At Trebia Hannibal had withdrawn his light infantry once they had driven in their Roman counterparts, and sent them to support his cavalry, adding to the discomfiture of the already tired and outnumbered Roman horse. It is possible that he did the same at Cannae, for later in his narrative, Livy tells us that 'at the beginning of the battle he [Paullus] had been seriously wounded by a slingstone’, although this was not to stop him from continuing to lead his men. Polybius does not mention this and in fact says that, when Paullus left the defeated right wing and went to join the struggle in the centre, he was unwounded. Perhaps he meant that the consul had not suffered an incapacitating wound, which might reconcile the account with Livy’s, or he had simply not heard this tradition, but it is also possible that the story was a later invention intended to add to the already heroic character of Paullus. Ultimately we cannot say precisely what happened to the Punic light infantry, or indeed the Roman velites, after each screen had withdrawn behind their main lines, but it seems probable that they continued to act in support of the formed troops.26
These heavily decorated boxes from Florence show scenes from Roman legends. On the left can be seen the figure of Marcus Curtius riding a horse and fully equipped as a cavalryman. His equipment would not have been out of place amongst the Roman and Allied cavalry at Cannae.
What is clear is that very early in the battle, and certainly before the main lines of infantry had clashed, Hasdrubal led his Spanish and Gallic horsemen in a direct charge against the Roman cavalry on the right. Normally cavalry combats were fast moving and fluid affairs. In a charge, the faster a squadron went the more its formation dissolved as faster horses with better riders outstripped the lesser mounts and less skilled horsemen and, even more importantly, the minority of bolder soldiers naturally pushed ahead of the majority of more timid men. Deep formations, with cavalry as with infantry, made it harder for men to run away, but it was very difficult to keep in close formation as the speed increased. A successful charge, especially when the victors gave in to their natural exhilaration and pursued the fleeing enemy, resulted in scattered men and tired horses. If the victors were then themselves charged by a fresh and well formed enemy squadron then it was very likely that they in turn would flee. In most cases, as with cavalry encounters in the eighteenth and nineteenth century AD, it was comparatively rare for the two sides to cross swords in a prolonged melee, since usually one or the other wheeled and fled before contact. Charge and pursuit was frequently followed by flight until the enemy were in turn driven back by fresh reserves and the squadron could reform. Victory normally went to the side which kept in hand a formed reserve on fresh horses after all the opposing cavalry had been committed.
The fighting at Cannae did not conform to this pattern and Polybius tells us that there was none of the normal 'wheeling about and reforming facing the original direction'. According to Livy this was because the fighting occurred in such a confined space, between the river and the flank of the Roman infantry, making it impossible for either side to outflank the other. Instead the Carthaginians attacked head on into the Roman cavalry, and the ensuing combat was described by Polybius as 'barbaric', clearly in the more general sense of the word as particularly brutal and unsophisticated rather than implying tactics peculiar to the tribal peoples. He says that many men dismounted and fought on foot as infantry. Livy's account is similar and he claims that, once the two sides had met, horsemen began to drag their opponents bodily from their seats. This was not the first occasion in the Second Punic War where our sources claim that cavalrymen dismounted to fight on foot, as Polybius and Livy both state that many riders had done this at Ticinus. It used to be thought that the ancient cavalryman's lack of stirrups gave him the most precarious of seats and as a result made him likely to fall off if he engaged actively in hand-to-hand combat. Recent trials with reconstructions of the four-horned saddle, probably already in use with all the horsemen at Cannae apart from the Numidians, have shown that in fact this provided a very secure seat and allowed a rider to deliver a range of blows, leaning to either side without losing his balance. The horsemen of this period were probably no more likely to fall off during a combat than cavalrymen equipped with stirrups. It was therefore not necessity that persuaded cavalrymen to fight on foot.27
Plutarch tells the story that Paullus' horse was wounded and the consul forced to dismount. His staff quickly followed suit and then all the rest of the Roman cavalry, assuming that this was a general order, also got off their horses. Seeing this, Hannibal is supposed to have said that the Romans might just as well have handed themselves over to him in chains like captives. It is not altogether clear to which phase of the battle this anecdote refers, although most probably it concerns the initial fighting on the Roman right wing. Livy tells much the same anecdote, but in this version it was weakness due to his own wounds which prevented the consul from staying on his horse and made both him and his cavalry bodyguard dismount. This occurs much later in the battle and assumes that some at least of the Roman cavalry were not swept away in the rout of the right wing and continued to follow the consul. Appian mentions another tradition which had Paullus dismounting near the end of the battle to fight to the death on foot with a group of survivors. The reliability of any of these stories is very difficult to judge and the tale of the Roman cavalry accidentally putting themselves at a disadvantage by dismounting may simply have been a Roman invention to excuse their defeat.28
These stories imply that the decision of some Roman cavalry to fight dismounted was either a mistake or a sign of desperation. As Hannibal's supposed comments make clear, it made little sense for cavalry to give up the mobility which was their chief advantage. What is clear from our sources is that the fighting between the Roman and Carthaginian cavalry was especially fierce and far less fluid than most cavalry combats. The Greeks and Romans associated determined, static fighting with foot rather than horse, and it is just possible that an account stating that the combat was more like an infantry than a cavalry melee is the source of these passages. Yet all our sources imply that the Roman cavalry did not advance any significant distance to meet the oncoming Gallic and Spanish horse and that their posture was essentially defensive. Cavalry have never been well suited to holding ground, for their advantages lie in speed and mobility. Unless very densely packed indeed and especially determined, a stationary mass of cavalry was always inclined to stampede to the rear when charged by enemy horse. There were many occasions in the ancient world when blocks of infantry were interspersed with cavalry squadrons. The foot provided firepower and, even more importantly, solid shelter for retreating squadrons to rally behind. It may be that at Cannae, some or all of the Roman horsemen were dismounted to act in this way or perhaps some detachments of ordinary infantry were interspersed with the cavalry squadrons, giving a stability to the wing which cavalry on their own would have lacked.29 The Romans needed their cavalry wings to stay in place for long enough to prevent the Punic cavalry from threatening the main assault by the legions in the centre. The Carthaginian cavalry was known to be better than their own horse and, as the enemy deployed, it must have been clear that the Romans were heavily outnumbered on this wing.
The reliefs on the Arch of Orange in southern France depict a fierce battle between Romans and Gallic tribesmen. Although dating to more than 200 years after Cannae, and showing the Roman army equipment of the later period, the Gauls shown here would not have been out of place in the Second Punic War. This whirling combat may give some idea of the fighting at Cannae, where Gauls formed the majority of Hannibal’s heavy cavalry and a considerable proportion of his infantry.
Mixing mounted with dismounted men offered the prospect of delaying a defeat which probably seemed inevitable, so that it could not affect the eventual outcome of the battle. If this was the consuls' plan, then it failed.
Livy says of the combat on the wing that ‘the fight was more fierce than of long duration, and the battered Roman cavalrymen turned their backs and fled'. It is always difficult to know what to make of such vague and relative statements of time, but the Romans appear to have broken not long after the infantry centres clashed. The Roman plan required their cavalry to hold out for as long as possible so that their overwhelming assault would have time to smash through the Punic centre.
Conversely Hannibal needed his left wing, where he had stationed at least two thirds of his mounted men, to rout their opponents as swiftly as possible and then return to the attack.
It was Hasdrubal rather than Paullus who was best able to achieve his objective and considerable credit must go to this officer for leading his men in such a furious charge. The Carthaginians had a great numerical superiority, somewhere between two and three to one, but were probably prevented from gaining much advantage from this due to the confined space. More importantly they were better motivated and more confident than their opponents, for throughout the early years of the Second Punic War Roman cavalry had lost virtually all the engagements which they had fought. Included in the cavalry at Cannae were almost certainly some of the survivors of the routs at Ticinus and Trebia and it is more than likely that the Roman horse had simply accustomed themselves to the idea of losing to their Carthaginian opponents. Roman horsemen were recruited from the wealthiest classes and it may be that the accounts of their determined resistance were exaggerated to please these influential citizens, something taken to an extreme by Appian who tells of them hurling back several assaults.30
The second phase of the battle. After a stiff fight. Hasdrubal routed the Roman cavalry and pursued them for a short distance along the river. The consul Paullus escaped and rode to join the heavy infantry in the centre. These had moved forward and come into contact with the advanced enemy centre. There the Gauls and Spanish were fighting hard to hold back the great columns of Roman infantry, both sides being urged on by their commanders.
For whatever reasons, the Roman right wing gave way and dissolved into rout. Hasdrubal's men pursued them, cutting down many of the fugitives, their flight made difficult by the shape of the river. Paullus, his staff, and perhaps some others went to join the infantry, but the majority of the Roman cavalrymen were killed or dispersed and would take no further part in the fighting. Hasdrubal exercised very tight control of his men. If. as suggested earlier, his troops had begun the battle in several lines, then only the first line may have actually been committed to the fighting and subsequent pursuit. The narrowness of the plain edged by the meandering river and the proximity of the smaller Roman camp may have helped to keep the pursuers together as well as hindering the Romans' escape. It was not long before the bulk of the Carthaginian left wing cavalry was re-formed and rested, ready to re-enter the battle.
On the opposite flank, where the Numidians faced Varro and the allied horse, the fighting had been far more tentative. These light horsemen fought in their traditional way, small groups closing to throw javelins, but rapidly retreating before the enemy could reach them with a charge. It is highly unlikely that these missiles caused significant casualties. Varro's men seem to have made little effort to drive the enemy back. Their role was simply to protect the infantry's flank and merely staying where they were achieved this satisfactorily. Livy claims that some disorder was occasioned at the very beginning of the battle when a group of 500 Numidians pretended to desert to the Romans. These men are supposed to have carried swords concealed under their body armour (which in fact the Numidians rarely, if ever, wore) and, once they were behind the Roman lines, had suddenly attacked them from the rear.
In Appian it is a group of 500 Celtiberians who employ the same ruse. It is unlikely that either story is true, but such tales of Punic treachery may have been current even in the immediate aftermath of the battle as attempts to explain the Romans’ overwhelming defeat as anything other than their simply having been outfought.31