The western Alps · autumn 218 BC
Which pass did the Carthaginian general Hannibal take when he led his army and its thirty-seven elephants over the Alps into Italy in the autumn of 218 BC?
What happened. Carthage, a city on the coast of what is now Tunisia, fought Rome three times for control of the western Mediterranean, and in the second of those wars its general Hannibal marched an army from Spain to Italy overland. In the autumn of 218 BC he took forty-six thousand men and thirty-seven elephants from the river Rhône over the Alps in fifteen days: the first snow was falling when the army camped for two days on the summit, from which Hannibal pointed his men to the plains of the Po, the flat land of northern Italy, and on the descent, where the previous winter's snow still lay under the new fall, a broken cliff path had to be built up before the animals could pass. The elephants came over, in Polybius's words, "in a wretched state from hunger", and three days later the army was on flat ground among the Taurini, the people around modern Turin. Two ancient historians tell the story: Polybius, a Greek who wrote within about seventy years of the crossing, questioned men who had been on it and walked the Alps himself, and Livy, a Roman who wrote two centuries later and drew, through a Roman annalist, on the same Greek account of the march.
Why it is a mystery. Neither historian names the pass, and the two readings that would fix it are damaged in the manuscripts: the Island, the fertile triangle of land between the Rhône and a second river that the army reached four days' march up the Rhône from its crossing, has the Isère, which joins the Rhône at Valence, as its second river only by a conjecture that the geographer Cluverius made in 1624, and Livy's "1,000 feet" for the height of the broken ledge on the descent is a Renaissance restoration of a corrupt line. The argument has run in print since Josias Simler raised it in 1574 and no candidate has a majority; the candidate passes lie along the whole frontier between France and Italy (each is a col, a gap in the chain low enough for a track to cross), and the sides divide first on the river: if the Island's river is the Isère, the army went up that valley and crossed into the Susa valley west of Turin by the Col du Clapier or by the Mont-Cenis beside it, and the Clapier has been the hedged answer of historians of the ancient world writing in English since Colonel Perrin proposed it in 1883 after seeing the view of the Po plain from it. If the Island lay further down the Rhône at Orange, as Livy's list of tribes implies, the army went east up the river Durance to the Montgenèvre, the easiest road into the Susa valley, to the Col de Larche further south, or to the Col de la Traversette, a high notch under Monte Viso at the head of the Po; the Traversette is the highest and hardest of the candidates and has the most recent support: the books of Gavin de Beer, the earth-science papers of a team led by William Mahaney (2007 to 2026), which cored a boggy meadow below the col and reported a layer churned by animals that it dated to 218 BC and identified as the passage of Hannibal's army, and a 2026 energetics study by Berti and Vollrath, a model of the energy cost of each route. The two St Bernard passes to the north, which come down into the Aosta valley, carry an old Roman tradition that Livy himself rejected.
How it was judged. Both texts were read in Greek and Latin with their manuscript variants; twelve routes over eight cols were measured on a terrain model with a 90 m grid against Polybius's day-counts and stades (the Greek unit of distance, about 180 m); each descent was profiled for ground steep and broken enough to match his cliff path; what each col can see of the Po plain was computed by casting sight-lines over the terrain; old snow, the previous winter's snow on the descent, was judged against the modern snowline and the glacier record, the dated advances of the Alpine glaciers; the setting of the Pleiades, the star cluster that Polybius says was going down at dawn during the summit halt, was computed for 218 BC; and every published radiocarbon date from the Traversette meadow was recalibrated. Fourteen lines of evidence were then given likelihood ranges for eight passes (for each line and each pass, an estimate of how likely that evidence would be if that pass were the one), argued item by item, and combined, with the nine items that come from Polybius at half strength because they share one author. The Clapier comes first at about 40 per cent, the Traversette second at about 30 and the Mont-Cenis third at about 15; the Traversette's physical evidence is consistent with any large body of animals at the meadow between about 380 and 150 BC and does not test the crossing. The ranking depends on the reading of the Island, since with the Island at Orange the Traversette leads at about 54, and a manuscript giving the Island's river another name, or a datable object from the army at any col, would move the whole ranking.
On the weights used here the Col du Clapier comes first at about 40 per cent, on condition that Polybius's Island lies between the Rhône and the Isère; the Col de la Traversette is second at about 30 and the Mont-Cenis third at about 15.
Every share is a judgement: fourteen likelihood ranges, argued item by item and combined in the evidence model. The order turns on where Polybius's Island lies, the Clapier leading on the Isère reading and the Traversette with the Island at Orange. Walbank, the principal modern commentator on Polybius, and de Beer, whose books make the principal modern case for the Traversette, were read only in part.
The Montgenèvre, the Larche and the two St Bernards share about 6 per cent between them, and about 9 is left for a col nobody has argued for. The ranking hangs on one reading: that Polybius's Island is the land between the Rhône and the Isère, so that the army marched up the Isère and met the Allobroges, the Gallic tribe that ambushed it on the first climb, at the entrance to the Maurienne, the valley of the Arc. Grant that, and the Clapier is the only col off the Isère with a view of the Po plain, a steep broken descent and a short march to the Taurini. The strongest case against it is de Beer's: put the Island at Orange and send the army up the Durance, as Livy's itinerary implies, and the Traversette leads at about 54 per cent. The manuscripts do not decide between the readings: none of them names the Isère. Three things answer de Beer. Polybius gives 1,400 stades from the Rhône crossing to the foot of the ascent, of which the last 800 are the ten days along the river from the Island; the 600 that remain for the four days to the Island fit only an Island at the Isère, though his day-rates can be turned the other way. Pliny, a Roman writer of the first century AD whose geography lists the peoples of the region, puts the Allobroges beyond the Isère, which matches Polybius's distinction between the Island's people and the Allobroges. Livy's itinerary by tribe, coherent only from Orange, uses names that do not occur in Polybius and has been treated by the Clapier's advocates as an intrusion into Livy from the annalists, the earlier Roman historians whose year-by-year accounts he also used. No object from the army has been found at any col, and the churned layer in the meadow below the Traversette is the only physical evidence claimed for any candidate.
The headline numbers, which the verdict rounds, come from the run of the evidence model in which the nine items from Polybius enter at half strength, because they share one source, and everything else at full strength: Clapier 40% (5 to 95 per cent Monte-Carlo band 20 to 60), Traversette 29% (13 to 48), Mont-Cenis 16% (6 to 31), other col 9%, Larche 3%, Petit St Bernard 1.5%, Montgenèvre 1.4%, Grand St Bernard 0.3%. With every item at full strength the Clapier is 70%; with each source family (the five groups of items by source: Polybius, Livy, the other ancient writers, the meadow and practicability) collapsed to one item the table is nearly flat, 15 to 18% for the top four, and the Larche leads at 18 because it is an easy pass and Livy's Durance suits it.
Dropping items one at a time shows what carries the ranking. Remove the practicability judgement, the estimate of how passable each col is for an army with elephants, and the Traversette leads, 45 to 32; remove the Allobrogian ambush and it leads 40 to 31; remove the Island, or the ten days along the river, and the two are level at about 35 each; remove Livy's Druentia, his name for the Durance, and the Clapier rises to 49. If Livy's itinerary by the Tricastini, a tribe on the Rhône south of the Isère, and the Durance is trusted in full, the Traversette leads at 46; if Mahaney's deposit is credited as its authors read it, at 55; if the Traversette is granted to be as practicable as any other col, at 51. If Polybius's summit view is a literary set piece, the Clapier still leads, but the Mont-Cenis rises to 24 per cent. The Clapier's odds over the Traversette are about seven to five.
From the headline run of the evidence model, with the 5 to 95 per cent Monte-Carlo band in brackets. When every range is sampled 20,000 times, the Clapier comes first in 65 draws out of 100 and the Traversette in 29. With every item at full strength the Clapier is 70%; with each source family collapsed to one item the top four sit at 15 to 18% each.
Two ancient narratives survive: Polybius, a Greek historian who wrote within about seventy years of the crossing and walked the Alps himself, in book 3 and in book 34, chapter 10, and Livy, a Roman who wrote two centuries later, in book 21. Both were read in Greek and Latin with their manuscript variants, though Livy's text came from Foster's edition of 1929 in the Loeb Classical Library as transcribed on Wikisource, with the Oxford edition's record of manuscript readings from scans. Several modern books were known only from abstracts, snippets or catalogue records: Walbank's commentary on Polybius (1957), de Beer's two Traversette books (1955 and 1967), Prevas (1998), who argues for the Traversette, Proctor (1971), Lancel (1995) and Hunt (2017), who follow Walbank to the Clapier, and Mahaney's Hannibal's Odyssey (2008).
Twelve routes over eight cols (a col is a gap in the chain low enough for a track to cross) were drawn as least-cost paths, the routes a computer finds easiest over the terrain, along the valley floors on the Copernicus GLO-90 elevation model, a terrain map with a 90 m grid, measured in stades (the Greek unit, about 180 m) and tested against Polybius's day-counts. Each descent was profiled for ground steep and broken enough to match his cliff path. What each col can see of the Po plain was computed by casting sight-lines over the terrain from the col, from the first 10 km of its descent and from the ground within 1.5 km of it.
Old snow on the descent, the previous winter's snow, was judged against the modern snowline and the glacier record; the setting of the Pleiades, a star cluster, was computed for 218 BC; and every published radiocarbon date from the Traversette meadow was recalibrated. Fourteen lines of evidence were given likelihood ranges for eight passes (how likely each item would be if that pass were the one) and combined in an evidence model whose headline run counts the nine items from Polybius at half strength because they share one author; the model was rerun with each item dropped in turn, under seven sceptical scenarios, and with every range sampled 20,000 times.
Figure 1 maps the candidate cols and the least-cost routes to them over a shaded-relief map of the Alps, from the two possible sites of Polybius's Island, the land between the Rhône and a second river, at the Isère and at Orange. Figure 2 gives the shares in the headline run: the Clapier at 40 per cent, the Traversette at 29 and the Mont-Cenis at 16, with the 5 to 95 per cent band. Figure 3 shows what each col sees of the Po plain: the Traversette across a 63-degree sector, the Clapier a two-degree sliver 80 km away, and the other cols none.
In the headline run the Col du Clapier comes first at 40 per cent, the Col de la Traversette second at 29 and the Mont-Cenis third at 16, figures the verdict rounds to about 40, 30 and 15. The ranking hangs on one reading, that the Island's second river is the Isère; with the Island at Orange the Traversette leads at about 54. The Traversette's physical evidence, the churned deposit in its meadow reported by Mahaney's earth-science team, is consistent with any large body of animals there between about 380 and 150 BC and does not test the crossing. Every share is a judgement, conditional on the items chosen and the ranges set.
Already established by others. That the Isère reading of the Island, the Allobrogian ambush at the mouth of the Maurienne and the Clapier's summit plateau fit Polybius best was argued by Perrin (1883) and by Azan (1902) and Colin (1904), who adopted the col after him, and was the conclusion, always hedged, of Walbank (1956 and 1957) and of the historians who followed him, Proctor (1971), Lazenby (1978), Lancel (1995), Hunt (2017) and Hornblower (2024), and of the reply that Kuhle and Kuhle made to Mahaney in 2012. That the Traversette fits the view, the snow and the steep descent was argued by its advocates Torr (1924), de Beer (1955) and Prevas (1998). That the Grand and Petit St Bernard fail on the Taurini is Livy's own argument. That the meadow's radiocarbon window, the span of years its dates allow, admits several armies was said by de Galbert (2016), a critic of the Traversette dating, and that the Traversette has no ground for a summit camp by Hunt (in EARTH magazine, 2016).
What historians have said. There is no consensus. A count by Peddie in 1997 of twenty-four modern writers split six Mont-Cenis, six Montgenèvre, five Clapier, four Petit St Bernard, three Traversette. Before and after that count, the Clapier has been the usual hedged answer of historians of the ancient world (Walbank "hesitatingly" in 1956, Lazenby 1978, Lancel 1995, Hunt 2017, Hornblower "probably" in 2024), the Montgenèvre the scholarly alternative, argued by Connolly (1981) and by Hoyos (2006 and 2015), who follows Livy's tribes to the Durance, and the Traversette the choice of de Beer and those who followed him (1955 to 1974), of Prevas (1998), who has crossed the col many times, of Mahaney's papers (2007 to 2026), which presented the meadow deposit as "the first direct evidence of the route", with "solid evidence" in the press coverage of 2016, and of Berti and Vollrath (in the journal PNAS, 2026), whose energetics model does not engage with the deposit. Three recent writers on Hannibal, Goldsworthy (2000), MacDonald (2015) and Lendering (2022), decline to choose, and the two most recent peer-reviewed contributions, Hornblower and Berti and Vollrath, point in opposite directions.
No earlier finding is overturned. The result agrees with the Clapier reading at the cautious level its advocates gave it, and agrees with them that the question remains undecided: weakening the Traversette's dating does not establish the Clapier, whose share rests on the reading of the Island. It moves the Traversette from "the first direct evidence of the route" (Mahaney and colleagues in the journal Archaeometry, 2017, part II; "solid evidence" in the press coverage of 2016) to a strong second, because the meadow's radiocarbon window admits any army from about 380 to 150 BC and nothing in the output of the published age model places the churned layer in 218 BC.