MHQ Comments
Winter 2009

Please send comments, which may be edited for length and clarity, with name and address to

In “Can the Counters Be Counted On?” (Autumn 2008), Adrian Goldsworthy defends the accuracy of the casualty figures reported by ancient historians, but when it comes to the numbers reported by Polybius for the battle of Cannae, he falls into a common error. It is not the 70,000-casualty figure of Polybius that is in error but the numbers given for the Roman forces, especially the cavalry.

After defeats at the Ticinus River, at Trebia, and at Lake Trasimenus, the Romans raised a massive army: eight legions, matched by an equal number from their Italian allies. They increased each legion from 4,000 to 5,000 soldiers, bringing the Roman infantry total to 80,000.

Since the Romans attributed their defeats to the advantage of the Carthaginian horse, it stands to reason that the Romans also increased the equestrians accompanying each legion, probably from 200 to at least 300 or even 400. Polybius states that “on occasions of exceptional gravity” the Romans increased the equestrian number to 300, and allied cavalry were required to be “three times as numerous as the Roman.” If we use the figure of 400 per Roman legion (or a total of 3,200 for the eight legions raised for the battle), and triple that number, counting the strength for each allied legion as 1,200 (giving a total of 9,600), we arrive at a combined cavalry of 12,800. The Romans not only outnumbered Hannibal’s 40,000 infantry two to one but also had more cavalry (Hannibal’s numbering 10,000). Even if we accept only an increment of 300, this would give us 2,400 (300 x 8) Roman and 7,200 (900 x 8) allied cavalry, or 9,600. The Romans had at least as many horsemen as Hannibal.

Taking these revised figures into account, Polybius’ cavalry numbers seem quite plausible. In view of the overall greater reliability of Polybius over Livy, it seems reasonable to accept that 70,000 Romans and a little over 5,000 of Hannibal’s men died on the plain of Cannae on August 2, 216 BC.

—Yozan Dirk Mosig, University of Nebraska, Kearney

Adrian Goldsworthy replies:
It is clear that either Polybius’ figures for the size of the Roman army or for the casualties they suffered are wrong, since they contradict each other. Absolute certainty is impossible, but the most natural reading is to see the casualty figures as a mistake. Polybius’ subsequent comment that Hannibal’s victory demonstrated that “in times of war it is better to give battle with half as many infantry as the enemy and an overwhelming force of cavalry than to be in all respects his equal” does show he believed the Roman horsemen to have been heavily outnumbered.

It is also worth noting that none of the other sources argue for such large numbers of Roman cavalry as suggested. Finally, the practicalities of raising such a large cavalry force must be considered. Roman and allied horsemen were drawn from a limited pool based on social and economic status. This class had already suffered significant casualties in the earlier encounters of the war and especially in 217.

An even greater restriction was the availability of mounts. Losses in the earlier battles and simple attrition during the campaign would have reduced the number of available cavalry horses even more significantly than the number of cavalrymen. These would be much harder to replace and train at short notice.

All in all, it is easier to believe that Polybius either made a mistake about the casualty figures or that the numbers subsequently became corrupted. The suggestion that the Romans wanted to conceal the scale of the disaster is possible, but if so then they did this to a very limited degree. Admitting losses of 50,000 men and some 18,000 to 19,000 prisoners still produced quite staggering totals.