Content uploaded by Jeffrey Royal
Author content
All content in this area was uploaded by Jeffrey Royal on Nov 30, 2017
Content may be subject to copyright.
© Journal of Roman Archaeology
The landscape of the naval bale
at the Egadi Islands (241 B.C.)
Sebastiano Tusa and Jerey Royal
The nal bale of the First Punic War between Rome and Carthage, the bale of the
Egadi Islands, took place in 241 B.C. Finds of multiple bronze warship rams, helmets, and
amphoras, destined for a Carthaginian garrison on Sicily, conrm the naval bale's general
location and dene its landscape.1 The nds suggest new lines of inquiry into the bale's
precise location, the resulting landscape of bale débris, the dimensions of rams and war-
ships in the 3rd c. B.C., and hypotheses for eet construction and composition during the
First Punic War. Subsequent publications will provide more detailed discussions of ram
manufacture and their morphology, as well as the inscriptions present on the rams.
Contesto storico
La prima guerra punica ebbe la durata di un’intera generazione. Individuarne precise
responsabilità rientra nella casistica tucididea dei προφσεις (pretesti). Fu certamente uno
scontro epocale tra civiltà che scoppia, ancorché non pianicato, nel momento in cui il con-
fronto sempre più ravvicinato tra i due imperialismi diventa inevitabile. Roma vinse grazie
al forte senso dello Stato della nobilitas romana strenuamente esaltata da Polibio.2 Tra le
fonti Polibio rappresenta la migliore; le altre fonti sono incomplete e prolisse. E’ Polibio
che correamente individua nell’“ambizione per il dominio universale” (1.52.4: φιλοτιμα
τν λων) della classe dirigente romana la forza principale che fece loro vincere la guerra.
I presupposti dello scontro nale possono iniziare dalla fulminea e fortunata azione di
Giunio Pullo che riuscì a occupare la sommità del monte Erice e le pendici del monte dalla
parte di Trapani (1.55.10: τν π Δρεπνων πρσβασιν). Tale azione fu determinante
per gli sviluppi successivi della guerra poiché tagliavano fuori il porto naturale di Trapani
dalla portata dei Cartaginesi. Fraanto c’è da ricordare che l’eparchia punica in Sicilia non
esisteva più e, pertanto, quando nel 247 Amilcare Barca sbarcò in Sicilia, non aveva posi-
zioni da tenere lanciandosi con grande astuzia in una guerra di posizione ecacemente
(1.57.1-2), notevolmente onerosa e senza risultati concreti per entrambe i contendenti.
Ma vi fu un altro faore decisivo per la vioria romana, dovuto all’allestimento di una
oa costruita su iniziativa pubblica, nanziata delle famiglie romane più facoltose, nel
243/242, composta da 200 quinquiremi su modello di una imbarcazione punica caurata
dai Romani ad Annibale Rodio nel tentativo di forzare il blocco romano streo aorno a
Lilibeo. La novità di questo tipo d’imbarcazione risiedeva nel nuovo sistema di voga con
5 rematori su ogni remo, che garantiva una maggiore velocità. Ciò fu possibile, secondo
L. Basch, con la realizzazione di balconate lungo le murate dell’imbarcazione, che consenti-
vano una maggior larghezza fuori acqua, e il restringimento della chiglia dando maggiore
ecacia alle linee d’acqua dello scafo.3 E’ per questo che alla ne del conio si arrivò
ad una spiccata standardizzazione dell’architeura navale dimostrata anche dalla identità
1 The project was conceived and is directed through the cooperative eorts of the present authors
and their institutions, the Soprintendenza Del Mare, Regione Siciliana, and RPM Nautical
Foundation.
2 De Sanctis 1916.
3 Basch 1987, 353-54.
S. Tusa and J. Royal
8
dei rostri rinvenuti che si aribuiscono ai Romani o ai Cartaginesi soltanto per l’iscrizione
sulla guaina.
La baaglia vera e propria, combauta il 10 marzo del 241 a.C.,4 fu vinta per un con-
corso di faori positivi per i Romani. Innanzituo non lasciarono ai Cartaginesi il controllo
delle isole di Favignana e Levanzo poiché da li avrebbero forzato con successo il blocco
dei porti di Lilibeo e Drepanum, com’era avvenuto in precedenza (g. 1).5 Fu per questo
che Annone, l’ammiraglio cartaginese, provenendo da Cartagine, con la sua oa di circa
700 navi in aiuto di Amilcare accampato sulle balze seentrionali del monte Erice, si fermò
in aesa a Mareimo. Da qui fu costreo a scegliere la roa più seentrionale possibile
per due ragioni: la prima perché doveva raggiungere la baia di Bonagia a Nord del porto
di Drepanum poiché da questo l’ascesa all’accampamento di Amilcare era sbarrata dalle
truppe romane (Polyb. 1.55.7-9; 1.58.2-3).6 Fu qui che Annone commise l’errore fatale di
non intuire che i Romani si fossero nascosti a levante dell’imponente mole della punta set-
tentrionale di Levanzo. Non sappiamo se ciò fu l’eeo di un’abile diversione informativa
operata dall’intelligence romana o la repentina intuizione di Lutazio Catulo, l’ammira-
glio romano, che, intuendo la roa di Annone, spostò repentinamente la oa a Levanzo
(1.60.4). A questo punto fu determinante la quinquereme poiché, grazie alla sua velocità,
i Romani furono in grado di piombare in tempi rapidissimi sul anco della oa cartagi-
nese impedendo l’organizzazione di un’adeguata ed ecace difesa poiché in preda ad una
notevole confusione che non permise l’operatività di una normale catena di comando.
Certamente dovee giocare anche molto l’aeggiamento che i Romani ebbero verso
questo conio. A dierenza dei Cartaginesi, combaevano con l’ardore di chi vedeva
nell’esito del conio una questione di vitale importanza. Vi era un’adesione sentita e par-
tecipata alla guerra. Di tale dierenza di aitudine al conio ne sono testimonianza le
iscrizioni sui rostri. Quelle romane rappresentano punzoni che garantiscono la correezza
della fusione e la partecipazione nanziaria alla loro manifaura; quella cartaginese è un
appello agli dei. Annone, per evitare il disastro, non poté che ordinare la ritirata favorita
dal vento che gli permise una veloce navigazione con vento in poppa. Svaniva così per
sempre ogni speranza cartaginese di continuare a dominare il Mediterraneo che, di li a
poco, sarebbe diventato per secoli mare nostrum. Ma lasciamo a Polibio (1.59.1-8) il compito
di spiegarci con la sua consueta lucidità il ‘segreto’ della vioria romana:
Allo stesso modo i Romani, combaendo disperatamente, benché avessero completamente
rinunciato, ormai da quasi cinque anni, alle operazioni sul mare a causa dei rovesci subìti e
per il fao che si erano convinti di decidere la guerra con le sole forze di terra, ora, vedendo
che l’impresa non riusciva secondo i loro calcoli, e soprauo per l’audacia del comandante
cartaginese, decisero per la terza volta di riporre le loro speranze nelle forze navali, rite-
nendo che solo così, grazie a questa decisione, se avessero posto mano all’impresa nel modo
opportuno, avrebbero potuto dare alla guerra una conclusione positiva. E questo, alla ne,
fecero … L’impresa fu, essenzialmente, una loa per la vita. Nell’erario, infai, non c’erano
più risorse per sostenere quanto si erano proposti: tuavia, grazie alla generosità verso lo
stato e alla nobiltà d’animo dei primi ciadini, si trovò ancora quanto serviva per la sua
4 La data è riferita da Eutropio (2.27.2), e riportata da Polibio (1.60-61), che indica la ne della
prima guerra punica sei giorni prima delle idi di marzo dell’anno 511 dalla fondazione dell’Urbe.
Tale data è unanimemente acceata dagli studiosi anche se ha un forte caraere convenzionale
e solleva qualche problema connesso sia alla data in sè, sia alla rispondenza tra i calendari.
5 Gullea 2005.
6 Filippi 2005. Archaeological surveys carried out on the W slope of Mt. Eryx identied a wide
area with Punic poery of the same period at San Cusumano near Bonagia Bay.
The landscape of the naval bale at the Egadi Islands 9
S. Tusa and J. Royal
10
The landscape of the naval bale at the Egadi Islands 11
realizzazione. A seconda della prosperità dei loro patrimoni, infai, uno per uno, due a due
o tre a tre si impegnavano a fornire una quinquiremi equipaggiata, a condizione che avreb-
bero recuperato la spesa, una volta che le operazioni fossero riuscite secondo i calcoli. In tal
modo, essendo state rapidamente approntate 200 navi a 5 ordini di remi, la cui costruzione
condussero sul modello della nave del Rodio, in seguito, eleo comandante Gaio Lutazio,
lo fecero partire all’inizio dell’estate.
Overview of the project, the bale zone and landscape
Investigation into the bale zone was prompted by the Carabinieri’s seizure of a bronze
ram (Egadi 1) in 2004 from a private collection in Trapani. It was reportedly collected by
a sherman while drag-net shing in proximity to Levanzo Island. Subsequently, S. Tusa
aempted to localize the bale zone by examining historical sources and making an analy-
sis of the environmental conditions of the area.7 Punic nds at Bonagia Bay and numerous
anchors discovered o the NE coast of Levanzo Island provided further clues.8 The present
authors teamed up in 2005 and established the survey project, which had amongst its goals
to locate other remains from the bale zone. Assuming that the Carthaginian eet’s destina-
tion was Bonagia Bay, on the coast north of Mt. Eryx, the range of likely routes from their
anchorage at Mareimo Island would pass just north of Levanzo Island (g. 1). Allowing
for the Roman eet to round the northern tip of Levanzo Island and get into formation,
we postulated that a clash of warships occurred northwest or west of Levanzo Island. In
order to increase the likelihood of nding material associated with the bale (as well as
sites of other eras), our survey area enveloped Levanzo Island and extended to the W side
of Favignana Island. The principal methods utilized in this ongoing project are multibeam
echosounder mapping combined with verication of anomalies using a Remotely Oper-
ated Vehicle (ROV). Our primary base of operations was the research vessel Hercules.9 The
project has completed 7 seasons of eldwork and mapped over 270 km2 of sea-oor. By
the end of the 2011 season the location of at least one deposition area from the bale was
conrmed.
The bale landscape incorporates varied sites associated with the event in its widest
context (g. 1). For the Bale of the Egadi Islands the relevant area extends from Mareimo
Island where the Carthaginians anchored, through the Egadi Islands, the anchorage site
o Levanzo Island, along the W coast from Marsala (Lilybaeum) from where the Roman
eet operated, to Bonagia Bay and Erice (Mt. Eryx). Historical records assist in dening the
bale landscape, but it is through the archaeological investigation of cultural material and
its spatial relationships that the locations of specic sites and zones are determined. Since
sea bales were spread out over wide areas, our survey hopes to dene multiple areas
where warships clashed (‘bale zones’), given that some vessels reportedly sank and oth-
ers lost material overboard. Fieldwork (2005-7) dened much of the sea-oor; it indicated
that dragnet shing had impacted the area, including the bale zone we have identied.
The sea-oor in the area directly west-northwest of Levanzo Island is rife with long fur-
rows from dragnet equipment; not only are artifacts scarce, but lile macrofauna remains.
Concentrations of artifacts were found in areas of rock outcrops (e.g., west and north of
Levanzo Island) where nets cannot enter or are caught and ripped. In 2006, the Levanzo 1
7 Tusa 2005a, 63-68; Ricordi 2005, 95-105.
8 Filippi 2005; Tusa 2005a.
9 For a more detailed discussion of the methods and equipment used during this project, see
Royal and Tusa 2012.
S. Tusa and J. Royal
12
wreck was discovered north of Levanzo Island, its survival due to surrounding outcrops.10
Concluding that the rocky zones might be unaected by dragnet operations, from 2008
we concentrated our search on such zones and located our rst nds associated with the
bale, namely the Egadi 2 ram and several Greco-Italic amphora fragments. In 2009, we
continued the search in the region where the Egadi 2 ram was found, working south with
few results.
In 2010, verication by the ROV revealed a concentration of cultural material with spa-
tial and chronological associations that clearly marks at least one bale zone. The nds
include 4 bronze warship rams (Egadi 3-6), at least 8 bronze helmets, and over 175 Greco-
Italic V/VI and Punic amphoras all dispersed within 1 km2 (not yet fully examined). This
sector (designated PW [Punic War] -A [as the rst of several sites in the area]) lies nearly 7
km west of Levanzo Island and over 6 km south-southeast of where the Egadi 2 ram, and
possibly Egadi 7, were found; this laer region was designated PW-B (g. 1), at the edge
of the rocky zone. PW-A extends from the edge of an open, sandy sector (where drag-
ging scars are prevalent) into a rocky area farther west. Although dragnets have removed
almost everything to the east, we suspect that more artifacts will be found in the area con-
tinuing farther west (g. 2). Discarded dragnets found near the rocks and a lack of drag
scars on the seaoor indicate that dragging was hazardous there and was discontinued.
The artifacts in sector PW-A are thus either where they were originally deposited or only
slightly disturbed. The artifact distribution (g. 2, inset) is a product of the area searched
to date. Material from east of sector PW-A or from between the two sectors had been col-
lected in dragnets and deposited c.6 km north in sector PW-B when the nets struck rocky
outcrops. Hence, we believe that sector PW-B represents secondary deposition of material
from the bale zone, which is indicated by the débris in sector PW-A.
The rams
General features of warship rams11 (g. 3)
Prior to our survey, warship rams were an extremely rare artifact, with only three
waterline rams known, only one of them recovered from an archaeological context.12 To
date, the project has added 7 new examples. To understand both the relationship between
the rams and their warships’ hulls as oensive and defensive weapons, and the relation-
ship of ram dimensions to warship size, we must rst say something about their design and
manufacture. The terminology used follows that laid out by J. Stey for the Athlit ram,13
supplemented by additional terms proposed here (g. 3). Warship rams were constructed
by the lost-wax method whereby they were molded directly onto the bow timbers of the
nished warship.14 Hence the shape of each ram’s interior is governed by the junction
10 Ibid. 31.
11 Individual ram measurements are provided in Table 1. Those for the Egadi 2-6 rams were taken
by J. Royal and for the Egadi 7 ram by S. Zangara and J. Royal; those for Egadi 1 were less com-
prehensive and required later emendments by J. Royal. As marine growth covers the rams, they
will be re-measured and weighed following conservation.
12 The Athlit ram was recovered from an archaeological context o the coast of Israel. The Bremer-
haven and Piraeus rams lack provenience. A fourth ram, the Acqualadroni, was recovered in
2008 o NE Sicily.
13 Stey 1991, 12, g. 2-7; 18, g. 2-13.
14 Stey 1991, 38-39; Oron 2001.
The landscape of the naval bale at the Egadi Islands 13
of the bow timbers. Thus interior shapes and the overall dimensions of each ram varied
according to the dimensions at the bow of the ship in question. The timber conguration at
each bow was designed to create a strong hull for withstanding the force of the sea, redis-
tributing the force of impacts throughout the hull and providing a single point of impact
for ramming aacks. Such waterline rams were designed for shaping the forces generated
from the ramming blow, both those delivered and those incurred, and for protecting the
bow timbers during impacts. As a result, each bronze waterline ram shares basic aributes
with the others; these include components, morphology, and aachment strategies.
Each ram has a basic design consisting of a driving center, a cowl, and a boom plate.
The driving center, which encases the ramming timber and wale timbers, initially transfers
the forces during a ramming event. It is comprised of 3 horizontal ns and the port and
starboard wale pockets. It extends from the forwardmost end of the ns at the head to the
aftmost extension of the inlet and n tails. From the inlet, the wale pockets recede from
the outer extent of the ns to form the n cavities, converging at the head to form the ver-
tical post that runs perpendicular to the center of the ns. Wale pockets have trapezoidal
cross-sectional shapes as they were formed over and encased the termini of the port and
starboard wales as well as the ramming timber; as such, the wale pockets serve to protect
these timbers and their scarfs. The forwardmost portion of the wale pockets that abut the
ramming timber join to form a solid metal mass. Each n increases to its maximum thick-
ness roughly one-quarter of the driving center’s length measured from the head. While the
upper and lower ns thicken more on their inner faces, the middle n thickens on both
faces, producing a lancelet shape. At their termination at the head, the upper n curves
upwards, the lower n downwards, while the central n largely remains horizontal. From
their thickened and ared ends at the head, the ns converge on the sides of the driving
center at the inlet to form the n plates. Fin widths at the head follow a paern whereby
the upper n is slightly wider than the middle n, which in turn is slightly wider than the
lower n; this results in a trapezoid-shaped forward end. Viewed from above in plan, the
ns have a constricted waist that forms the narrowest, and often shortest, part of the driv-
ing center. Ribs of c.0-3 – 1.0 cm3 run centrally along the port- and starboard-side faces
of each n, from the head to just before the termini of the n tails. Fin tails are shaped as
narrow projections with rounded ends on all the rams, except for the middle ns of Egadi
S. Tusa and J. Royal
14
6 and 7. All the rams are hollow, except for c.5-10 cm of solid metal at the head where the
wale pockets converge and the cowl nosing and boom plate constrict and align with the
vertical post.
The cowl nosing is the forward cowl face that covers an internal chock timber; this tim-
ber sits forward of the scarf that joins the stem and ramming timber. From the upper n,
the cowl sweeps upwards from the head in an increasingly steep curve. Both sides of the
cowl extend aft to a point even with the inlet. Since a ram was modelled over a warship’s
stem and chock that had a trapezoidal cross-sectional shape, the cowl has a narrowed
forward face (cowl nosing) that widens at its aft opening. The ram’s cowl sheathed and
protected the warship’s stem and chock, as well as the important points where timbers
joined one another. One of the most critical joins was that of the stem and the ramming
timber/deadwood, formed by an intricate scarf and aachments. From the head, the bot-
tom plate extends at an angle towards the inlet; in some instances it extends beyond the
after-end of the inlet to form a tailpiece. This angle typically is not perfectly linear but
forms a subtle concave curvature as it was molded to encase and protect the keel and ram-
ming timber, as well as the essential scarf that joined them.
The rams were aached to their warships with bronze spikes. In every case, except for
Egadi 1 and 7, the aachment paerns on the rams’ driving centers, cowls, and boom
plates are similar. Two bronze spikes were used on each side of the driving center on the
n plate, near the inlet, and in line with the two n cavities, to secure the ram to the wales.
Three bronze spikes were normally used on each side of the cowl towards its aft edge to
aach it to the stem, but Egadi 1 and 7 used two per side.15 The aachment of the boom
plate typically included at least one spike centrally located, near the plate’s aft terminus.
Although the rams are not fully cleaned, where discernible additional aachments exist in
most cases on the starboard and/or port sides of the rams’ boom plates.
Egadi 1 warship ram (g. 4)
A previous publication of the Egadi 1 ram provides some of the dimensions utilized
here.16 This ram has no provenience other than reportedly being recovered by a sher-
man from the sea north of Levanzo Island; even if this is accurate, the precise place where
the ram was picked up by the net, as well as the course of the vessel before the ram was
removed from the net, remain unknown.17 The ram is intact with no major damage or defor-
mation. The only discernible damage are a small section aft of the upper, starboard cowl
aachment, a small section aft of the lower port cowl aachment, and a small indention on
the starboard side of the upper n at the head. No construction marks were noted on the
interior or exterior surfaces. Egadi 1 has the shortest driving center of the Egadi rams, being
nearly identical in length to Egadi 5, yet has the longest tailpiece and the highest mid-length
height and width. The reduction from the head to constricted waist is slightly greater than
from the inlet. Given the ram’s signicant increase in height from its constricted waist, it
possesses the third tallest and second widest head. Its large head combined with a short
driving center gives this ram a stubbier design than the others.
15 Further cleaning of the Egadi 7 ram revealed an additional aachment on the cowl nosing.
16 Tusa 2005a, 2005b and 2010; Buccellato and Tusa 2007. However, Royal was able to supplement
these in 2012.
17 Without knowledge of its provenience, method of aachment (lost prior to its seizure), initial
state of marine growth or associated nds, lile can be deduced about the ram’s deposition.
Thus it is not conclusive that this ram is associated with the events of the bale of 241.
The landscape of the naval bale at the Egadi Islands 15
In addition to the 4 cowl aachments, there was a single aachment through the cowl
nosing driven into the chock. There are no aachments on the boom plate to ax the ram
directly to the keel. The arrangement of the locations of spikes and the presence of decora-
tive features on the n plate are also unique. Two distinctive ‘rosee’ elements formed by
a central rounded knob surrounded by 5 petals are located on the n plates of each side;
they are set in line with the n cavities so that the spike aachments are displaced to just
above and aft for the upper spike, and just below and aft for the lower spike aachments.
A Latin stamp was discovered on the lower cowl nosing (see below).
Egadi 2 (Catherine D18) warship ram (g. 5)
The Egadi 2 ram was discovered in 2008 within sector PW-B c.10 km northwest of
Levanzo Island. A thorough search of the sea-oor in a 50-m radius yielded no other arti-
facts. A number of broken Greco-Italic type V/VI amphoras were found amongst rock
outcrops within 500 m to the north-northeast. The ram lay inverted on a at seabed,
roughly one-third buried in soft sand. Excavation around the ram to a depth of c.15 cm to
ascertain its structural integrity revealed no artifacts. The surrounding matrix consisted of
sand, small shells and stones, with a substrate of hard-packed sand below. This was not its
18 Each ram recovered during our survey or delivered to authorities after collected from the Egadi
Islands is designated with a running numerical designation (Egadi #). Those rams discovered
during our own survey operations receive an additional designation of a name.
S. Tusa and J. Royal
16
original place of deposition, to judge from the minimal amount of marine growth on the
exposed boom plate and lower n (i.e., its upper surfaces when discovered). Its original
location had lasted much longer, as indicated by the greater amount of marine growth
on its upper half. Some 40% of the upper n in the aft portion and most of the cowl were
missing; none of the breaks were fresh and marine growth covered portions of them. As
discovered, the ram’s upper half was almost entirely buried in sand, which had killed the
marine growth that had originally formed there. Since most of the damage was on this
upper half and since the breaks were covered with growth, several conclusions are possi-
ble: a) the ram had sat in an upright position upon its initial deposition; b) the damage to
the upper half could not have occurred when it was repositioned upside down in the spot
where we discovered it; and c) this damage must have occurred prior to the ram’s original
deposition, the most likely cause being a violent collision with another warship during the
bale. This conclusion is supported by the wood remains trapped between the starboard
side of the lower and middle ns at the head, a result of ramming an enemy hull.
Because the ram is split, the cowl nosing from the head extends c.16.0 cm aft, where it
joins with the surviving section of the cowl. The cowl nosing widens from 0.5 cm at the
head to 5.5 cm at a point c.7.0 cm aft. A gentle upward angle is preserved in the surviving
portion of the cowl, along with a small preserved portion of the join between the starboard
side of the cowl and the upper n. This indicates it was probably of the same general shape
as the other rams. The forward corners on the starboard side of the middle n and port side
of the lower n are missing, but the ram’s boom plate and short tailpiece are undamaged.
The ram’s n tails are longer and come to a narrower rounded point than on other Egadi
rams. Groove-and-ridge paerns are found on the interior of the wale pockets, a product of
shaping the wax model on the ship’s bow timbers. This ram has the third smallest driving
center and its head’s height and width dimensions are the smallest of the rams found thus
far. The head has only a moderate amount of are. Since the inlet is wider than the head, it
has an asymmetrical shape in plan.
As the ram is incomplete, it is impossible to determine the locations of all spike aach-
ments that axed it to its warship’s bow timbers. Two spike connections survive on the port
and starboard n plates. An additional bronze spike (1.07 cm diam.) was driven through
the starboard side of the boom plate at c.2.0 cm from the join with the boom n; none was
located on the port side. One of the surviving spikes, located between the lower-middle ns
on the starboard n plate, extends more than 6 cm into the ram’s interior with no defor-
mation or bending, and specically none towards the inlet end. There is no indication of
deformation on the ram’s driving center or boom plate that would indicate that the war-
ship’s timbers and ram had been pulled apart forcefully; thus the ram must still have been
aached to the wale timbers when it was deposited. Initially deposited in an upright posi-
tion, only exposure to teredo can account for the absence of timbers as well as the lack of
deformity to the aachments left in place. Here too a heavy layer of marine growth formed
on the upper half of the ram (including the damaged edges of the missing upper n and
cowl section). After being dragged from its original place of deposit (possibly from the E
edge of our sector PW-A) within recent decades, the ram was deposited in an inverted posi-
tion with its damaged and growth-covered upper half now buried. In its new location the
marine growth on the now-upright boom plate has been minimal.
Egadi 3 (Vincenzo T) warship ram (g. 6)
The ram was found in 2010 in sector PW-A lying on its port side at a depth of 81 m. No
intrusion into the shallow sediment was noted and no other surface deposits of artefacts
The landscape of the naval bale at the Egadi Islands 17
were observed within a radius of 3-4 m. Starting at a distance of 5 m, several amphoras
included Greco-Italic and Punic types. Rocks, shells and other débris, caused by octopus
nesting, lled and spilled from the ram’s cavity on the seaoor. There are no major miss-
ing portions or damage other than broken n tips at the head, slight damage on the aft
end of the boom plate and n tail, and a series of V-shaped notches across all three ns
at the head midway between their port sides and the vertical post connection. The notches
are vertically aligned, decrease in depth from top to boom of the ns, and make a crum-
pled paern formed by a forward impact. All of the damage is covered with identical
amounts of growth present on the rest of the ram’s surfaces, which indicates damage pre-
deposition. In addition, the structured light scanning (SLS) and X-ray uorescence (pXRF)
examination indicates the corrosive period, or the time exposed to seawater, for all the
damaged areas is equivalent to the time of exposure for the undamaged portions of the
ram’s exterior. Both sets of evidence are in accord and place the timing of the damage prior
to deposition.19 The nature of this damage is enigmatic, although it is clear that the ram
struck a V-shaped metal object head-on. Since the V-shaped notches match the angle of the
vertical post at a ram’s head, one possibility is a ram-to-ram collision. In this instance the
ram could have struck a ram with a greater head height, such that each n made contact
with the opposing vertical post.20 It is also possible that the ram struck some other metal
19 Hassebrook et al. 2011. SLS utilized photographic images from dierent angles with various
light paerns, while pXRF scans the surface area of the object in a three-phase scan to deter-
mine elements in three spectral ranges. Subsequent analysis provides the chemical composition
of the object’s surface, micro-fracture information, and produces a three-dimensional model.
20 The lack of damage on the ram’s vertical post, presumably where the opposing ram’s ns
would have struck, may be due to the greater structural strength of vertical posts formed by
thick solid metal than of the ns. Hence, the opposing ship’s ram also suered damage to its
ns, but not its vertical post.
S. Tusa and J. Royal
18
object of a similar shape, or the corner of an object with a square cross section axed to an
opposing ship.
Compared to the others, this ram has moderate dimensions which include an aver-
age length driving center, average head width and height (Table 1). It has no tailpiece, is
symmetrical in plan, and shows lile change in height from the head to central waist. The
exterior of its boom plate is covered with groove-and-ridge paerns running perpendicu-
lar to the plate’s length along its starboard and port sides, while the marks run lengthwise
on its boom face. They were undoubtedly made when the wax mould was shaped by a
tool, yet no eort was made to smooth the exterior surface of the boom plate prior to cast-
ing. An inscription in Punic script of the 3rd c. B.C. (see below) is located on the ram’s cowl
nosing; it was made by incision with a sharp instrument that cut into the wax layer of the
mould prior to casting. An experimental analysis21 of the Egadi 3 ram provided an indica-
tion of its composition on its outer, corroded surface. The unusually high percentages of
lead (27.0-36.3) and tin (12.7-20.2) as compared to copper (37.7-41.2) is partly explained by
the leaching of copper. Analysis of bronze cores is required to determine original relative
percentages, but it is already clear that a surprisingly high percentage of lead is present.
TABLE 1
SELECT RAM FEATURES, MEASUREMENTS, AND CALCULATIONS
Egadi 1 Egadi 2 Egadi 3 Egadi 4 Egadi 5 Egadi 6 Egadi 7
FEATURES
Find location Unknown PW-B PW-A PW-A PW-A PW-A Unknown
Cultural aliation manufacture Rome ? Carthage Rome ? Rome Rome
Cowl curve rise Vertical Vertical? Vertical Vertical ? Vertical Vertical
Cowl join location Set back Set back Set back Set back Set back Set back Set back
Cowl nosing X X X X X X X
Waisted ared ends X X X X X X X
Lanceolate ns X X X X X X X
Tail n paern Points Points Points Points Points Indent Indent
Tail n extension Moderate Short None Short None Short None
Molded decoration X ? - X ? X X
Inscription Latin Punic Latin Latin Latin
Driving center aachments 4 4 4 4 4* 4 4
Cowl aachments 4 + 1 - 6 6 - 6 4
Boom plate aachments 0 1 2 1 3 2 or 1 2 or 1
MEASUREMENTS
Max l. 84.0 76.5 85.0 93.5 64.0 100.9 74.0
Driving center l. 58.8 66.3 74.4 83.4 59.5 85.9 69.0
Head h. 22.2 19.9 21.5 21.8 21.5 24.0 24.0
Average head w. 40.5 31.1 37.4 35.6 32.2 40.6 38.0
Average inlet w. 39.7 38.4 37.9 35.5 31.5 44.5 39.0
Fin w., central constriction 33.3 27.7 31.6 27.2 28.5 30.6 33.6
Fin h., central constriction 21.0 17.7 20.2 18.0 17.0 20.3 20.0
Cowl vertical h. 38.8 - 36.9 35.6 - 30.0 38.0
Length along cowl face ? - 91.2 96.0 - 94.0 88.5
Th. at inlet 1.5 1.1 0.9 0.5 0.8 0.9 0.9
21 X-ray uorescence was chosen by the researchers as it was a non-destructive method that pro-
vided rapid elemental composition of corrosion and metallurgy for the two bronze artifacts:
Hassebrook et al. 2011.
The landscape of the naval bale at the Egadi Islands 19
Egadi 1 Egadi 2 Egadi 3 Egadi 4 Egadi 5 Egadi 6 Egadi 7
Th. of boom plate aft 0.7 1.0 1.1 0.5 1.1 0.9 0.8
Length of tail piece 24.5 15.5 0.0 18.0 0.0 15.1 0.0
Upper n th., mid length 5.3 3.7 4.2 3.2 3.0 3.6 3.5
Middle n th., mid length 5.0 3.2 3.8 2.9 3.4 3.4 3.1
Lower n th., mid length 4.5 3.5 3.6 2.9 3.3 3.7 2.9
Avg. n th., forward bulge 5.6 4.6 4.9 4.1 4.6 4.4 4.0
Rib th. range 0.6-0.8 0.8-1.0 0.6-1.0 0.7-0.9 0.7-1.0 0.6-1.0 0.3-0.8
Rib extension range 0.6-0.8 0.8-1.1 0.7-1.2 0.8-1.0 0.8-1.0 0.6-1.0 0.4-0.8
Outer cowl w. at top, aft 15.7 - 21.1 13.0 - 17.5 16.5
Outer cowl w. at top, fore 9.5 - 11.5 8.5 - 10.6 10.0
Spike head diam. n plate 1.1 1.7 - 1.3-16 1.4-1.8 1.3 1.3
Max. inner boom plate 15.3 x 9.1 12.0 x 12.0 9.0 x 13.5 10.0 x 10.4 17.1 x 12.4 14.4 x 14.5 11.7 x 11.4
Max. inner wale pocket 17.3 x 14.4 17.1 x 8.0 15.0 x 10.0 15.0 x 9.0 12.7 x 9.0 17.7 x 11.5 21.1/14.7
x 11.8
CALCULATIONS
Est. keel molded x sided 13.3 x 11.1 16.8 x 14.0 18.6 x 15.5 14.9 x 12.4 17.3 x 14.4 19.8 x 16.5 16.1 x 13.4
Est. keel cross-section (cm2) 147.6 235.2 288.3 184.76 249.1 326.7 215.7
Stem max. molded x sided 18.0 x 16.5 - 22.0 x 20.0 17.0 x 15.0 - 20.0 x 18.0 22.0 x 20.0
W. reduction inlet to waist 16% 27.9% 16.6% 23.4% 9.5% 31.2% 13.8%
W. reduction head to waist 17.7% 10.9% 15.5% 23.6% 11.5% 24.6% 11.6%
H. reduction head to waist 5.4% 12.1% 6.0% 17.4% 20.9% 15.4% 16.7%
Head w. to inlet w. coecient 1.0 0.8 1.0 1.0 1.0 0.9 1.0
Striking surface (cm2) 70.8 57.9 68.6 65.6 61.6 74.0 69.5
Striking area (cm2) 899.1 618.9 804.1 776.1 692.3 974.4 912.0
Head aspect coecient 0.55 0.64 0.57 0.61 0.67 0.59 0.63
H./h. = height, W./w. = width, th. = thickness, Avg. = average, Max. = maximum, Est. = estimated
The ram was axed to the bow with aachments additional to those of a typical fasten-
ing paern; they include a spike driven on each side of the cowl’s base near the inlet edge,
and an additional bronze spike driven through the starboard side of the boom plate near
the aft edge of the tailpiece. One of the spikes (0.8 cm max. diam.) through the starboard
n plate remained intact; it extends nearly 3.0 cm into the ram’s interior; it is not deformed
and is clearly in its original orientation. Three bronze spikes with round, tapered shafts in
the cowl also remained in place, extending into its interior, although their lengths may be
truncated. None of these aachments seemed to be deformed and their orientation was
parallel to the stem’s sided face.22 Both iron and bronze intact aachments were found
loose within the ram’s interior cavity. One loose iron nail (PW10-0002-01, length 7.2 cm.,
diam. of head 1.7 cm; g. 6) was straight; another (PW10-0002-02, length c.14 cm, diam. of
head 1.2 cm; g. 6) was bent. These iron nails were not associated with the ram aachment
but were used in timber aachments inside the warship. In view of the iron nails and the
undeformed aachments in its interior, it is evident that the ram was deposited while still
aached to the bow timbers. The timbers were subsequently eaten away by teredo worms,
leaving the aachments in place. The pXRF analysis indicates the interior of the ram has
endured less corrosion time than the exterior, for the entire inner surface was lled and
22 Sided refers to the faces of ships’ timbers that are parallel to the vertical plane; the sided dimen-
sion is the distance between these faces. Molded faces of timbers are those parallel to the hori-
zontal plane; the molded dimension is the distance between them.
S. Tusa and J. Royal
20
blocked from the ow of water for a period after its deposition.23 This implies that the ram
remained ed on the bow of a warship at the time of its deposition. The small stones,
ceramic sherds, bronze nails and iron nail concretions dragged into the cavity by an octo-
pus show that the ram remained near a wreck site, probably that of its own warship, after
the timbers had roed away.
Egadi 4 (Claude D) warship ram (g. 7)
The ram was recovered in 2011 in sector PW-A at depth of 80 m and c.100 m south of
a rocky area that extends slightly farther east. Here it was largely protected from dragnet
operations running north-south. The ram was found inverted within a slight depression,
with minimal burial, siing above an intact, half-buried Greco-Italic V/VI amphora (PW11-
0011). Excavation around ram and amphora to free them for recovery revealed no other
cultural material other than st-sized stones, two of which were raised for analysis; 6 addi-
tional stones were found inside the ram’s cavity (g. 7, inset). These smooth stones appear
intrusive to this region and are of sizes consistent with ballast stones.24 Also found within
the cavity were 3 tile fragments, a poery sherd, and an iron nail concretion from the hull.
This cultural material was in the innermost section of the cavity, at the head; near the
entrance to the cavity were primarily shells and rough fragments of rock, which again sug-
gest that the cultural items were pulled into the ram by octopuses after the wood inside it
had roed, and that the ram remained near the site of the wreck. A small section of the keal
was preserved in the forward section of the boom plate.
The primary damage to the ram was the missing front corner portion on the starboard
side of the upper n; the corrosion and marine growth indicate that this was an old break.
23 Hassebrook et al. 2011.
24 Several stones are undergoing petrological analysis by M. Aulinas Juncà and G. Alias (Depart-
ment of Geology, University of Barcelona).
The landscape of the naval bale at the Egadi Islands 21
The straight angle of the break, without associated deformation, indicates that the corner
was cleaved o by a direct impact on the head. This kind of sheer break requires a great
concentrated force, a force realized during a ram-to-ram collision in which only the top n
of the ram’s starboard side bore the initial hit. The port side of the upper n is deformed
upwards, a possible result of contact later in the impact event; occurring later, the force
would have diminished when the port n was struck (a split due to casting imperfections
is unlikely, given that rams were cast head down and weaker planes within the casting
were perpendicular to the long axis).
This is the second longest ram overall and has the second-longest driving center (Table
1). In plan it has a symmetrical shape, with a signicant amount of head are (the most
in its width). Rectangular chaplet ends, made ush, that were placed into the mould for
casting are visible running along both n cavities on the port and starboard sides. On the
cowl nosing is a molded decoration and inscription (see below); no other decorations or
construction marks appear on the visible surfaces. All the bronze aachments on the cowl
and driving center were preserved to some degree; they extend into the interior of the
ram. Each aachment was either straight or curved in a manner consistent with aach-
ments being driven into wood. The center bronze spike on the starboard side of the cowl
was completely straight and intact through to the opposing face of the cowl. It appears the
spikes were purposefully angled, presumably with pilot holes, aftward into the stem; this
would have been prudent so as not to risk spliing the stem along its thinner outer face.
In order for these bronze spikes to be preserved in this manner, the ram must have been
deposited on the sea-oor while aached to the bow timbers. Further evidence for the
presence of the bow timbers can be seen in the remains of wood within the starboard wale
pocket and boom plate, preserved due to the leaching of copper.
Marine growth was heavier on the central portion of the driving center’s starboard
side than towards the head and inlet; the greatest amount was on the underside of the
boom n and boom plate. A moderate growth was seen on the port side of the driving
center and the cowl, but the starboard side of the middle n at the head was largely free
of growth. This paern is consistent with the ram still being aached to the warship when
it came to rest on the seaoor. The length of hull aft of the ram would have been over 6-8
m in order to counterbalance the ram’s weight and prevent it from tipping forward. If the
ram had tipped forward while aached to its bow, the starboard side of the lower n and
boom plate would have sunk into the sand and no growth could have accumulated, nor
would the ram have fallen over in an inverted position. Once degraded by teredo, the
ship’s bow timbers weakened suciently so that the ram fell free because of its weight. It
struck and rolled o the amphora below it, aided by its heavy cowl, coming to rest on its
starboard side. Given the missing starboard portion of the upper n at the head, the mid-
dle n would have nested into the sand, bearing the ram’s weight at the head. The ram
remained in this position until its discovery, inverted with its inlet extending above the
amphora, and only slightly buried. Here it rested on the upper portion of the starboard
side of the inlet end driving center as well as the upper-starboard tip of the cowl and on
the starboard side of the middle n at the head. Marine growth adhered to the exposed
portions as it slowly seled into the shallow sediment layer, but the growth was less where
it rested on the sea-oor and almost completely absent on the starboard side of the mid-
dle n at the head. Additional support for this depositional sequence is found on a stain
across the forward starboard quarter of the ram that was created through proximity to an
S. Tusa and J. Royal
22
iron object25 that oxidized, the angle of which matches the plane of the sea-oor in relation
to the ram as discovered and shows that the iron object was lying beneath the ram where
its head came to rest. As iron oxidizes quickly in seawater, the ram will have come to rest
on its starboard side shortly after the warship reached the sea-oor. This is consistent with
the degradation of the hull timbers and the lack of marine growth on the starboard side of
the middle n at the head.
Egadi 5 (Rachael R) warship ram (g. 8)
The Egadi 5 ram was found in 2011 on the E edge of site sector PW-A, c.110 m southeast
of Egadi 4 at a depth of 84 m. Its ndspot was slightly beyond the westward extent of the
central at sandy zone, at a point where the amphora scaer ends and dragmarks begin;
here is the apparent limit of the area protected from N–S dragnet operations. The ram was
found inverted with a small piece of static net beneath it. Static net operations, which took
place within sector PW-A through 2011 and have the ability to move lighter artifacts small
distances, could not move an intact ram, but this ram is fragmentary, having lost most of
its cowl and its port side. Some 30% of its upper n remains, along with most of the boom
plate. No decorations or writing were cast or inscribed on the surviving portion, but the
groove-and-ridge paern, similar to that found on Egadi 2 and 3, is present along the bot-
tom plate, again from unsmoothed tool marks in the wax layer prior to casting.
Although this ram has the shortest driving center and one of the smaller head widths,
its head height is on a par with that of the other Egidi rams. Its keel dimensions (as indi-
cated by the width and height of the channel of the boom plate) are larger than all the
others except Egadi 6, which has similar dimensions. This ram has a symmetrical shape
in plan, with the greatest decrease in height from head to the mid-length waist yet lile
reduction in width over this same length; as a result it has one of the stubbier shapes. The
short driving center reects a short ramming timber on its warship’s bow, yet the keel
dimensions indicate a vessel of similar dimensions as is associated with the other rams.
Thus the ram was modeled over the short ramming timber that required a suciently
wide head and increasing width at the inlet as the ram extended onto the widening hull.
25 Common iron objects aboard ships included the fastenings in the hull timbers, anchors, and
tools; given that this was a warship, other possible objects include weapons and ings of
various kinds.
The landscape of the naval bale at the Egadi Islands 23
The short distance between head and inlet, as well as the large internal timbers, eectively
limited the degree of narrowing of the ram’s width at the waist. With lile forward pro-
jection from the ship, there was insucient length to expand fully the width at the ram’s
head; thus in order to aain a sucient striking area at the head, the boom n splayed
downwards at a greater angle at the head, which accounts for the relatively high reduction
in n height from the head to the mid-length waist.
Marine growth was heavy on the ram’s starboard side, the boom plate, most of the
remaining top n, and throughout its interior. Growth was observed on all of the broken
edges. The paern suggests that this portion of the ram was exposed for a long period.
However, two c.2-cm areas on the broken edge of the top n are devoid of growth, exhib-
iting a green color characteristic of bronze corrosion; here small fragments were recently
separated, possibly due to the action of the static net. Both bronze spikes on the starboard
n plate have their heads (diam. 1.4 cm) in place, but only a nub of each of their interior
length survives. Two spikes on the underside of the boom plate are missing, but a por-
tion of the one on the starboard side survives (diam. of head 1.4 cm), curving forward c.4.0
cm into the ram’s interior, which indicates that this piece of the ram was still aached to
the keel when it was deposited on the sea-oor. The ram’s port side and cowl suered
great damage; a complete sheering along its longitudinal axis evidently occurred prior to
deposition. To judge by the boom plate aachment, at least part of this badly damaged
ram remained xed to the warship’s bow timbers; then after degradation of the timbers
separate pieces fell away. The portion recovered was evidently snagged in a static net and
moved a short distance, to be deposited in an inverted position; the rest of this ram is prob-
ably still to be discovered. Given the keel aachment, it is not possible that this piece of
ram fell from the warship while it was on the surface: it is not feasible that a ship’s keel at
S. Tusa and J. Royal
24
this junction with the wales at the bow, or any part of the bow, could become separated
from the rest of the hull (see below).
Egadi 6 (Cece) warship ram (g. 9)
This ram was found in 2011 at the W edge of sector PW-A at a depth of 72 m, nearly
500 m west of rams 4 and 5. Completely intact, the ram was siing upright at the base
of a large outcrop, listing to starboard. This location is inaccessible to dragnets, while its
weight prevented it from being disturbed by static nets. Thus there is no indication that it
has moved after deposition on the seaoor. Due to the extremely thin sand cover, heavy
marine growth adhered to nearly every surface except the corner of the boom n on the
starboard side at the head, where the n supported the ram’s weight. Nesting by octopuses
brought numerous nds into the ram’s cavity, all near the head: iron nails, a tile fragment,
two bronze cheekpieces for Montefortino helmets (PW11-0034-001 and PW11-0034-002;
g. 9), and a bronze hinge (PW11-0034-003; g. 9) that connected a cheekpiece to a helmet
(see below). Five of the iron nails range in length from 7.4 to 12.1 cm (g. 9, inset) with
square heads (diam. 1.0-1.3 cm) and shafts that taper (from 0.5-0.6 cm diam. near the head
to 0.1-0.3 cm at their tips). These nails were clearly used for timber aachments within the
hull and are not associated with aachment of the ram to the hull. They suggest that the
ram has remained near the original wreck site.
There was lile damage to the ram other than bending of the upper and lower starboard
ns at the head, almost certainly due to it having rammed into the hull of an enemy ship;
wood trapped between the upper and middle ns and the middle and lower ns at the
starboard side of the head aests to this. The cowl nosing bears a molded decoration and
inscription (see below). All of the ram’s surfaces were smoothed of manufacturing marks.
A distinctive design feature is the concave prole between its upper and lower tail ns at
the inlet. It has the longest driving center (it is slightly longer than Egadi 4) and the longest
tailpiece of all the rams. Its wide mid-length dimension ares to the head and inlet, giving
a slightly asymmetrical shape in plan. The head has a high degree of horizontal splaying
that is not oset by the increase in height; as a result, it has the widest shape of the Egadi
rams. The inner dimensions of the wale pockets and boom plate channel indicate slightly
larger wales and keel, respectively, than for all other rams except Egadi 5.
Four bronze aachments on the cowl and two on the n plates extend from 3.0 to 7.2 cm
into the ram’s interior at varying angles consistent with their being driven into wood from
the outside and not bent outward. The other two n-plate aachments have only surviving
heads, but taken together they aest to the ram being aached to the bow timbers when
it reached the boom. Further evidence for the presence of the bow timbers comes in the
signicant wood remains within the wale pockets and boom plate, and the remnants of
the ramming timber at the head, all preserved by the leaching of copper. With the upper
n cleaned, at least two rectangular chaplets, made ush, were noted on its upper surface
towards the inlet.
Egadi 7 (Maltese) warship ram (g. 10)
The Egadi 7 ram is without a veriable provenience: reportedly it was pulled from the
northwest of Levanzo Island by shermen. Although there is uncertainty as to whether it
was found within the bale zone, it is likely that it is associated with the event. A molded
decoration and a Latin inscription extend down the upper portion of cowl’s nosing (see
below) and, like Egadi 6, it has a concave indention between the upper and lower tail ns
at the after end of the n plate. A series of initial measurements allow for a comparison
The landscape of the naval bale at the Egadi Islands 25
with the other rams.26 In general its shape, size, and features are consistent with the others:
its driving center length, width of inlet, thickness of n, and height of driving center are
values in the middle range, but it has some of the larger head dimensions. Thus it is one of
the stubbier rams, with a relatively short length along the cowl face. It is the widest ram at
the waist, but maintains a symmetrical shape in plan. Damage is slight, with similar miss-
ing portions located just aft of the upper, starboard cowl aachment. Small notches on the
upper n at the head, particularly the port side, seem old and worn smooth. Perhaps this
ram was ed on a warship that had seen a long period of service. The most conspicuous
damage is the possible complete removal of the face of the molded head, the ragged bound-
ary and lack of any other deformation suggesting deliberate removal in antiquity.
Helmets (Table 2)
Helping to conrm sector PW-A as the bale zone are at least 8 bronze helmets, scat-
tered mostly in a concentrated area (see below). Two cheekpieces (both for the left side,
so representative of two helmets) and one hinge connection were also found within
the Egadi 6 ram. Seven helmets lay in a limited area of c.25 m2; the eighth was found
in the rocks c.200 m to the southwest, within 30 m of the location of Egadi 6. All were
lying on the boom and not buried to any signicant degree. A further helmet (now in
the Favignana Maritime Museum) was reportedly raised from sector PW-A before 2005
by shermen with static nets. Four helmets, all of the Montefortino type, were raised in
2011 (g. 11, indicated by PW11 numbers); sections of the dome were missing on two
(PW11-0010 and PW11-0031), while the other two were intact. Even though there is no
standardization in their size, these four helmets exhibit a number of similar characteristics:
26 Measurements and photographs are by S. Zangara (Soprintendenza del Mare) and J. Royal.
S. Tusa and J. Royal
26
manufacture from cast bronze that was then worked into a thin dome; a dome with a
solid crest knob at its apex (missing on PW11-1101); a thickened lower rim projecting
on the back side as a neck guard; a rope motif running around the lower rim and neck
guard (g. 11); and, inside the rope paern, at least two incised lines (dicult to dis-
cern on PW11-0010) around the rim and neck guard. The same decorative features are
present on the example in the Favignana museum. After cleaning of PW11-0010, PW11-
0011 and PW11-0032, the rope paern is seen to be twisted in opposite directions on the
left and right sides until meeting at the center line. An additional basket-weave paern
runs inside the rope paern and below the incised lines of the neck guard on PW11-0010.
All of the crest knobs seem to be cast and appear seamlessly integrated with the helmets;
TABLE 2
SELECT MEASUREMENTS AND DESIGN ELEMENTS OF BRONZE HELMETS
PW11-0010 PW11-0030 PW11-0031 PW11-0032
Find location PW-A PW-A PW-A PW-A
Missing dome 35% - 45% -
Max. h. 23.4 21.2 26.0 27.1
Max. diam: neck guard to brim 29.6 24.3 27.3 27.2
Max. diam: ear to ear 25.0 19.8 23.8 23.1
Max. diam: brim to neck guard 26.1 21.8 24.8 24.5
L. of neck guard at center 3.5 3.1 3.1 3.0
Distance ear to ear at mid h. (interior) 21.0 18.4 19.4 20.2
Th. neck guard at center 0.3 0.4 0.4 0.5
Th. brim center 0.4 0.4 0.7 0.7
Diam. of neck guard at brim mid h. (int.) 23.4 19.8 21.6 22.2
Knob h. - 2.3 2.8 3.3
Knob max. diam. - 2.4 2.7 2.9
Th. at break at mid h. 0.1 - 0.2 -
Percent encrusted 90% 80% 95% 95%
Rope decoration X X X X
Incised line paern X? X X X
Rope extension from ear 0.4 0.7 1.2 0.5
Approximate neck guard angle 40˚ 40˚ 15˚ 15˚
they are solid metal in their upper portion and do not have holes for aachments. Where
discernible, the crest knobs are decorated with a molded paern of a central circular ele-
ment with outward radiating spokes. What appears to be possibly a semicircular chaplet is
located at the center front edge of helmut PW11-0010; no aachment evidence is found on
the interior opposite this feature.
Both bronze cheekpieces (g. 9) were thin (from 1.0 mm at their edges to c.2.0 mm at
their centers) and both have two forward projections for guarding the cheekbone and jaw.
The more intact (PW11-034-002) is nearly 15.0 cm long and 8.5 cm wide, with forward
projections c.3.0 cm long. The bronze hinge (PW11-034-003), c.11.0 cm long, found within
the Egadi 6 ram, aligns perfectly with the opposite hinge projections on the cheekpieces.
The hinge itself is a aened bar with two at projections at each end curled on their
extremity; the angle of the curl when found was c.180°. Projections for hinge connections
at the top of cheekpiece PW11-034-002 were formed from similarly-folded at projections,
and the angle of their curl was also c.180°. At some point the cheekpieces and hinge were
separated from their respective helmets, when the curled projections pulled open to some
degree; originally they were folded by more than 180° to secure them tightly. Centrally
at the top of the hinge piece, on the opposite side from the projections, is a portion of an
The landscape of the naval bale at the Egadi Islands 27
upward-projecting ring (inner diameter c.0.4-0.5 cm) to hold the fastener connecting the
cheekpiece to the helmet; this connector allowed the cheekpiece to swing and maintain a
roughly perpendicular angle to horizontal blows. Given the angle of the hinge aachments
and the shape of the cheekpieces, their lower portion would have hung slightly canted
forward so as to protect both the ear and jaw. After cleaning, the hinge connections on the
helmets are clearly discernible on the interior where part of the hinge survived. There are
two hinge connections on each side, situated forward of the midline so as to align with the
jaws forward of the ears. As there are two aachment points for each cheekpiece on these
helmets, they dier from the helmet-hinge aachment for PW11-034-003 that has a single
aachment point. An additional hinge aachment was found at the center point of the neck
guards of each helmet. Here a single pin held some type of hinge on the underside of the
neck guard, where an additional guard was aached.
Italy of the 4th c. B.C. saw wide variations in bronze weaponry, including helmet styles,
which included the Aic and Chalcidian types in Etruria, the Negau type in N Italy, and
the Apulo-Corinthian type in S Italy. A signicant change in Roman weaponry occurred
with the development of the maniple at the end of the 4th c. B.C. This included adoption of
the Montefortino type of helmet (galea or cassis), an Italo-Celtic hybrid originating in what
is now France and Austria during the 5th c. B.C. This helmet style is the most prominent
in Spain, Gaul and Italy during the 3rd c.27 It is unclear what types of armor Carthagin-
ian marines or infantry utilized, but we know they employed Celto-Iberian mercenaries
as well as mercenaries from S Italy. Therefore it is not possible to be sure of the cultural
aliation of the helmets based solely on their style, but a grato incised over the molded
decorative paern on the crest knob of helmet PW11-0030 (g. 11), which appears to form
a leer corresponding to a Punic “he” (H) or Celto-Iberian “E”,28 may give a clue.
The helmets from sector PW-A derive from an artifact assemblage with a known context
in the mid-3rd c., whereas other examples of Montefortino helmets are rare outside funer-
ary or votive deposits (excavations on land or bale sites typically do not yield helmets in
context29) and as a result their chronology is less secure. Many examples come from 4th- to
2nd-c. graves, such as the Alfedena necropolis, burials at Chieti (Contrada Sant’Anna), Vil-
lafonsina and Villamagna in Chieti, and Pretoro, or from the votive deposit at the sanctuary
of Pietrabbondante. Those helmets may be spoils from Roman forces or their Campanian-
Sabellian allies. Comparative examples are found at the Museo Guarnacci (V54), dated to
the 4th c., and in the British Museum (dated to the era of the Punic Wars); there are also
some 3rd-c. examples from Perugia and Bologna (g. 11).30 The form of the Egadi cheek-
pieces is comparable to the examples shown in g. 11; they are also of a design common to
later nds known from the Agen and Port areas, which came to typify that used through-
out the Republic.31 However, the comparative examples of g. 11 have a more rounded
dome than the somewhat pointed and taller domes of the Egadi helmets.
27 Goldman (forthcoming) discusses the history of Republican-era armor and the archaeological
evidence.
28 P. Schmi kindly reviewed this grato. There does not appear to be an analogous Latin char-
acter after the 6th c. B.C., nor any such Greek leer-form.
29 Bishop and Coulston 2006, 65.
30 The example from the British Museum (inv. AN 14094001, reg. num. 1975,0603.1, Greek and
Roman Antiquities cat. no. Bronze 2726) is 21.7 cm in diameter and 19.2 cm high, with a slightly
shorter crest knob. See other comparative examples in g. 11.
31 Bishop and Coulston 2006, 65.
S. Tusa and J. Royal
28
a:
b:
c: d:
e:
The landscape of the naval bale at the Egadi Islands 29
Amphoras
In 2009, three Greco-Italic type V/VI rim-neck-handle fragments (PW09-0004, 0005 and
0006) were collected from sector PW-B because of their 3rd-c. B.C. date and their proxim-
ity to the Egadi 2 ram, and another Greco-Italic V/VI amphora (PW09-0001) was collected
c.1.75 km south of sector PW-B. All were found directly west of Levanzo Island near the
junction where the rocky area meets the sandy area covered with drag-net scars. In 2010,
however, many amphoras were noted near the Egadi 3 ram in sector PW-A, and one Punic
amphora (PW10-0001) lying 146 m southwest of that ram was collected. In 2011, further
ROV operations in sector PW-A revealed a eld of amphoras, including various Punic
types and Greco-Italic V/VI. Limited investigation within an area of 1 km2 documented the
locations and depths of 177 amphoras, nearly 70% of them being intact. The search paern
covered c.30-40% of the seabed within the 1 km2; the random search strategy employed
indicates that amphoras remain to be located in the unexplored areas; the full extent of
sector PW-A is also still to be determined.
All the amphoras are distributed in a scaered paern with no concentrated group-
ings of the kind typically associated with merchantman wrecks (g. 2). This distribution
was probably created by amphoras spilling from ships at the surface and/or while mov-
ing downward from the surface through the water column. Five were raised for analysis:
4 Greco-Italic V/VI (PW0011-0001, 0002, 0003, 0011) and one Punic (0004). The Greco-Italic
V/VI amphoras from 2009 and 2011 are identical. To judge by the heavy N-S dragnet scars
that delimit the E edge of sector PW-A, dragnets must have moved and broken apart
amphoras from somewhere east of sector PW-A towards the north-northwest, where the
nets struck rocks and dumped the material in our sector PW-B. Many other amphoras from
east of sector PW-A were undoubtedly collected and retrieved in nets, an occurrence well
aested in the area that accounts for many of the nds turned in to local authorities by
shermen. The same phenomenon accounts for the Egadi 2 ram being moved to its more
northerly location in sector PW-B; as with the amphoras, it was probably originally depos-
ited to the east of sector PW-A. Likewise, Egadi 7 was collected in nets some years ago.
This means that the original eastern extent of the artifact distribution in sector PW-A (i.e.,
part of the conict area) is severely disturbed; it may indeed have extended a kilometer or
more farther east.
Punic amphora types (Table 3)32
Amphora PW10-0001 (g. 12) is a Ramon type T-9.1.1.1/2 with a production date
in the second half of the 3rd c. B.C. and a distribution that increases during the follow-
ing century. These amphoras were produced in the area of Cadiz and on Ibiza, and are
common nds in Punic Sicily, Malta and at Carthage.33 Amphora fragments PW11-0004
and 0005 (g. 12) are Ramon type T-3.2.1.2; additional examples of this type were noted
on the sea-oor (see, e.g., g. 12, lower left). Production areas for the type include N
Africa (perhaps around Carthage) and W Sicily. The type circulated in the W Mediter
ranean around the mid-3rd c. B.C. and particularly in the last quarter, and is a common
nd on the coast of Spain.34 A small fresh break on an older broken edge of PW11-
0004, as well as breaks at both ends of fragment PW11-0005, revealed a red fabric with
32 All Punic amphora identications were corroborated by R. Docter (Ghent University) based on
images and measurements.
33 Ramón Torres 1995, 289-91; Di Stefano 1993, 40
34 Ramón Torres 1995, 183 and 519-20; Ramón et al. 1998; Gómez et al. 2006.
S. Tusa and J. Royal
30
The landscape of the naval bale at the Egadi Islands 31
TABLE 3
RAISED PUNIC AMPHORAS: 2009-11
Aribute PW10-0001 PW11-0004 PW11-0005
Condition intact frag. handle
Typology T-9.1.1.1/2 T-3.2.1.2 T-3.2.1.2
Max. ht. 36.5 45.8 10.0
Max. diam. 23.3 19.8 3.8
Handle shape Ear Ear Ear
Handle ht. 10.7 13.4 (13.5)
Handle l. 14.6 19.4 (20.0)
Handle section - M 3.0 x 2.2 3.5 x 2.1 3.5 x 2.0
Rim diam. - O 13.2 - -
Rim diam. - I 10.1 - -
Rim ht. 0.7 - -
Rim type Flared - -
Neck ht. N/A N/A -
Body th. - 0.6-0.8 -
Base diam. 12.6 - -
Base indent depth 2.8 - -
Fabric color outer 2.5YR 3/4 - -
Fabric color inner - 10R 2.5/6 10R 2.5/6
numerous small black and sparse white inclusions. Comparanda include those from
the necropolis at Lilybaeum,35 Xlendi harbor on Malta,36 and various sites in Spain.37
35 Di Stefano 1993, 59, pl. 4.3.
36 Gambin 2002-3, 21, g. 3.
37 Ramón Torres 1995, gs. 156-57 and 244.
S. Tusa and J. Royal
32
Greco-Italic V, V/VI amphoras (Table 4)38
All of Greco-Italic amphoras collected from both sectors PW-A and PW-B (g. 13) have
thick triangular rims, upward conical necks, strongly carinated shoulders, tapered bodies,
and a short toe. All but one rim slope downward from the inner to outer rim edge; amphora
PW11-0001 has less of a slope and forms a slightly aer rim. Each amphora features one
handle having slightly more curvature than the other (note, e.g., the particularly asym-
metrical amphora PW09-0005). Two general handle proles are seen: those that are slightly
more vertical (PW09-0001, -0006, PW11-0011) and those with greater curvature (PW09-
0004, -0005, PW11-0001, -0002, -0003). Typically the joins of the handle to the neck and
shoulder are not worked to a smooth nished surface. A slight ridge runs along the mid-
line of the handles’ outer face on PW09-0004 and PW11-0002. Amphoras with intact bases
feature a small toe (between c.4 and 6 cm in length) solid in its boom third; one example
(PW11-0001) ares slightly at the terminus while on others (PW11-0002, -0011) it remains
straight. Where there was a fresh chip or an old break that could be cleaned, the fabric color
was consistently 2.5YR 4/6 with a few black and a moderate number of white inclusions.39
Construction marks include ridging on the interior necks of PW09-0004, PW11-0003, and
possible nger indentations associated with the joining of the base of at least one handle to
the shoulder on PW09-0001, -0005, and PW11-0002. An incised line is seen around the base
38 All Greco-Italic amphora identications were corroborated by G. Olcese (Università La Sapi-
enza) based on images and measurements.
39 Thin-section analysis is planned to determine production areas for the amphoras. Petrological
analysis should allow dierentiation of production between the Bay of Naples and individual
regions in Sicily or N Africa. Greco-Italic amphoras from Ischia are typically very similar in fab-
ric, the light beige color tending to pink (5YR 6/6-5YR 7/3) or brown (10YR 7/3); among the inclu-
sions visible to the naked eye are grains of volcanic material and akes of mica (Olcese 2004, 179)
since the clays from Ischia have volcanic inclusions and limestone with microfossil inclusions
(Thierrin-Michael and Galei 1996). Greco-Italic amphoras are also found in local fabrics at
Portus, Sabratha and Carthage; the fabrics from N Africa are ne-grained and typically have
numerous quar inclusions, sometimes with limestone too.
TABLE 4
RAISED GRECO-ITALIC AMPHORAS (2009-11)
Aribute PW09-
0001
PW09-
0004
PW09-
0005
PW09-
0006
PW11-
0001
PW11-
0002
PW11-
0003
PW11-
0011 Avg.
Condition intact frag. frag. frag. intact intact intact intact
H. with toe - - - - 68.5 64.5 - 76.1 69.7
Body h. without toe 59.0 - - - 59.3 59.0 61.2 67.7 61.2
Diam. 38.0 - 38.5 - 35.0 36.2 39.4 37.5 37.4
Handle h. 13.0 - 12.6 13.0 13.5 13.0 12.8 14.0 13.1
Handle section - M 3.2 x 2.8 3.0. x 2.4 3.0 x 2.7 3.0 x 2.6 3.1 x 2.0 3.3 x 2.5 3.9 x 2.7 4.1 x 2.3 3.3 x 2.5
Rim diam. - O 17.9 18.0 18.0 18.0 17.3 18.5 17.0 17.2 17.7
Rim diam. - I 11.8 12.0 12.5 12.0 11.2 11.9 11.8 11.7 11.9
Rim h. 2.8 2.4 2.7 2.6 2.0 2.5 2.9 2.5 2.6
Carination h. - - - - 47.0 50.0 - 54.2 50.4
Neck h. 9.8 10.5 10.0 9.5 10.0 11.0 10.2 11.1 10.3
Neck diam. - M 11.0 11.5 9.5 11.0 11.6 10.2 10.8 11.1 10.8
Body th. - 0.9-1.2 0.8-1.1 0.9-1.1 - - - - 1.0
Toe length - - - - 9.2 11.0 - 8.4 9.5
Toe diam. - M - - - - 4.5 5.5 - 3.8 4.6
The landscape of the naval bale at the Egadi Islands 33
of the neck on PW09-0004, PW09-0006, and PW11-0003, and possibly PW09-0005, while an
incised line runs around the top of the neck on PW11-0002. Around the bases of PW11-0002
and 0003 near the toe join are 1-2 incised lines. No stamps were noted on any handles or rims,
but a Punic grato is located on the shoulder of PW11-0003 (g. 14) that reads “Resh Tau”.40
The Greco-Italic amphora type41 is a relatively long-lived form that was widely
40 P. Schmi (Eastern Michigan University) kindly reviewed the grato and leer-forms.
41 Also known as type MGS V and VI amphoras (Van der Mersch 1994), previously categorized
as Will A1 and A2. Will (1982) provided a summary of work on Greco-Italic amphoras and
the history of its typology up to the early 1980s. Subsequent eldwork in Sicily and Italy pro-
duced numerous nds of this type and it was proposed that Greco-Italic amphoras were man-
ufactured in Magna Graecia/Sicily between the 5th and 2nd c. B.C. (Van der Mersch 1994),
which gave rise to the ‘MGS’ designation referring to a production area (see, e.g., Caravale
S. Tusa and J. Royal
34
produced and distributed throughout the Mediterranean. The numerous amphoras from
sector PW-A are the Greco-Italic type V to V/VI, demonstrating a 3rd-c. form at the tran-
sition between V and V/VI.42 Type IV Greco-Italic amphoras (mid-4th c.) have a more
elongated shape, longer handles, a greater amount of shoulder slope, a straight conical
neck, and a more aring rim. Type V amphoras are dated by contexts at Ischia to the late
4th through mid-3rd c., while types V and V/VI range from the end of the 4th into the
2nd c. B.C.43 At Ischia, the type V amphoras have a rim that is not as wide as in the earlier
phase, a slightly wider and more ‘V’-shaped body, and a tronco-conical neck compared to
the earlier cylindrical neck.44 The rim form, angle of neck, and body shape of the type V-V/
VI amphoras from sector PW-A are consistent with a date in the mid-3rd c. B.C. They were
once thought to have been produced only in Italy and Sicily, but discoveries of kiln sites
coupled with petrological analyses now point to additional production areas. Earlier work
suggested Spain and Sicily as the primary source areas, but newer evidence indicates S
Italy, the central Tyrrhenian coast from Cosa to Pompeii including Portus,45 Velia, as well
as southern France (Pech Maho) as production areas. Production occurred on Ischia from
the late 4th to 3rd c. for types IV and V, and for type V/VI to the end of the 3rd and into the
2nd c. B.C.46 Greco-Italic amphoras are also found in N Africa at Sabratha and Carthage
made in local fabrics.47 The grato points to a Carthaginian cargo originating in Tunisia,
but the planned thin-section analysis may modify this.
Closely comparable amphoras were carried on the Montecristo A shipwreck o Tus-
cany, variously dated to the early to mid-3rd c. B.C.48 Other comparable examples around
Sicily include the Secci di Capistello wreck49 (rst quarter of the 3rd c.), the Roghi di Pan-
area wreck50 (recently re-dated to the second quarter of the 3rd c.), and the Filicudi B
wreck51 (end of the 3rd c.), all found around Sicily. Comparable examples are found at
Lilybaeum.52 From the Cala del Diavolo wreck o the Tuscan coast come early to mid-3rd
c. examples with a height range of 65-70 cm and thick, triangular rims.53 An early 3rd-c.
example also comes from Barbarossa at Porto Azzuro on the same coast.54
and Toolei 1997, 82-89, who dierentiated Greco-Italic amphoras by the generally-accepted
chronological categories of I through VI, but maintained the MGS designation).
42 The primary designation for the types follows Olcese (2007, 64, g. 2). Van der Mersch (1994)
classies the types as V and VI = Will A1 and A2.
43 Olcese 2004, 177; ead. 2007, 70; Caravale and Toolei 1997, 88-89.
44 Olcese 2007, 64-65.
45 Most of the study examples were in contexts of the late 3rd to 2nd c. B.C.: Grace 1952; Will 1982;
Peacock and Williams 1991, 85. For Sicily see Alaimo et al. 1997, 62. For Pompeii, see Mannoni
1984. For Velia, see Liko 2001a and 2001b. For France and the Tyrrhenian coast, see Hesnard et
al. 1989 and Thierrin-Michael 2000. For Campania, see Olcese 2004. For kiln sites at Cádiz and
in the Straits region, see Sáez Romero and Díaz Rodríguez 2007; Sáez Romero 2008.
46 Olcese 2004, 176-79; ead. 2007, 70.
47 See Roman Amphorae: a digital resource (University of Southampton, 2005), hp://archaeology
dataservice.ac.uk/archives/view/amphora_ahrb_2005/petrology.cfm?id=141&CFID=0&CF
TOKEN=4BA6E221-2C63-4CFD-BDC6495EAE7C0E1A (viewed Oct. 24, 2011).
48 Martelli 1982, 65-68; Corsi 1998, 137-43.
49 Frey et al 1978.
50 Bernabò Brea and Cavalier 1985, 69-70. The amphoras are re-dated in Olcese 2010, 248.
51 Bernabò Brea and Cavalier 1985, 88-89.
52 Di Stefano 1993, 59, pl. 4.1.
53 Maggiani 1982, 67.
54 Rossi 1982, 85, g. 76.
The landscape of the naval bale at the Egadi Islands 35
Artifacts were found within three of the intact amphoras from sector PW-A (g. 14).
Amphora PW11-0001 contained a large amphora body sherd, while PW11-0002 contained
several coarseware sherds, including one rim fragment. This material is consistent with
wares common to Punic N Africa and W Sicily in the 3rd c. Amphora PW11-0003 contained
the rim of a small coarseware container, a tile fragment, and two iron nail concretions (12
and 21 cm long) clearly from ship timber aachments, like those found within the Egadi
3, 4 and 6 rams, while the tile fragment is similar to fragments found within the Egadi 4
and 6 rams. As these three amphoras were found in a small triangular area (16 x 19 x 34 m)
over 200 m from the ndspots of the Egadi 3-6 rams, possibly they were close to a dierent
wreck from the wrecks associated with the rams which have been documented thus far.
Discussion
The larger context of the archaeological evidence
The study of human behavior in the past is a process by which hypotheses are put for-
ward on the best available evidence to date and then critiqued for their ability to explain
that evidence. New evidence often prompts novel ideas to be put forth. The Egadi rams
and the other evidence from the bale zone oer important new data sets for addressing
questions relating to warships. In his substantive contribution to this eld of inquiry, J. S.
Morrison provided an honest assessment of the evidence on which warship reconstruc-
tions had been based to date:
A naval architect aempting to reconstruct ancient Mediterranean warships is oered
scanty literary and iconographical evidence on which to draw.55
The problems of interpreting literary and iconographical sources are well documented;
likewise, the limits of the archaeological data must be addressed. The material associated
with the Bale of the Egadi Islands is pertinent to warships and their operation in the 3rd
c. B.C. only. The cultural material from the Egadi Islands provides evidence for the loca-
tion of the events associated with the bale, the depositional fates of the warships from this
bale, and the size of the warships that engaged in this specic conict. Comparisons can
be made to remains from dierent eras, such as the Athlit ram or Augustus’ Victory Monu-
ment at Nicopolis, but the results will be limited to general diachronic changes.
For the class sizes of the rams lost during the bale we have an inconclusive literary
record. The nomenclature is not likely to have been consistent over more than seven centu-
ries or between the dierent cultures using it and it is unwise to assign the same meaning
to references for “threes” or “ves” in the 4th c. B.C. and those in the 1st c. A.D. There
was undoubtedly evolution and innovation in warship design and their armament: dier-
ences are already seen in the archaeological record between the 3rd-c. rams from the Egadi
Islands and those of the 2nd c. B.C. We will focus on likely sizes from design considerations.
From conict area to bale landscape
In view of the range of potential routes for the Carthaginian eet travelling from Maret-
timo Island towards Bonagia Bay, and for the possibility that the Roman eet rounded the
N tip of Levanzo Island to meet them (Polyb. 1.60), the naval engagement took place west
of Levanzo (g. 1). Although only a small percentage of the known artifact distribution
(c.1 km2) has been examined thus far, the dispersion of artifacts in sector PW-A supports
55 Morrison 1996, 297.
S. Tusa and J. Royal
36
this narrative of events. This convergence between the archaeological and the historical
record is in itself an important result. Given the hundreds of warships engaged in the
bale, actual conict would initially have extended over perhaps 7-10 km, and probably
expanded as the conict unfolded. Unfortunately, material east of sector PW-A has been
removed or displaced by dragnets, making it impossible to ascertain the original extent
of the bale zone hereabouts. Additional nds will be made to the degree to which sector
PW-A extends to the west, and if other distinct bale zones are undamaged. Nevertheless,
the undisturbed nature of the material in sector PW-A already permits a discussion about
a bale landscape.
Intensive recording methodologies to detect small objects, in addition to the larger rams
and amphoras, will be necessary to comprehend it beer. Precise mapping of all artifacts
could allow inferences to be made about the nature of ship-to-ship conicts, the expanse of
the conict, and what material from a naval bale can be expected to survive in the archae-
ological record. Through a combination of the possibly associated evidence of anchorage
on Levanzo Island, the nds at Bonagia Bay, the excavations at Marsala, and ongoing mari-
time investigation around the Egadi Islands, a bale landscape is coming to light. It may
eventually be possible also to consider strategic and tactical operations, ports and anchor-
ages, and land support.
Depositional fates of warships
The fate of warships becoming swamped is well-aested in literary sources. A fresco of
a warship wrecked along the shore in the Temple of Isis at Pompeii is evidence that such
a ship swamped and oated ashore. However, a common misconception about ancient
warships is that they could not, and did not, sink, even though thousands of ancient ships
of all eras did sink. Instead, it is proposed that warships simply oated on the surface
swamped with water and awash until they were towed away by friend or foe or crashed
into a nearby shore in every instance.56 Statements from ancient authors are only moder-
ately helpful as they typically refer, in an abbreviated or imprecise manner, to ‘destroying’
a vessel.57 Yet Lucan (Phars. 3.627-33) distinctly mentions a warship sinking beneath the
surface, and Polybius (1.60-61.6) uses the word katedusan which seems to mean “to sink into
the water”.58 The underlying basis used to argue that warships could not sink is that wood
oats, but this simplistic observation fails to understand the nature of ships or the changing
state of the wood in their hulls. Most (but not all) wood species have a specic gravity less
than 1.0 (that of water) and therefore oat, but the specic gravity for wood is not a con-
stant. Over time the wood in ships’ hulls becomes waterlogged through immersion, spray
and humidity; water is absorbed into the wood cells up to the ber’s saturation point and
then is absorbed in liquid form into the wood’s cell cavities and pores, which increases its
mass and raises its specic gravity.59 As the specic gravity of the wood in hulls increases,
its ability to carry additional mass decreases in an inversely proportional manner.60 Metal
fasteners, lead sheathing, rigging gear, anchors and other metal xtures aached to ships’
56 For such an example, see Strauss 2004, 198-208.
57 E.g., Diod. 11.24.1-2.
58 W. M. Murray kindly brought these references to our aention.
59 In the analysis of the Acqualadroni ram, the specic gravity of most wood samples equalled or
exceeded the value of water: Romagnoli 2009, 6, Tables 1-2.
60 Although ships may have been regularly hauled out and degraded hull sections replaced,
water remained to some degree; further, during naval operations the water absorption acceler-
ated greatly and raised a hull’s mass.
The landscape of the naval bale at the Egadi Islands 37
hulls increased their overall mass.61 Shipwrights have understood the nature of wood for
millennia: they designed hulls to aain buoyancy through the displacement of water (a
ship), rather than relying on the buoyancy (specic gravity) of the parent construction
material (a raft). All ships’ hulls are waterlogged to some degree, and some more than
others; in the 4th c. B.C., warships in the Athenian eet with less waterlogged hulls were
considered the best and fastest ships (ταχυναυτοῦσαι), but this implies that there were also
warships with more waterlogged hulls that were in a less ideal state for bale.62
A further assumption lying behind the ‘non-sinking’ premise is that warships were
virtually empty shells during their operation, carrying no shipboard items, cargo or bal-
last on-board. But the notion that any vessel that operated by both oar and sail propulsion,
with numerous individuals on board, and in conict situations, did so without equipment,
stores and ballast on board is contrary to a basic understanding of how a ship operates. The
ship’s own tools and equipment, containers, weapons (both of the ship and personal), and
other equipment belonging to those on-board must be taken into account. It is this heavy
equipment necessary for long-distance navigation that ancient authors note is removed
when possible before entering into an engagement with enemy ships. As with Renais-
sance war galleys,63 ballast was mandatory to counter the forces placed on the vessel by
the sails, and weight must have been concentrated below the waterline in warships’ hulls,
particularly as they were long and narrow. Thus the stones found at the site of the ship that
carried the Egadi 4 ram are not surprising.
The evidence from our survey shows that entire warships sank to the seaoor along
with their rams. This is contrary to the commonly accepted belief that warships never
sank when destroyed, but oated in a swamped condition at the water’s surface. In certain
conditions, however, it now seems likely that warships could and did sink, taking their
rams and associated bow timbers to the boom. Such a conclusion is really not so radical.
Ancient warships were often required to carry heavy loads of supplies, siege machinery,
and so on. In the case of this bale, the Carthaginians were sailing their warships laden with
supplies to the Egadi Islands when the Romans suddenly engaged them (Polyb. 1.61.4-6;
cf. Livy 36.43.6-7 for an instance of laden warships in 191 B.C.). The dispersion of amphoras
and nds of armor in sector PW-A accord well with Polybius’ account of the Carthaginian
ships being heavily laden with supplies and troops. The vessels operating at the Bale of
the Egadi Islands were warships, not war-rafts; once breached by ramming aacks during
the bale, their hulls lled with water. The hulls lost the buoyancy aained from displace-
ment and the waterlogged hulls had less ability to carry the additional mass; the rigging,
ship’s equipment, armaments, ballast and particularly the cargoes in the Carthaginian
warships all raised the overall mass to a point where the warships sank to the boom.64
The discovery since 2008 of 9 waterline rams provides evidence to counter the belief
that rams broke o their warships’ bows along with the bow timbers, and that, while the
61 Hence, it is not uncommon to encounter wooden hulls on the seaoor that are not also encum-
bered by a cargo. Examples directly familiar to the authors include sites TK06-AD (Royal 2008a,
93-95; Royal and McManamon 2010, 331-39), AB10-AB and AB10-AC (hp://www.rpmnautical.
org/albania2010.html), Tantura A and Tantura B (Kahonov and Royal 1996; Kahanov et al. 2002),
and the Marsala Punic wreck (Frost 1974).
62 Cf. Gabrielsen 1995, 235 n.6 and IG II2 1611, lines 97, 147, 151, 157.
63 D’Agostino et al. 1997; Royal 2008a and 2008b; Royal and McManamon 2009 and 2010.
64 Coates (1995, 133-34) already put forward this logical observation.
S. Tusa and J. Royal
38
rams sank due to their weight, the bow-less warships remained aoat.65 First, such a belief
is at odds with the archaeological evidence for ship construction, in which the keel, main
wales, ramming timber, and stem timbers at the bow (on which rams were ed) came
together to form one of the strongest unions in a hull.66 Analyses of the timbers inside the
Athlit and Acqualadroni rams reveal a complex system of scarfs, bevels, rabbets, tenon
and nail aachments that resulted in a fully integrated unit at the bow. In addition, heavy
knees, frames, and deadwood contributed to the hull’s strength at the bow. A hull’s bow
requires heavy construction to absorb intensive shock forces from waves. The sheering o
of a warship’s hull at the bow aft of its ram would involve the complete breaking of the
keel, both wales, both garboards, the stem, numerous strakes, deadwood, and any chocks
or frames present. J. R. Stey noted the implausibility of bows becoming separated from
the rest of the hull in his study of the Athlit ram:
The ram and bow timbers were not wrenched from the ship; all aftward terminations are
natural, eroded separations, and the unit shows no signs of violent distortion. It is, in fact,
unthinkable that such a substantial assembly of timbers could be ripped away from the bow
by any force existing in antiquity.67
The Egadi rams too remained aached to their bow timbers. The only damage and defor-
mation is limited to their heads, linear damage, which points to ram-to-hull or ram-to-ram
collisions.68 The fragmentary rams show how rams were typically sheared from the bow.
It is hard to imagine shipwrights repeating design aws that would allow such a valuable
item as a ram to be easily removed and lost at sea.69
Geological factors aecting various depositional environments can help explain the
relative preservation of wooden hull remains. Shipwreck sites close to shore are generally
found in thick layers of sediment resulting from run-o and uvial transport, in which
degradable material such as wood can rapidly become buried — a prerequisite for its sur-
vival in the archaeological record. Such was the case with the Athlit and Acqualadroni
rams. Further oshore, accumulation of sediment is comparatively much less and occurs
at a slower rate, so that the soft layers are typically thinner. Excavation at the Levanzo 1
wreck (which lies an equal distance from Levanzo Island as our sector PW-A), as well as
of the sediment around Egadi rams 2 and 4, showed that the soft upper layer was 5-15 cm
deep.70 The underlying hard substrate restricts cultural material from seling below this
upper soft layer. Without a protective sediment cover, warship timbers in sector PW-A
65 Morrison et al. 2000, 222-23. Only one of these previously known rams, the Athlit, had a prove-
nience. It presented evidence contrary to the present hypothesis but it was the only example.
66 The term ‘bow’ refers to the forward portion of ship’s hull, typically where the sides begin to
curve inwards toward the stem. It is not in itself a specically constructed section separate from
the remainder of the hull or demarcated by a particular set of structural elements. Hulls are
single integrated structures; timbers such as the keel, strakes, and wales, for example, run from
bow to stern. Descriptive terms such as ‘bow’, ‘stern’, ‘port’, and ‘starboard’ provide spatial ori-
entation on-board a ship or when referring to areas of its hull.
67 Stey 1991, 29.
68 Damage to the rams provides supporting evidence for the use of frontal ramming tactics into
the 3rd c. B.C. (Murray 2011). The wood trapped at the head of the Egadi 6 ram suggests that
ramming a hull could produce damage beyond simply opening planking seams.
69 See Murray’s (1985, 145-47) analysis of ram values based on the records by the curators of the
shipyards at Athens: for example, during the 4th c. B.C., Athenian trierarchs were expected to
return to the yards rams from old ships or ships damaged in storms.
70 Royal and Tusa 2012.
The landscape of the naval bale at the Egadi Islands 39
were left exposed and were destroyed by teredo, resulting in the rams’ aachments being
left in place and the iron nail found in the interiors of both rams and amphoras. Some
warships from this bale were destroyed at a distance of more than 7 km from the nearest
shore on Levanzo Island, where the water was over 80 m deep and where of course there
was no interplay with a shoreline. Not only did their heavily-laden condition prevent these
ships from swamping or remaining aoat; if they had, the ships would have drifted far
from the bale area where the cargos and armament were spilled. As the rams were found
amidst the other materials, the ships must have sank to the boom at the same time as the
amphoras and armor. Wood remains, nail concretions, marine growth paerns and pXRF
data from the Egadi rams (as well as the Acqualadroni and Athlit rams71) indicate they
were aached to their warships’ bows upon deposition. Hull aachments, tile fragments,
ballast stones and coarseware from sector PW-A, as well as the amphoras themselves,
point to highly degraded and scaered wrecksites. Clearly it was possible for the warships
to sink, depending on their condition and the specics of the materials carried on board; as
a result, warship rams are found on the seaoor, in some cases with the bows partially pre-
served. This is not to say that warships sank in every instance, but, as with all other ships it
is possible that they could sink. Whether they sank, swamped, or washed ashore depended
on the interplay of specic warship operation/mission and the taphonomic processes spe-
cic to the depositional site.
Ram size and warship size
A striking characteristic of the Egadi rams is their small dimensions. The most useful
measurement for comparing ram lengths is their driving centers, since the lengths of tail-
pieces vary and exaggerate the overall length of the ram. The driving center length of the
Athlit ram is 1.575 m, of the Acqualadroni ram 0.950 m (both perhaps a century earlier or
more); the Bremerhaven ram is 0.260 m and the Piraeus ram is 0.536 m (both probably at
least a century earlier).72 The driving center lengths of the Egadi rams fall within the range
of 0.590 to 0.859 m. While a more detailed comparison of ram morphology and dimen-
sions is slated for a later publication, it is clear that 10 of these 11 rams have a driving
center length between roughly 0.5 and 1.0 m.73 The Athlit ram is of exceptionally large size
amongst warship rams found to date, yet is consistent with timber sizes of merchantmen of
the 2nd c. B.C. as well as those found within the almost-contemporary Acqualadroni ram.
Several preliminary observations can be made regarding the dimensions of the war-
ships on which the Egadi rams were ed: the cross-sectional dimensions for the warships’
wales can be ascertained from the rams’ wale pockets, their keel dimensions calculated
from the rams’ boom plates, and an indication of stem dimensions deduced from cowl
measurements. Boom plate-tailpiece shapes indicate that the keels of the warships were
slightly trapezoidal at the bow, and were shaped oddly in that they had nearly the same
molded and sided dimensions. Stey, who noted similarly odd dimensions (molded 10.5
71 Romagnoli 2009; Stey 1991, 29.
72 For the Athlit ram, see Stey 1991; Acqualadroni ram, Tusa et al. 2012; Bremerhaven ram,
Bockius forthcoming; Piraeus ram, Steinhauer 2006, 36-38 (measurements are both given and
interpolated).
73 Although the Bremerhaven ram is comparatively short, it has head dimensions and overall
width dimensions analogous to these 10 ‘typical’ rams. A twelfth waterline warship ram, the
Follonica ram, recently seized by Italian police in Rome, has no published dimensions but
appears similar in size to that of the Egadi rams.
S. Tusa and J. Royal
40
cm, sided 20.0 cm) for the keel timber preserved within the Athlit ram, surmised that
the molded dimension increased aft of the ramming timber to a molded dimension of
c.25.0 cm;74 this resulted in an aspect ratio (molded-to-sided dimension) of 1.25, which
is in accord with the structural demands required of the keel for ancient hulls.75 If we
allow for a similar increase in the Egadi keels’ molded dimensions to aain a 1.25 aspect
ratio, it is possible to estimate keel dimensions for the warships (Table 5) and calculate
the cross-sectional area of the keels. A cursory review of ancient merchant ships indicates
that the keel cross-sectional area and overall vessel length are strongly correlated (Table
5). Given that warships have a high length-to-beam ratio, they had a greater requirement
for robust longitudinal support timbers, particularly the keel. The main wales at the war-
ship’s water line were other important longitudinal timbers within the hull, which sat
in line with the ramming timber. The waterline was the area of stress in the hull for it
is here that it gained and lost the support of the water for holding up the ship’s mass;
it was at this point that hulls were most likely to ex.76 The sturdiness of a warship’s
main wales was crucial since they helped redirect impact forces throughout the hull.
Both the actual dimensions and cross-sectional area for keel and wale timbers in the
Egadi warships were relatively small by comparison with those from merchantmen. Keel
sizes for our warships were similar to the 14-m long Kyrenia ship (c.300 B.C.) and the 13-m
74 Stey 1991, 17 and 34.
75 The value of the keel aspect ratio (“molded/sided dimension”) in sea-going merchantmen of the
later Roman period was typically between 1.0 and 1.75: Royal 2002.
76 Stey 1994, 49-50. They were also at the waterline in the Athlit ram: Stey 1991, 30.
TABLE 5
WRECKSITES DATA FOR KEEL-VESSEL LENGTH CORRELATION
Wreck Site Date Est. L. Keel molded x sided (cm) Keel cross section (cm3)
Kyrenia1c.300 B.C. 14 10-13 x 20.3 233.5
Cavaliere2100 B.C. 13 11.0 x 16.0 176
Madrague de Giens31st c. B.C. 37.6 33 x 40 1320
Procchio42nd c. A.D. 20 25 x 28 700
Bourse de Marseilles52nd c. A.D. 23 17-28 x 29 652.5
Laurons 262nd c. A.D. 15 20 x 16 320
1 Stey 1994, 42-52.
2 Charlin et al. 1978, 60-93.
3 Rival 1991, 147-239.
4 Zecchini 1982, 138-68, gs. 49-50.
5 Gassend 1982, 75-76.
6 Gassend et al. 1984, 93-94.
The landscape of the naval bale at the Egadi Islands 41
Cavaliere wreck (c.100 B.C.).77 Similarly, the wales on our warships were of relatively small
dimensions (15-21 cm in width, 9-14 cm in thickness), comparable to those on the Kyrenia
ship and the Bourse de Marseilles vessel (2nd c. A.D.) at 23 m in length.78 Larger vessels
such as the Madrague de Giens (1st c. B.C.) merchant vessel, estimated to be 37.6 m long,
not only incorporated a large keel (40 x 35 cm), but also had a double hull totalling 10 cm
in thickness, a thickness analogous to the wale timbers of our warships.79 This is not to
say that Egadi warships were less than 15.0 m in length, rather to illustrate the diminutive
dimensions of the timbers and the general limits they dictate for hull length. As with the
Athlit ram, the Egadi warships probably featured planking between the keel and wales
that was thicker than would be found on contemporary merchant vessels. Given the likely
construction method for warships that featured much heavier boom strakes up to the rst
wale, there was additional longitudinal support. This would allow the construction of war-
ships c.25-28 m in length, a size concomitant with a “three”.
We have limited literary evidence for warships of the 3rd c. B.C. Although quinqueremes
are the only class of warship aested specically for this bale (Polyb. 1.59.6-8, 1.63.5) and
generally emphasized for the First Punic War as a whole, triremes (“threes”) and lemboi
were also components of the eets at this time (Polyb. 1.20.9, 1.53.9). In light of the new
archaeological evidence, questions arise concerning the dimensions and conguration of
the warships and how they relate to statements in the literary record. Previous hypoth-
eses, based on literary and iconographic evidence, do not conform to the material from the
Egadi Islands, which points to relatively small ships.80 Polybius implies that quinquiremes
(“ves”) were involved in the bale. If his implication is correct, then the "ves" in this bat-
tle were surprisingly small and would conict with his observation (1.63.8) that the "ves"
of the First Punic War were much larger and more impressive than the "threes" of the Per-
sian Wars. It is also unlikely that the 7 Egadi rams are a skewed sample of warship sizes
within a given class in view of the probable population of rams deposited at this bale site.
They most likely represent the actual, and typical, dimensions of a single class of warship
in this period. Larger, or smaller, rams may be discovered, but these would undoubtedly
represent another class of warship.
Given the size of the timbers, and thus warships, idicated by the Egadi rams, it is incum-
bent upon us to oer a hypothesis on the class of warship they represent. As noted, Polybius
only relates the presence of “ves” at this bale. For the complement of Roman "ves"
crossing from Sicily to Africa in 256 B.C., Polybius (1.26.7) enumerates 300 rowers and 120
marines.81 It is clear that in the 3rd c. “ves” were large warships, and specically larger
than the warship remains from the Egadi Islands. L. Casson proered the idea of a sin-
77 For the Kyrenia ship, see Stey 1994, 43. For the Cavaliere wreck, Charlin et al. 1978, 60-93.
78 The main wales on the Kyrenia ship were 8.0 cm wide and 21.5 cm thick (Stey 1994, 49). Those
on the Bourse de Marseilles wreck were 6.0 cm wide and 14.0 cm thick (Gassend 1982, 76-77).
79 For the Madrague de Giens vessel, see Tchernia et al. 1978, Pomey 1982, and Rival 1991.
80 For discussions, see Casson 1995, 100-3; Coates 1995, 138-39; Morrison 1995, 68-69; Murray
2012, 23-30. Here priority is given to the archaeological record, in which the Egadi rams are the
only direct evidence for warships of the 3rd c. B.C, apart from Athlit which may be late 3rd c.
81 L. Casson (1995, 105) adjusted this to 270 rowers, 30 ocers and crew, and 120 marines, still
totaling 420 individuals. Whether or not 120 marines was the typical complement or not, the
total number of individuals ascribed by the limited references to “ves” is well over 350. For
additional discussion of “ves” having 300 rowers plus marines and crew, along with the nec-
essary large size of such a vessel, see Morrison 1996, 270-71 and 285-94.
S. Tusa and J. Royal
42
gle-banked “ve” in the First Punic War supported by iconographic examples;82 yet these
“ves” would also require much larger hulls than those associated with the Egadi rams.
The Egadi warships best accord with the dimensions and aributes of “threes”; however,
the timber dimension evidence indicates smaller overall dimensions for “threes” than has
been previously suggested. Possible support for the presence of threes may be found in the
account of the aftermath of this bale. Subsequent to the naval engagement, 70 captured
Carthaginian warships were reported to have a combined crew of nearly 10,000 men taken
as prisoners, which provides an untenable average of c.142 men per warship if all captured
warships were “ves” (Polyb. 1.61.6-8). With 420 individuals aboard a “ve”, the expected
number of prisoners would be nearly 30,000. A portion of the unaccounted-for Carthag-
inians from these 70 warships will be the number who died during the conict, thus not
included in the total of prisoners. A second factor in this conspicuously low number may
also be that the captured warships included “threes”, which had about half the total crew
size of a “ve”. Also in support of “threes” present at the bale of the Egadi Islands is Dio-
dorus 24.11.1 quoting Philinus, a source for Polybius, who notes mixed vessel types (that
is, makrais nausi, ploiois, and poreiois, or ‘warships’, ‘merchant ships’ and ‘transports’, a
higher number of ship totals, and a larger number of sunken ships for both sides.
This begs the question why “ves” were the only warships in this bale mentioned by
Polybius. It is possible that only the “ves” were mentioned as he is stressing the special
building project and larger, more prominent ships. Given the success of the “ves” for the
Romans and the overall high losses by the Carthaginians, the losses of “threes” may have
been less important for the historians to mention. In fact, Polybius (1.63.7-8) stresses the
greater importance of these engagements compared to the earlier naval bales of Antig-
onus, Ptolemy and Demetrius, on the basis that the naval bales of the First Punic War
featured a large number of “ves” versus the “threes” of the earlier periods.83
Iconographic and epigraphic evidence
Like many state-manufactured items requiring a large investment, the Egadi rams dis-
play decorative features, albeit to varying degrees. A decorative feature present on all is
the stylized shape of their ns. Unlike the cast sword handles on the n plates of the Athlit
and Acqualadroni rams that depict each n as the blade of a sword, the Egadi ns come
together at the n plates to form a shape akin to a trident (gs. 4-10). Tridents are com-
monly depicted in Greco-Roman antiquity, usually associated with Poseidon/Neptune.
Although less is known of the iconography associated with Punic sea deities, it seems
the trident was familiar at Carthage by the 3rd c. B.C. Other molded decorative elements
were included in the rams’ casting. The sole decorative elements on Egadi 1 are small
rosees cast on its n plates (g. 4). Winged Victories are cast on the upper cowl nosings
of Egadi 4 and 6 (gs. 7 and 9); both face to their right dressed in an ankle-length gown,
and carry a laurel crown in their right hand, with a palm frond in their left. C. Halle and
T. Hölscher note that crowns and palm branches were introduced at Rome for victors in
293 B.C. (see Livy 10.47.3) and the Victory motif on the two rams is similar to that on
Roman didrachms minted between 265 and 241 B.C. (Crawford, RRC no. 22/1). Hölscher
points out the frequency of this motif on 3rd-c. coins of Sicily and S Italy, the geo-political
sphere where Rome begins its display of power, in part, by the issue of coinage. A general
82 Casson, ibid. 105-6 n.41, 156, 418, gs. 120-23.
83 We thank W. M. Murray for ideas and discussion.
The landscape of the naval bale at the Egadi Islands 43
military context for Victoria in 3rd-c. Rome is given by the temple to Victoria dedicated
on the Palatine in 294 by Q. Postumius Megellus to commemorate the victory in the Sam-
nite Wars.84 The Egadi 7 ram has a warrior head wearing a Montefortino helmet emerging
from the upper cowl nosing, and its cheekpieces are identical to those found inside Egadi
6; the warrior head may be Roma personied, Minerva, or simply a soldier (identication
is dicult as it appears the face was damaged in a purposeful fashion). Atop the helmet
are three feathers aached to a ring siing on its upper dome, a conguration described by
Polybius (6.23.12-13).
The inscriptions on the Egadi rams are a valuable historical resource. A full exploration
of the inscriptions and in-depth analyses of leer- and word-forms is slated for a later pub-
lication, but the preliminary analysis provides some useful results. Inscriptions were noted
on the cowls of rams 1, 3, 4, 6 and 7, all in scripts of the 3rd c. B.C., and read from the star-
board side. The Punic inscription on Egadi 3 aests to Carthaginian manufacture. A partial
segment of the inscription has been analyzed with two possible translations. The rst is:
“We pray to Baal that this ram will go into the enemy ship and make a big hole”.85 Subse-
quently, P. Schmi, based on extensive comparanda, reads: “. . . tn̊t k̊ bn̊ bʿlm̊ l hwn spt ršp
h̊mm lʾbn tḥ[t . . .]”, translated as: “... Tanit, for in it are its ocers. Blow, gales of Reshep! and
build the surge/overowing/wave und[er (?) ... ]”.86 This preliminary translation would
reect the Carthaginians invoking the gods in hope of assistance for success, while the ref-
erences to winds underscore this crucial aspect of sailing for successful warship missions.
The Latin inscriptions on Egadi 1, 4, 6 and 7 aest to Roman manufacture, but they
reect dierent contexts. The inscription on the cowl nosing of Egadi 1 was engraved in 4
lines after the ram was cast. The rst published transcription was: C SESTIO P F / Q SALONIO
Q [F] / SEX VIROEN[-?--] / PROBAVE[-?--].87 J. Prag oers an expanded text: C(aios) Sestio(s)
P(ublii) f(ilios) / Q(uintos) Salonio(s) Q(uinti) [f(ilios)] / SEX VIROEN[-?--] / probave[re]. The
text records 6 ocials (sex vir(i)) overseeing the quality control of the rams’ castings. A por-
tion of the inscription is illegible on the right side and E N remains enigmatic, so the exact
meaning is still uncertain. Prag suggests that the leers following sex vir(i) or sex viro(m)
may be the abbreviated title of the sevirate’s function (compare the Gracchan termini).88 In
84 The iconography was reviewed by C. Halle (Univ. of California at Berkeley) and T. Hölscher
(Univ. of Heidelberg). Halle further observes that the hairstyle appears to be Early Hellenistic
(Melonenfrisur).
85 An initial reading was provided by G. Garbini (Università di Sapienza, Roma).
86 For “ocers” Schmi notes bʾlm. (nom. masc. pl.) can possibly, but less likely, be translated as
‘lords/masters’. The semantic uncertainty is whether bʾl refers to human military commanders,
to citizenry, or to divine beings. The Phoenician word bʾl generally means ‘owner, master, lord’
(Dictionary of North-West Semitic Inscriptions 183 s.v. bʾl1); in that sense it serves as an epithet of
the Storm god also known as Haddad. But the word can also designate human beings who are
husbands (ibid. 183 s.v. bʾl1), landowners and/or citizens (ibid. 183-84), and a military or naval
ocer (ibid. 184). The phrase bʿl m ym, ‘master of the waters of the sea’ (CIS I 86 B = KAI5 37
B) appears to have been a title of naval command at Kition in the Achaemenid period. We can
speculate that the blm in this text might also be naval commanders. Another possible word
grouping oered by Schmi would provide “. . . tn̊ tk̊bn̊ bʿl m̊ l” and translate as “tn May you O
Ball extinguish it, opposite …”.
87 See Gnoli 2011 for publication; slightly modied transcription here by J. Prag (Univ. of Oxford).
88 Gnoli 2011; Prag points out the missing missing ‘f’ in line 2 and the incomplete ending of
S. Tusa and J. Royal
44
any event, the forms of the script are consistent with those of the mid-3rd to early 2nd c.
B.C. and conrm the ram’s manufacture by Rome. This inscription alone cannot be used
to support a particular eet building program or the eet’s construction in 242 for the sur-
prise aack at the Egadi Islands.
Egadi 4 and 6 feature an identical text in their cast inscriptions, which are arranged in
two lines on their cowl nosing: M·POPULICIO·L·F·Q·P / C·PAPERIO·TI·F vac., but the names
are listed in reverse order on each ram (gs. 7 and 9). The leers Q.P. are aligned with the
end of M· POPULICIO· L·F on the top line on Egadi 4, but at the end of both names between
the upper and lower lines on Egadi 6. A preliminary reading is: M(arcos) Populicio(s)
L(ucii) f(ilios) / C(aios) Paperio(s) Ti(berii) f(ilios) / Q(uaestores) p(robavere). It indicates
that the rams are from the same building program and probably celebrate the work of two
quaestors in nancing the eet construction.89 Prag notes that, with lile epigraphic evi-
dence for quaestors during the 3rd c., these rams provide the chief positive evidence for the
quaestorship at this date and are particularly important from an institutional perspective.
The neatly-engraved inscription on Egadi 7 (g. 10) reads F QVAISTOR· PROBAVET. The
curved but still open P diers from those on Egadi 4 and 6 and is more similar to that on
Egadi 1.90 The AI diphthong in QVAISTOR points to earlier Latin and is plausible for the 3rd
c., as are the leer-forms. Here again a quaestor is celebrated prominently on the cowl nos-
ing; his name is lost, but the available space and the presence of F(ilius) makes its original
presence a certainty.
The dierent types of inscriptions possibly indicate the rams are from distinct build-
ing programs by Rome, but it is not possible to assign them an absolute chronology or
to a specic building program. The inscriptions aest dierent meanings and contexts
in their production. The Egadi 1 ram received an inscription after casting, aesting to
quality control by a board of six, and it was not featured in a prominent manner. This
ram displayed no deities, decoration being limited to simple rosees on the n plates.
Whereas the inscription on Egadi 1 is wholly bureaucratic in nature, those on Egadi 4,
6, and 7 celebrate the eorts of specic ocials, the quaestors, who oversaw eet nanc-
ing. Since over the course of the 3rd c. milestones began to record the specic eorts
of ocials (aediles and consuls);91 perhaps the quaestors were following a similar pat-
‘probave’ in line 4. Gnoli considers the form “VIRO” to be erroneous, amending to “VIR<I>”.
However, Prag points out that “VIRO” at this date is an acceptable abbreviation of the geni-
tive plural (virom), which would not be unusual for the name of a college, and that Gnoli’s sug-
gestion of EN[BOLUM], based on a single instance in Petronius, is speculative. One possible
resolution for SEXVIROEN (which Prag owes to M. H. Crawford) is sex vir(i) o(peri) e(xigendo)
n(avali), but clearly other possibilities can be imagined, depending on how many leers may be
lost after EN.
89 Reading provided by Prag, who rightly points out that the inscriptions do not require any com-
mand functions by the quaestors. A rst reading was made by F. Oliveri (Soprintendenza del
Mare, Sicily).
90 Reading provided by Prag, who notes that probavet is merely an archaic variant for probavit, and
that the damaged area which precedes the F has space for c.8-10 leers, enough for the usual
abbreviated praenomen, a short nomen, and abbreviated patronymic. Prag is reasonably certain
from autopsy that the patronymic can be read as C(aii).
91 Later in the 2nd c., coinage increasingly featured the family and names of the moneyers in an
eort to increase their prestige and further careers: Crawford 1975, 47-65. Examples of mile-
stones include ones found on the Via Appia at Posta di Mesa (CIL I2 21 = ILLRP 448) and one at
The landscape of the naval bale at the Egadi Islands 45
tern. If so, then Egadi 7, which has the name and oce engraved only after casting, may
date to a eet building program earlier than that of Egadi 4 and 6, whose cowl nosings
feature prominent quaestors’ names in relief leers. If indeed the “threes” are under-
reported in Polybius, then several of the Egadi rams with Roman inscriptions may
have been under Roman command at this bale. In any account of losses the Cartha-
ginians’ were greater, so the greater number of ram nds of Roman manufacture is curi-
ous. One possible solution may explain both the high numbers of Roman ram nds and
the dierence in iconography/inscriptions. According to Polybius, Rome had 5 warship
construction phases from 255 until the nal bale in 241. Some 140 warships were built in
two projects during 255, c.110 were built in two phases in 250, and at least 200 were con-
structed for the bale that took place in 241 B.C. (Polyb. 1.30-61). Although classes are not
always specic for these eet constructions, we know that “threes” and “ves”, as well as
lemboi, were the fundamental classes used in this war (Polyb. 1.20.9, 1.53.9). If we propose
a hypothetical seriation of building programs for the rams based on their inscriptions,
Egadi 7 could be associated with one of the building programs in 255, while the Egadi 4/6
rams could be part of the eet construction in 250. Perhaps the Egadi 1 ram was axed to
one of the 200 ships built in 242, as Gnoli suggests.92 Regardless of the specic building
program, the rams 4, 6, and 7 were probably part of the earlier eet building programs.
The Carthaginians had captured over 100 Roman warships during several conicts in 249
B.C. (Polyb. 1.60.1-2), some or many of which were undoubtedly “threes”. An impression is
clearly given that the Carthaginians were scrambling to rush a eet into service in 241 B.C.
(Polyb. 1.61.4-6) and consequently the captured Roman ships (built in 255 and 250) formed
part of the eet sent to the Egadi Islands in 241. In this scenario, it is not unthinkable that
the Roman iconography of the deity/warrior’s face on the Egadi 7 ram was vandalized by
its new owners. Since the Carthaginians were forced to rush naval preparations for the
relief eort in 241 (Polyb. 1.61.4-5), the lack of decoration and unnished tool marks on
the Carthaginian-manufactured Egadi 3 ram (and possibly Egadi 2 and 5) is also consist-
ent with the story provided to us. Evidently the Carthaginian eet that appeared o the
Egadi Islands in 241 included older Roman-built ships (perhaps up to a quarter of the eet)
along with new warships rushed into service, all carrying the supplies and equipment for
the relief mission. These captured warships were older, their hulls in poorer condition, and
likely did not fare as well in the conict. Some of these ships were “threes”, and we have
located a sector where numerous “threes” sank. This helps explain why the Egadi 4 ram of
Roman manufacture was lying atop an amphora with a Punic grato.
Conclusions
The nds from the Egadi Islands provide archaeological evidence for the key naval
bale described by Polybius that ended the First Punic War in 241 B.C., a crushing defeat
for the Carthaginians which launched Rome on its path of expansion and empire. It is the
sought-after convergence of the archaeological and historical records. The nds represent
a new body of evidence for the study of the First Punic War and its naval engagements.
Warship rams, helmets and amphoras discovered in situ in a dened area aest to the
Corleone, Sicily (CIL I2 2877 = ILLRP 1277), both dating to the 250s B.C.
92 Gnoli 2011. The clear dierences in iconography, inscriptions and overall shape, combined with
its unknown provenience, make an association of the Egadi 1 ram with the events of the First
Punic War somewhat problematic.
S. Tusa and J. Royal
46
location of ship-to-ship conicts. The rams more than double the corpus of waterline rams
that survive from antiquity. This assemblage from the bale landscape is the best data set
available for assessing this naval bale, techniques of ram manufacture, warship construc-
tion, and the depositional fates of warships of the 3rd c. B.C. Helmets from the bale zone
provide some of the earliest contextually-dated evidence for the study of armor in the 3rd
c. B.C. Further helmets to be collected and the probable presence of additional armor and
weapons in the bale zone should enhance our understanding of military equipment used
by central- and western-Mediterranean states at this time. The amphora nds aest to the
particular mission of the warships in this conict, while the entire bale zone gives elo-
quent testimony to the high stakes and deadly forces involved in ancient naval warfare.
sebtusa@archeosicilia.it Soprintendenza del Mare — Regione Siciliana
jroyal@rpmnautical.org RPM Nautical Foundation, Key West, FL
Acknowledgements
We thank Stefano Zangara (Soprintendenza del Mare, Sicilia) for assistance in the eld and docu-
mentation of the rams, Giovanni Garbini (Università di Roma La Sapienza) for analysis of the Egadi 3
ram’s Punic inscription, Jonathan Prag (University of Oxford) for comments and ongoing analysis of
the Latin inscriptions, and Philip Schmi (Eastern Michigan University) for comments and ongoing
analysis of the Punic inscriptions. Tommaso Gnoli kindly provided a series of papers concerning the
Egadi 1 inscription. Much appreciation is oered to A. Goldman (Gonzaga University) for provid-
ing material on Roman armor, M. Pagano (Soprintendente per I Beni Archeologici dell’Umbria) for
permission to use images of helmets in the Museo Archeologico Nazionale di Perugia, L. Minarini
for the images of helmets in the Museo Civico Archeologico di Bolgna, and C. H. Halle (University
of California at Berkeley) and T. Hölscher (Univ. of Heidelberg) for their review of the iconography
on the Egadi 4 and 6 rams. Gloria Olcese and Roald Docter kindly oered their opinions on the
amphora material and were extremely helpful in furthering the research. The discussion of wood
properties beneed greatly from consultation with Brian Jordan. The authors and editor of JRA
oer their special thanks to William Murray who reviewed the article in detail and oered many
valuable suggestions.
Bibliography
Alaimo, R., G. Montana and R. Giarusso 1997. “Le ceramiche comuni di Agrigento, Segesta e Termini
Imerese: risultati archeometrici e problemi archeologici,” in S. Bianchi and B. Babbri (edd.), Il con-
tributo delle analisi archeometriche allo studio delle ceramiche grezze e comuni: il rapporto forma/funzione/
impasto (Bologna) 46-69.
Basch, L. 1987. Le musée imaginaire de la marine antique (Athens).
Bernabò Brea, L. and M. Cavalier 1985. Archeologia subacquea nelle Isole Eolie (Boll. d’Arte Suppl. 29;
Archeologia Subacquea 2) 11-127.
Bishop, M. and J. Coulston 2006. Roman military equipment (Oxford).
Bockius, R. forthcoming. “News from the Bremerhaven ram,” Tropis IX Conference, Cyprus 2005.
Buccellato, C. and S. Tusa 2007. “Il rostro della baaglia delle Egadi,” in M. L. Famà (ed.), Antiche
testimonianze di pace e guerra dal mare di Trapani (Trapani) 37-40.
Caravale, A. and I. Toolei 1997. Anfore antiche: conoscerle e identicarle (Formello).
Casson, L. 1995. Ships and seamanship in the ancient world (Princeton, NJ).
Charlin, G., J.-M. Gassend and R. Lequément 1978. “L’epave antique de la Baie de Cavaliere (Le
Lavandou, Var ),” Archaeonautica 2, 9-93.
Coates, J. F. 1995. “The naval architecture of oar systems of ancient galleys,” in Gardiner 1995, 127-41.
Corsi, L. 1998. “Isola di Montecristo,” in G. Poggesi and P. Rendini (edd.), Memorie sommerse. Archeo-
logia subacquea in Toscana (cat. mostra, Santo Stefano–Grosseto 1997) 136-41.
Crawford, M. H. 1975. Roman Republican coinage (Cambridge).
D’Agostino, M., M. Bondioli and L. Fozzati 1997. “Lago di Garda, Lazise (VR). Relio di nave lunga
veneziana: il relazione preliminare,” ArchMed 24, 147-53.
De Sanctis, G. 1916. Storia dei Romani III. L’età delle guerre puniche, Parte 1 (Torino).
The landscape of the naval bale at the Egadi Islands 47
Di Stefano, C. 1993. Lilibeo punica (Rome).
Filippi, A. 2005. “Le forticazioni militari sul monte Erice durante la prima guerra punica,” in Tusa
2005, 83-94.
Frey, D., F. Hentschel and D. Keith 1978. “Deepwater archaeology: the Capistello Wreck excavation,
Lipari, Aeolian Islands,” IJNA 7, 279-300.
Frost H., 1974. “The Punic wreck in Sicily, Second season of excavation,” IJNA 3.1, 35-54.
Gabrielsen, V. 1995. “The Athenian navy in the fourth century B.C.,” in R. Gardiner (ed.), The age of
the galley: Mediterranean oared vessels since pre-Classical times (Conway’s History of the Ship) 234-40.
Gambin, T. 2002-3. “The harbours of ancient Gozo,” Malta Arch. Rev. 6, 20-26.
Gardiner, R. (ed.), The age of the galley: Mediterranean oared vessels since pre-classical times (Conway’s
History of the Ship; London).
Gassend, J.-M. 1982. L’épave antique du Lacydon (Marseilles).
Gassend, J.-M., B. Liou and S. Ximenes 1984. “L’épave 2 de l’Anse de Laurons,” Archaeonautica 4,
75-105.
Gnoli, T. 2011. “La baaglia delle Egadi: a proposito di ritrovamenti recenti,” RivStorAnt 41, 47-86.
Goldman, A. forthcoming. “Weapons and the army,” in J. DeRose Evans (ed.), The archaeology of the
Roman Republic (Oxford).
Gómez, L., M. Galán, J. Pérez and S. Lillo 2006. Estudio histórico-arqueológico de la ciudad de Carteia (San
Roque, Cádiz) 1994-1999, vol. 1 (Seville).
Grace, V. 1952. “Timbres amphoriques trouvés à Délos,” BCH 76, 514-40.
Gullea, M. I. 2005. “Navi romane fra gli Specola Lilybitana e le Aegades Geminae? Note per una
ricostruzione topograca della baaglia della Egadi,” in Tusa 2005, 71-82.
Hesnard A. et al. 1989. “Aires de production des gréco-italiques et des Dr. 1,” in Amphores romaines et
histoire économique: dix ans de recherche (CollEFR 114) 21-65.
Hassebrook, L, E. Crane, C. Begley and P. Campbell 2011. “Surface analysis of Egadi artifacts,” avail-
able at www.rpmnautical.org/egadi3/egadi3.pdf
Kahanov, Y. and J. Royal 1996. “The INA/CMS Joint Expedition to Tantura Lagoon Excavation
Report: Tantura A – Hull Construction,” CMS Newsleer, 21-23.
Kahanov, Y., J. Royal and J. Hall 2002. “The Tantura A and B wrecks: an overview of the discoveries
and their place in ancient Mediterranean shipbuilding,” in F. Hocker and C. Ward (edd.), Contri-
butions in honor of Professor J. Richard Stey (College Station, TX) 113-27.
Liko, H. 2001a. “Die Produktion großgriechischer Amphoren in Velia. Eine Untersuchung der
Amphoren (4.-1. Jh. v.Chr.) aus der Insula II in Velia,” in F. Blakholmer and H. Szemethy (edd.),
Akten des 8. Österreichischen Archäologentages am Institut für Klassiche Archäologie der Universität
Wien (Wiener Forschungen zur Archäologie) 491–95.
Liko, H. 2001b. “Zur Entwicklung graeco-italischer Amphoren in Velia,” RCRFActa 37, 265-68.
Maggiani, A. 1982. “Isola di Montecristo–Cala del Diavolo,” in M. Martelli (ed.), Archeologia subacquea
in Toscana (Boll. d’Arte Suppl. 4; Archeologia Subacquea) 65-68.
Mannoni, G. 1984. “Caraerizzazioni mineralogico-petrograche di alcune classi di reperti,” in
M. Bonghi Jovino (ed.), Ricerche a Pompei: l’Insula 5 dela Regio VI dalle origini al 79 d.C. (Rome)
346-51.
Martelli, M. 1982. Archeologia subacquea in Toscana. Archeologia subacquea (Boll. d’Arte Suppl. 4) 37-86.
Morrison, J. S. 1995. “Hellenistic oared warships: 399-31 BC,” in Gardiner 1995, 66-77.
Morrison, J. S. 1996. Greek and Roman oared warships (Oxford).
Morrison, J. S., J. Coates and N. Rankov 2000. The Athenian trireme. The history and reconstruction of an
ancient Greek warship (Cambridge).
Murray, W. 1985. “The weight of trireme rams and the price of bronze in fourth-century Athens,”
GRBS 26, 141-50.
Murray, W. 2011. The age of Titans. The rise and fall of the great Hellenistic navies (Oxford).
Olcese, G. 2007. “The production and circulation of Greco-Italic amphorae of Campania (Ischia/Bay
of Naples). The data of the archaeological and archaeometric research,” SKYLLIS, Zeitschrift für
Unterwasserarchäologie 7 [2005/6] 60-75.
Olcese, G. 2004. “Anfore greco-italiche antiche: alcune osservazioni sull’origine e sulla circolazione
alla luce di recenti ricerche archeologiche e archeometriche,” in E. C. De Sena and H. Dessales
(edd.), Archaeological methods and approaches: industry and commerce in ancient Italy (BAR S1262;
Oxford) 173-92.
Oron, A. 2001. The Athlit ram: Classical and Hellenistic bronze casting technology (M.A. thesis, Texas
A&M Univ., College Station).
Peacock, D. P. S. and D. F. Williams 1991. Amphorae and the Roman economy (London).
S. Tusa and J. Royal
48
Pomey, P. 1982. Le navire romain de la Madrague de Giens (Paris).
Ramón Torres, J. 1995. Las ánforas fenicio-púnicas del Mediterraneo central y occidental (Barcelona).
Ramón Torres, J., J. Grego, D. Vilaro and J. Ponce (edd.) 1998. Les facies ceràmiques d’importació a la
costa ibèrica, les Balears i les Pitiuses durant el segle III a.C. i la primera meitat del segle II a.C. (Arqueo
Mediterrània 4, Barcelona).
Ricordi, P. 2005. “Studio della baaglia delle Egadi: ipotesi tecniche sulla dinamica,” in S. Tusa (ed.),
Il mare delle Egadi, regione siciliana (Palermo) 95-105.
Rival, M. 1991. La charpenterie navale romaine (Paris).
Romagnoli, M. 2009. “Aggiornamento. Analisi rostro di Acqualadroni: identicazione dei legni”
(Dipt. di Tecnologie, Ingegneria e Scienze dell’Ambiente e delle Foreste, Universita degli Studi
della Tuscia, September 23, 2009).
Rossi, D. 1982, “Barbarossa,” in P. Pelagai and E. Riccomini (edd.), Archeologia subacquea (Boll.
d’Arte Suppl. 4) 85.
Royal, J. 2002. The development and utilization of ship technology in the Roman world in late antiquity: third
to seventh century A.D. (Ph.D. diss., Texas A&M Univ., College Station).
Royal, J. 2008a. “Description and analysis of the nds from the 2006 Turkish coastal survey: Marma-
ris and Bodrum,” IJNA 37, 88-97.
Royal, J. 2008b. “The discovery of a medieval galley in Turkey: a response,” IJNA 38, 388-90.
Royal, J. and J. McManamon 2010. “At the transition from Late Medieval to Early Modern: the archae-
ology of three deepwater nds from Turkey,” IJNA 39, 327-44.
Royal, J. and J. McManamon 2009. “Three Renaissance wrecks from Turkey and their implications for
maritime history in the Eastern Mediterranean,” J. Maritime Arch. 4.2, 103-29.
Royal, J. and S. Tusa 2012. “The Levanzo I wreck: a 4th-century CE merchantman in the service of the
annona?,” IJNA 41, 26-55.
Sáez Romero, A. 2007. “El fenómeno del estampillado anfórico en el alfar tardopúnico gadirita de
Torre Alta. Balance historiográco y novedades,” Actas del III Encontro de Arqueologia do Sudoeste
Peninsular, Vipasca (2006) = Arqueologia e História 2 (Aljustrel) 243-53.
Sáez Romero, A. and J. Díaz Rodríguez 2008. “La producción de ánforas de tipo griego y greco-itálico
en Gadir y el área del Estrecho. Cuestiones tipológicas y de contenido,” Zephyrus 60, 195-208.
Stey, J. R. 1994. Wooden ship building and the interpretation of shipwrecks (College Station, TX).
Stey, J. R. 1991. “The ram and bow timbers: a structural interpretation,” in L. Casson and J. R. Stey
(edd.), The Athlit ram (College Station, TX) 6-39.
Steinhauer, G., 2006. The monuments and the Archaeological Museum of Piraeus (Athens).
Strauss, B. 2004. The Bale of Salamis (New York).
Tchernia, A., P. Pomey and A. Hesnard 1978. L’épave romaine de la Madrague de Giens (Var) (campagnes
1972-1975) (Gallia Suppl. 24).
Thierrin-Michael, G. 2000. “Amphores gréco-italiques de l’Oppidum de Pech-Maho (III siècle avant
J.-C.). Détermination d’origine par l’étude pétrographique,” SFECAG, Actes du Congrès de Libourne
(Marseille) 225-31.
Thierrin-Michael, G. and G. Galei 1996. “Amphores vinaires italiques: deux problèmes rencontrés
lors de l’utili-sation de groupes de références pour l’aribution de tessons d’origine inconnue,” in
G. Galei (ed.), L’archéométrie dans les pays européens de langue latine et l’implication de l’archéométrie
dans les grands travaux de sauvetage archéologique (Actes du Colloque d’Archéométrie 1995; Suppl.
à la Revue d’Archéométrie) 119-24.
Tusa, S. (ed.) 2005. Il mare delle Egadi, Storia, itinerari e parchi archeologici subacquei (Palermo).
Tusa, S. 2005a. “Sintesi storico-archeologica e potenzialità della ricerca,” in id. 2005, 55-70.
Tusa, S. 2005b. “Il rostro della baaglia,” Archeo 21.3, 10-12.
Tusa, S. 2010. Archeologia e storia nei mari di Sicilia (Udine) 219-28.
Tusa, S. and C. Buccellato 2012. Il rostrum Acqualadroni.
Van der Mersch, C. 1994. Vins et amphores de Grande Grèce et de Sicile, IVe-IIIe s. av. J.-C. (Naples).
Will, E. 1982. “Greco-Italic amphoras,” Hesperia 51, 338-56.
Zecchini, M. 1982. Relii romani dell’Isola d’Elba (Lucca).
