APPENDIX 1: Numbers


1. The March to Italy


Detachments and Losses:


Spring 218 setting out from New Carthage

12,000 cavalry & 90,000 infantry + 37 elephants

Polybius 3. 35. 1 & Livy 21. 23. 1 Appian Hariri. 4

Summer 218 left in Eastern Spain

1000 cavalry and 10,000 infantry at least 11,000 other less reliable troops sent home

Polybius 3. 35. 5-6 Livy 21. 23. 3-6

Summer 218 after crossing the Pyrenees

9,000 cavalry and 50,000 infantry

Polybius 3. 35. 7-8

Summer 218 after crossing the River Rhone

8,000 cavalry and 38,000 infantry

Polybius 3. 60. 5

Late autumn 218 in Northern Italy after crossing the Alps

6,000 cavalry (mixed Numidian and Spanish) and 20,000 infantry (12,000 Libyan and 8,000 Spanish)

Polybius 3. 56. 4

Of these figures Polybius specifically attributes the last set to an inscription erected by Hannibal on the Lacinian Peninsula (3. 56. 4). He does not appear to have derived the other numbers from such a reliable source, and this has sometimes led to these being questioned by historians.1 Ultimately it is impossible to know whether the numbers given for the early stages of the expedition are accurate or not. If they are correct, then Hannibal detached around 22,000 men before leaving Spain and lost around 4,000 cavalry and 50,000 infantry through combat, desertion, disease and attrition in the five months it took him to march from New Carthage to Italy. The losses may be broken down for each stage of the journey as follows:

1. New Carthage to the Pyrenees 21,000 men

2. Pyrenees to the Rhone 13,000 men

3. Rhone to Italy 20,000 men

Polybius comments that Hannibal had lost nearly half of his army in the last phase of the march, mostly in crossing the Alps (3. 60. 5). Losses amongst baggage animals had been proportionally even higher (3.56. 2). However, it is noticeable that throughout the march his cavalry had suffered a lower percentage loss than his infantry, 50% compared to 88%. This is surprising, since horses will usually break down before men. The cavalry were very much the elite of Hannibal's army and it is probable that he took particular care of them. This, perhaps along with higher morale, may explain this marked difference.

Whether or not these figures are correct, all of our sources believed that Hannibal’s army suffered very heavily on the march to Italy, especially during the passage of the Alps. Probably the bulk of such losses were as a result of physical weakness or disease which made it impossible for men to keep pace with the column, or through desertion. If Hannibal’s army was initially as large as Polybius believed, then the overwhelming bulk of its manpower most probably consisted of recently recruited Spaniards. Tribal warfare did not require the same stamina as such a long march and probably did little to prepare warriors for its rigours.

Polybius provides us with the most plausible estimate of Hannibal's strength and losses. However, Livy says that there was a very wide range of numbers given by his sources. Lucius Cincius Alimentus was one of Rome’s first historians, a senator who fought in these campaigns and was at one stage taken prisoner by the Carthaginians. He claimed that he had heard Hannibal say that he had lost 36,000 men and an enormous number of horses and baggage animals after crossing the Rhone. Alimentus estimated that Hannibal had 10,000 cavalry and 80,000 infantry on arrival in Italy. However, as Livy points out, this figure is of little value for it included the many Gallic and Ligurian tribesmen who would rally to Hannibal's cause after his crossing of the Alps (21. 38. 2-5).

2. The Campaigns in Italy, November 218-spring 216 BC

(a) The battle of Trebia, December 218 BC:

(i) Overall total - 10,000 cavalry (6,000 Spanish and Numidian + 4,000 Gauls) 28,000 infantry (8,000 skirmishers and 20,000 close order infantry of whom at least 12,000 were African, 8,000 Spanish and 8,000 Gauls).

Sources - Polybius 3. 72. 7-9, Livy 21. 55. 2-4.

(ii) Losses - unspecified, but relatively light amongst the Africans and Spanish, and heavier amongst the Gauls. However in the winter months many men and horses died along with all of the army's elephants (Polybius 3. 74. 10-11).

(b) The battle of Lake Trasimene, June 217 BC:

(i) Overall total - unspecified, but it is clear that Hannibal had been joined by a significant number of Gallic tribesmen before he left Cisalpine Gaul. He must have had at least the 50,000 men present at Cannae since he received no reinforcements before that battle.

(ii) Losses - 1,500, mostly Gauls and 30 senior officers (Polybius 3. 85. 5). 2,500 in the battle and 'many' subsequently of their wounds (Livy 22. 7. 3).

(c) Gerunium, autumn 217 BC (i) Overall total - not stated

(ii) Casualties - some of Livy's sources claimed that 6,000 Carthaginians were killed (22. 24. 14). Polybius says vaguely that many were killed, but also claims that the greater part of Hannibal's army was not present (3. 102. 8). It is probable that the figure of 6,000 is hugely exaggerated.

(d) The battle of Cannae, August 216 BC:

(i) Overall total - 10,000 cavalry (maximum of 6,000 Numidians and Spanish, and the remainder Gauls) 40,000 infantry (perhaps 8,000 skirmishers and 32,000 close order foot: absolute maximum of 12,000 Africans and 8,000 Spanish (and probably fewer) and the remainder Gauls Sources - Polybius 3. 114. 5, Livy 22. 46. 6.

B: The Roman Army at Cannae

1. Numbers:

(i) Polybius' version (3. 107. 9-15, 113. 5):

Eight legions each of 5,000 infantry and 300 cavalry, supported by the same number of allied foot and more cavalry.

Total = 80,000 infantry and 6,000 cavalry (2,400 Roman, if all legions were at full strength, and the remainder supplied by Latin and Italian allies).

(ii) Livy's alternatives (22. 36. 1—S):

(a) 10,000 new soldiers enlisted as replacements - total c. 50,000-55,000.

(b) Four new legions formed to add to the four already at Gerunium (each legion either 4,000 foot and 200 horse or 4,200 foot and 200 horse)- total c. 64,000-67,200 infantry and 4,800 cavalry.

(c) A variation on (b) - four exceptionally strong legions enrolled, consisting of 5,000 foot and 300 horse, and supplements sent to bring the existing legions up to the same strength. Twice as many cavalry and an equal number of infantry also supplied by the allies - total c. 80,000 infantry and 7,200 cavalry.

Livy's narrative clearly assumes that estimate (c) was correct, but some scholars have preferred the lower estimate. Brunt stated that the lower estimate is to be preferred '... because the success of Hannibal's tactics at Cannae is unintelligible if the Roman forces outnumbered his own by two to one’. As we have seen in the main text, the opposite is true, for the deep formation adopted by the Roman centre makes no sense if the Romans had roughly the same number of infantry as the enemy. ^

2. Identity of the legions in 216 BC

Assuming that Polybius was correct and there were eight legions at Cannae, half had been raised in late 217 or early 216 and the other four were the troops formerly commanded by the dictator. These consisted of:

1. The two legions formed by Fabius Maximus (Livy 22. 11. 2-3) in 217.

2. The two legions formerly commanded by Servilius Geminus. The cavalry of this army had been wiped out in Centenius' disaster in the days after Trasimene. Geminus had taken command in March 217 of half the army which had re-formed after Trebia. Flaminius took over the two legions commanded by Sempronius Longus and Servilius Geminus took those of Scipio (Livy 21. 63. 1, Appian Harm. 8). One of these legions had been stationed in Cisalpine Gaul in 218 under the command of the praetor Lucius Manlius Vulso. The other had originally been raised for Scipio's expedition to Spain, but was sent instead to Cisalpine Gaul when the Boii rebelled.

1 For discussion see J. Lazenby, Hannibal's War (Warminster, 1978), pp. 33-48, H. Delbriick (trans. W. Renfroe), History of the Art of War l (Nebraska, 1975), pp. 357-362, B. Caven, The Punic Wars (London, 1980), pp. 105-6, and J. Peddie, Hannibal's War (Gloucestershire, 1997), pp. 100-108.

2 For a discussion see F. Walbank, Polybius 1 (Oxford, 1970), pp. 439-440, Lazenby (1978), pp. 75-6 and Delbriick (1971), pp. 325-7. G. de Sanctis, Storia dei Romani vol. 3 (Turin-Florence, 1953), ii pp. 131-5 and P. Brunt, Italian Manpower (Oxford, 1971), pp. 418-419 argue for the lower figure.

If you find an error or have any questions, please email us at Thank you!