Archaeometry 52, 6 (2010) 1096–1109
doi: 10.1111/j.1475-4754.2010.00526.x
HANNIBAL’S INVASION ROUTE: AN AGE-OLD QUESTION
REVISITED WITHIN A GEOARCHAEOLOGICAL AND
PALAEOBOTANICAL CONTEXT*
W. C. MAHANEY
Quaternary Surveys, 26 Thornhill Avenue, Thornhill, Ontario, Canada, L4J 1J4
and
Department of Geography, York University, 4700 Keele Street, North York, Ontario, Canada, M3J 1P3
P. TRICART
Laboratoire de Géodynamique des Chaînes alpines, University of Grenoble, Observatoire des Sciences de l’Univers,
38041 Grenoble, France
C. CARCAILLET, O. BLARQUEZ and A. A. ALI
Centre for Bio-Archeology and Ecology (UMR5059 CNRS) and Paleoenvironments and Chronoecology (PALECO, EPHE),
Institut de Botanique, University Montpellier 2, 163 rue Broussonet, 34090 Montpellier, France
J. ARGANT
LAMPEA-UMR 6636, Institut Dolomieu, Grenoble, France
and
ARPA, Université Claude Bernard, Lyon 1, France
R. W. BARENDREGT
Department of Geography, Lethbridge University, Lethbridge, Alberta, Canada, T1K 3M4
and V. KALM
Institute of Ecology and Earth Science, Tartu University, Tartu, Estonia 51014
The point of Hannibal’s departure from New Carthage in Iberia, in 218 BC, and his subsequent
march along the Mediterranean coast to the Pyrénées and on to the Rhône Basin, has been
reconstructed by ancient historians with considerable accuracy. The latter 400-km phase
through the Alps, however, has been the subject of some controversy as to whether the Punic
Army followed a southern versus a northern invasion route, or some intermediate variant.
What is certain from the ancient texts is that Hannibal was trapped by Gallic tribes in a large
defile—a gorge large enough to hold the entire army—along the approach to the high col of
passage on to the Po River Plains of northern Italia. The entrapment involved an enfilade
attack planned by an unknown Gallic commander, a military operation that nearly decimated
the Punic Army. Previous arguments as to the location of the defile have hinged on inconclusive topographic, geological and geomorphic assessments. New data from palaeobotanical
reconstruction of the northern approach route show the Gorges de la Bourne and the Gorge
du Bréda, astride the Isère River, to have been forest covered during the invasion, which would
*Received 5 July 2009; accepted 29 January 2010
© University of Oxford, 2010
Hannibal’s invasion route
1097
have made the Gallic assault impossible. The existing evidence argues for a southern route,
the approach through the narrow defile of the Combe de Queyras, with passage over the Col
de la Traversette, as argued by Sir Gavin de Beer nearly a half century ago. Narrowing the
approach route focuses on sites worth geoarchaeological exploration.
KEYWORDS: HANNIBALIC INVASION, 218 BC, METHODS, GEOARCHAEOLOGY,
PALAEOBOTANY, UPPER TREELINE
INTRODUCTION
The actual route followed by the Punic Army into Italia in 218 bc has been the subject of
considerable controversy for millennia (Larauza 1826; Dodge 1891; de Montholon 1905; Brown
1963; Guillaume 1967; Hart 1967; de Beer 1969; Connolly 1978, 1981; Cottrell 1992; Renaud
1996; Lazenby 1998; Bagnall 1999; Lancel 1999; Appian, trans. White 2002; Mahaney 2008,
2010; Mahaney et al. 2010), a mystery clouded over by the generic descriptions provided in the
ancient literature by Hannibal’s two chief chroniclers: Livy and Polybius (Livy, trans. de Sélincourt 1972; Polybius, trans. Scott-Kilvert 1979).
Three main invasion routes (Fig. 1 (a)) have been proposed by numerous authorities (Wilkinson 1911; Walbank 1956, 1990; Proctor 1971; Seibert 1993; Healy 1994; Mahaney 2004, 2008;
Mahaney et al. 2008a,b) after Hannibal left his re-supply position at the ‘Island’ on the Rhône
River. The northern route (Fig. 1 (a)) follows the Rhône to its confluence with the Isère River,
along the Isère to the Arc River and thence to the Col du Mont Cenis or Col du Clapier. A second
possibility is a middle route from near the present-day city of Grenoble, past the Pelvoux Massif
to rejoin the Durance and march over the Col du Mont Genèvre. A third possibility is the southern
route along the Rhône to the Drôme and its junction with Le Bez River leading to the Gorges des
Gaz, over the Col de Grîmone towards the city of Gap and the Durance Basin, and thence through
the Combe de Queyras to the Col de la Traversette and the upper Po River country.
Both Livy and Polybius document that the Gauls planned two attacks against the Punic Army.
The first was a blocking manoeuvre in which the Gauls held the high ground, presumably a col,
where Hannibal would have had to deploy and engage them to effect passage. If Hannibal took
the northern or middle routes, this engagement must have occurred near present-day Grenoble; if
along the Drôme-Le Bez catchments, near either Col de Grîmone or Col de Cabre. Of these two
cols, the Col de Grîmone is preferred, as the topography closely resembles what Polybius (III, 50)
described, where Hannibal deployed his army in the lower valley, giving the impression that he
was camping for the night. After his scouts reported that the enemy had departed the col at night,
retiring to a nearby town, Hannibal formed a special column and led a night action, taking the col
unopposed. When the Gauls reappeared the following morning, he surprised them and routed
their columns. The nearby town has been referred to by many sources either as Gap, Grenoble
(which did not exist in any size in Hannibal’s time), Mons Seleucus or Les Lusettes. If this action
took place at the Col de Grîmone, the nearby town within walking distance would have to be at
or near Les Lusettes (Mahaney 2008).
More importantly, perhaps, and following this night action somewhere on the approach route
to the high col, Hannibal was attacked by Gauls who were intent on pillaging his baggage train
and/or halting the invasion altogether. The exact location of this attack depends on the existence
of a prominent defile with steep cliffs, from which the Gauls could roll or fling rocks, producing
a considerable rock avalanche, a veritable ‘missile’ barrage. Polybius (III, 53) is clear that the
defile in question is a ‘gorge’, an entrenched valley of considerable size, one capable of containing the entire army, and located somewhere near the final col leading into the Po River country.
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
1098
W. C. Mahaney et al.
(a)
(b)
Figure 1 (a) A satellite image of southern France and the Western Alps, showing the three possible routes that Hannibal
may have followed in 218 BC (from Mahaney 2008). Of the possible approach routes, only three exhibit steep defiles:
Combe de Queyras [1], Gorges de la Bourne [2] and Gorge du Bréda [3]. (b) The geology of the internal and external
arcs of the French and Italian Alps (from Mahaney 2008; Mahaney and Tricart 2008).
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
Hannibal’s invasion route
1099
The disposition of the Punic Army was such that the cavalry led, followed by the baggage train
with heavy infantry, under Hannibal’s command, making up the rear echelon. His heavy infantry
covered the first assault and repulsed the Gauls. However, an engagement quickly followed,
developing into an enfilade attack along the Punic flanks with devastating effect (see Mahaney
and Tricart 2008). Whatever planning went into the Gallic offensive, the effect was severe enough
to cause considerable casualties, with perhaps as many as 10 000 soldiers, principally mercenaries, dying in one 24-hour period. Hannibal was forced to spend the night with only half his force
near a ‘certain bare rock which offered some protection’ (Polybius, III, 53; Mahaney 2008),
indicating substantial rock outcrops that might provide cover for Hannibal and his staff/
bodyguard. Following the onslaught, Polybius states that ‘Hannibal was able to rejoin the cavalry
and baggage animals and advance towards the top of the pass’, which indicates the col was within
a reasonably short distance (i.e., ~30 km).
To assess and reconstruct the approach route, it is necessary to analyse the topography and the
geology/geomorphology of the proposed approach routes. The defile needs to have the geometry
to contain the entire Punic Army, which consisted of some ~35 000–40 000 men and 10 000–
15 000 animals at this point. Moreover, the valley-side slopes need to be steep enough to roll/drop
boulders on to the soldiers making their way along the valley bottom; hence, the slopes need to
be free of timber and within an easy trek of the high col.
Thus, the only significant defiles that exist on the southern, middle and northern routes are
analysed.
REGIONAL GEOLOGY/TOPOGRAPHY
The regional geology (Fig. 1 (b)) shows a series of overlapping thrust sheets oriented from
north-east to south-west, through which valleys have been cut over Tertiary and Quaternary time,
first by water flowing down the regional slope and later by meltwater from a succession of glacial
advances during the ice ages (Billard and Orombelli 1986; Bini and Zuccoli 2004; Ehlers and
Gibbard 2007). The bedrock geology upon which drainage lines have been imprinted is the
culmination of several episodes of plate tectonics, resulting in first the opening and then the closure
of a small oceanic domain called the Alpine Tethys. The ridge and trough structures that make up
the Alps today mainly derive from this vanished ocean and its European margin. It is through these
valleys, ringed with high ridges, that Hannibal marched his army into Italia in 218 bc.
The Alps straddle the plate boundary between Europe and Africa—or, more precisely, a
northern African promontory, called Adria. Along this boundary, the first predominant plate
divergence engendered continental rifting (Early–Middle Jurassic) followed by ocean spreading
(Late Jurassic – Cretaceous). A kinematics reversal induced subduction of the ocean (Late
Cretaceous – Eocene) and then a continental collision that is still active today (e.g., Tricart 1984).
Large-scale overriding of the Adriatic palaeo-passive margin on to the European one was early
described by Argand (1911), who gave the Alps their structural celebrity. The Western Alpine arc
(Fig. 1 (b)) developed along a 300-km length, with a width ranging between 150 and 200 km.
Globally, the resulting relief displays a marked asymmetry, the inner, eastern slope being much
steeper than the outer, western one, a point reiterated by historians who have studied the routes
(Lazenby 1998; Bagnall 1999). This orogen asymmetry is directly inherited from the structural
asymmetry of the subduction to collision wedge, resulting from repeated outward-directed
tectonic transport. Viewed on the map, curvature resulted from unequal shortening of the European palaeomargin at the western tip of the Adria promontory, acting as an indenter during the
Oligocene paroxysm of the collision.
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
1100
W. C. Mahaney et al.
Outside the arc, in the external (or Helvetic–Dauphinois) zone, the main mountainous relief
forms are contorted (flexed) between NE–SW in the north and NW–SE in the south. They
underline the main fold–thrust structures developed in the Palaeozoic crystalline basement (e.g.,
the Mont-Blanc, Belledonne or Argentera massifs) or in the Mesozoic ante-rift to post-rift
sedimentary cover (e.g., the Chartreuse or Vercors massifs, the latter being one of the gorge
systems where Hannibal is thought to have been ambushed by the Gauls; see Lazenby 1998).
Between these two massive ensembles, a major tectonic uncoupling is identified by Triassic
evaporites or in the thick overlying Liassic marls of Mesozoic age. In this external arc, gentle
Cainozoic shortening is identified by inverted Mesozoic rift-related fault blocks, the trend of
which also underlines the curved geometry of the outer mountain belt. Transverse valleys may
correspond to syn- or post-folding transcurrent faults, but also to much older faults interpreted as
transfer faults in the palaeorift structure.
In the south-west, original east–west elongated relief features such as Dévoluy and Diois
correspond to a train of regional-scale east–west folds. These are not strictly Alpine folds
developed in Late Cretaceous to Eocene times, but are directly related with Provence–Pyrénées
tectonics, when the Iberia microplate collided with Europe.
This contrasts with the core of the Western Alpine arc, which displays a thick pile of metamorphic nappes; in other words, allochothonous rock sheets moved on a horizontal surface.
These nappes derive from both an uplift shoulder and the contiguous deepest part of the rifted
structure when referring to the Early to Middle Jurassic palaeogeography, and derive from the
middle/distal part of the passive margin, and from the oceanic realm, when referring to the
following period. The original stack of nappes with its recumbent folds bearing flat-lying
foliations was mainly emplaced in an accretionary wedge, as the ocean and its stretched margin
was subducted. During the subsequent collision, several severe shortening phases resulted in
post-nappe folds and thrusts. The latest ones are directed towards the core of the arc (the
so-called backfolds and backthrusts) and underline the curved geometry of the inner mountain
belt. From the Late Cainozoic onwards (the Miocene?—see the discussion in Tricart et al.
2006), this complex polystage structure corresponds to the most thickened part of the Alpine
crust, which underwent gravitational collapse. It resulted in extensional and transtensional
faulting, parallel and transverse to the mountain belt; that is, roughly tangential and radial to the
arc. This late Alpine fault net, which remains partly active, is directly responsible for the
location of the main valleys, the approach routes that Hannibal followed to reach the high col
or passage into Italia. It explains why relief appears more distorted in the internal arc than in
the external arc.
Hannibal would not have been interested in the geology as much as the topography, but when
the attack came in the approach defile, he certainly must have been acutely aware of the ease with
which Gallic sappers removed large boulders from the upper cliffs, hurling them with great
precision on to his columns of troops. While many outcrops contain schist, which in tabular form
is not easily moved, the Gallic attackers proved adept at picking subround boulders as ‘missiles’
from among the mixed lithologies left behind by Late Pleistocene glaciers.
The middle route (Fig. 1 (a)) proposed by Connolly (1981), along the Drac River to the
Romanche River, comprises drainages lacking shear vertical cliffs. All steep slopes lie
within the upper limit of forest species in Hannibal’s time, as indicated by palynological
reconstructions.
Comparing the rubble volume in the northern gorges with the southern Combe de Queyras, the
latter arguably contains the far greater mass of rock debris, some piles probably a vestige of the
2200-year-old assault. More importantly, the great piles of rock rubble all along the Combe de
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
Hannibal’s invasion route
1101
Queyras (Mahaney 2008; Mahaney and Tricart 2008) provide fertile ground for geoarchaeological exploration with ground-penetrating radar (GPR), and metal detecting and/or magnetometer
equipment.
Several geophysical methods are available that can be used to probe the shallow subsurface of
the rubble piles, hearths and stone wall systems outlined herein and in Mahaney (2008), but these
tests should be undertaken by someone with a license to carry out exploratory archaeology. These
methods include GPR, metal detectors and magnetometers.
Ground-penetrating radar, as with metal detectors and magnetometer surveys, offers nondestructive geophysical techniques for imaging the subsurface of key sites associated with the
invasion. The pathways into Italia from the Combe de Queyras, the hearths in the upper Guil
River, and the two-tier rockfall and the regrouping area in the Upper Po Basin are choice localities
to image with GPR. The GPR instrument images a cross-section of a site and provides a profile
of underground features including buried objects, bedrock and deposits of various compositions.
The method relies on transmitted pulses of ultra-high-frequency radio waves (microwavelength
electromagnetic energy) into the ground through an antenna (transducer), the transmitted energy
reflected from underground structures or objects being received and stored in a digital control
module. As the radiowave signal passes through the ground, it contacts strata with changing
moisture contents and, hence, electrical conductivities and dielectric constants. Only part of the
radar waves reflect off the object; the rest pass to the next interface, strata or object. Reflected
signals return to the antenna and pass into the digital control module, which measures the
two-way travel time in nanoseconds and amplifies signals in the process. The output signal
voltages are plotted as a profile showing different colour bands for each change in strata or
interface that produced the reflected wave.
The system is ideal for imaging deep structures, and for most applications here the depth
required is less than 1.0 m in most instances—or in the regrouping area, where debris flow
material covers important sites, the depth is expanded to 2.0 m, well within the range of the
instrument. For most shallow imaging, high-frequency radar is suitable and uses antennae in the
300–1500 MHz range (Conyers 2005).
Along with GPR, researchers might consider conducting a survey of key sites using a metal
detector, as outlined in several technical manuals (i.e., Ostdiek and Bord 2008). A metal detection
survey might reveal a wealth of buried metal objects, including knives, spear points, swords, belt
buckles, shovels, pry bars, picks, equestrian materials, coins and so on, all of inestimable value
in reconstructing the military culture of ancient Carthage. An outline of the method may be
obtained by consulting ASTM (American Society for Testing and Materials) manual D7046-04
(ASTM 2004), which summarizes the equipment, methods employed in the field and interpretation necessary for the assessment of subsurface materials. Both ferrous (iron and steel) and
non-ferrous (aluminium and copper) metals are detected by inducing eddy currents in conductive
objects. Metal detectors use either frequency domain (continuous frequency or wave) or time
domain (pulsed) systems.
A vast range of metal detectors is commonly available, although the Fisher M-Scope 1212-X
is widely used. In contrast, if a magnetometer is used, it will detect only ferrous materials. A
widely used instrument is the Schonstedt Model GA-52Cx (see Schonstedt 2001).
PALAEOBOTANICAL RECONSTRUCTION
If Hannibal advanced along the Isère River to the Arc River (Fig. 2 (a)), he would not have
encountered a gorge with steep cliffs and he could have marched either side of the river without
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
1102
W. C. Mahaney et al.
(a)
(b)
Figure 2 (a) The broad expanse of the Isère River Valley. (b) Gorges de la Bourne.
fear of a major Gallic ambush. If, on the other hand, he deviated from the Isère into the Gorges
de la Bourne, astride the Vercors Massif near Grenoble, and later through the defile between La
Rochette and Pontcharra (the Gorge du Bréda, immediately to the east of the Isère Valley; Fig. 2
(b)), he would have rejoined the Isère River near its confluence with the Arc. To do this he would
have had to pass through short defiles with steep forested cliffs on either side (see Lazenby 1998).
Despite the presence of large talus cones on either side of streams in the Gorges de la Bourne and
the La Rochette – Pontcharra defile, a kind of buffer zone that would absorb the impact of hurled
or dropped boulders, it is important to establish whether forest laced the valley sides during the
advance. If forested, the attack as described by Polybius and Livy, would have been impossible
on the northern approach routes.
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
Hannibal’s invasion route
1103
The forest-limit position 2200 years ago (218 bc), and eventually the treeline, are critical
components to establish if one were to argue that the decline or advance of forest into the alpine
tundra influenced the ease with which the Gauls were able to dislodge boulders on to the Punic
soldiers in the lower valley. Climatic reconstructions (Neumann 1992; von Grafenstein et al.
1998; Davis et al. 2003; Ortu et al. 2006, 2008) of the temperature/precipitation pattern for the
Alps during the Neoglacial (200 bc) suggest that the climate was similar to that of the mid-20th
century, without the advanced global warming under way in the first decade of the 21st century
(Mahaney 2008).
The Combe de Queyras (Figs 3 (a) and 3 (b)), the only major defile along the southern route,
has almost vertical cliffs along the 21-km stretch of the Guil River, which nullifies the importance
of timberline fluctuations, large or small. However, timber fluctuations in the Gorges de la Bourne
and the Gorge du Bréda, major tributaries of the Isère River, are of some importance in assessing
the suitability of the defile for the attack outlined in the ancient literature.
The palaeobotanical records based on macroremains of woody plants (Fig. 4), the most precise
treed vegetation proxy (Birks and Birks 2000), show that at the time of Hannibal’s invasion, the
upper treeline was at 2260 m a.s.l., at least. Isolated evidence indicates that trees, mainly cembra
pine (Pinus cembra) and European larch (Larix decidua), grew at elevations as high as 2670 m
(Ali et al. 2005) until the Middle Ages (ad 700–900). All sites contain continuous records of tree
remains until ad 100, when trees disappeared from the Lac Cristol (2254 m), and then Lac du
Lait (2196 m) experienced an abrupt drop in the abundance of tree remains at ad 900 (Fig. 4).
The Lac du Thyl, which was forested until 2000 bc at least (Genries et al. 2009b), shows little
tree remains over the past 2000 years, with their total disappearance from sediment at c. 400 bp
(ad 1550). The other lakes (L. Loup, L. Perso) have kept trees on their lakeshores until nowadays,
but the total tree abundance has certainly been very low compared to what occurred between 6000
and 1000 bc (c. 8000–2950 yr bp) (not shown; see details in Blaquez et al. 2010). The upper
treeline in this region obviously moved down to reach the modern treeline between ad 500 and
1500, but the tree abundance was already low between 1000 and 0 bc compared to the middle
Holocene, ~5000–1000 bc (c. 7000–3000 yr bp) (Nakagawa et al. 2000; Genries et al. 2009a),
suggesting that forest clearing had already begun at most sites.
The generalized stratigraphy for the Gorges de la Bourne, shown in Figure 5, depicts the
lithostratigraphy extending from the early Neoglacial (3000 bc; 5000 yr bp) with coarse sedimentation (presumably a deteriorating climate in the Subboreal transitioning to the Subatlantic
Chronozone at ~800 bc (2800 yr bp). Prior to the Subatlantic, a period of weaker sedimentation
is followed by renewed frost riving and coarse sedimentation, culminating in more extensive frost
riving near the top of the section; that is, during the Little Ice Age.
The generalized pollen data in Figure 5 show variations in arboreal and non-arboreal pollen
that together reflect cultural-induced forest clearing prior to 800 bc (2800 yr bp), followed by a
rise in arboreal pollen (i.e., the Hallstatt epoch). Hence, the elevated forest limit may have lasted
for two to three centuries following the ingress of fir, which might relate to Neoglacial cooling.
After 800 bc (2800 yr bp), slight alternations in hazelnut and boxtree species indicate a quasistable forest limit with only minor variations in elevation. It is clear that the present forest limit
more or less approximates the forest limit that Hannibal would have encountered had he ventured
into Italia along the northern route advocated by Lazenby (1998).
If the forest limit has not significantly fluctuated since Hannibal’s time (Genries et al. 2009b),
the snowline has remained relatively constant with the exception of the Little Ice Age (ad
1500–1800; 450–150 yr bp). Previous mention of a drop in the average snowline by Connolly
(1978) to the order of 1000 m in Hannibal’s time is a great over-exaggeration. Today, the
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
1104
W. C. Mahaney et al.
(a)
(b)
Figure 3 (a) The entrance to the Combe de Queyras from the lower Guil River (from Mahaney et al. 2008b). (b) The
upper gorge near La Maison du Roi (from Mahaney 2008).
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
Hannibal’s invasion route
2450
1105
2670 m
2400
Dated charcoal
from soil or
from travertine
2350
2300
L. Cristol, 2254 m
2250
L. Miroir, 2214 m
2200
L. Lait, 2196 m
2150
2100
L. Thyl, 2035 m
2050
L. Loup, 2026 m
2000
L. Perso, 2000 m
1950
1000
3000
0
Age yr BC/AD
2000
1000
1000
Age cal. yr BP
2000
0
Figure 4 A synthesis of all the spatially precise palaeobotanical records from the region around the Col de la
Traversette plotted against elevation (m a.s.l.). The data are from six lakes distributed between 2000 and 2240 m, and
from dated charcoal from soil or travertines (= tuff). The lake palaeobotanical data are the total accumulation rate of tree
macroremains (mainly Pinus cembra, Pinus mugo/uncinata, Larix decidua, Abies alba, Betula sp., Alnus sp. and Sorbus
sp.). The macroremains accumulation rates are first rescaled (a minimax transformation computed over the whole
Holocene, where total remains generally peaked between 6000 and 4000 BC), then smoothed to suppress the noise due
to chaotic records of plant macroremains. The scales of the curves are identical. The dated charcoal from soil are Pinus
cembra, Larix decidua, Salix and Betula. The vertical grey line corresponds to the date of the Hannibalic invasion.
snowline fluctuates between 1800 m a.s.l. in winter to ~2900 m a.s.l. in summer, with variations
of 1300 m depending on aspect (for a discussion of microclimate on permafrost distribution in
the Western Alps, see Mahaney et al. 2007). The average fall snowline position (October–
November) would be 2600 m a.s.l. 1200 m today, similar to Hannibal’s time. For satellite
analysis of snowline positions, see Droz and Wunderle (2002). The forest-limit elevation would
be within the plus-or-minus margins or around 2500 m a.s.l. on the French side, and about
1800 m a.s.l. on the Italian flank of the Alps, the contrasting elevations reflecting aspect, nearness
to maritime sources and prevailing wind. In part, the modern lower treeline elevation on the
Italian side is due to human activity (transhumance, charcoal production).
DISCUSSION
The environmental evidence for the second attack described by Polybius argues against the
Gorges des Gaz in the Dauphiné Alps. Not only is the gorge heavily wooded up close to the Col
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
1106
W. C. Mahaney et al.
Figure 5 A synthetic palynological curve from the Choranche, a rock shelter in the Bourne Valley, close to its confluence
with the Isère River (after Bintz and Girard in Bui Thi Mai et al. 1987).
de Grîmone, but it is not of a size compatible to hold the entire Punic Army as one massive arm.
The gorge is also nearly 200 km from the high col of either the Mont Genèvre or the Traversette,
neither of which lie within easy marching distance as described by Polybius. While the Gorge de
Gaz – Col de Grîmone fits the description of the first attack described by Polybius, it is hardly
convincing as the venue for the second attack.
The Combe de Queyras, picked as the most likely candidate for the second attack by de Beer
(1967, 1969) and Prevas (1998), is favoured here, if only due to its geometry, geological setting
and closeness to the high col. The Combe de Queyras is 21 km long, <100–300 m wide in most
places, and is consistently flanked with steep, almost vertical cliffs, which afford ample material
to be used as ‘missiles’. The volume of the defile is of a size that could hold the entire Punic
Army, with the elephant cavalry corps up front, followed by the baggage train buttressed with
mercenaries and the heavy infantry as the rear guard. Both in situ carbonate and siliceous bedrock
could be made ready to drop into the valley, and boulders of varying dimensions left behind by
ice during the last glaciation could also be used as ‘missile’ material. More than this, the distance
to the high col is of the order of 25–30 km, approximately a two-day march.
The Gorges de la Bourne, a tributary to the Isère River, narrows in places to a point where a
crossing to the upper Isère is possible. The gorge itself is quite steep, as pointed out by Lazenby
(1998), but with large accumulations of talus on the valley sides. To argue for this as the site of
the first attack, one needs to question why Hannibal would deviate from the Isère River through
a defile that would take him into the Vercors, extremely rugged terrain offering no tactical
advantage. If Hannibal did take the northern route, surely his cavalry would have informed him
that the Isère River offered the fast route to the high passes. In addition, to consider the Gorge du
Bréda as the venue for the second attack (cf., the alternative to the Combe de Queyras), one needs
to consider that the wide talus deposits act as a buffer zone between boulder-launch sites and the
main valley where the soldiers would be deployed. Any ‘missiles’ fired off the ridge crest would
come to rest on the talus piles, resulting in little, if any, damage to the army below. The forest
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
Hannibal’s invasion route
1107
cover today, at the narrowest cross-valley profile of the gorge (see Plate III in Lazenby 1998),
stretches up the valley sides and talus cones to near the top of the gorge. If this is the site of the
second ambush, one would have to explain how the Gauls managed to hurl/slide boulders through
the forest on to the valley bottom. Moreover, one needs to explain why Hannibal would deviate
from the wide valley of the Arc River (Fig. 2 (a)) to enter the Gorge du Bréda, an alternative route
that would not save any time.
Contrasting the two areas—Combe de Queyras versus the Gorges de la Bourne—the site
depicted by Polybius as the location of the vaunted ambush that nearly decimated Hannibal’s
forces would have to be along the southern route on the Guil River, below the Col de la
Traversette, as first suggested by Sir Gavin de Beer (1969).
CONCLUSIONS
Of the two approach route possibilities, the Combe de Queyras provides a near-perfect match for
the venue for the second attack. Its sheer cliffs, ample supply of easily detached bedrock, loose
regolith on the high ramparts astride the Guil River and closeness to the high col combine to make
it the most likely candidate of the two proposed sites. The enormous accumulation of rock rubble
adjacent to the Guil River, some of which has been affected by engineering and flood control
engineering works, is probably a residual mass dating to the actual attack some 2200 years ago.
Weathered scars visible on the high walls of the gorge in many places have similar light tonal
contrast with adjacent heavily altered, oxide-rich rock outcrops, along the flanks of the gorge.
While it might be impossible to find an artefact record in the rock rubble using metal detectors,
it may be worth the effort to attempt geoarchaeological exploration of the rubble sheets using
GPR. Certainly any new finds of weapons or accoutrements employed by the Punic Army would
shed considerable light on the military culture of ancient Carthage. The presence of numerous
ancient hearths in the subalpine forest of the upper Guil River and stone walls in the regrouping
area on the lee side of the Alps also invite geoarchaeological exploration.
To the north, the Gorges de la Bourne and the Gorge du Bréda are covered with trees on the
valley sides along most of the approach route to the high col, making it impossible to engineer
rockfall of sufficient magnitude to inflict maximum damage on Hannibal and his army. Where
steep cliffs are found, large forest-covered talus cones lie between the ridge and valley flat,
providing a buffer zone that would absorb the bulk of a ‘missile’ attack without causing substantial damage to the army marching below. Palaeobotanical records indicate the upper forest limit
of today is similar to Hannibal’s time of ~2200 years ago, although the upper treeline drop occurs
later, between ad 500 and 1500.
ACKNOWLEDGEMENTS
This research was made possible by funds from Quaternary Surveys, Toronto. The palaeobotanical records (Fig. 4) were funded by the Centre National de la Recherche Scientifique (CNRS,
France) and the Ecole Pratique des Hautes Etudes (EPHE, France). We thank Barbara Kapran for
assistance in the field.
REFERENCES
Ali, A. A., Carcaillet, C., Talon, B., Roiron, P., and Terral, J.-F., 2005, Pinus cembra L. (arolla pine), a common tree in
the inner French Alps since the early Holocene and above the present tree line: a synthesis based on charcoal data from
soils and travertines, Journal of Biogeography, 32, 1659–69.
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
1108
W. C. Mahaney et al.
Appian (Appianus), 2002, Roman history, vol. 1, trans. H. White, Loeb Classical History, Harvard University Press,
Cambridge, MA.
Argand, E., 1911, Les nappes de recouvrement des Alpes occidentales et les territoires environnants, Matériaux Carte
Géologique Suisse, 27.
ASTM, 2004, ASTM D7046-04, Standard guide for use of the metal detection method for subsurface investigation,
ASTM International, West Conshohocken, PA.
Bagnall, N., 1999, The Punic Wars, Pimlico, London.
Billard, A., and Orombelli, G., 1986, Quaternary glaciations in the French and Italian piedmonts of the Alps, in
Quaternary glaciations in the Northern Hemisphere (eds. V. Sibrava, D. Bowen and G. M. Richmond), 407–11,
Quaternary Science Reviews, 5(1/4).
Bini, A., and Zuccoli, L., 2004, Glacial history of the southern side of the central Alps, Italy, in Quaternary glaciations—
extent and chronology (eds. J. Ehler and P. L. Gibbard), 195–200, Elsevier, Amsterdam.
Birks, H. H., and Birks, H. J. B., 2000, Future uses of pollen analysis must include plant macrofossils, Journal of
Biogeography, 27, 31–5.
Blaquez, O., Carcaillet, C., Bremond, L., Mourier, B., and Radakovitch, O., 2010, Trees in the subalpine belt since
11 700 cal. bp: origin, expansion and alteration of the modern forest. The Holocene, 20(1), 139–46.
Brown, J. E. T., 1963, Hannibal’s route across the Alps, Greece and Rome, 2nd Series, 10 (March), 38–46.
Bui Thi Mai, Girard, M., Bintz, P., and Vital, J., 1987, Végétations, variations climatiques et évolution
culturelle du Tardiglaciaire à l’Holocène à Choranche (Vercors, Isère), Revue de Paléobiologie, Genève, 6(2), 411–
31.
Connolly, P., 1978, Hannibal and the enemies of Rome, MacDonald, London.
Connolly, P., 1981, Greece and Rome at war, MacDonald, London.
Conyers, L., 2005, Ground penetrating radar for archaeology, AltaMira Press, Lanham, MD.
Cottrell, L., 1992, Hannibal: Enemy of Rome, Da Capo Press, New York.
Davis, B. A. S., Brewer, S., Stevenson, A. C., Guiot, J., and data contributors, 2003, The temperature of Europe during
the Holocene reconstructed from pollen data, Quaternary Science Reviews, 22, 1701–16.
de Beer, G., 1967, Hannibal’s march, Sidgwick and Jackson, London.
de Beer, G., 1969, Hannibal, The Viking Press, New York.
de Montholon, J. F. T., 1905, Memoires . . . a Sainte Hélène, vol. IV, 277–81, Paris.
Dodge, T. H., 1891, Hannibal, Houghton Mifflin, NewYork.
Droz, M., and Wunderle, S., 2002, Snow line analyses in the Alps based on NOAA–AVHRR data spatial and temporal
patterns for winter and springtime in 1990, 1996 and 1999, in Proceedings, EARSeL–LISSIG Workshop Observing our
Cryosphere from Space, Bern, Switzerland, 11–13 March 2002, 149–54.
Ehlers, J., and Gibbard, P. L., 2007, The extent and chronology of Cenozoic global glaciation, Quaternary International,
164/165, 6–20.
Genries, A., Mercier, L., Lavoie, M., Muller, S. D., Radakovitch, O., and Carcaillet, C., 2009a, The effect of fire
frequency on local cembra pine populations, Ecology, 90, 476–86.
Genries, A., Muller, S. D., Mercier, L., Bircker, L., and Carcaillet, C., 2009b, Fires control spatial variability of
subalpine vegetation dynamics during the Holocene in the Maurienne valley (French Alps), Ecoscience, 16, 13–
22.
Guillaume, A., 1967, Annibal franchit les Alpes, Editions des Cahiers des Alpes, Montfleury.
Hart, B. H. L., 1967, Strategy, Penguin, London.
Healy, M., 1994, Cannae 216 BC: Hannibal smashes Rome’s army, Osprey Military Campaign Series, 36, Osprey
Military, London.
Lancel, S., 1999, Hannibal, Blackwell, Oxford.
Larauza, M. J.-L., 1826, Critical history of the crossing of the Alps by Hannibal, Dondey-Dupré père et fils, Paris, 222
pp. and map.
Lazenby, J. F., 1998, Hannibal’s war, University of Oklahoma Press, Norman, OK.
Livy, 1972, The war with Hannibal, trans. A. de Sélincourt, Penguin, London.
Mahaney, W. C., 2004, A geological/topographical reconnaissance of Hannibal’s invasion route into Italia in 218 bc, in
Studies in military geography and geology (eds. D. R. Caldwell, J. Ehlen and R. S. Harmon), 67–78, Kluwer
Academic, Dordrecht.
Mahaney, W. C., 2008, Hannibal’s Odyssey: environmental background to the Alpine invasion of Italia, Gorgias Press,
Piscataway, NJ.
Mahaney, W. C., 2010, The Alpine invasion of Italia in 218 bc: geological reconnaissance of the Alpine route, American
Journal of Ancient History, New Series, 6, in press.
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109
Hannibal’s invasion route
1109
Mahaney, W. C., and Tricart, P., 2008, Hannibal’s debacle in the Combe de Queyras in 218 bc: the unknown Gallic
commander, in Military geography and geology: history and technology (eds. C. P. Nathanial, R. J. Abrahart and R.
P. Bradshaw), 87–96, Land Quality Press, Nottingham.
Mahaney, W. C., Kalm, V., and Dirszowsky, R., 2008a, The Hannibalic invasion of Italia in 218 bc: geological/
topographic analysis of the invasion routes, in Military geography and geology: history and technology (eds. C. P.
Nathanial, R. J. Abrahart and R. P. Bradshaw), 75–86, Land Quality Press, Nottingham.
Mahaney, W. C., Miyamoto, H., Dohm, J., Baker, V., Cabrol, N., Grin, E., and Berman, D., 2007, Rock glaciers on Mars:
Earth-based clues to Mars’ recent paleoclimatic history, Journal of Planetary and Space Sciences, 55, 181–92.
Mahaney, W. C., Kapran, B., Kalm, V., Tricart, P., Carcaillet, C., Blarquez, O., Milner, M. W., Barendregt, R. W., and
Somelar, P., 2010, The Traversette (Italia) rockfall: geomorphological indicator of the Hannibalic invasion route,
Archaeometry, 52, 156–72.
Mahaney, W. C., Milner, M. W., Kapran, B., Barendregt, R. W., Tricart, P., Krinsley, D. H., Dorn, R., Trapido-Lurie, B.,
Boccia, S., Sodhi, R. N. S., Kalm, V., Beukens, R., and Carcaillet, C., 2008b, Historical archaeology of the Hannibalic
invasion of Italia: technical applications, Technical Briefs in Historical Archaeology, 3, 39–47.
Nakagawa, T., Edouard, J. L., and de Beaulieu, J. L., 2000, A scanning electron microscopy (SEM) study of sediments
from Lake Cristol, southern French Alps, with special reference to the identification of Pinus cembra and other Alpine
Pinus species based on SEM pollen morphology, Review of Palaeobotany and Palynology, 108, 1–15.
Neumann, J., 1992, Climatic conditions in the Alps in the years about the year of Hannibal’s crossing (218 bc), Climatic
Change, 22, 139–50.
Ortu, E., Brewer, S., and Peyron, O., 2006. Pollen-inferred paleoclimate reconstructions in mountain areas: problems and
perspectives, Journal of Quaternary Science, 21, 615–27.
Ortu, E., Peyron, O., Bordon, A., Beaulieu, J. L. de, Siniscalo, C., and Caramiello, R., 2008, Lateglacial and Holocene
climate oscillation in the south-western Alps: an attempt at quantitative reconstruction, Quaternary International,
190, 71–88.
Ostdiek, V. J., and Bord, D. J., 2008, Inquiry into physics, Thomson Brooks/Cole, Belmont, CA.
Polybius, 1979, The rise of the Roman Empire, trans. I. Scott-Kilvert, Penguin, London.
Prevas, J., 1998, Hannibal crosses the Alps, Sarpedon, Rockville Center, NY.
Proctor, D., 1971, Hannibal’s march in history, Oxford University Press, Oxford.
Renaud, J. P., 1996, L’itineraire transalpine d’Hannibal. La fin d’une enigma, Archeologia. Prehistoire et Archeologie—
Dijon, 324, 48–55.
Schonstedt Instrument Co., 2001, Instruction manual, Model GA-52Cx, Kearneysville, WV.
Seibert, J., 1993, Hannibal, 75–134, Wissenschaftliche Buchgesellschaft, Darmstadt.
Tricart, P., 1984, From passive margin to continental collision: a tectonic scenario for the Western Alps, American Journal
of Science, 284, 97–120.
Tricart, P., Lardeaux, J.-M., Schwartz, S., and Sue, C., 2006, The late extension in the inner Western Alps: a synthesis
along the south-Pelvoux transect, Bulletin de la Société Géologique de France, 177, 299–310.
von Grafenstein, U., Erlenkeuser, H., Muller, J., Jouzel, J., and Johnsen, S., 1998, The cold event 8200 years ago
documented in oxygen isotope records of precipitation in Europe and Greenland, Climate Dynamics, 14, 73–81.
Walbank, F. W., 1956, Some reflections on Hannibal’s Pass, Journal of Roman Studies, 46, 37–45.
Walbank, F. W., 1990, Polybius, University of California Press, Berkeley, CA.
Wilkinson, H. S., 1911, Hannibal’s march through the Alps, Clarendon Press, Oxford.
© University of Oxford, 2010, Archaeometry 52, 6 (2010) 1096–1109