# Hannibal's Alpine pass (218 BC): scholarship and testimonia dossier

Scope: (1) who has argued for which pass, with what argument, and where each puts Polybius' "Island"; (2) the ancient testimonia outside Polybius 3 and Livy 21; (3) the case for and against each candidate pass; (4) the state of the debate 2010-2026 with "nothing has happened since" checks; (5) the access log. Raw texts are under `texts/scholarship/` and `texts/testimonia/`.

**Status labels used throughout.** FULL = complete text obtained and the cited pages read. PART = complete OCR obtained, only the cited chapters read (or a substantial extract). SNIP = Open Library search-inside snippets only (no page numbers unless a running head or index entry appears inside the snippet). ABS = Crossref/publisher abstract or metadata only. SEC = the author's verdict is known only from a named secondary source. NONE = not obtained. Anything from memory rather than a fetched source is marked "(unverified)".

**Headline.** No ancient text names Hannibal's pass. The modern literature has never converged: the Isère–Arc–Col du Clapier family (Perrin 1883/87, Azan, Colin, Wilkinson, Walbank "hesitatingly", Proctor, Lazenby, Huss, Lancel, Hunt, Kuhle & Kuhle, Hornblower 2024) is the most-favoured single answer among ancient historians since 1956; the Montgenèvre has a persistent minority (Hennebert, Caspari, Bullock Hall, Connolly 1981, Hoyos 2006/2015, W. T. Arnold's "best Durance pass"); the Col de la Traversette is the de Beer line (Torr 1924, de Beer 1955-69, Bagnall, Prevas 1998/2017, Mahaney 2007-2026, Seibert reportedly, Berti & Vollrath 2026); the Petit-Saint-Bernard survives among French regional scholars (Bocquet, Coninck) and in older German and English work (Mommsen, Lenschau, Law, Hyde); the Mont Cenis (Napoleon, Larauza, Maissiat, Osiander, Jullian) and the Col de Larche/Argentière (Freshfield) have lost their advocates; and a substantial group of recent syntheses declines to decide (Goldsworthy 2000, MacDonald 2015, Lendering 2022; Lancel 1995 calls the whole exercise "plus chimérique que de prétendre trouver le vrai site d'Alésia"). The label "resolved" is not available for any pass on the basis of the scholarship.

---

## 1. Summary table: advocates, passes, arguments, the Island

Chronological. "Island" = Polybius' Nesos (3.49.5) at the confluence of the Rhône and the Skaras/Skoras (MS readings; Livy's manuscripts give Arar / Ibisara / Bisarar, printed as "Isara" since Schweighäuser and Lemaire: Azan 1902 pp. 31-32).

| # | Advocate (work) | Pass | Rhône crossing / Island | Principal argument | Principal objection (by whom) | Status; where the verdict was seen |
|---|---|---|---|---|---|---|
| 1 | J. Whitaker, *The Course of Hannibal over the Alps ascertained*, 2 vols (1794) | Great St Bernard (Poeninus) | not extracted | the "Poeninae = Punic" tradition (Livy 21.38.6-7, Pliny 3.123) | Livy 21.38.6-9: the Poeninus leads to the Libui/Salassi, not the Taurini; Servius ad Aen. 10.13: named "a Poenina dea" | PART: OCR of both vols; verdict confirmed from Proctor 1971 snippet ("crossed the Alps into Italy by the Great St. Bernard. John [Whitaker]") and Torr 1924 p. v |
| 2 | J.-A. de Luc, *Histoire du passage des Alpes par Annibal* (1818) | Petit Saint-Bernard ("l'Alpe grecque, ou du Petit Saint-Bernard", OCR l. 11613) | Roquemaure; Island = Rhône–Isère; Mont du Chat, Lémenc (de Beer 1956 table snippet: "J. A. de Luc, 1818. 1 Roquemaure, 2 Isère, 3 Mont du Chat, 4 Lémenc…") | Livy's tribal itinerary and Coelius' Cremonis iugum read as the Little St Bernard | no view of Italy; 75 English miles from the col to the plain (Torr para 55, p. 30) | PART |
| 3 | H. L. Wickham & J. A. Cramer, *Dissertation on the Passage of Hannibal over the Alps* (1820; 2nd ed. 1828) | Little St Bernard (OCR l. 514: "[from] Séez, crossed the Little St. Bernard") | Rhône–Isère (unverified) | as de Luc, with fuller use of Polybius' distances | as above | PART (both editions' OCR) |
| 4 | J.-L. Larauza, *Histoire critique du passage des Alpes par Annibal* (1826) | Mont Cenis (OCR l. 7529: "Annibal a passé le mont Cenis"), written against de Luc | Rhône–Isère (Azan 1902 pp. 45-46 lists him under the Mont Cenis system) | Napoleon's judgement; the shortest road to Turin | no view from the Mont Cenis plateau (Azan pp. 58-61; Wilkinson 1911 p. 8 n. 1) | PART |
| 5 | G. Long, *The March of Hannibal from the Rhone to the Alps* (1831) | not extracted from OCR | — | — | — | PART (short tract; verdict not extracted) |
| 6 | R. Ellis, *A Treatise on Hannibal's Passage of the Alps* (1853); *Ancient Routes between Italy and Gaul* (1867) | Little Mont Cenis (Azan 1902 ch. IX n. 2: "Ellis adopte un itinéraire passant par le Petit Mont Cenis"; Law 1866 vol. 2 OCR l. 8521: "Mr. Ellis, who sends Hannibal over the Little Mont Cenis") | Rhône–Isère (unverified) | a col with a view near the Little Mont Cenis; the Polybian distances | Law 1866 (2 vols) is a book-length rebuttal | PART (1853 OCR) |
| 7 | W. J. Law, *The Alps of Hannibal*, 2 vols (1866) | Little St Bernard (vol. 2 OCR l. 232: "in favour of the Little St. Bernard") | Rhône–Isère (unverified) | Coelius' Cremonis iugum (Livy 21.38.7) = Cramont/Little St Bernard; Livy's Graian Alps; Polybius' Allobroges | Torr paras 55, 57 (distance; no view); Wilkinson p. 38 | PART |
| 8 | J. Maissiat, *Annibal en Gaule* (1874) | Mont Cenis (OCR l. 10475: "non-seulement qu'Annibal a passé au mont Cenis") | reads Araros (Saône) in Polybius (Azan p. 32) | — | — | PART |
| 9 | Col. J.-B. Perrin, *Marche d'Annibal des Pyrénées au Pô* (1883); *Étude sur Annibal* (1887) | **Col du Clapier** (first proposer) | Rhône–Isère; right bank of the Isère to Saint-Rambert and Saint-Genix-d'Aoste, Chambéry, the Maurienne (Colin 1904 p. 320) | the panorama of the Po plain from the col matches Pol. 3.54.2-3 (Perrin p. 74, quoted by Azan p. 93: "réunit toutes ces conditions, c'est le col du Clapier") | Freshfield 1914 preface: "a lofty and difficult by-pass near the Mont Cenis"; Connolly 1981 p. 165 (no view from the col itself; iced north face) | NONE; SEC: Azan 1902 p. 93 ("En 1887, M. le colonel Perrin publia un système nouveau… c'est lui du moins à qui revient tout le mérite d'avoir découvert le col"), Colin 1904 pp. xv-xvii ("c'est seulement vers 1880 que le colonel Perrin a signalé le col Clapier. C'est le pas le plus considérable… depuis deux mille ans"), Wilkinson 1911 p. 8 ("In 1884 Colonel Perrin published his account of the Col du Clapier"); titles from Bradford's and Prevas's bibliographies (snippets) |
| 10 | Col. E. Hennebert, *Histoire d'Annibal* (1870-91) | Mont Genèvre (Wilkinson 1911 p. 38: "a Hennebert the Mont Genevre") | — | a priori military-geographical reconstruction (claimed to have read "plus de cent cinquante mémoires", Azan p. 45) | Azan 1902 pp. 62-82: "il a eu le grand tort de ne pas aller dans les Alpes" (p. 81) | SEC |
| 11 | W. Osiander, *Der Hannibalweg* (1900) | Mont Cenis (OCR ll. 8578-8625: "ist der Genèvre der Pompeiuspass, so ist der Cenis der Hannibal[pass]… die Cenisroute diejenige Hannibals gewesen ist"; section "Kleiner Bernhard oder Montcenis?") | Rhône–Isère; "Skaras oder Skoras… ist der älteste nachweisbare Name der Isère" (Osiander p. 28 n. 7, quoted by Azan p. 32 n. 6) | Varro's list: Genèvre = Pompey, so Cenis = Hannibal | Azan pp. 58-61 (view; "Turina") | PART |
| 12 | P. Azan, *Annibal dans les Alpes* (1902) | **Col du Clapier** (p. 93: "Le col par lequel Annibal a passé les Alpes est à notre avis le col du Clapier"; adopts Perrin's col but not his itinerary, p. 100 n. 1) | Island = Rhône–Isère "provisoirement" (pp. 32-33: 204 km from the sea in eight marches; Delta shape between Rhône, Isère and the Chambéry mountains); "l'itinéraire d'Annibal doit longer l'Isère" then the Arc (p. 100); first battle at the Grand Cucheron (ch. XII) | statistics (Delacroix 1835 via Bouché-Leclercq 1874, p. 46): Grand St-Bernard 19, Petit St-Bernard 33, Mont Cenis 11, Mont Genèvre 24, Mont Viso 3 | Azan's part II ("Une conjecture": the Rhône once ran through the Chambéry valley) not followed by anyone I found | FULL OCR; chs. I-XI read |
| 13 | Capt. J. Colin, *Annibal en Gaule* (1904) | **Col du Clapier** (pp. xiv-xviii; 318-321), descending by the granges de Tuglia (Quatre-Dents) along the Ambin glacier rather than Perrin's path (p. xvii) | Rhône crossed near Fourques/Arles (p. 321); Island = the Vaucluse plain between Rhône, Durance and Sorgues, the name surviving at l'Isle-sur-Sorgues (pp. 344-347); then up the Rhône and the Isère, the anabolē at the bec de l'Échaillon (pp. 318-321) | Polybius' three distances (1600 Emporion–Rhône, 1400 to the ascent, 1200 across) fit only three itineraries, and only the Clapier also satisfies the view (p. xiv-xv: "ce col Clapier, qui a le monopole exclusif de répondre à toutes les conditions voulues") | Freshfield 1914; Torr 1924 (Isère route "far too much") | PART (avant-propos and part II ch. IV read) |
| 14 | E. Hesselmeyer, *Hannibals Alpenübergang* (1906) | not determined from OCR | — | — | — | PART |
| 15 | C. Jullian, *Histoire de la Gaule* I (1908), p. 479 | Mont Cenis via the Maurienne ("Hannibal, sans doute sur le conseil des Boïens, avait choisi cette dernière [la route de la Maurienne et du mont Cenis]… plus courte… menait sans détour au carrefour de Turin") | Rhône–Isère to the end of Allobrogian territory; first ambush at the bec d'Aiton (fr.wiki n. 38) | shortness; Boian advice | — | PART |
| 16 | S. Wilkinson, *Hannibal's March through the Alps* (1911) | **Col du Clapier** (pp. 8-9: "There is no other pass between the Lake of Geneva and the sea from which it would be possible for an army… to see the plain of Italy… The nature of the view from the Col du Clapier is the crucial evidence"; p. 48) | crossing at Fourques after Colin (p. 15); Island = Rhône–Sorgues, "Isle-sur-Sorgues" (p. 19); then Rhône and Isère to St-Nazaire, Grenoble, the Arc (pp. 24, 48) | the view; rejects the assumption that Hannibal used one of the four modern high-road passes (pp. 45-46: "more than thirty passes crossed by regular mule-tracks") | Freshfield 1914; Connolly 1981 p. 165 | FULL |
| 17 | D. W. Freshfield, articles 1886/1899 (unverified dates; Mahaney 2026 cites both) and *Hannibal Once More* (1914) | **Col de l'Argentière (Larche)** via Isère–Drac–Durance–Ubaye ("the Vars-Argentière route", ch. III pp. 31-57) | Island = Rhône–Isère (p. 31: "the generally received interpretation") | Varro's five passes taken in geographical order from the sea: coast, Argentière (Hannibal), Mont Genèvre (Pompey), Mont Cenis (Hasdrubal), Little St Bernard (Graian) (pp. 20-21); Pompey's "new road" shows Hannibal's pass ≠ Mont Genèvre (pp. 11-12); the 12 Cottian tribes absent from the Tropaeum (p. 12) | Wilkinson pp. 45-46; Torr para 69 (no view from the Argentière); W. T. Arnold's letter (Freshfield p. 12 n. 3): "the evidence converges on a Durance pass. The best Durance pass is certainly the Mont Genevre. But Varro and Sallust point rather to the Argentière" | FULL |
| 18 | C. Torr, *Hannibal Crosses the Alps* (1924; 2nd ed. 1925 not obtained) | **Col de la Traversette** (paras 68-71, pp. 37-38: Durance to Mont Dauphin, up the Guil, gorge below Château-Queyras, summit, landslip, Crissolo, Saluzzo) | Rhône crossed at Tarascon (paras 5, 71); Skoras = Durance (paras 31, 65); Island at the Rhône–Durance confluence | Polybius' 100 miles "along the river" = Strabo's 99-mile Roman road up the Durance to Embrun (para 29); Varro and Strabo make Hannibal's pass the next after the coast road (para 24); Pompey's = Mont Genèvre from Appian BC 1.109 and Strabo 4.6.5 (paras 25-26); the view only from the Traversette among southern passes (para 69); shortness with autumn advancing (para 70); Cremonis iugum possibly the Dora Riparia pass misread by Livy (para 57); Nepos' Graian Alps ambiguous (para 58) | Freshfield p. 29 ("no sound argument in favour of any of them is as yet forthcoming"); Connolly p. 164 (Combe de Queyras: "a maniac"); Lancel (narrow breach; elephants) | FULL (1924 ed.) |
| 19 | W. W. Hyde, *Roman Alpine Routes* (1935) | Petit Saint-Bernard (fr.wiki: "l'Américain W.W. Hyde" among its advocates) | — | — | — | NONE; SEC (Wikipedia finding aid; unverified) |
| 20 | Napoleon (Montholon, *Mémoires*, 1823) | Mont Cenis | Roquemaure (de Beer 1956 table snippet: "Napoleon, 1816. 1 Roquemaure, 6 Mont Cenis") | military judgement | Wilkinson p. 8 n. 1 (no view) | SEC: Proctor 1971 snippet ("regarded as the author of the Mont Cenis route is Napoleon. In 1823 General de Montholon…"), Lazenby 1998 n. 34, Peddie 1997, Fields 2010. The 1836 *Précis des guerres de César* obtained here does not contain the Hannibal discussion |
| 21 | Th. Mommsen (RG; CIL V p. 809) | Little St Bernard; Varro's 2nd pass = Great St Bernard, 4th = "Col de Monte Viso" or "by Val Chisone" (Freshfield p. 21 with n. 1) | first ambush at the Col du Chat (fr.wiki n. 37) | opulence of Allobrogian lands (fr.wiki) | Freshfield pp. 20-21 (must impute Livy's "vulgar error" to Varro) | SEC |
| 22 | Th. Arnold, *History of Rome* iii | Little St Bernard (Wilkinson p. 38, quoting Arnold: Polybius "so unscientific… persons who have gone over the several Alpine passes… can still reasonably doubt whether they were meant to apply to Mont Genevre or Mont Cenis or to the Little St. Bernard") | — | — | — | SEC (OL snippets of Arnold obtained, not read for this) |
| 23 | Caspari (Enc. Brit. 11th ed., "Hannibal"); Lenschau (RE, "Hannibal") | Mont Genèvre (Caspari); Little St Bernard, "probability decidedly favours" (Lenschau) | — | — | — | SEC: H. L. Jones, Loeb Strabo II (1923) p. 293 n. 5 |
| 24 | F. W. Walbank, "Some Reflections on Hannibal's Pass", JRS 46 (1956) 37-45 (= Selected Papers 1985, 107-19); *Historical Commentary on Polybius* I (1957) on 3.47-56 (pp. 382-90: page numbers from Lazenby's and MacDonald's notes, "Walbank I, 382, on Polybius 3.48.12", "Walbank I, 384", "vol. 1, 390") | Clapier, hesitantly. Walbank's own retrospect (2002 essays, snippet): "hesitatingly came down in favour of the Col du Clapier (which was also Huss' choice)". Proctor 1971 (snippet): Walbank "brackets [the Mont Cenis] with the Col du Clapier without declaring a preference". Connolly p. 165: Walbank "claims that the route to the Taurini can only be the Mont Cenis, the Clapier or the Montgenèvre" | Rhône–Isère (Walbank 2002 snippet: "when Hannibal leaves the so-called 'Island' to march up the Isère along the foot of the Alps, Polybius…") | rejected de Beer's Traversette in 1956 (so en.wiki "Col de la Traversette" citing JRS 46; en.wiki "Little St Bernard" cites pp. 44-45); Hunt 2017 n.: "Walbank says that too much ink has been [spilled]…" | — | NONE for JRS (JSTOR client challenge; DOI 10.2307/297963) and Comm. I (archive.org historicalcommen0001fwwa restricted, HTTP 401); SNIP for Walbank 2002 |
| 25 | Sir G. de Beer, *Alps and Elephants* (1955; US 1956), *Hannibal's March* (1967), *Hannibal: Challenging Rome's Supremacy* (1969), *Hannibal: the Struggle for Power* (1974) | **Col de la Traversette** ("strongest competitor", 1956 snippet; 1969 snippet: "That leaves the Col de la Traversette (2,950 metres), accessible through the valley [of the Guil]") | Skaras = Aygues (medieval forms Equeris 1278, Icaris 1321, Yquarum 1393, Yguaris 1414 per fr.wiki); Island = Rhône–Aygues at Orange (1969 pp. 132-5 per Lancel's note; 1956 snippets: Casaubon "calmly altered Skaras into Arar"); Drôme, Col de Grimone, Durance, Guil | (a) Varro's list plus Pompey's Mont Genèvre put Hannibal's pass south of the Genèvre (1969 snippet: "the pass which Pompey crossed, the Mont-Genèvre, was not the pass that Hannibal crossed… leave only the Ubaye and the [Guil]"); (b) Mont Genèvre "too low and too easy"; (c) Traversette alone has the view, old snow, the tree-line; (d) Cremonis iugum = Col de Grimone (Lazenby n. 39: "de Beer's neat identification"); (e) Col de Larche "probably the pass which Hannibal intended to cross" (1956 snippet) | Walbank 1956; Proctor 1971 ("De Beer thinks Hannibal crossed further downstream…"); Lazenby ("almost certainly wrong to locate Hannibal's pass between the Montgenevre and the coast"); Connolly p. 164; Lancel | SNIP (1956; 1969 x2; 1974 x1); archive.org lending copies restricted (alpselephantshan0000debe, hannibalchalleng0000debe: 401) |
| 26 | M.-A. de Lavis-Trafford (1956; *Mémorial* 1962) | Col de Savine-Coche, the Clapier's twin | — | névé in shade, screes, promontory with a view of Turin, lake basin for a camp (fr.wiki §Savine-Coche; Lancel snippet: "a depression which was a twin to that of the Clapier… Savine-Coche"; E. Meyer 1964 pp. 90-101 "abandoned the Clapier for Savine-Coche") | — | SEC (Blache 1963, RGA 51: 171-2, first page on Persée; Lancel; fr.wiki) |
| 27 | J. Hoyte, *Trunk Road for Hannibal: With an Elephant over the Alps* (1960) | Col du Clapier, tested in 1959 with the elephant Jumbo (British Alpine Hannibal Expedition); the party crossed by the Mont Cenis when the Clapier proved impassable for the animal | — | practical feasibility | — | NONE (archive.org trunkroadforhann0000john restricted); SEC: MacDonald 2015 p. 94 ("to take an elephant named Jumbo over the Col du Clapier to prove the feasibility of that particular route"); de.wiki ("überquerten mit einem Elefanten namens Jumbo den Pass am Mont Cenis") |
| 28 | D. Proctor, *Hannibal's March in History* (1971) | **Col du Clapier** (pp. 210-16 per Hornblower 2024 n. 54; snippet: the Mont Cenis "or its variant, the Col du Clapier… the most popular routes"; distances "215 kilometres by the Mont Cenis, 200 by the Col du Clapier, 265 by the long route over [the Little St Bernard]", "190 by the short middle route over [the Montgenèvre]") | Rhône–Isère (Mahaney 2026 p. 3: "the Isère matched against the Rhône, as some (Proctor, 1971) have asserted") | chronology (late departure, snow), Polybius' distances, the view | Raepsaet-Charlier, AC 41 (1972) 394-5 (first page only) | SNIP (140 snippets); archive.org hannibalsmarchin0000proc restricted (403) |
| 29 | J. F. Lazenby, *Hannibal's War* (1978; 2nd ed. 1998) | **Col du Clapier** (snippet: "fulfil all the requirements — the Col de Clapier. This is 2482 metres high, has a wonderful [view]…"; pp. 45-6 with plate IV per Hornblower; pp. 34-48, 275-7 per Goldsworthy; pp. 42, 47 per fr.wiki) | Rhône–Isère ("Hannibal here left the Rhone to march on up the Isère"); first battle at the gorge above Pontcharra, "approximately 120 kilometres" from the Clapier | distances; the view; old snow only on a high pass ("true of the high passes like the Col de Clapier"); on Coelius, "right to identify… 'the pass of Cremo' (Cremonis iugum) with a pass leading into the territory [of the Libui]" (snippet cut) | Connolly p. 165 ("Those who have argued that Hannibal crossed the Clapier must explain why he did not use the Mont Cenis which is both lower and easier") | SNIP (154 snippets); archive.org hannibalswarmili00laze restricted (401) |
| 30 | P. Connolly, *Greece and Rome at War* (1981), pp. 164-166 | **Col du Montgenèvre** ("the pass was almost certainly the Montgenèvre", p. 165), reached from the Isère at Grenoble by the Col Bayard (1,248 m) or Col de Manse (1,263 m) and the Durance | Island = Rhône–Isère ("one obvious candidate", p. 164; the Eygues' old name "Ekaris… similar to Skaras. But… Isara… is just as similar") | Polybius knows only one road to the Taurini, which must be the road of Hercules = Montgenèvre; seven passes tested against five conditions (pp. 164-5); 1979 Royal Engineers survey under Lt R. A. M. S. Melvin: "virtually impossible to get elephants up the Traversette because of the scree or down the other side" (p. 165); from the Clapier "from no point on the summit is it possible to see the Po valley" (p. 165) | Mahaney 2026 (the "mid N-S route"); Torr para 68 (nine days from the Durance to a summit ten miles away) | FULL (OCR of an English scan uploaded under an Italian title) |
| 31 | E. Bradford, *Hannibal* (1981) | Clapier slightly over Traversette (pp. 68-69: "the choice seems to lie between the Clapier and Traversette passes, with the odds slightly in favour of the Col du Clapier") | discusses de Beer's Skaras (pp. 58-59) | — | — | SNIP |
| 32 | E. C. Pyatt, *The Passage of the Alps* (1984) | follows de Beer ("Hannibal crossed the Aygues to the Island"; "Only the Col de Malaure and the Col Clapier are comparable [to] the Col de la Traversette") | Rhône–Aygues | — | — | SNIP |
| 33 | A. L. F. Rivet, *Gallia Narbonensis* (1988) | (a survey of the Gallic peoples; Prevas 1998 lists Rivet with "Hoyte (1960), Proctor (1971), Lazenby (1978)… and Coninck (1992)" as Clapier/Mont Cenis authors — snippet) | — | — | — | SNIP; archive.org gallianarbonensi0000rive restricted (401) |
| 34 | N. Bagnall, *The Punic Wars* (1990) | Traversette ("convincingly identified by Gavin de Beer… as being the Col de la Traversette", pp. 49-50) | de Beer's | — | — | SNIP |
| 35 | J. Seibert, *Forschungen zu Hannibal* (1993a) pp. 191-213 (per Hoyos 2015 n. 5); *Hannibal* (1993b) pp. 106-113 (per de.wiki) | army divided over two passes (Walbank 2002 snippet: "J. Seibert has boldly suggested that Hannibal divided his army into two parts, which crossed [by different passes]"; de.wiki: "hält auch eine Aufteilung des Heerzuges auf mehrere Pässe für möglich"); Mahaney 2026 p. 3: "Seibert (1993) favored the Traversette but provided little confirming evidence outside of elevation" | — | — | — | NONE; SEC |
| 36 | S. Lancel, *Hannibal* (1995; Eng. tr. 1998, pp. 57-80) | Clapier / Savine-Coche family by the Isère and the Arc (fr.wiki: second ambush "la haute vallée de l'Arc pour Serge Lancel", Lancel 1995 pp. 121-136); with the caveat "parcourir les Alpes avec son Tite-Live et son Polybe à la main est une entreprise plus chimérique que de prétendre trouver le vrai site d'Alésia" (1995 p. 122, per fr.wiki) | Rhône–Isère (Eng. snippets: "Hannibal arrived near Cularo, the site on which Grenoble is based"; rejects de Beer's Skaras = Aygues) | rejects the Traversette ("rather than a pass, a narrow breach"; elephants — fr.wiki n. 87) and the Montgenèvre (the "Herculean Way" known to Scipio) | — | SNIP (146 snippets of the 1998 translation); Desanges's review (RH 1997) metadata only |
| 37 | J. Peddie, *Hannibal's War* (1997) | tallies "twenty-four modern-day" scholars: "[Little St Bernard] most likely; six, Mont Cenis; five, Col du Clapier; six, the Mont Genevre; and three, la Traversette" (snippet, first figure lost); own verdict not recovered | Roquemaure ("the northern passage from Roquemaure") | — | — | SNIP |
| 38 | J. Prevas, *Hannibal Crosses the Alps* (1998; 2001) and *Hannibal's Oath* (2017) | **Col de la Traversette** (1998 pp. 157-169; 2017: "The author has gone over the Traversette… well over twenty [times]"; "a remote pass like the Traversette must have been completely unknown to [the Romans]") | Drôme route, first ambush at the Col de Grimone or Col de Cabre (pp. 117, 134), then Durance and Guil; discusses the Skaras/Isère resemblance (p. 104) | year-round snow on the Traversette; Roman ignorance of the pass | Hunt 2017 p. 284 n.; Kuhle & Kuhle | SNIP (124 + 94 snippets); archive.org hannibalcrossesa0000prev restricted (401) |
| 39 | D. Hoyos, "Crossing the Durance with Hannibal and Livy: the Route to the Pass", Klio 88 (2006) 408-465; notes in Yardley & Hoyos, Livy, *Hannibal's War* (OWC 2006/2009) pp. 620-30; *Mastering the West* (2015) pp. 105-106 | **Montgenèvre**, "or one of its neighbors such as the Larche — each a fairly easy climb up from Gaul, followed by a steep descent down to Italy. Other candidates (the best known are the Col du Clapier and the Traversette, both farther north) are much more difficult to fit into what evidence we do have" (2015 p. 106); Hornblower n. 54: Hoyos "prefers the Col de Montgenèvre" (OWC pp. 626-9) | Rhône–Isère: "along the Rhone to the Isère, over to the Drac valley, and from there to the high Alps and the pass"; the Allobrogian town "may be today's Grenoble" (2015 pp. 105-6); Livy's Tricastini, Vocontii, Tricorii accepted as Silenus/Sosylus material | Livy's tribal itinerary; supply; "a zigzag route… plausible for an army aiming to elude Roman observers" | — | 2015 FULL; 2006 metadata only (no abstract on Crossref); en.wiki "Tricorii" cites Hoyos 2006 pp. 442-4, 447 |
| 40 | A. Goldsworthy, *The Punic Wars* (2000; repr. *The Fall of Carthage* 2007) | no verdict: "the nature of our sources makes it impossible to resolve these disputes. I do not intend to discuss this topic" (ch. "Hannibal's march", section "The March to Italy"); "which pass is perhaps more fiercely debated than any other aspect of his route" | — | — | — | FULL (page numbers not recoverable from this scan) |
| 41 | G. de Galbert, *Hannibal en Gaule* (Grenoble 2005) | Isère route with a short cut across Belledonne by the Pas de la Coche and the Col de la Croix-de-Fer into the Maurienne (fr.wiki, n. 58 = de Galbert pp. 140-1); first ambush at the cluse de Voreppe (n. 34 = p. 11); the final col cited at pp. 141-147 but not recovered | Rhône–Isère (implied) | mediaeval use of the short cut | Lancel: the shortest is not the best in mountains (fr.wiki) | NONE; SEC (fr.wiki reference list) |
| 42 | W. C. Mahaney (Geoarchaeology 22, 2007; Geology Today 24, 2008; *Hannibal's Odyssey*, 2008; Archaeometry 52 (2010) 156-72; 55 (2013) 1196-204; MAA 14 (2014) 77-94; MAA 16 (2016) 97-105; Archaeometry 59 (2017) 164-178 and part II; 2019; 2023; LJRHSS 2026) | **Col de la Traversette**: the two-tier rockfall on the Italian side is "the only rockfall capable of blockading Hannibal" (2026 p. 7); firn: "the Traversette (~3000 m asl) being the only pass meeting these requirements [Pol. 3.55] in Hannibal's time and ours up to about 1990" (p. 6); Clostridia/horse "facies" in the mire below the col with ¹⁴C c. 2168 BP (2017; en.wiki summary) | Rhône crossed near Arles; Island = Rhône–Aygues at Orange (2026 p. 3, after de Beer 1967); first pass = Col de Grîmone/Cabre | Polybius as "probably the first stratigrapher in history" (2026 p. 1) | Kuhle & Kuhle 2011, 2014; Hunt 2017 p. 284 n.; MacDonald 2015 n.; Lancel; Connolly's 1979 survey | 2026 FULL (open PDF, 13 pp.); 2008 Geology Today and 2010 Archaeometry ABS; the rest metadata |
| 43 | M. Kuhle & S. Kuhle, Archaeometry 54 (2012 [online 2011]) 591-601; 57 (2015 [online 2014]) 759-771 | **Col du Clapier** ("For a long time now, historians have considered the way across the Col de Clapier as the most likely invasion route"); the two-tier rockfall "which they believe is mentioned in the ancient texts… a critical reassessment of the ancient texts and a closer look at the geomorphological situation clearly show that this interpretation is unwarranted" | — | philology of Pol. 3.54-55 ("Lost in Translation") | Mahaney 2013 reply | ABS (2011 abstract on Crossref; 2014 metadata) |
| 44 | E. MacDonald, *Hannibal: A Hellenistic Life* (2015) | undecided: "opinion balanced between the Col du Clapier and Col de la Traversette" (p. 95; index "Col du Clapier 94, 95; Col de la Traversette 95") | — | cites Hoyos 2006: 634-5 n. 35, Walbank I 390, Mahaney 2010 | — | SNIP |
| 45 | P. Hunt, *Hannibal* (2017); "Hannibal Barca", Oxford Bibliographies (2026) | **Col du Clapier–Savine Coche** ("summit camp possibly at the broad Clapier-Savine Coche saddle around 8,500 feet", p. 70; 284 n.); "many possible arguments against Col de la Traversette, not even including philology, as argued by Walbank, 'Hannibal's Pass', 37-45, esp. 43-45"; "the Col de la Traversette might not even be that ancient geologically" | not recovered (an Isère–Arc route is implied — unverified) | the view; Stanford Alpine Archaeology Project field seasons 2006-7 (fr.wiki n. 63, 488) | — | SNIP (130 snippets); de.wiki ("favorisiert den Col de Clapier"); OBO 2026 metadata only |
| 46 | J. Lendering, *Hannibal in de Alpen* (2022) | no pass: the two primary sources "provide too little accurate information and too much conflicting information" (en.wiki summary) | — | — | — | NONE; SEC |
| 47 | P. Freeman, *Hannibal: Rome's Greatest Enemy* (2022) | not recovered | — | — | — | NONE (Prevas's review, The Historian 2021, DOI 10.1080/00182370.2021.2077005, metadata only) |
| 48 | J.-M. Kötter, *Hannibal* (Beck 2024) | not recovered | — | — | — | NONE (Crossref chapter metadata only) |
| 49 | S. Hornblower, *Hannibal and Scipio: Parallel Lives* (CUP 2024) | **Col du Clapier**, "probably": "The likeliest mountain pass must be one whose Italian end is closest to Turin, where he is first heard of in action after his descent. That probably indicates that the best candidate for the pass is the Col de Clapier" (p. 86, OCR page marker ±1; n. 54 cites Walbank 1985 [1956], Proctor 1971: 210-16, Lazenby 1978: 45-6, and notes Hoyos's preference for the Montgenèvre) | Rhône–Isère: "He clearly left the Rhône to follow its tributary the Isère, but there agreement ends" | proximity to the Taurini | Hoyos, HZ 2026 review (metadata only) | FULL |
| 50 | E. Berti & F. Vollrath, "Energy costs of Hannibal's alpine crossing", PNAS 123 (2026), DOI 10.1073/pnas.2612764123 | **Traversette**: "Among the possible alpine passes, the route crossing at the Col du Clapier was considered the most likely candidate. Recent philological and geomorphological analyses suggest instead a route via the Col de la Traversette… Our analysis favors the route of the Traversette" | not stated in the abstract | energy-landscape modelling of the crossing for war elephants | — | ABS |
| 51 | French and Italian work found on Persée (first pages/abstracts only) and in Hunt's notes: E. Davin, "Le passage du Rhône et des Alpes par Hannibal", Bull. Budé 1958: 65-80; R. Bailet, Bull. Budé 1960: 377-388; P. Pédech, "Sur les pas d'Hannibal", REG 79 (1966) 727-729 (review of Marquion 1965, *Sur les pas d'Hannibal*); P. Marquion, "Le passage du Rhône par Hannibal", Provence historique 18 (1968) 13-38; J. Blache, "L'ancienne voie du Cenis", RGA 51 (1963) 171-2 (on the Lavis-Trafford *Mémorial*, 1962); G. Barruol, Provence historique 13 (1963) 6-15 (Memini/Cavares); M.-Th. Raepsaet-Charlier, AC 41 (1972) 394-5 (review of Proctor); J. Dumont, REA 75 (1973) 146-8 (review of Walbank II); R. S. Fisher, "Polybius III, 46", RBPh 60 (1982) 41-50 (the elephant rafts); C. Jourdain-Annequin, DHA 25 (1999) 101-127 and *Quand les Grecs et les Romains découvraient les Alpes* (2011); P.-E. Klingbeil, "La marche d'Hannibal: ravitaillement et stratégie", Ant. afr. 36 (2000) 15-38; R. Halleux, "Sur le prétendu vinaigre", CRAI 151 (2007) 529-534; F. Artru, DHA 39 (2013) 237-263 and *Sur les routes romaines des Alpes Cottiennes. Entre Mont-Cenis et col de Larche* (ISTA 2016); C. Gauchon, "les plus anciens tunnels des Alpes", CTHS 2002: 267-282 (Traversette tunnel); Italian: Gastaldi, "Annibale e i Taurini" (pp. 116-21) and F. Landucci Gattinoni, "Annibale sulle Alpi", Aevum (both via Hunt 2017 notes, snippets); G. Granzotto, *Annibale* (1980) p. 310 (via Hunt); Brizzi, *Scipione e Annibale* (archive.org 401) | Verdicts: Marquion 1965/1968 = Rhône crossing and route to the Mont Cenis/Clapier group (Proctor's bibliography lists it; Pédech's review first page gives no verdict); Lavis-Trafford = Savine-Coche (Blache); Jourdain-Annequin = the mythic/Heraclean frame explains the ancients' uncertainty (abstract: "le modèle héracléen… permet… de mieux comprendre les incertitudes de l'historiographie quant à la voie suivie par le Barcide"); Halleux = the vinegar is a literary motif; Artru = Roman roads of the Cottian passes (Mont-Cenis, Montgenèvre, Larche), not a Hannibal verdict; Fisher, Klingbeil, Gauchon = no pass. None of the Persée items could be read beyond the first page (docAsPDF 403) | — | — | — | Persée first page / Crossref abstract; SNIP for the Italian titles |
| — | Also noted, not obtained: A. Bocquet (Petit-Saint-Bernard; Allobrogian coin argument; rejects the view criterion because of Po-plain fog — fr.wiki §Petit-Saint-Bernard); F. de Coninck 1992 (Petit-Saint-Bernard per fr.wiki; Prevas's snippet groups him differently); E. Pais, J. Debelmas (Petit-Saint-Bernard, fr.wiki); T. A. Dodge 1891 (Little St Bernard per en.wiki; "even into Switzerland" per Mahaney 2026 — SNIP files exist, not read); W. Huss 1985 p. 305 (Clapier, per Walbank 2002 and Lancel's note); R. Dion 1962, E. Meyer 1958/1964, H. Ferrand 1908, G. Barruol 1996 (Clapier, per fr.wiki list); L. Cottrell 1960 (narrows to Traversette, Malaure, Clapier, pp. 79-80; final choice not recovered — SNIP); N. Fields 2010 (SNIP; no verdict recovered); R. Miles 2010 (no relevant snippet); Brizzi (archive.org scipioneeannibal0000briz restricted); Le Bohec 1996 (not obtained) | | | | | | |

---

## 2. The ancient testimonia outside Polybius 3 and Livy 21

Texts saved under `texts/testimonia/` with the edition noted in each file header. Translations are the Loeb's where one was obtained (Jones for Strabo, Rolfe for Sallust and Ammianus, Cary for Zonaras); otherwise my own, marked (tr. mine).

### 2.1 Varro apud Servius, ad Aen. 10.13 (Thilo–Hagen II, 1884; text from la.wikisource, `Servius_ad_Aen_10.13_latin_wikisource_Thilo.txt`)

> *quas Hannibal post bella Hispaniae, quae xvii annis confecit, ante exustas aceto infuso rupit: Iuvenalis "et montem rupit aceto". denique loca ipsa quae rupit, Poeninae Alpes vocantur. quamvis legatur a Poenina dea, quae ibi colitur, Alpes ipsas vocari. sane omnes altitudines montium licet a Gallis Alpes vocentur, proprie tamen iuga montium Gallicorum sunt. quas quinque viis Varro dicit transiri posse: una, quae est iuxta mare per Ligures; altera, qua Hannibal transiit; tertia, qua Pompeius ad Hispaniense bellum profectus est; quarta, qua Hasdrubal de Gallia in Italiam venit; quinta, quae quondam a Graecis possessa est, quae exinde Alpes Graiae appellantur.*
>
> (tr. mine) "…which Hannibal, after the Spanish wars that he finished in seventeen years, broke open, having first burned them and poured on vinegar (Juvenal: 'and broke the mountain with vinegar'). Indeed the very places he broke open are called the Poenine Alps — though one reads that the Alps themselves are so named from the goddess Poenina who is worshipped there. …Varro says they can be crossed by five roads: one, which is beside the sea through the Ligurians; a second, by which Hannibal crossed; a third, by which Pompey set out for the Spanish war; a fourth, by which Hasdrubal came from Gaul into Italy; a fifth, which was once held by Greeks and is therefore called the Graian Alps."

What it says: Varro (1st c. BC) knew five Alpine roads and distinguished Hannibal's from Pompey's, from Hasdrubal's and from the Graian (Little St Bernard) road. It does not name Hannibal's pass. Servius himself, in the same note, places the burned rocks in the *Poeninae Alpes* (Great St Bernard), i.e. he shares the "vulgar error" Livy 21.38.6-9 rejects.

How scholars read Varro's order:
- *Geographical, south to north* (Freshfield 1914 pp. 20-21; Torr 1924 para 24; de Beer 1956/1969; Prevas): coast road; Hannibal's; Pompey's = Mont Genèvre; Hasdrubal's = Mont Cenis; Graian = Little St Bernard. On this reading Hannibal's pass lies between the Corniche and the Montgenèvre: Col de Larche (Freshfield), Col de la Traversette (Torr, de Beer). Freshfield: "Is it not then reasonable to assume that Varro in his catalogue took all the five passes in their geographical order, beginning from the sea? Is it not the only plausible assumption?" (p. 20). Torr adds that the list "agrees with Strabo's quotation from Polybios in making Hannibal cross the next pass to the coast-road" (p. 12).
- *Not geographical* (Mommsen, CIL V p. 809, and Strachan-Davidson, as reported by Freshfield p. 21): Varro's second = Great St Bernard, i.e. Varro shared Livy's "vulgar error"; Mommsen could not place the fourth ("Col de Monte Viso" or "by Val Chisone").
- *Sceptical* (Wilkinson 1911 pp. 45-46): Freshfield "assumes that these five passes correspond to the five modern high-road passes", but "between Mont Blanc and the sea there are more than thirty passes crossed by regular mule-tracks or by footpaths"; Varro's list cannot fix a col. Torr (paras 24, 59, 62) also warns that Varro, as quoted, makes Hannibal and Hasdrubal cross different passes, whereas Livy 27.39.7 ("per munita pleraque transitu fratris, quae antea invia fuerant") and Appian Hann. 52 make Hasdrubal use Hannibal's; either Servius's quotation is interpolated or Varro assigned the Cenis to Hasdrubal.
- Lazenby (snippet) judges de Beer "almost certainly wrong to locate Hannibal's pass between the Montgenevre and the coast" on the strength of Varro.

### 2.2 Sallust, Hist. 2.98.4 Maurenbrecher (Pompey's letter to the Senate, winter 75/74 BC; `Sallust_Hist_2.98M_Epistula_Pompei_latin_Ahlberg.txt`, checked against the Loeb OCR)

> *hostisque in cervicibus iam Italiae agentis ab Alpibus in Hispaniam submovi; per eas iter aliud atque Hannibal, nobis opportunius, patefeci.*
> "…[I] had driven the enemy, who were already at the throat of Italy, from the Alps into Spain; and over those mountains I had opened for you another and more convenient route than Hannibal had taken." (Rolfe, Loeb 1921, p. 415)

Which pass is "Pompey's": the Mont Genèvre, by near-universal agreement since Mommsen — Freshfield pp. 11-12 ("reasons which are accepted by almost all scholars from Mommsen onwards"); Torr paras 25-26: Appian BC 1.109 puts Pompey's road near the sources of the Rhône and the Po, which Strabo 4.6.5 corrects to the sources of the Durance and the Dora, "enough to show that Pompey's pass was Mont Genèvre"; and Strabo 4.1.3's Roman road Tarascon–Cavaillon–Durance–Embrun–Briançon–Cesanne "must thus have crossed the Alps by Mont Genèvre, and presumably was Pompey's". Marindin (CR 13, 1899) tried "Mont Genèvre plus the Col de Lautaret" (Freshfield p. 12). The consequence everyone draws: Hannibal's pass was *not* the Mont Genèvre (Freshfield p. 11: "if the Mont Genevre had been already known as Hannibal's Pass in Polybius's time, Pompey would hardly have ventured to assert that by crossing it he had opened a new route"; de Beer 1969 snippet; Torr para 71). The Montgenèvre advocates (Connolly p. 165; Hoyos 2015) must therefore read *iter aliud* as a new road-line rather than a different col — W. T. Arnold's formulation, quoted by Freshfield p. 12 n. 3: "The Romans had a clear idea of the main Alpine wall, and would not rank a new way over secondary obstacles as a new pass 'per Alpes'… 'nobis opportunius' distinctly pointed to the low and easy Mont Genevre as opposed to a rather more difficult pass… not far off."

### 2.3 Strabo 4.6.5-7 and 4.6.12 (Meineke; `Strabo_4.6.5-12_greek_Meineke.txt`; Jones, Loeb II 1923 pp. 271-295, `Strabo_4.6.5-12_english_Loeb_Jones_1923.txt`)

4.6.12: *τέτταρας δ᾽ ὑπερβάσεις ὀνομάζει μόνον· διὰ Λιγύων μὲν τὴν ἔγγιστα τῷ Τυρρηνικῷ πελάγει, εἶτα τὴν διὰ Ταυρίνων, ἣν Ἀννίβας διῆλθεν, εἶτα τὴν διὰ Σαλασσῶν, τετάρτην δὲ τὴν διὰ Ῥαιτῶν, ἁπάσας κρημνώδεις.* — "But he [Polybius] only names four passes over the mountains: the pass through the Ligures (the one that is nearest the Tyrrhenian Sea), then that through the Taurini, which Hannibal crossed, then that through the Salassi, and the fourth, that through the Rhaeti — all of them precipitous passes" (Jones p. 293). Jones's n. 5: "Polybius (3. 56) does not say where Hannibal crossed the Alps… Both the ancient writers and modern scholars differ"; Caspari (EB) inclines to the Mont Genèvre, Lenschau (RE) to the Little St Bernard.

What "through the Taurini" does and does not settle: it excludes the Salassi passes (both St Bernards) and the coastal road, and puts Hannibal's pass among those debouching on Turin — Mont Cenis, Clapier, Montgenèvre (Walbank's three, per Connolly p. 165) and, for the de Beer school, also the Traversette (Saluzzo is 60 km from Turin; Torr para 30 measures to Saluzzo). Torr p. 11 notes Strabo's "through the Taurini" is not exactly what Polybius says (3.56.3, descent "to the Insubres"), and Freshfield p. 11 uses it to argue that Polybius counted the fighting with the Taurini as part of the passage. Since this is Strabo's paraphrase of Polybius, it is the strongest early witness that the *Taurini* pass, not the Graian or Poenine one, was Hannibal's.

4.6.5: *μετὰ δὲ Ὀυοκοντίους Ἰκόνιοι καὶ Τρικόριοι καὶ μετ᾽ αὐτοὺς Μέδουλλοι, οἵπερ τὰς ὑψηλοτάτας ἔχουσι κορυφάς· τὸ γοῦν ὀρθιώτατον αὐτῶν ὕψος σταδίων ἑκατὸν ἔχειν φασὶ τὴν ἀνάβασιν, κἀνθένδε πάλιν τὴν ἐπὶ τοὺς ὅρους τοὺς τῆς Ἰταλίας κατάβασιν.* — after the Vocontii come the Iconii and Tricorii, then the Medulli "who have the highest peaks", with a hundred-stade ascent and a like descent to the Italian border; up in hollows a large lake and two springs, one feeding the Druentia to the Rhône, the other the Duria to the Po; the Medulli lie "above the confluence of the Isar and the Rhodanus" (Jones p. 271-3). This is the passage from which the Medulli are put in the Maurienne (Arc), and it makes Strabo's Vocontii–Tricorii–Medulli sequence run from the Drôme/Durance uplands northwards — relevant to Livy 21.31.9 (Tricastini, Vocontii, Tricorii) and Ammianus 15.10.11.
4.6.6: the Taurini, *Λιγυστικὸν ἔθνος*, on the Italian side, with the lands of Donnus and Cottius; then the Salassi; above them the Ceutrones, Caturiges, Varagri, Nantuates. 4.6.7 and 4.6.11: the two roads from the Salassi/Lugdunum side, the wagon-road through the Ceutrones (Little St Bernard) and the short steep Poeninus (Great St Bernard) — the roads Livy 21.38 rejects for Hannibal.

### 2.4 Cornelius Nepos, Hannibal 3.4 (Fleckeisen/Halm; `Nepos_Hannibal_3_latin_Fleckeisen-Halm.txt`)

> *ad Alpes posteaquam venit, qua Italiam ab Gallia seiungunt, quas nemo umquam cum exercitu ante eum praeter Herculem Graium transierat (quo facto is hodie saltus Graius appellatur), Alpicos conantes prohibere transitu concidit, loca patefecit, itinera muniit, effecit ut ea elephantus ornatus ire posset, qua antea unus homo inermis vix poterat repere.*
> (tr. mine) "When he came to the Alps, which separate Italy from Gaul, and which no one before him had ever crossed with an army except the Greek Hercules (whence that pass is today called the Graian), he cut down the Alpine peoples who tried to bar his crossing, opened up the ground, built roads, and made it possible for a harnessed elephant to go where before a single unarmed man could scarcely crawl."

What it says: Nepos does not say Hannibal used the *saltus Graius*; the relative clause explains the name. Read as an implication that Hannibal's pass was the Graian (Little St Bernard) by the Little St Bernard school (Law, Lenschau); Torr para 58 (pp. 31-32) shows the term was elastic ("Ptolemy iii.1.35-36 makes the Graian Alps include Briançon and Embrun. Tacitus (Hist. ii.66) speaks of a legion marching from Turin across the Graian Alps, clearly meaning the Genevre or Cenis passes, as the Little St Bernard is not accessible from Turin"). Pliny 3.123 and 3.134 (below) attach Hercules to the Graiae and the Poeni to the Poeninae. The "Hercules" road is also the basis of Connolly's argument for the Montgenèvre (the Herculean way = the Durance road, so Diod. 4.19.3-4; Hornblower 2024 p. 85-86 n. 52 notes Diodorus does not mention the Alps).

### 2.5 Ammianus Marcellinus 15.10.2-11 (`Ammianus_15.10_latin_LatinLibrary.txt`; Rolfe, Loeb I 1935 pp. 175-187, `Ammianus_15.10_english_Loeb_Rolfe_1935.txt`)

15.10.2-7: the Cottian road from Segusio up to *stationem nomine Martis* (Ad Martis = Oulx) seven miles on, then the steep climb *ad Matronae … verticem* (Mont Genèvre) and down to Brigantia, built by King Cottius; 15.10.9: the Poenine Alps and the Graian named from Hercules; 15.10.10-11 (Latin as printed in the Loeb OCR, agreeing with the Latin Library text):
> *Quae Hannibal doctus a perfugis, ut erat expeditae mentis et callidae, Taurinis ducentibus accolis, per Tricasinos et oram Vocontiorum extremam, ad saltus Tricorios venit. Indeque exorsus, aliud iter antehac insuperabile fecit; excisaque rupe in immensum elata, quam cremando vi magna flammarum acetoque infuso dissolvit, per Druentiam flumen, gurgitibus vagis intutum, regiones occupavit Etruscas.*
> "But Hannibal learned of this from deserters, and being of a nimble and crafty wit, came, under the guidance of natives from among the Taurini, through the Tricasini and the extreme edge of the Vocontii to the passes of the Tricorii. Starting out from there, he made another road, where it hitherto had been impassable; he hewed out a cliff which rose to a vast height by burning it with flames of immense power and crumbling it by pouring on vinegar; then he marched along the river Druentia, dangerous with its shifting eddies, and seized upon the district of Etruria." (Rolfe p. 187; his notes identify Ad Martis = Oulx and Matrona = Mont Genèvre)

What it says: Ammianus (4th c., drawing on Timagenes for the Alps, unverified) reproduces Livy's tribal sequence (Tricastini, Vocontii, Tricorii), adds Taurini guides, and has Hannibal reach the Durance *after* the Tricorii and the rock-splitting — i.e. a Durance-basin pass, which Torr (para 61) and the de Beer school stress and which the Isère school treats as Livian confusion. Note the two Latin texts differ at one word: Latin Library *in solidam solvit*, Loeb *dissolvit* — check Seyfarth before citing.

### 2.6 Appian, Hannibalica 4 (= 1.4 Mendelssohn; `Appian_Hannibalica_1-2_greek_Mendelssohn.txt`)

> *…κακοπαθῶν χιόνος τε πολλῆς οὔσης καὶ κρύους, τὴν μὲν ὕλην τέμνων τε καὶ κατακαίων, τὴν δὲ τέφραν σβεννὺς ὕδατι καὶ ὄξει, καὶ τὴν πέτραν ἐκ τοῦδε ψαφαρὰν γιγνομένην σφύραις σιδηραῖς θραύων, καὶ ὁδὸν ποιῶν ἣ καὶ νῦν ἐστιν ἐπὶ τῶν ὀρῶν ἐντριβὴς καὶ καλεῖται δίοδος Ἀννίβου.*
> (tr. mine) "…suffering badly from the deep snow and the cold, cutting and burning the timber, quenching the ash with water and vinegar, and breaking with iron hammers the rock that was thus made crumbly, he made a road which even now is well trodden on the mountains and is called 'Hannibal's pass'."

What it says: in the 2nd c. AD a road over the mountains was popularly called *diodos Annibou*; Appian does not locate it. Mahaney 2026 (p. 4) cites it for the southern route; it is equally consistent with any col. Appian Hann. 52 (Hasdrubal used the road Hannibal had opened, *hodopoiēmena proteron hypo Annibou*, per Torr para 62) and Iberica 13-14 (the march to the Alps; sixth month from New Carthage, Hann. 4) were checked in the Perseus files and add no pass.

### 2.7 Zonaras 8.23 (epitome of Cassius Dio; Cary, Loeb Dio II 1914, `Zonaras_8.23_Dio_Loeb_vol2_Cary_1914_pp70-95_region.txt`)

"For the mountains there are exceedingly precipitous, and the snow, which had fallen in great quantities, was driven by the winds and filled the chasms, and the ice was frozen very hard… many of them perished by reason of the cold and lack of food; many also returned home. There is a story to the effect that he himself would also have turned back but for the fact that the road already traversed was longer and more difficult than the portion remaining before him." No pass, no tribe, no landmark; useful only for the snow/ice tradition.

### 2.8 Pliny, NH 3.117, 3.123-125, 3.133-138 (Mayhoff via Perseus; `Pliny_NH_3.117-138_latin_Mayhoff.txt`; 3.31-37 from the Latin Library)

- 3.117: *Padus, e gremio Vesuli montis celsissimum in cacumen Alpium elati finibus Ligurum Bagiennorum visendo fonte profluens…* — the Po rises on Monte Viso, "the loftiest summit of the Alps", in Bagienni territory. Cited by the Traversette school (de Beer, Mahaney) because the Traversette descends straight to the Po's source; Pliny says nothing of Hannibal here.
- 3.123: *Salassorum Augusta Praetoria iuxta geminas Alpium fores, Graias atque Poeninas — his Poenos, Grais Herculem transisse memorant* — "they relate that the Carthaginians crossed by the Poenine [gate], Hercules by the Graian." This is the *vulgar* Poeninus tradition (Great St Bernard), the same one Livy 21.38.6-9 and Servius record; Pliny reports it without endorsement (*memorant*). 3.124: *Vercellae Libiciorum ex Salluis ortae* — the Libui/Libici at Vercelli, whom Livy 21.38.7 says the Poenine and Cremonis routes would have reached. 3.125: the Caturiges, *Insubrum exsules*, extinct.
- 3.133-134: Alpine length and breadth per Coelius, Timagenes, Nepos, Livy; the Lepontii "left from Hercules' company, with limbs frost-bitten in the crossing of the Alps", and the Graii "of the same army" as inhabitants of the Graian Alps.
- 3.137-138: the Tropaeum Alpium list — *SALASSI · ACITAVONES · MEDVLLI · VCENNI · CATVRIGES · BRIGIANI · SOGIONTI · BRODIONTI · NEMALONI · EDENATES · VESVBIANI…* — and *Non sunt adiectae Cottianae civitates XV, quae non fuerant hostiles, item adtributae municipiis lege Pompeia.* Freshfield (p. 12) uses the exemption of the fifteen Cottian *civitates* to argue that Pompey's road, and later Cottius's, was the Mont Genèvre.
- 3.36-37 (Narbonensis): Vocontii a federated *civitas* with two capitals, Vasio and Lucus Augusti; the Tricastini (*Augusta Tricastinorum*) — the tribes of Livy's itinerary (not quoted here at length; see the saved file).

### 2.9 Caesar, BG 1.10.3-5 (`Caesar_BG_1.10_latin_LatinLibrary.txt`, checked against the Perseus XML)

> *…qua proximum iter in ulteriorem Galliam per Alpes erat, cum his quinque legionibus ire contendit. Ibi Ceutrones et Graioceli et Caturiges locis superioribus occupatis itinere exercitum prohibere conantur. Compluribus his proeliis pulsis ab Ocelo, quod est oppidum citerioris provinciae extremum, in fines Vocontiorum ulterioris provinciae die septimo pervenit; inde in Allobrogum fines, ab Allobrogibus in Segusiavos exercitum ducit.*

In 58 BC the "nearest road" from Cisalpina into Further Gaul ran from Ocelum (Val di Susa; Avigliana or Drubiaglio, unverified) to the Vocontii in seven days, past Ceutrones, Graioceli and Caturiges holding the heights — i.e. the Mont Genèvre/Durance corridor (Torr, Freshfield) or, for the Cenis school, the Mont Cenis (Jullian). It fixes the Vocontii as the first Transalpine people reached from the Cottian passes and shows that the Durance road was the ordinary military road twenty years after Pompey.

### 2.10 Silius Italicus, Punica 3.477-556 and 4.1-60 (`Silius_Punica_book3/4_latin_LatinLibrary.txt`)

Epic amplification of Livy: perpetual snow, no road, the Graian Hercules as the only predecessor, the descent to the Taurini. No topographical datum independent of Livy; not used as evidence by any modern advocate I found except for colour.

### 2.11 Ptolemy 3.1.30-36 (Müller 1883, OCR in `editions/`)

Taurini "under the Salassi", Augusta Taurinorum; the Graian Alps extended to include Briançon and Embrun (Torr para 58 citing 3.1.35-36); Ptolemy 3.1.30-32 places Aosta and Ivrea with the Salassi and Vercelli and Lomello with the Libici (Torr para 20). Used only to gloss Livy's Libui and Nepos' Graian pass.

### 2.12 Roman road distances (Itinerarium Antonini and Burdigalense, Parthey–Pinder 1848 OCR: `Itinerarium_Antonini_and_Burdigalense_Alpine_routes_Parthey-Pinder_1848_OCR.txt`; numerals must be checked against the scan)

- Itin. Ant. 339.6-341.5 Wess., *a Mediolano Arelate per Alpes Cottias*: … Taurinis – Fines xviii – Segusione xxxiii – Ad Martis xvi – Brigantione xviii … (Mont Genèvre road). Itin. Burdig. 555-556: *inde incipiunt Alpes Cottiae … Byrigantum, inde ascendis Matronam, Gesdaone, Ad Marte, Secussione, inde incipit Italia.*
- Itin. Ant. 344.3-346.3, *per Alpes Graias*: Augusta Praetoria – Arebrigium – Bergintrum – Darantasia … (Little St Bernard road).
- Itin. Ant. 356-358: Ramae – Eburoduno – Caturrigas – Vapinco – Monte Seleuco – Luco – Dea Vocontiorum – Augusta – Valentia – Ursolis – Vienna (the Durance–Gap–Drôme road), the line Torr (para 29) and Strabo 4.1.3 (Tarascon 63 m.p. to the ascent, 99 to Embrun, 71 to Cesanne) use to match Polybius' 1,400 stades "along the river".
- The Peutinger Table is cited only through Azan p. 33 (the Fossae Marianae–Arles–Avignon–Orange–Valence–Vienne road, 204 km from the sea to the Isère confluence) and Bocquet (fr.wiki: a Graian route Aosta–Isère cities); not obtained. The Vicarello goblets were not obtained (they give the coastal Gades–Rome road, no inland Alpine pass — unverified).

### 2.13 Timagenes, Justin 32.3, Diodorus 25

Timagenes survives for the Alps only through Ammianus 15.9-12 (cf. 15.9.2, not re-read here; unverified). Justin 32.3 (saved, `Justin_32_latin_LatinLibrary.txt`) concerns the Gauls' migrations and Philip/Perseus, nothing on Hannibal's pass. Diodorus 25 (fragments) was not obtained (Perseus Diodorus book 25 not under the tried paths) — nothing is expected there beyond the Hercules road of 4.19.

### 2.14 Coelius Antipater's *Cremonis iugum* (Livy 21.38.6-7 — outside this remit, but the readings belong here)

Livy: the "vulgar" view is the Poeninus; "Coelius says he crossed by the Cremonis iugum"; both would have led him to the Libui, not the Taurini, so Livy rejects both. Readings: (a) Little St Bernard / Mont Cramont (Law 1866; Lenschau; Livy's own inference — Torr para 20 p. 10, para 57 pp. 31); (b) a pass at the head of the Dora Riparia mistaken by Livy, who like Strabo 4.6.5 confused the two Duriae (Torr para 57, p. 31: "Livy may have made the same mistake and fancied that Hannibal would go down the Dora Baltea… when Coelius really meant the pass at the source of the Dora Riparia"); (c) Col de Grimone on the Drôme road (de Beer; Lazenby n. 39 "de Beer's neat identification of the Cremonis iugum with the Col de Grimone"; de Beer 1969 snippet dismisses Mont Cramont as "not worth any [consideration]"); (d) Lazenby (snippet, cut): Livy "right to identify… with a pass leading into the territory [of the Libui]". Since Coelius drew on Silenus, who was with Hannibal (Cic. Div. 1.24.49, per Torr para 57), this is the one ancient datum that could name the pass, and it is irrecoverable.

### 2.15 The Gallic peoples (Barruol 1969 not obtained; Rivet 1988 snippets; primary texts above)

Allobroges: Rhône–Isère, up to the Grande Chartreuse (Pol. 3.49.5-13; Caes. BG 1.10.5; Hoyos 2015 p. 106). Tricastini: the Tricastin, Saint-Paul-Trois-Châteaux (Pliny 3.36; Livy 21.31.9; Amm. 15.10.11). Vocontii: Vaison and Luc-en-Diois/Die (Pliny 3.37; Strabo 4.6.5; Caes. BG 1.10.5). Tricorii: between the Vocontii and the Medulli (Strabo 4.6.5), i.e. the Drac/Trièves–upper Durance uplands (Hoyos 2015 p. 106: "on the other side of the Vercors massif"; en.wiki "Tricorii" citing Hoyos 2006 pp. 442-4). Caturiges: Chorges/Embrun (Pliny 3.137; Caes. 1.10.4; Itin. Ant. Caturrigas). Medulli: the Maurienne, "highest peaks" (Strabo 4.6.5; Pliny 3.137). Ceutrones: Tarentaise, Little St Bernard road (Strabo 4.6.6-7, 11; Caes. 1.10.4; Pliny 3.135). Graioceli: Caes. 1.10.4 only; usually the upper Maurienne/Mont Cenis (unverified). Segovii/Segusini: Segusio (Pliny 3.123; Amm. 15.10.3). Taurini: Ligurian, on both sides of the Po around Turin (Strabo 4.6.6; Pliny 3.123; Pol. 3.60.8; Livy 21.38.5). Insubres: Milan (Pliny 3.124-125; Pol. 3.56.3). Salassi: Aosta valley (Strabo 4.6.7; Pliny 3.123, 134). Libui/Libici: Vercelli (Pliny 3.124; Livy 21.38.7). Guy Barruol's *Les peuples préromains du Sud-Est de la Gaule* (1969) remains the standard reference (unverified — not obtained; only his 1963 Memini article first page on Persée).

---

## 3. The case for and against each pass, in the advocates' words where available

**Col du Clapier (2,482 m; Savine-Coche and Petit Mont-Cenis variants).** For: the view of the Po plain from a promontory that "would hold several thousand men" (Wilkinson pp. 8-9; Colin p. xv: "Ce n'est pas un lambeau de plaine qu'on aperçoit, mais bien… 'toute l'Italie'"); Polybius' distances from the Rhône (Colin pp. 318-321: Isère–Arc–Clapier fits 1,400 + 1,200 stades, "à condition de passer, non pas par le mont Cenis, mais par le Clapier"); the Italian end nearest Turin (Hornblower p. 86); old snow lying on a high pass (Lazenby); the "acropolis" impression (Wilkinson p. 8). Against: "a lofty and difficult by-pass" (Freshfield preface); Connolly p. 165: from the col itself no view ("Only by trekking along the hillside to Monte Aria is this possible"), the north side "completely iced up" if old snow lay on the south side, and "Those who have argued that Hannibal crossed the Clapier must explain why he did not use the Mont Cenis which is both lower and easier"; Torr para 69: "Hannibal had no motive for going so far north"; Bocquet: the Ambin glacier and Po-plain fog (fr.wiki); Mahaney: no rockfall, no firn. Advocates: Perrin, Azan, Colin, Ferrand, Wilkinson, Dion, Meyer, Lavis-Trafford (Savine-Coche), Hoyte, Walbank (hesitantly), Proctor, Huss, Lazenby, Barruol, Lancel, Kuhle & Kuhle, Hunt, Hornblower; slightly preferred by Bradford.

**Col de la Traversette (2,947 m).** For: Varro's order and Pompey's Genèvre put Hannibal south of the Genèvre (Freshfield, Torr, de Beer); Polybius' "along the river" = the Durance road (Torr para 29); the view (Torr para 69; Connolly concedes it is "the only pass from which the plain of the Po can be seen", p. 164); year-round snow (Prevas); the two-tier rockfall, firn and the Clostridia mire (Mahaney); elephant energetics (Berti & Vollrath 2026). Against: the Combe de Queyras gorge, 15 km "with precipitous sides" ("Hannibal would have had to have been a maniac to take his army this way", Connolly p. 164; Azan pp. 62-3 quoting Roman on "l'horreur de cette gorge"); altitude "more than 100 m higher than the Montgenèvre" and the lower Col de la Croix on the way (Connolly p. 164); the 1979 Royal Engineers' verdict on elephants and scree (Connolly p. 165); "rather than a pass, a narrow breach" (Lancel); Walbank 1956 (not read); Kuhle & Kuhle on the philology of the "rockfall"; Hunt ("might not even be that ancient geologically"). Advocates: Torr, de Beer, Cottrell?, Bagnall, Pyatt, Prevas, Seibert (reportedly), Mahaney, Berti & Vollrath.

**Col du Montgenèvre (1,854 m).** For: the lowest and easiest pass, the "road of Hercules", the only road to the Taurini Polybius knew (Connolly p. 165); Livy's Tricastini–Vocontii–Tricorii–Durance itinerary leads there (Hoyos 2015 p. 106; Hoyos 2006 Klio); Caesar's 58 BC road; W. T. Arnold: "the evidence converges on a Durance pass. The best Durance pass is certainly the Mont Genevre." Against: Pompey's *iter aliud* (Freshfield pp. 11-12; de Beer; Torr para 71); "too low and too easy" and no view (de Beer; Azan p. 63 quoting Macé: "Du col du mont Genèvre… on n'aperçoit que des [gorges]"); nine days from the Durance to a summit ten miles from it (Torr para 68); Lancel: the Herculean way was exactly what Scipio expected. Advocates: Gibbon 1763 (de Beer table), Hennebert, Bullock Hall 1898 (Freshfield p. 12), Caspari, Connolly, Hoyos, Yardley–Hoyos.

**Col du Mont-Cenis (2,083 m; Petit Mont-Cenis 2,182 m).** For: Napoleon's military judgement; the shortest road to Turin (Jullian p. 479); easier than the Clapier (Connolly). Against: no view from the plateau ("un cirque de montagnes enferme complètement", Colin p. xv; Azan pp. 58-61; Wilkinson p. 8 n. 1). Advocates: Simler, Larauza, Maissiat, Ellis (Little Mont Cenis), Osiander, Jullian, Napoleon; no advocate since Hoyte's expedition crossed it faute de mieux (1959).

**Col de Larche / Argentière (1,991 m).** For: Varro's second pass in geographical order (Freshfield pp. 20-21); the Vars–Argentière route with villages and cornfields "all the way" (p. 28); an easy col; "probably the pass which Hannibal intended to cross" (de Beer, who then diverts him to the Traversette). Against: no view (Torr para 69); 40 miles to the plain (Torr para 55); Wilkinson pp. 45-46; Freshfield's own W. T. Arnold preferred the Genèvre; Hoyos 2015 keeps it only as the Genèvre's "neighbor". Advocates: Freshfield alone in the 20th c.; Hoyos as an alternative.

**Petit-Saint-Bernard (2,188 m).** For: Coelius' Cremonis iugum, Nepos' saltus Graius, the Ceutrones wagon-road (Strabo 4.6.11), the opulence of Allobrogian lands (Mommsen, Bocquet), Livy's "turn to the left" from the Rhône. Against: Livy 21.38.7-8 (leads to the Libui); no view and 75 miles to the plain (Torr para 55; Colin p. 321: "On ne voit de là que d'affreux précipices"); Strabo 4.6.12 "through the Taurini". Advocates: de Luc, Wickham & Cramer, Long?, Law, Arnold, Mommsen, Lenschau, Hyde, Pais, Dodge, Coninck, Debelmas, Bocquet.

**Grand-Saint-Bernard (Poeninus), Col Agnel, Col de la Croix, Col de Malaure.** The Poeninus is the "vulgar" ancient view (Livy 21.38; Pliny 3.123; Servius) held by Whitaker 1794 and refuted by Livy's own argument; Torr para 23 ("incredible that Hannibal should have crossed a pass so far eastward"). Agnel and Malaure are named as by-ways only (Freshfield p. 28; Wilkinson p. 9 on Malaure's view "from a narrow ridge on which no body of troops could be stationed"; Cottrell's short list); the Col de la Croix is on Connolly's road to the Traversette (p. 164). No advocate found for any of them.

---

## 4. How the debate stands in 2026, and the "nothing has happened since" checks

**Exact quotations held (for the report).**
- Walbank 1957 (Commentary I) is not obtained; his own retrospect, *Polybius, Rome and the Hellenistic World* (2002), snippet: "…hesitatingly came down in favour of the Col du Clapier (which was also Huss' choice). A slight variant…"; and "More recently J. Seibert has boldly suggested that Hannibal divided his army into two parts, which crossed…"; and "Hence when Hannibal leaves the so-called 'Island' to march up the Isère along the foot of the Alps, Polybius…". Proctor 1971 (snippet) on the Commentary: "Walbank brackets it [the Mont Cenis] with the Col du Clapier without declaring a preference." Hunt 2017 (snippet): "Walbank says that too much ink has been [spilled]…"; "not even including philology, as argued by Walbank, 'Hannibal's Pass,' 37-45, esp. 43-45."
- Lazenby 1998 (snippets): "…fulfil all the requirements — the Col de Clapier. This is 2482 metres high, has a wonderful [view]…"; "the Isere, moreover, and Hannibal here left the Rhone to march on up the Isere, as those…"; "The distance between the gorge above Pontcharra and the Col de Clapier is approximately 120 kilometres, and this…"; "[de Beer] was almost certainly wrong to locate Hannibal's pass between the Montgenevre and the coast"; "de Beer's neat identification of the 'Cremonis iugum' with the Col de Grimone on RN 5".
- Hoyos 2015, *Mastering the West* pp. 105-106: "This zigzag route is plausible for an army aiming to elude Roman observers: along the Rhone to the Isère, over to the Drac valley, and from there to the high Alps and the pass. Such a route would best lead to the Montgenèvre pass or one of its neighbors such as the Larche — each a fairly easy climb up from Gaul, followed by a steep descent down to Italy. Other candidates (the best known are the Col du Clapier and the Traversette, both farther north) are much more difficult to fit into what evidence we do have."
- Goldsworthy 2000/2007, "The March to Italy": "Even in Livy's day, there was a fierce dispute over which pass the Carthaginians had taken over the Alps… Their conclusions have varied enormously and unfortunately the nature of our sources makes it impossible to resolve these disputes. I do not intend to discuss this topic since it would be impossible to do it justice in the framework of a treatment of all three Punic [Wars]"; and "which pass is perhaps more fiercely debated than any other aspect of his route."
- Hornblower 2024, p. 86: "Hannibal's own exact route across the Alps is disputed to this day. He clearly left the Rhône to follow its tributary the Isère, but there agreement ends. The likeliest mountain pass must be one whose Italian end is closest to Turin, where he is first heard of in action after his descent. That probably indicates that the best candidate for the pass is the Col de Clapier. Astonishingly, all his elephants got through"; n. 54: "Walbank 1985: 107–19 [orig. 1956]; for the Col de Clapier, Proctor 1971: 210–16; Lazenby 1978: 45–6 with plate iv; 302 map 3. But Hoyos in Yardley 2006: 620–30 at 626–9 prefers the Col de Montgenèvre."
- de Beer 1956/1969 (snippets): "the pass which Pompey crossed, the Mont-Genevre, was not the pass that Hannibal crossed. If he had crossed it…"; "…Isere basin and the knowledge that Hannibal did not cross the Mont-Genévre, leave only the Ubaye and the [Guil]…"; "That leaves the Col de la Traversette (2,950 metres), accessible through the valley [of the Guil]…"; "identify the Cremonis jugum with the Mont Cramont, an outlying spur of the Mont-Blanc range, is not worth any [consideration]"; "…in thinking that by Cremonis jugum Coelius meant the Petit Saint-Bernard; Coelius, whose testimony because…"; on Skaras: "he calmly altered Skaras into Arar to make it agree with Livy"; the Col de Larche "probably the pass which Hannibal intended to cross"; the Traversette the "strongest competitor".
- Kuhle & Kuhle 2011 (abstract): "For a long time now, historians have considered the way across the Col de Clapier as the most likely invasion route of Hannibal's army across the Alps… a critical reassessment of the ancient texts and a closer look at the geomorphological situation clearly show that this interpretation is unwarranted."
- Berti & Vollrath 2026 (abstract): "Among the possible alpine passes, the route crossing at the Col du Clapier was considered the most likely candidate. Recent philological and geomorphological analyses suggest instead a route via the Col de la Traversette… Our analysis favors the route of the Traversette."
- Mahaney 2026, p. 8: "…the great captain crossed either Mt Cenis or the Col du Clapier, no matter the evidence against it. The facts compiled by Polybius, Sir Gavin de Beer and myself not only argue otherwise, but the rockfall itself, a one-of-a-kind landform on the lee of the Alps, certifies the route."

**Timeline 2005-2026 (verified items only).** de Galbert 2005 (Belledonne short cut; fr.wiki). Hoyos 2006, Klio 88: 408-465 (Durance route to the pass; Montgenèvre). Halleux 2007, CRAI 151: 529-534 (the vinegar is a literary motif; first page on Persée). Mahaney et al. 2007 (Geoarchaeology 22: burnt schist outcrops), 2008 (Geology Today; *Hannibal's Odyssey*), 2010 (Archaeometry 52: 156-72, the Traversette rockfall). Kuhle & Kuhle 2011 (Archaeometry 54: 591-601, "Hannibal gone astray?"). Jourdain-Annequin 2011 (*Quand les Grecs et les Romains découvraient les Alpes*; earlier DHA 1999). Hunt 2012 (EAH "Alps"). Mahaney 2013 (Archaeometry 55: 1196-204); Artru 2013 (DHA 39: 237-263, Cottian roads) and 2016 (ISTA monograph, Mont-Cenis to Larche); Kuhle & Kuhle 2014/15 (Archaeometry 57: 759-771, "Lost in Translation"); Mahaney et al. 2014 (MAA 14: 77-94, "Polybius's 'previous landslide'"); Ancel 2015 (Archéologie médiévale, the Traversette tunnel); MacDonald 2015; Hoyos 2015; Mahaney 2016 (MAA 16: 97-105) and 2016/17 (Archaeometry 59: 164-178 + part II: biostratigraphy, Clostridia, ¹⁴C c. 2168 BP); Allen 2016 (The Conversation; BBC "Dung clue", 5 April 2016); Hunt 2017 (*Hannibal*); Prevas 2017; Oakley 2019 (GIF Bibliotheca: Livy 21.32.6-33.1 and Pol. 3.50.1-51.13); Mahaney et al. 2019 (upper Po reconnaissance); Lendering 2022; Freeman 2022; Mahaney 2023 (Polybius as stratigrapher, per 2026); Hornblower 2024; Kötter 2024; Berti & Vollrath 2026 (PNAS); Hoyos 2026 (HZ review of Hornblower); Hunt 2026 (Oxford Bibliographies); Mahaney 2026 (LJRHSS).

**Checks on "nothing since" claims.**
1. "No one has examined the Traversette geomorphology since Mahaney": false — Kuhle & Kuhle 2011 and 2014 reassessed both the texts and "the geomorphological situation" (abstract), and Berti & Vollrath 2026 modelled the pass energetically.
2. "Classicists have ignored the geoarchaeology": partly false — MacDonald 2015 (n. citing Mahaney 2010), Hunt 2017 (p. 284 n., citing Kuhle & Kuhle against Mahaney), Hoyos 2015 (n. 5, "Mahaney 2010" listed among "notable modern discussions") engage it; Hornblower 2024 n. 54 does not cite Mahaney at all, which is itself a datum on the mainstream's reception.
3. "The Clapier is the consensus": overstated — it is the most-cited single answer (Kuhle & Kuhle: "for a long time now, historians have considered the way across the Col de Clapier as the most likely"; Berti & Vollrath: "was considered the most likely candidate"; Hornblower "probably"), but Hoyos (Montgenèvre), the French Petit-Saint-Bernard school (Bocquet), and the Traversette school are live, and Goldsworthy, MacDonald and Lendering refuse a verdict.
4. "No new pass or route has been proposed since 1971": false — Savine-Coche (Lavis-Trafford 1956/Meyer 1964, revived by Lancel and Hunt), de Galbert's Belledonne short cut (2005), Seibert's divided army (1993), Bocquet's Petit-Saint-Bernard revival, Mahaney's Grîmone–Aygues–Guil line.
5. "Nothing has been done on the Roman roads of the Cottian Alps": false — Artru 2013, 2016 (Persée).
6. "No one has done a physical/quantitative model of the crossing": false — Berti & Vollrath 2026 (PNAS) did, favouring the Traversette; earlier the 1979 Royal Engineers survey (Connolly p. 165) and Hoyte 1959 tested feasibility with an elephant.

**Majority view, stated carefully.** Among professional ancient historians writing in English 1956-2024 the Isère–Arc–Clapier/Savine-Coche route is the modal answer, always hedged (Walbank "hesitatingly", Hornblower "probably", Lancel "chimérique"); the Montgenèvre is the principal scholarly alternative (Hoyos); the Traversette is favoured by the earth-science literature and by popular authors in the de Beer line, and now by a PNAS energetics paper; the Petit-Saint-Bernard is a French regional position. No candidate commands a majority of all authors, and the two most recent peer-reviewed contributions (Hornblower 2024; Berti & Vollrath 2026) point in opposite directions.

---

## 5. Access log

Full log with timestamps: `texts/scholarship/ACCESS_LOG.tsv` (41 OK, 19 FAIL lines for the download helper; the many search/API calls that produced the snippet, Persée and Crossref files are not logged in it).

**Obtained in full (public-domain OCR from archive.org unless noted; filenames under `texts/scholarship/`):** Whitaker 1794 (2 vols), de Luc 1818, Wickham & Cramer 1820 and 1828, Larauza 1826, Brockedon 1828 vol. 1 and 1833, Long 1831, Ellis 1853, Chaix 1855 (JRGS), Wiedemann 1856, Law 1866 (2 vols + Google copy), Maissiat 1874, Ball 1898, Hall 1898, Osiander 1900, Azan 1902, Colin 1904, Hesselmeyer 1906, Coolidge 1908, Jullian 1908 vol. 1, Wilkinson 1911, Freshfield 1914, Torr 1924, Napoleon 1836 *Précis* (not the relevant text), Connolly *Greece and Rome at War* (English OCR), Goldsworthy 2000/2007, Hoyos 2015, Hornblower 2024, Mahaney 2026 (open-access PDF). Testimonia: Strabo (Perseus Greek; Loeb Jones OCR), Sallust (Loeb OCR), Nepos (Perseus), Servius (la.wikisource Thilo; Thilo vol. 2 OCR in `editions/`), Ammianus 15 (Latin Library; Loeb Rolfe OCR), Appian Hann. and Iber. (Perseus), Zonaras/Dio (Loeb Cary OCR), Pliny NH 3 (Perseus Mayhoff; Latin Library), Caesar BG 1 (Latin Library; Perseus), Silius 3-4 (Latin Library), Justin 32 (Latin Library), Ptolemy (Müller 1883 OCR), Itin. Antonini/Burdigalense (Parthey–Pinder 1848 OCR).

**Obtained in part or by snippet only (Open Library search-inside; no page numbers):** de Beer 1956, 1969 (x2), 1974; Proctor 1971; Lazenby 1998; Lancel 1998 and Carthage 1995; Prevas 1998 (x2) and 2017; Hunt 2017; MacDonald 2015; Connolly 1998 ed.; Bradford 2000 ed.; Pyatt 1984; Bagnall 1990/2008; Peddie 1997/2005; Fields 2010; Cottrell 1960-92; Miles 2010 (no hits); Putnam 2008; Rivet 1988; Walbank 2002 essays; Toynbee 1965; Caven 1980; Picard 1991; Scullard; Cary; Mommsen; Arnold; Dodge; de Sanctis; Dirkzwager. Persée first pages/abstracts only: Raepsaet-Charlier 1972 (Proctor review), Dumont 1973 (Walbank II review), Davin 1958, Bailet 1960, Pédech 1966 (REG), Marquion 1968, Blache 1963, Barruol 1963, Halleux 2007, Artru 2013 and 2016, Jourdain-Annequin 1999, Klingbeil 2000, Fischer 1982, Gauchon 2002 and others (37 files in `persee/`). Crossref abstracts: Mahaney 2008 (Geology Today), Mahaney et al. 2010 (Archaeometry), Kuhle & Kuhle 2011, Law 2014 (CUP reprint), Mahaney 2026, Berti & Vollrath 2026, Jourdain-Annequin 1999. Wikipedia en/fr/de/it crossing articles and tribe/pass pages, used only as finding aids (`wikipedia_surveys/`).

**Not obtained at all (with the failure):** Walbank JRS 46 (1956) — JSTOR "client challenge" page; Walbank Commentary I (1957) — archive.org lending copy 401; Perrin 1883/1887; Hennebert; Hyde 1935; Hoyte 1960 — archive.org 401; Seibert 1993a/b; Brizzi — archive.org 401; Le Bohec 1996; Lancel 1995 French original; Freeman 2022; Kötter 2024; Lendering 2022; de Galbert 2005; Bocquet; Hoyos 2006 Klio — de Gruyter/DOI 403 (metadata only); Kuhle & Kuhle 2014 full text — Wiley 403; Mahaney 2010/2013/2014/2016/2017/2019 full texts — Wiley/Zenodo not reachable (429/403); Oakley 2019 — Brepols 403; Hunt 2026 OBO — OUP 403; Berti & Vollrath 2026 full text — PNAS not attempted beyond Crossref (abstract obtained); Torr 1925 2nd ed.; Freshfield's 1886/1899 journal articles; Marindin CR 13 (1899); Barruol 1969; Persée full PDFs (docAsPDF 403); Gallica (403 "Accès interdit"); Google Books previews (429/403); HathiTrust (403); Wayback Machine availability API (429 on every attempt). Web search engines were effectively unusable through the proxy (Google consent redirect; Bing returning results for the first word only; DuckDuckGo/Brave/Startpage failures), so the 2010-2026 bibliography rests on Crossref, Persée, Open Library, archive.org and Wikipedia reference lists and is certainly incomplete.

**Files added for this dossier:** `texts/scholarship/*.txt` (32 full OCR texts + `Mahaney_2026_…pdf/.txt`), `texts/scholarship/OL_search_inside_snippets/` (c. 130 files), `texts/scholarship/persee/` (37), `texts/scholarship/wikipedia_surveys/` (24), `texts/scholarship/ACCESS_LOG.tsv`, `texts/testimonia/` (18 files + `editions/`), and this dossier. `texts/scholarship/Allen_2016_TheConversation_…txt` contains the wrong extraction (navigation text) and should be discarded in favour of `texts/allen2016_conversation_horse_dung_bacteria.txt`.

---

## 6. Items I could not verify (carried as "(unverified)" above)

The dates of Freshfield's journal articles (1886, 1899); Whitaker's own argument beyond the pass name; the Island for de Luc, Wickham & Cramer, Ellis, Law; Hesselmeyer's and Long's verdicts; Hyde 1935's argument; the Ocelum identification; the Graioceli's location; Timagenes as Ammianus' source; the Vicarello goblets' content; Barruol 1969's standing; Seibert's exact two-pass proposal; Hunt's Island; de Galbert's final col; Freeman's and Kötter's verdicts; Walbank's page numbers in Commentary I (taken from Lazenby's and MacDonald's notes); the Ammianus 15.10.11 variant (*dissolvit* / *in solidam solvit*); Bocquet's publication details; the precise figures in Peddie's tally (snippet garbled).