concrete technology
rapidly spread outside Italy in the wake of Rome’s gaining
control over the entire Mediterranean basin, from the
Augustan period onwards, and was made available to
Rome’s client kings, such as Herod of Judaea, for the
construction of the harbour of Sebastos at Caesarea.
ere, Roman technology and pozzolana imported from
the Bay of Naples were used to create an artiﬁcial harbour
using ﬂoating caissons that were ﬁlled with concrete
and sunk, creating an outer harbour basin of 20 ha, with
breakwaters wide enough to support warehouse space,
and ﬂushing channels to prevent silting (Figure 2.21).84 e
construction of moles or breakwaters suciently wide to
accommodate loading and unloading facilities meant an
extension of the area where large ships could dock, since
they could unload anywhere around the harbour basin
and were not merely limited to the landward side.
79 Cf. Wilson et al., forthcoming.
80 Goddio et al. 1998; Goddio and Bernand 2004; Goddio
and Fabre 2008: 266–74 for a synopsis of recent ﬁndings.
81 Philo of Byzantium Limenopoeica (now lost); Vitruvius De
Architectura 5.12; cf. Blackman 2008: 643.
82 Oleson 1988; Blackman 2008: 644–9; cf. for recent work
on Roman hydraulic concrete: Oleson et al. 2004; Brandon
et al. 2008.
83 Schörle, this volume (Chapter Five).
84 Oleson 1988: 152.
85 Testaguzza 1970. Cf. Keay et al. 2005.
86 Bartoccini 1958: 12–13.
Figure 2.19. Plan of the Punic cothon at Mahdia, Tunisia. (©
Yorke et al., Cambridge Expedition to Sabratha 1966 plan 6.)
2: Developments in Mediterranean shipping and maritime trade
Figure 2.21. Reconstruction drawing of the harbour of Sebastos, Caesarea Maritima. (© 1992 A. Raban, ‘Sebastos: the royal harbour
at Caesarea Maritima—a short-lived giant’, International Journal of Nautical Archaeology 21 (2): 122, ﬁg. 18. John Wiley & Sons Ltd.)
Figure 2.20. Plan of the eastern harbour at Alexandria. (© IEASM.)
way limiting the cargo capacities that could be easily
handled. For smaller ships, such unloading through the
shallows continued in the Roman period at some sites—a
mosaic from Sousse shows the unloading of logs of wood
which are being weighed on the beach (Figure 2.23),87 and
beaching like this may have been common for smaller
ships in good weather at sites along the gently shelving
coast of Tunisia which lacked long moles like those at
apsus and Leptiminus. Such practices were in fact
common around the Mediterranean and even on the
coasts of Britain well into the nineteenth century.88
e introduction of concrete changed this: long moles
could be constructed on this shelving coastline running
for hundreds of metres from the shore out to where a
depth of two or three metres, sucient for medium to
large cargo ships, could be attained. At Sullecthum the
mole is c. 350 m long (Figure 2.24 published here for the
ﬁrst time), while that at Leptiminus the mole is c. 560 m
long, with an angled end forming a rectangle whose sides
provided some 300 m of mooring space (Figure 2.25).89 At
apsus, parts of the mole were visible as recently as 1987,
until a ﬁshing port was built on top; originally the mole
extended for 1,000 m from the shore, in places 100 m
wide, and is the longest known ancient artiﬁcial mole
(Figure 2.26).90 Long artiﬁcial moles were constructed
at least ﬁve sites along the Tunisian coast (Table 2.1),
vastly improving the access to large shipping for sites
e impact of the new Roman concrete technology
is seen very clearly along the Tunisian coast, where a
gently shelving bottom means that shallows extend
hundreds of metres from the shore and prevent large
ships approaching. Cothons, presumably of Punic origin,
are known at Ruspina (Monastir) and Mahdia (Figure
2.19); but many of the other Punic towns along this coast
were largely unsheltered and ships must have been drawn
up on the beach or unloaded through the shallows, either
Figure 2.22. View of the harbour quayside at Lepcis Magna,
with steps and holes for mooring posts (foreground),
and warehouses on the quayside behind the colonnade.
(Photo: A. I. Wilson.)
87 Foucher 1960: 77–8 no. 57.169 and pl. XLIa. Cf. Wilson
et al., forthcoming, for the interpretation of the cargo as
wood (ﬁrewood?) rather than lead ingots.
88 Houston 1988: 560–1.
89 Yorke et al. 1966: 16 (privately circulated, now available at:
1966.pdf, last consulted 1 June 2011; Davidson 1992.
90 Yorke et al. 1966: 14–16 (privately circulated, now avail-
able at: http://www.societies.cam.ac.uk/cuueg/Archives/
Sabratha_1966.pdf, last consulted 1 June 2011); Slim et
al. 2004: 152.
2: Developments in Mediterranean shipping and maritime trade
Figure 2.23. Mosaic from a tomb at Hadrumetum (Sousse, Tunisia), showing unloading of ﬁrewood through the shallows,
and weighing on the beach. (From Nieto 1997: 159.)
Even villas had port or harbour facilities in the
Roman imperial period, sometimes larger than those
of Republican towns.91 In Istria, for example, where
numerous wealthy villas existed, many with multiple
presses for the production of large marketable surpluses
of olive oil, there are many small artiﬁcially constructed
ports of Roman date, sheltered by projecting moles or
which became key centres of export for the olive oil and
salted ﬁsh production of the Tunisian Sahel in the Roman
period. It is not an exaggeration to say that this concrete
port technology played a signiﬁcant part in facilitating
the development of the central and southern part of
the Roman province of Africa Proconsularis as a major
exporter of oil and salt ﬁsh.
91 Schörle, this volume (Chapter Five).
Figure 2.26. Plan of the Roman port mole at Thapsus, Tunisia.
(© Yorke et al., Cambridge Expedition to Sabratha 1966 plan 3.)
Figure 2.25. Plans of the Roman port moles at Acholla and Leptiminus, Tunisia. (© Yorke et al., Cambridge Expedition to Sabratha
1966 plans 4 and 5.)
Figure 2.24. Plan of the Roman mole at Sullecthum
(Salakta, Tunisia). (© Yorke et al., Cambridge Expedition to
Sabratha 1966, previously unpublished.)
equipped with jetties. Some of these were demonstrably
associated with villas, as at Medolino and Val Catena,
while others were probably or possibly so, as near Isola
(over 0.53 ha), or at Punta di San Simone (0.84 ha).92
Other supporting elements of Hellenistic and Roman
technology enhanced this harbour infrastructure and
facilitated the continuous arrival, unloading and loading
of cargoes.93 Cranes, with multiple pulleys, enabled the
loading and unloading of cargoes of large stone, heavy
timbers, and even large amphorae such as the Tunisian
and Tripolitanian olive oil and salted ﬁsh amphorae with
capacities of over 60 l. Vitruvius, writing in the late ﬁrst
century BC, mentions slewing cranes and cranes mounted
on a rotating base to enable them to swing loads between
ships and the quayside;94 he does not specify whether
such cranes were mounted on the ships or the quayside,
or both. e normal mooring arrangement in Roman
harbours was prow-on, which required unloading to be
done by manual porterage down a gangplank over the
forequarters.95 Since unloading by crane will have usually
required ships to moor broadside-on to the quay, special
parts of harbours must have been reserved for this. e
upstream river port in the ninth region of Rome, where
wine from the Tiber Valley was unloaded, was known as
Ad ciconias nixas (literally, ‘at the place of the straining
storks’, for the resemblance of the cranes to storks) after
the dockside cranes there for the unloading of large wine
barrels.96 Cuttings in the rock surface at the early Christian
quarries at Aliki on asos, adjacent to the sea, seem to
have been intended as ﬁxing points for the masts and stays
of cranes for loading the marble onto ships.97
To keep a harbour functioning, dredging was required
and the excavations of the ancient harbour at Marseilles
have revealed three Roman dredging boats of the ﬁrst
century AD each with a slot in the hull for the dredging
mechanism (Figure 2.27).98 Evidence for dredging during
the Greco-Roman and Byzantine period has also been
recognised from gaps in the chronological sequence of
stratigraphy in Tyre’s northern harbour.99 At Ephesus,
silting of the harbour was a considerable and recurrent
problem;100 Tacitus records a major dredging operation
by the proconsul of Asia, Marcius Barea Soranus, in
AD 61,101 and in the early second century AD the prytanis
C. Licinius Maximus Iulianus contributed 2,500 denarii
to the city towards the costs of dredging the harbour.102
Hadrian diverted the river Cayster to the north to try
to tackle the source of the problem, for which he was
thanked by the city in a decree of AD 129.103 Under
Alexander Severus, between AD 222 and 238, the asiarch
M. Aurelius [. . .] spent 20,000 denarii on dredging the
harbour.104 Dredging the harbour at Side in Pamphylia
in southern Turkey was a proverbial never-ending task,
which had to be started again as soon as it was ﬁnished,
like painting the Forth Bridge.105 River ports also needed
dredging; the river-bed at Antioch was dredged for the
river port there in the reign of Antoninus Pius and again
Byzantine and early Medieval harbours
e construction of new ports, or the enlargement of
existing ones, continued in some places through the
92 Degrassi 1962: esp. 833–8, 860–1, 864–70; Stokin et al.
2008: 64–7; Auriemma et al. 2008.
93 Rougé 1966: 160–66; Casson 1971: 369–70. On cranes
generally, Wilson 2008: 342–4.
94 Vitruvius De Architectura 10.2.10.
95 As shown, for example, on a second-century relief from
Narbonne, a third-century painting of the Isis Geminiana
from Ostia, and the relief of the tabularii from Portus: Tch-
ernia 1997: 117, 119 and 127.
96 Rougé 1957, discussing CIL VI.1785 = 31931, a third- or
fourth-century AD tariff for the handling charges per barrel
of wine, detailing charges for unloading by crane, trans-
port to the state warehouses in the Templum Solis, receipt
and opening. Cf. LTUR s.v. Ciconiae.
97 Sodini et al. 1980: 113–4, 119–22.
98 Pomey 1995: 463–9; Hesnard et al. 1999: 46–9.
99 Marriner and Morhange 2006.
100 Strabo Geographica 14.1.20; Zabehlicky 1995: 204–5.
101 Tacitus Annales 16.23.
102 IvE VII.1 3066.
103 IvE II.274.
104 IvE VII.1 3071.
105 Bean 1968: 100; PECS p. 835.
106 Rougé 1975: 182.
2: Developments in Mediterranean shipping and maritime trade
Table 2.1: Ports with long moles on the Tunisian coast.a
Site Length (m) Width (m) Platform at
Leptiminus 560 15 rectangular
apsus 1000 100 in places
33 rectangular 100
x 70 m
Gigthis 140 17 semicircular
a Data from Constans 1916: 70; Yorke et al. 1966 Report:
7, 11–12, 14–16 (privately circulated, now available
bratha_1966.pdf ladt accessed 1 June 2011); Slim et al.
2004: 105–6, 138, 152, 154.
and during the course of the second half of the ﬁrst
millennium AD some of the existing harbours fell out
of use or began to silt up. Justinian did build two new
harbours on the Bosphorus, at Heraeum and Eutropius,
whose construction is described by Procopius.108 Some of
the larger Byzantine harbours were however maintained,
at least until the sixth and seventh centuries: the dredging
of the ‘portus Iuliani’ at Constantinople under Anastasios
in AD 509 was thought worthy of record in the Chronicon
of Marcellinus Comes,109 and the port of the Neorion at
Constantinople was dredged in 698, perhaps when it
became the main base for the city’s naval ﬂeet, which
was stationed in it by 715.110 But even at Constantinople,
the capital of the Byzantine empire, the eodosian
harbour silted up through a combination of progressive
sedimentation from the Lykos river and several violent
storm events which deposited thick layers of sediment in
the mid-sixth, the eighth/ninth, and the late tenth/early
eleventh centuries. It was not eectively dredged and by
the end of the twelfth century the harbour was usable
only by small ﬁshing boats and small coasting vessels.111
Soon after Egypt fell to the Arabs, the Caliphate de-silted
the port of Clysma at the head of the Red Sea in AD 642
or 643, though this suggests the degree to which it had
fallen into disrepair in the Byzantine period.112
e northern mole of the harbour at Caesarea
Maritima, which had sunk during the Roman period
as a result of tectonic movements, was restored under
Anastasios I in AD 502, but the eect was short-lived.
Gertwagen argues that the repairs were not made using
hydraulic concrete, but using wooden caissons ﬁlled
with non-hydraulic mortared rubble; when the caissons
rotted and disintegrated, the rubble spilled out and the
mole collapsed.113 ere is debate over whether hydraulic
concrete continued to be used in Byzantine harbour
construction or whether the technique had already
been lost by the reign of Justinian.114 e Justinianic
port of Anthedon had quays built of concrete using
powdered ceramics, which would have had hydraulic
properties, but this was not used underwater, and the
Muslim construction of the port of Akko in the ninth
century used caissons ﬁlled with non-hydraulic mortared
rubble; as a result the lime deteriorated and the mole
disintegrated to form a reef.115 Similarly, the fourteenth-
century Venetian harbour works at Candia (Herakleion,
fourth century AD, notably at Constantinople whose
supply needs continued to grow at this period. As the two
harbours on the Golden Horn were proving insucient,
the emperor Julian built a new port, the portus magnus,
also known as the portus Iuliani or ‘port of Sophia’, in
AD 362, the length of whose quay Mango estimates at
some 1000 m. Towards the end of the fourth century a
second new port, the eodosian harbour, was built.107
But after the fourth century AD we do not ﬁnd the
construction of major new harbour works on the same
scale as in Hellenistic and early Roman Imperial periods;
107 Mango 1985: 38–40.
108 De Aediﬁciis 1.11.18–20. Cf. Hohlfelder 1988.
109 Chronicon of Marcellinus Comes. MGH, Chronica Minora
2.97 (under the year 509): Portus Iuliani undis suis rotalibus
machinis prius exhaustus caenoque effosso purgatus est.
‘The Portus Iulianus, ﬁrst drained of its waters by lifting
machines and newly dug out was cleaned.’ Cf. Nollé 1993:
350, n. 89.
110 Mango 1985: 55–6.
111 Kocabaş 2008: 32–4.
112 Mayerson 1996: 125.
113 Gertwagen 1988: 149.
114 Hohlfelder 1988 (arguing for continued use of the tech-
nique); Gertwagen 1988: 150–1 (arguing that the tech-
nique was already lost). Procopius De Aediﬁciis 1.11.18–20,
mentions the use of caissons in Justinianic harbour works
on the Bosphorus, but does not state whether or not the
concrete, whose use is implied, was hydraulic.
115 Gertwagen 1988: 150–1.
Figure 2.27. Wreck of a Roman dredger of the ﬁrst or second
centuries AD, from the Place Jules Verne, Marseilles, with a slot
in the bottom of the hull for the dredging machinery. (Centre
Camille Jullian, CNRS. Pomey and Rieth 2005: 50.)
Recent years have seen some debate over whether the
dominant pattern of maritime trade in antiquity was
cabotage, by which anglophone historians tend to mean
speculative coastal tramping from port to port, selling a
bit of cargo here and there, and picking up other wares
to sell on further down the line, or direct shipping
between major principal ports or emporia.119 e
argument is important because it carries implications
about the overall scale of trade and levels of information
about markets; tramping is speculative, opportunistic,
and relatively small-scale, while emporia trading
relationships imply organised, often regular trac, and
relatively good information about markets at the other
end, often facilitated by agents or diaspora trading
communities in remote ports. Several anglophone
ancient historians have argued that cabotage was the
normal trading mechanism in the Roman period, an
idea given recent prominence by Horden and Purcell
(who do nevertheless acknowledge the importance of
‘le grand commerce maritime’);120 and which persists in
some more recent writers.121 A contributory factor to
the longevity of this view may be confusion between
the concepts of coastal sailing and coastal tramping,
owing to misunderstanding of how the word ‘cabotage’
is used in French and Italian, where its primary meaning
is ‘coastal sailing’. One of the best discussions of the issue
in fact remains that of Rougé, who gives a good, nuanced
analysis, distinguishing between speculative coastal
tramping, more or less regular coastal trading between
a succession of ports which the shipper comes to know
well, and ‘le grand commerce’, conducted between
Crete) used non-hydraulic mortared rubble, and soon
disintegrated. Byzantine and Venetian moles were not
generally laid on rubble foundation layers, and they thus
lacked protection against undermining of the seabed
beneath them by currents and wave action.116
In the western Mediterranean one has the impression
of far less activity in harbour maintenance between the
ﬁfth and eleventh centuries, though it must be admitted
that archaeological research on Late Antique and Early
Medieval port facilities is still very limited. Even from
the twelfth century onwards, despite impressive works
by the Venetian maritime empire in port construction,
which included arsenals (shipyards) and warehouses,
most Medieval harbours did not reach the size of the
large artiﬁcial Hellenistic and Roman projects, and the
smaller volumes of trac that they handled may be
gauged not only from their size, but also from the fact
that after the fall of the western empire, harbour cranes
are not documented in Europe again until the thirteenth
century—the earliest being at Utrecht in 1244.117
e practice of beaching small ships, known
throughout classical antiquity and the Roman period and
discussed above for parts of the Tunisian coastline (Figure
2.23), seems to have become more common again with
the increased prevalence of smaller vessels in the early
Middle Ages. Accounts of early Medieval voyages also
suggest a predominant pattern of coastal sailing, putting
into shore to spend the night on land; this eectively
halved the speed of sea travel by comparison with
24-hour sailing. Round-the-clock sailing did, however,
necessarily persist on some long open-water routes, such
as that across the Adriatic, and seems to have revived
from the ninth century onwards.118
116 Gertwagen 1988.
117 AD 1244 in Utrecht, 1263 in Antwerp, 1288 in Brugge and
1291 in Hamburg: Matheus 2001: 345; Matthies 1992:
542–3. In England the treadwheel is not recorded again
until 1331: Matthies 1992: 524. Jörns (1979: 121) inter-
prets a stone base measuring 2.2 x 2.5 m at the Caro-
lingian river port of Zullenstein as possibly the base for a
crane or a feature for tying up ships. McCormick (2001: 9)
prefers to see it as a crane, but a problem with this idea
is that, for ancient cranes at least, the diagnostic traces
are generally cuttings into stone surfaces to take upright
poles or attach stays, rather than built stone bases. Only
the foundations survive, and other interpretations for this
stone base could also be imagined.
118 McCormick 2001: 481–500.
119 The Oxford English Dictionary deﬁnes ‘cabotage’ as
‘Coasting; coast-pilotage; the coast carrying trade by sea’;
it is also a specialised term in the transport industry, where
coastal trafﬁc between ports in a single country, or be-
tween airports in a single country, may usually be operated
by domestic companies. By contrast, in French ‘cabotage’
is usually used to mean ‘coastal navigation’, often without
implying anything about the trading strategy underlying
such coastal sailing (Dictionnaire de l’Académie française,
4th edn 1762: ‘Cabotage. s. m. Terme de Marine. Navigat-
ion le long des côtes, de cap en cap, de port en port.’). Cf.
Arnaud, this volume (Chapter Three) on the terminological
confusion thus caused between the anglophone and fran-
120 Woolf 1992: 287; Horden and Purcell 2000: 143–52; 365–
121 E.g. Bang 2008: 141–2, who believes, implausibly, of cit-
ies up to c. 10,000 inhabitants in the Roman world, that:
‘These markets were normally served by a system often
referred to as cabotage: small merchant ships would more
or less casually tramp along the coast from harbour to har-
bour in search of a good bargain’, though he does con-
cede that more organised and directed trade was required
to supply the very largest cities of the empire.
2: Developments in Mediterranean shipping and maritime trade
e Hellenistic and Roman peak in Mediterranean
trading activity that we may deduce from ship size and
from harbour infrastructure, and, with due caution
and some uncertainty as to its end, from the shipwreck
graphs, was not simply a result of the maritime
technology I have outlined. Technological factors
enabled, but did not drive, the process; more important
in this regard were institutional developments.129
Laws of contract, maritime loans and sea laws existed
from the Classical period on and provided a necessary
framework for the organisation of large-scale maritime
trade. e Roman integration of the Mediterranean
under a single political system, and the virtual
eradication of piracy by Pompey, together with the
use of a single currency in nearly all of this area except
Egypt, all greatly reduced transaction costs in supplying
what had now became a vast pan-Mediterranean
market. e Roman state introduced some incentives
for shipbuilding, such as exemption from munera for
annona contractors; although we tend to think of this
as rewards for people who were ship-owners anyway,
it is highly likely that such measures in fact encouraged
other elite landowners to invest in shipping in order
to escape the heavy ﬁnancial burdens of civic munera.
With the breakdown of the Roman empire, many of
these institutions also disappeared.
A combination, therefore, of institutional factors (the
political integration of the Mediterranean, the greater
integration of circum-Mediterranean markets and the
development of legal institutions and ﬁscal instruments
encouraging trade) and technological advances (bilge
pumps, harbour construction to accommodate large
ships, and cranes for cargo handling) enabled the
emergence of large merchant shipping in the Late
Republic and High Empire. With the disintegration of
this political system the institutions that it had created
weakened or disappeared entirely, and levels of trade
fell steeply. As a result, technologies changed or even
vanished, to suit the lower levels of investment in
both shipping and harbour technology justiﬁed by the
smaller trade volumes of the time. Maritime trade never
disappeared from the Mediterranean, but the world of
the seventh to ninth centuries was a world of coastal
voyaging by small lateen-rigged craft, between harbours
that were often an inherited infrastructure from previous
ages. It was not until the twelfth or thirteenth centuries
major ports or emporia.122 As he emphasises, speculative
coastal tramping was rare in the ancient Mediterranean;
most captains knew their routes and markets.123 Parker’s
study of the distribution of wrecks with cargoes of
particular amphora types likewise suggests directed bulk
trac along certain routes between major ports, rather
than coastal tramping.124
e idea that most Roman trade took the form of coastal
tramping sits ill even with the ancient written sources,
and is refuted by the archaeological evidence. Eratosthenes
of Cyrene, writing in the third century BC, saw coastal
voyaging as a practice of mythical antiquity, practised by
Jason and the Argonauts, whereas in his day open-water
sailing was the norm for long-distance merchant voyages.125
e archaeological evidence from shipwrecks shows
that mixed cargoes were normal, but these do not imply
tramping; mixed cargoes enabled more ecient utilisation
of hold capacity (e.g., with crates of pottery or sacks of
nuts loaded in the space above a part-cargo of amphorae).
Where cargo disposition within a single wreck can be
studied it is clear that the large majority of mixed cargoes in
Greek and Roman shipwrecks were loaded in a single go at
one port, and must represent heterogenous cargoes picked
up at emporia, being traded either to another distant major
port, or being redistributed to lesser ports in the coastal
foreland of an emporium.126
e cargo evidence from shipwrecks, the distribution
of traded goods around the Mediterranean, the levels of
investment in port infrastructure, and the evidence of
resident trading communities in ports, all combine to show
that commerce in the Hellenistic and Roman periods was
emphatically not largely a matter of coastal tramping.127
Instead, large merchant ships conveyed sizeable cargoes
between principal ports or emporia; smaller vessels then
loaded heterogeneous cargoes at these emporia and
conveyed them to secondary ports in the economic foreland
of the primary port. e coastal shipping of the Roman
period was primarily engaged in supplying an emporium
from smaller ports in the surrounding coastal zone, and
in coastal redistribution towards those ports, as part of an
organised system of trade. It was only after the collapse of
intensive long-distance trade in the early Middle Ages that
cabotage tramping again became a signiﬁcant mechanism
for trade in the Mediterranean. e patterns of trade and
travel analysed by McCormick for the sixth to ninth centuries
show an overwhelmingly dominant pattern of coastal
voyaging, in relatively small ships, in strong contrast to the
maritime world of the Hellenistic and Roman periods.128
122 Rougé 1966: 415–21.
123 Ibid.: 418.
124 Parker 1990b.
125 Eratosthenes frag. 1.B.8 Berger = Strabo Geographica
1.3.2 C.43 (cited by Arnaud, this volume, Chapter Three).
126 Parker 1990a: 342–3 (on mixed cargoes); 1992a: 20–22;
Rougé 1966: 415–21; Tchernia 1997: 124–7; Nieto 1997;
Jézégou 2007; Arnaud, this volume (Chapter Three). Con-
tra: McCann and Oleson 2004: 55–7, 120, 203, 207–8.
127 Nieto 1997; Tchernia 1997; Arnaud, this volume (Chapter
Three); Wilson et al. forthcoming.
128 McCormick 2001: 481–500.
129 Cf. Arnaud, this volume (Chapter Three).
(OXREP). I am particularly grateful to Julia Strauss for her
work on updating the corpus of ancient shipwrecks, and
to Ben Russell for providing data on wrecks of ships that
carried stone cargoes. Michael McCormick, Candace Rice
and Katia Schörle all made helpful comments on earlier
versions of this paper, and provided useful references.
Bob Yorke kindly gave permission to publish here for the
ﬁrst time the plan of the mole at Salakta in Fig. 2.24.
that large square-rigged merchantmen of several hundred
tons began to ply the open water routes again with
anything approaching the frequency with which they had
in Classical antiquity.
Much of the research for this paper was carried out as part
of the AHRC-funded Oxford Roman Economy Project
2: Developments in Mediterranean shipping and maritime trade
C, M.-B. (2007). ‘Les pompes de cale et l’évacuation
de l’eau de sentine sur les navires antiques’, in J.-P. Brun
and J.-L. Fiches (eds), Énergie hydraulique et machines
élévatrices d’eau durant l’antiquité. Napoli: 51–66.
C, M. B. and J, M. P. (1984). ‘Pompes à
chapelet sur des navires de l’antiquité et du début du
moyen âge’, Archaeonautica 4: 115–43.
C, L. (1956). ‘Fore-and-aft sails in the ancient world’,
e Mariner’s Mirror 42: 3–5.
C, L. (1971). Ships and seamanship in the ancient
C, F., F, N., V, T. and C, F. (2008).
‘A quantitative look at Mediterranean lateen- and
square-rigged ships (Part 1)’, International Journal of
Nautical Archaeology 37: 347–59.
C, L. A. (1916). Gigthis. Étude d’histoire et
d’archéologie sur un emporium de la Petite Syrte. Paris.
D, D. P. (1992). ‘Survey of underwater structures’,
in N. Ben Lazreg and D. J. Mattingly (eds), Leptiminus
(Lamta): a Roman port city in Tunisia, Report no. 1.
Journal of Roman Archaeology Suppl. 4. Ann Arbor:
D, J. K. (2006). ‘Hellenistic economies’, in G. R. Bugh
(ed.), e Cambridge Companion to the Hellenistic
World. Cambridge: 73–91.
D C, F. (2005). ‘e Graeco-Roman economy
in the super long-run: lead, copper and shipwrecks’,
Journal of Roman Archaeology 18: 361–72.
D, A. (1962). ‘I porti romani dell’Istria’, in Scritti
vari di antichità 2. Rome: 821–70.
E, R. (1923). e problem of the obelisks. London.
F, D. and G, F. (2010). ‘e Development
and Operation of the Portus Magna in Alexandria—
An Overview’, in D. Robinson and A. Wilson (eds),
Alexandria and the North-Western Delta. Joint
Conference Proceedings of Alexandria: City and Harbour
(Oxford 2004) and e Trade and Topography of Egypt’s
North-West Delta (Berlin 2006), Oxford Centre for
Maritime Archaeology Monograph 5. Oxford: 53–74.
A, P. (forthcoming). ‘L’homme le temps et la mer:
continuité et changement des routes maritimes de
et vers Portus’, in S. Keay (ed.), Rome, Portus and
the Mediterranean, Archaeological Monographs of the
British School at Rome. London.
A, R., D, V., D, P., G, D., M, S.
O, F. and R, D. (2008). ‘Terre di mare:
paesaggi costieri dal Timavo alla peninsola muggesana’,
in R. Auriemma and S. Karinja (eds), Terre di Mare.
L’ archeologia dei paesaggi costieri e le variazione climat-
iche. Atti del Convegno Internazionale di Studi, Trieste,
8–10 novembre 2007. Trieste: 75–211.
B, P. (2008). e Roman Bazaar. A comparative study
of trade and markets in a tributary empire. Cambridge.
B, R. (1958). Il porto romano di Leptis Magna,
Bollettino del Centro studi per la storia dell’architettura
Suppl. 13. Rome.
B, L. (1989). ‘e way to the lateen sail’, e Mariner’s
Mirror 75: 328–32.
B, G. E. (1968). Turkey’s southern shore: an archaeo-
logical guide. London.
B, F. (1961). Fouilles sous-marines. L’épave du Grand
Congloué à Marseille, Gallia Suppl. 14. Paris.
B, J. (2005). A reassessment of the ancient sailing
season: the case for wintertime seafaring on the Graeco-
Roman Mediterranean. DPhil. thesis, University of Oxford.
B, D. J. (2008). ‘Sea transport, Part 2: Harbors’,
in J. P. Oleson (ed.), Engineering and Technology in the
Classical World. Oxford: 638–70.
B, G. C. (1957). Roman Silchester: e Archaeology of
a Romano-British Town. London.
B, R. L., Jr. (1956). ‘e earliest lateen sail’, e
Mariner’s Mirror 42: 239–42.
B, C., H, R. L. and O, J. P. (2008).
‘e Concrete Construction of the Roman Harbours
of Baiae and Portus Iulius, Italy: e ROMACONS
2006 ﬁeld season’, International Journal of Nautical
Archaeology 37: 374–92.
C, I. C. (1995). ‘e lateen sail in world history’,
Journal of World History 6: 1–23.
2600 ans d’histoire. Aix-en-Provence.
H, F. M. and W, C. A. (2004). e philosophy
of shipbuilding. Conceptual approaches to the study of
wooden ships. College Station.
H, R. and W, D. (1983). Mohammed,
Charlemagne and the origins of Europe: Archaeology
and the Pirenne thesis. London.
H, R. (1988). ‘Procopius, De Aediﬁciis, 1.11.18–
20, Caesarea Maritima and the building of harbours in
late antiquity’, Mediterranean Historical Review 3 (2):
H, K. (1980). ‘Taxes and trade in the Roman Empire
(200 BC–AD 400)’, Journal of Roman Studies 70: 101–25.
H, P. and P, N. (2000). e corrupting sea. A
study of Mediterranean history. Oxford.
H, G. W. (1988), ‘Ports in perspective: Some
comparative materials on Roman merchant ships and
ports’, American Journal of Archaeology 82 (4): 553–64.
J, M.-P. (2007). ‘L’épave Ouest-Embiez 1: proposi-
tion d’un modèle de reexportation de produits verriers
et du vin à la charnière des IIe/IIIe siècles après J.C.’,
in J. Pérez Ballester and G. Pascual (eds), Comercio,
redistribución y fondeaderos. La navegación a vela en el
Mediterráneo. Valencia: 451–60.
J, W. (2007). ‘Gibbon was right: the decline and
fall of the Roman economy’, in O. Hekster, G. De
Kleijn and D. Slootjes (eds), Crises and the Roman
Empire. Proceedings of the Seventh Workshop of the
International Network Impact of Empire (Nijmegen, June
20–24, 2006). Leiden: 183–99.
J, W. (1979). ‘Zullenstein. Ein Beitrag zur Kontinuität
von Bauwerken’, in Deutsche Königspfalzen. Beiträge
zu ihrer historischen und archäologischen Erforschung
3, Veröentlichungen des Max-Planck-Instituts für
Geschichte, 11.3. Göttingen: 111–35.
K, Y. and R, J. G. (2001). ‘Analysis of hull
remains of the Dor D Vessel, Tantura Lagoon, Israel’,
International Journal of Nautical Archaeology 30 (2):
K, Y., R, J. G. and H, J. (2004). ‘e Tantura
wrecks and ancient Mediterranean shipbuilding’, in F.
M. Hocker and C. A. Ward (eds), Conceptual approaches
to the study of wooden ships. College Station: 113–27.
K, S., M, M., P, L. and S , K. 2005.
Portus. An archaeological survey of the port of imperial
Rome, Archaeological Monographs 15. Rome.
K, U. (2008). ‘Life at the eodosian Harbour,
wrecks and a rapid silting’, in U. Kocabaş (ed.), e ‘Old
Ships’ of the ‘New Gate’ / Yenikapı’nın Eski Gemileri 1.
L, F. C. (1964). ‘Tonnages, medieval and modern’,
Economic History Review, Second Series 17: 213–33.
L, F. C. (1966). ‘Venetian shipping during the
F, E. W. B. (1998). ‘Preface’, in L. Saguì (ed.), La
ceramica in Italia: VI-VII secolo: atti del convegno in onore
di John W. Hayes: Roma, 11–13 maggio 1995, Biblioteca di
archeologia medievale 14 (1). Florence: 5–6.
F, E., F, S., H, R. B. A P,
P. (2004). ‘Accounting for ARS: Fineware and sites in
Sicily and Africa’, in S. E. Alcock and J. F. Cherry (eds),
Side-by-side survey; comparative regional studies in the
Mediterranean World. Oxford: 147–62.
F L, F. (1984). ‘New views on bilge pumps
from Roman wrecks’, International Journal of Nautical
Archaeology 13: 85–93.
F L, F. (1989). ‘e problem of the bilge
and the pump in antiquity’, in H. Tzalas (ed.), 1st
International Symposium on Ship Construction in
Antiquity. Athens: 91–4.
F, L. (1958). ermes romains des environs
d’Hadrumète, Notes et documents, nouvelle série 1. Tunis.
F, L. (1960). Inventaire des mosaïques. Feuille no.
57 de l’Atlas Archéologique: Sousse. Tunis.
F, L. (1967). ‘Un voilier antique’, Antiquités
Africaines 1: 83–98.
G, G. (1996). ‘Roman ﬂanged pump bearings: further
ﬁnds in the harbour of Ponza (Pontine Islands, Italy)’,
International Journal of Nautical Archaeology 25:
G, R. (1988). ‘e Venetian Port of Candia,
Crete (1299–1363): Construction and Maintenance’,
Mediterranean Historical Review 3 (2): 141–58.
G, D. (2001). ‘Shipwrecks and Hellenistic trade’, in
Z. H. Archibald, J. Davies, V. Gabrielsen and G. J. Oliver
(eds), Hellenistic economies. London: 273–312.
G, F. and B, A. (2004). Sunken Egypt.
G, F. and F, D. (eds) (2008). Egypt’s sunken
G, F., B, A., B, E., D, I., K, Z.
and Y, J. (1998). Alexandria. e submerged royal
G L R, L. (1956). ‘Fore and aft sails in the
ancient world’, e Mariner’s Mirror 42: 238–9.
H, L. and V, C. (2000). Die unsterblichen
Obelisken Ägyp tens, Zaberns Bildbände zur Archäologie.
Mainz am Rhein.
H, J. W. (1972). Late Roman pottery. London.
H, J. W. (1998). ‘Introduction: e study of Roman
pottery in the Mediterranean 23 years after Late
Roman Pottery’, in L. Saguì (ed.), La ceramica in Italia:
VI–VII secolo: atti del convegno in onore di John W.
Hayes: Roma, 11–13 maggio 1995, Biblioteca di archeo-
logia medievale 14 (1). Firenze: 9–21.
H, A., M, M., C, F. and B, M.
(1999). Parcours de villes. Marseille: 10 ans d’archéologie,
2: Developments in Mediterranean shipping and maritime trade
O, T. J. (1982). ‘e chain pump: an 18th century
example’, International Journal of Nautical Archaeology
O, T. J. (1996). Ships’ bilge pumps. A history of
their development, 1500–1900, Studies in Nautical
Archaeology 2. College Station.
O, J. P. (1984). Greek and Roman mechanical water-
lifting devices: the history of a technology, Phoenix
Suppl. 16. Toronto.
O, J. P. (1988). ‘e technology of Roman harbours’,
International Journal of Nautical Archaeology 17:
O, J. P. (2000). ‘Ancient sounding weights: con-
tr ibution to the history of Mediterranean navigation’,
Journal of Roman Archaeology 13: 293–310.
O, J. P. (2008). ‘Testing the Waters: e Role of
Sounding-Weights in Ancient Mediterranean
Navigation’, in R. L. Hohlfelder (ed.), e Maritime
World of Ancient Rome, Memoirs of the American
Academy in Rome Suppl. 6. Ann Arbor: 119–76.
O, J. P., B, C., C, S. M., C,
R., G, E. and H, R. L. (2004). ‘e
ROMACONS Project: a contribution to the histor-
ical and engineering analysis of hydraulic concrete in
Roman maritime structures’, International Journal of
Nautical Archaeology 33 (2): 199–229.
P, A. J. (1990a). ‘Classical antiquity: the maritime
dimension’, Antiquity. A Quarterly Review of
Archaeology 64 (June): 335–46.
P, A. J. (1990b). ‘e pattern of commerce
as evidenced by shipwrecks’, in T. Hackens and M.
Miró (eds), PACT 27, Le commerce maritime romain
en Méditerranée occidentale / El comercio maritimo
romano en el Mediterráneo occidental: 147–68
P, A. J. (1992a). Ancient Shipwrecks of the Med-
iterranean and the Roman Provinces, British
Archaeological Reports International Series 580. Oxford.
P, A. J. (1992b). ‘Cargoes, containers and stowage:
the ancient Mediterranean’, International Journal of
Nautical Archaeology 21 (2): 89–100.
P, A. J. (2008). ‘Artifact distributions and wreck
locations: the archaeology of Roman commerce’, in
R. L. Hohlfelder (ed.), e Maritime World of Ancient
Rome, Memoirs of the American Academy in Rome
Suppl. 6: 177–196.
P, H. (1937). Mahomet et Charlemagne. 11th edn.
P, H. (1952). Medieval cities: their origins and the
revival of trade. Princeton.
P, P. (1982). ‘Le navire romain de la Madrague de
Giens’, Comptes Rendus de l’Académie des Inscriptions et
P, P. (1995). ‘Les épaves grecques et romaines de
la place Jules-Verne à Marseille’, Comptes Rendus de
l’Académie des Inscriptions et Belles-Lettres: 459–84.
commercian revolution’, in Venice and History. e
Collected Papers of Frederic C. Lane. Baltimore: 3–24.
L, B. A P, P. (1985). ‘Direction des recherches
archéologiques sous-marines’, Gallia 43: 547–76.
MC, A. M. and O, J. P. (eds) (2004). Deep-water
shipwrecks o Skerki Bank: the 1997 Survey, Journal of
Roman Archaeology Suppl. 58. Rhode Island.
MC, M. (1998). ‘Bateaux de vie, bateaux de
mort. Maladie, commerce, transports annonaires et
le passage économique du bas-empire au Moyen Age’,
in Morfologie sociali e culturali in Europa fra tarda
antichità e alto medioevo, Settimane di studi del Centro
italiano di studi sull’alto medioevo 45. Spoleto: 35–122.
MC, M. (2001). Origins of the European economy.
Communications and commerce AD 300–900.
M, C. (1985). Le développement urbain de
Constantinople (IVe–VIIe siècles), Travaux et mémoires
du Centre de Recherche d’Histoire et Civilisation de
Byzance Monographies 2. Paris.
M, E. (2002). L’outre et le tonneau dans l’Occident
romain, Monographies Instrumentum 22. Montagnac.
M, N. and M, C. (2006). ‘Geo-
archaeological evidence for dredging in Tyre’s ancient
harbour, Levant’, Quaternary Research 65: 164–71.
M, M. (2001). ‘Mittelalterliche Hafenkräne’, in
U. Lindgren (ed.), Europäische Technik im Mittelalter
800–1400. Tradition und Innovation—Ein Handbuch.
M, A. L. (1992). ‘Medieval treadwheels: Artists’
views of building construction’, Technology and Culture
M, P. (1996). ‘e Port of Clysma (Suez) in trans-
ition from Roman to Arab Rule’, Journal of Near Eastern
Studies 55 (2): 119–126.
M, S. (2004). De rebus nauticis. L’arte della navigazione
nel mondo antico. Roma.
M, S. (2008). ‘Le atrezzature veliche nel mondo
antico. La vela a tarchia, la vela latina e altre tipol-
ogie minori’, in J. Pérez Ballester and G. Pascual (eds),
Comercio, redistribución y fondeaderos. La navegación a
vela en el Mediterráneo. Valencia: 79–111.
M, R. (1973). Roman Ostia. 2nd edn. Oxford.
M, N. (2007a). ‘e Early Roman Empire:
Distribution’, in W. Scheidel, I. Morris and R. Saller (eds),
e Cambridge Economic History of the Greco-Roman
world. Cambridge: 570–91.
M, N. (2007b). Trade in Classical Antiquity, Key
emes in Ancient History. Cambridge.
N, X. (1997). ‘Le commerce de cabotage et de
redis tribution’, in P. Pomey (ed.), La navigation dans
l’antiquité. Aix-en-Provence: 146–59.
N, J. (1993). Side im Altertum. Geschichte und
Zeugnisse, Inschriften griechischer Städte aus Kleinasien
43 (1). Bonn.
T, A. (1997). ‘Le commerce maritime dans la
Méditerranée romaine’, in P. Pomey (ed.), La navigation
dans l’antiquité. Aix-en-Provence: 116–45.
T, A., P, P., H, A., C, M.,
G, M.-F., G, M., H, E., L,
F., L, F., C-G, A., C, A.,
G, J.-M., R, G. and R, M. (1978).
L’épave romaine de La Madrague de Giens (Var), Gallia
Suppl. 34 (Campagnes 1972–1975). Paris.
T, O. (1970). Portus. Illustrazione dei porti di
Claudio e Traiano e della città di Porto a Fiumicino.
T, P. (1972). ‘Romans on the sea’, in G. F.
Bass (ed.), A history of seafaring based on underwater
archaeology. London: 65–78.
T, J. MI. and S, A. G. (1999). ‘e
Syracusia as a giant cargo vessel’, International Journal
of Nautical Archaeology 28: 105–25.
U, G. (1950). Le navi di Nemi. Rome.
W B, S. E. A. (1926). Cleopatra’s Needles and
other Egyptian obelisks. London.
W, D. (1965). ‘Forum Ware. A distinctive
type of early medieval glazed pottery in the Roman
Campagna’, Medieval Archaeology 9: 55–63.
W, R. J. (2008) Maritime technological change
in the Ancient Mediterranean: e invention of the
lateen sail. PhD thesis, University of Southampton.
W, C. R. (1989). ‘Amphorae and trade’, in M.
Lenoir, D. Manacorda and C. Panella (eds), Amphores
romaines et histoire économique. Dix ans de recherche.
Actes du colloque de Sienne (22–24 mai 1986) organisé
par l’Università degli Studi di Siena, l’Università di Studi
di Roma-La Sapienza, Le Centre National des Recherches
Scientiﬁques (RCP 403), et l’École française de Rome,
Collection École française de Rome 114. Rome: 537–9.
W, C. (2005). Framing the Early Middle Ages:
Europe and the Mediterranean, 400–800. Oxford.
W, A. I. (2001). ‘Ti. Cl. Felix and the date of the
second phase of the East Baths’, in L. M. Stirling, D. J.
Mattingly and N. Ben Lazreg (eds), Leptiminus (Lamta):
a Roman port city in Tunisia. Report no. 2, Journal of
Roman Archaeology Suppl. 41. Ann Arbor: 25–8.
W, A. I. (2008). ‘Machines’, in J. P. Oleson (ed.),
Handbook of Engineering and Technology in the Classical
World. Oxford: 337–66.
W, A. I. (2009). ‘Approaches to quantifying
Roman trade’, in A. K. Bowman and A. I. Wilson
(eds), Quantifying the Roman Economy: Methods and
Problems, Oxford Studies on the Roman Economy 1.
W, A. I., S, K. and R, C. (forthcoming).
‘Roman ports and Mediterranean connectivity’, in
S. Keay (ed.), Rome, Portus and the Mediterranean,
Archaeological Monographs of the British School at
P, P. (1997). ‘Les navires’, in P. Pomey (ed.). La navig-
ation dans l’antiquité. Aix-en-Provence: 60–101.
P, P. (2004). ‘Principles and methods of construction
in ancient naval architecture’, in F. M. Hocker and C.
A. Ward (eds), Conceptual approaches to the study of
wooden ships. College Station: 25–36.
P, P. and R, E. (2005). L’archéologie navale. Paris.
P, P. and T, A. (1978). ‘Le tonnage maximum
des navires de commerce romains’, Archaeonautica 2:
P, J. H. (1988). Geography, technology and war.
Studies in the maritime history of the Mediterranean,
R, A. (1992). ‘Sebastos: the royal harbour at Caesarea
Maritima—a short-lived giant’, International Journal of
Nautical Archaeology 21 (2): 111–24.
R, C. (1901). ‘Notes on the plant-remains of Roman
Silchester’, Archaeologia 57: 252–6.
R, P. (1995). Trade in the western Mediterranean,
AD 400–700: the ceramic evidence, British Archaeologic al
Reports International Series 604. Oxford.
R, M. (1991). La charpenterie navale romaine.
Matériaux, méthodes, moyens. Paris.
R, J. (1957). ‘Ad ciconias nixas’, Revue des Études
Anciennes 59: 320–8.
R, J. (1966). Recherches sur l’organisation du
commerce maritime en Méditerranée sous l’empire
romain, Ecole pratique des hautes études. VIe section.
Centre de recherches historiques 21. Paris.
R, J. (1975). La Marine dans l’antiquité. Paris.
S. J H, W. H. and F, G. E. (1898). ‘Excavations
on the site of the Roman city at Silchester, Hants, in
1897’, Archaeologia 56: 103–26.
S, L. (1998). ‘Il deposito della Crypta Balbi: una testi-
monianza imprevedibile sulla Roma del VII secolo?’, in
L. Saguì (ed.), La ceramica in Italia: VI-VII secolo: atti
del convegno in onore di John W. Hayes: Roma, 11–13
maggio 1995, Biblioteca di archeologia medievale 14
(1). Firenze: 305–30.
S, E. (1969). e Alexiad of Anna Comnena. London.
S, H., T, P., P, R. and O, A. (2004).
Le littoral de la Tunisie. Études d’Antiquités africaines
Étude géoarchéologique et historique. Paris.
S, J.-P., L, A. and K, T. (1980). ‘Les
carrières de marbre d’Aliki à l’époque paléochrétienne’,
in Aliki 1, Études thasiennes 9. Athènes: 79–137.
S, J. R. (1994). Wooden Ship Building and the
Interpretation of Shipwrecks. College Station.
S, M., G, A., K, S. and E, M. (2008).
‘Archaeological research of maritime infrastructure of
Roman settlements on the Slovenian coast of Istria
(1993–1997)’, in R. Auriemma and S. Karinja (eds), Terre
di Mare. L’archeologia dei paesaggi costieri e le variazione
climatiche. Atti del Convegno Internazionale di Studi,
Trieste, 8–10 novembre 2007. Trieste: 56–74.
2: Developments in Mediterranean shipping and maritime trade
integration of the Roman economy’, World Archaeology
23 (3) (February): 283–93.
Z, H. (1995). ‘Preliminary views of the Ephesian
harbor’, in H. Koester (ed.), Ephesos: Metropolis of
Asia, Harvard eological Studies 41. Cambridge,
W, A. (2000). ‘How the obelisks reached Rome:
evidence of Roman double-ships’, International Journal
of Nautical Archaeology 29: 273–83.
W, A. (2003). ‘Supplementary remarks on the
Roman obelisk-ships’, International Journal of Nautical
Archaeology 32: 121–3.
W, G. (1992). ‘Imperialism, empire and the