We do not know how long it took for the two armies to march from their camps and deploy for battle, but at the very least it must have taken several hours. Throughout this process, each army's officers, especially the Roman tribunes who seem to have had a particular responsibility for overseeing the army's deployment, needed to be very active, closely regulating the columns and then ensuring that each unit ended up in the right place and correct formation. At the end of this process, something like 126,000 men and at least 16,000 horses were packed into a few square kilometres of the narrow plain between the Aufidius and the high ground near Cannae. In summer the Apulian soil is dry and the tread of so many feet and hoofs must have thrown up great clouds of fine, sandy-coloured dust to be whipped around by the sudden gusts of the Volturnus wind. A fragment of the Roman poet Ennius, who composed his epic verse history of Rome not long after the Second Punic War, appears to refer to the dust of Cannae.22
Each army marshalled its line behind a screen of light infantrymen and it was these troops who opened the fighting, closing to skirmish with each other. Javelins could be thrown perhaps as far as 30-40m, although their effective range is likely to have been less. Slings and bows - and there may have been a few archers at Cannae though none are specifically attested - had a range of nearer 200m, but it is much more difficult to estimate their effective range. The distance and accuracy of fire was determined far more by the skill of the individual slinger or archer than by the technological limitations of his weapon. Unlike firearms, where the missile is projected by chemical energy, a sling or bow transfers the physical strength of the operator to its projectile. Skirmish combats in this period were conducted at ranges of less than a few hundred metres and usually considerably closer. Most battles in the classical world began with such encounters, but these were very rarely described in any detail in our sources. Cannae is no exception, and we are simply told that the light infantry screens met without either side winning a significant advantage. In ideal circumstances skirmishers were supposed to drive back their opposite numbers and then begin to weaken the enemy's main line, but such successes were exceptionally rare. Even those close order troops who lacked body armour or helmets usually carried large shields which gave very good protection against thrown javelins, arrows or sling stones. It was also extremely dangerous for the light troops to get too close to a formed line for they were highly vulnerable to a sudden charge, especially if unsupported by close order infantry or cavalry of their own.23
It is improbable that many casualties were inflicted on either side during combats between skirmishers. Thrown and shot missiles could be delivered with accuracy and some force, but were also highly visible in flight - this was also true, though to a lesser extent, of sling bullets - and therefore comparatively easy for the target to dodge or catch on a shield. Skirmishers operated in a very loose order, with wide gaps between men to ensure that they could easily move to avoid an incoming missile. Even if a man was wounded, and the vast majority of injuries caused by such missiles would not have been fatal, then the distances involved usually ensured that he could be carried away to the rear by his comrades. Skirmish fights seem to have been able to go on for several hours, or even all day, with very few men on either side being killed and no clear result. This is a little difficult for us to imagine, although very similar to some of the long range musketry duels of the eighteenth and nineteenth centuries AD. Modern studies suggest that relatively few soldiers, even in the best trained units, actively aim at and seek to kill the enemy in combat, most firing their weapons wildly and some not even firing at all. Certainly the ratio between the number of rounds fired and the number of casualties inflicted on the enemy in the well documented combats of the last few centuries has been staggeringly low, usually at least several hundred to one. It is unlikely that, in the pressure of combat when the target was firing back, the archers, slingers and javelinmen of the ancient world did much better. The loose and fluid formation employed by skirmishers, where there were no set places in rank or file and each man was allowed great freedom to advance and retire at will, made it difficult to force men to fight properly, for it prevented an officer and his comrades from knowing precisely what a man was doing. A minority of soldiers went close to the enemy and sought to make use of their weapons as effectively as possible, inflicting most, if not all, of the casualties. The majority did enough to appear eager, periodically going forward to perhaps within extreme range of the enemy and throwing or shooting a missile, but being more concerned to avoid being hit themselves than to harm the enemy. A minority probably stayed as far in the rear as possible, rarely if ever coming within range. The tentative nature of the fighting between the scattered skirmishers and the ease of avoiding missiles whose flight was readily visible help to explain the indecisiveness and low number of casualties in such encounters.24
According to our estimates for the size of the armies at Cannae, Hannibal had at least 8,000 light infantryman and the Romans perhaps as many as 20,000. It is distinctly possible that the first figure is too low and the second too high, but even so the Romans ought to have had a significant numerical advantage and we need to ask why this does not seem to have brought them more success. One reason might be that when, as discussed above, only a minority of soldiers fought effectively sheer numbers were not of decisive importance. Another possibility is that the battlefield was too small for so many loose order troops to deploy and made it impossible for the numerically superior Romans to outflank their opponents. Probably the most important reason was the greatly superior quality of Hannibal's light infantry. These included the renowned Balearic slingers, Spanish caetrati (warriors with light equipment and the small round shields from which they derived their name), and probably, Libyans and Numidians. The combination of slings and javelins made the Punic skirmishers effective at both long and short range and they seem to have been well trained, specialist troops. In contrast the Roman velites consisted of those too young to fight with the heavy infantry or too poor to afford the necessary equipment. Nearly all were armed with javelins, although it is just possible that there was also a small contingent of archers, but they do not appear to have received much training for their role. At Telamon in 225 BC the velites had performed very well, although significantly they were not opposed by many enemy light troops on this occasion. In the early second century BC Roman velites proved themselves markedly superior to the skirmishers in eastern armies, displaying a notable willingness to close and fight hand to hand. There was no trace of similar aggression in the early years of the Second Punic War. Some have suggested that it was only after the legions changed the equipment and training of their light infantry in a major reform in 211 that these began to become effective troops, but there is no good evidence for this. Far more probably it was a case of the greater experience derived from service in the war with Carthage which produced the high quality velites of the early second century, and we should note that in every respect the legions fielded in these years were far better than their predecessors. In 216 the Roman light infantry were mostly inexperienced and had received little or no training.25
The first phase of the battle. Skirmishers ran forward from the main lines to fight a desultory skirmish in the centre. In spite of a significant numerical superiority, the Romans failed to gain any advantage. On the Punic left, nearest the river, Hasdrubal led his massed squadrons of Gallic and Spanish cavalry in a ferocious charge against the Roman cavalry.