ON 2 AUGUST 216, Hannibal won his greatest victory in the plain (north of the small, hilltop town of Cannae in southern Italy. By the end of the day his outnumbered mercenaries had enveloped and massacred the greater part of the largest army Rome had ever fielded, turning this into one of the bloodiest battles ever fought, rivalling even the industrialized slaughter of the twentieth century. For the Romans Cannae became the yardstick by which other defeats were measured, never surpassed and only once or twice equalled in the next six centuries. Cannae remains one of the most famous battles ever fought, frequentiy alluded to in modern military writing, and Hannibal's tactics are still taught in the military academies where today's officers are trained. The UN commander in the Gulf War, General Norman Schwartzkopf, claimed to have employed principles based on study of Hannibal's campaigns and Cannae in particular in the planning and control of his own brief and devastatingly effective campaign. In the nineteenth and early twentieth centuries the disciplines of Ancient History and Archaeology were overwhelmingly dominated by German scholarship, and it was perhaps a reflection of this that the study of ancient warfare was taken very seriously by the Prussian and German military. Von Schlieffen, the architect of the plan for the invasion of France in AD 1914, was obsessed with Cannae, studying it in incredible detail throughout his life, and attempting in his war plan to achieve just such a total victory. Cannae became the shorthand term for a complete success for many German generals. In AD 1941, as Rommel drove the British army back towards Tobruk, he wrote in his diary that 'a new Cannae is being prepared', and a year or so later in December 1942 during the Stalingrad campaign, the commander of the 6th Panzer Division produced a boastful report of a successful day's fighting around the obscure village of Pakhlebin, calling the engagement 'the Cannae of Pakhlebin'.1 Yet in spite of having fought 'the perfect battle' Hannibal ultimately lost the war and, having achieved no other victories as great as this, but still undefeated in battle, was forced to evacuate Italy twelve years later. How and why the Romans were able to survive this disaster is the theme of this chapter.
If we are to understand the Cannae campaign and the battle itself, we must constantly remind ourselves that at the time no one could have guessed at its outcome and that, even during the battle itself, there were several stages when things might have turned against Hannibal for all the brilliance of his tactics. Livy's account of the preliminaries to the battle is dominated by a sense of impending disaster, as once again an impulsive Roman commander recklessly led the army to defeat, ignoring the advice of his more experienced colleague. The sense of inevitable catastrophe pervading the narrative is utterly false.
The Roman Senate had resolved on mounting a major effort for the campaigning season in 216. The magistrates for the year were a distinguished group, with one of the consuls and three out of four praetors all having held the consulship before, whilst the other consul and all the praetors had held the praetorship in an earlier year. For the first time ever each consul was given a double-sized army of four legions which were expected to fight together. In 225 the consuls had also each commanded four legions, but the participation of both armies in the battle of Telamon had been coincidental. The legions themselves were each to be larger than normal, expanding to the size the Romans felt appropriate for the current crisis, so that each mustered 5,000 infantry as well as the usual complement of 300 horsemen. We do not know the size of the allied alae attached to each army, but can safely assume that their infantry were roughly equal in number to the Roman foot, whilst their cavalry was more numerous. Sometimes the unprecedentedly large size of this army has been doubted, especially since Livy mentions that there were a variety of traditions about the number of troops enrolled in this year. However, Polybius clearly believed that there were eight legions in the united army of the two consuls and there is no good reason to doubt him. This was not the only Roman army to be fielded in this year. In addition to the forces in Spain and Sicily, an army of two legions was sent north to face the tribes of Cisalpine Gaul which remained in open rebellion. This expedition was commanded by Lucius Postumius Albinus, who had been twice consul in 234 and 229 and was now probably nearly 60 years old.2
The incoming consuls given command of this, the largest army Rome had ever put into the field, were Caius Terentius Varro and Lucius Aemilius Paullus. The latter was the grandfather of Scipio Aemilianus and therefore receives a very favourable treatment from Polybius and all subsequent historians. This was his second consulship, as in 219 he had fought successfully against the Illyrians and celebrated a triumph, although he seems to have been involved in the scandal associated with the campaign which had led to the retirement from politics of his colleague, Marcus Livius Salinator. Paullus was to fall in battle, but, unlike Flaminius, he was a member of a wealthy and well-established aristocratic family, who were more than capable of defending his reputation in later years. A convenient scapegoat was found in the person of his colleague, who survived the fighting, but was a new man and vulnerable to the propaganda of such powerful families. Varro's descendants continued to gain membership of the Senate, but none had distinguished careers and the family never gained admittance to the core of senatorial families who dominated the senior magistracies till the end of the Republic.
Livy portrays Varro as a demagogue much like Flaminius. This conforms with his theme that it was the radical, popular politicians who caused most of the disasters to befall the state, when the mob ignored the wise leadership of the experienced aristocrats in the Senate. He tells us that Varro's family was undistinguished, that it was said his father was a butcher. Such an accusation is typical of the exaggerated invective that was a normal feature of Roman political debate and should not be taken seriously. Varro is supposed to have been one of the main supporters of Metilius' bill to grant Minucius equal power to Fabius Maximus the year before, but otherwise, even by Livy's account, his career had not been a radical one. As with Flaminius, he must have had considerable support from the wealthier classes in the Comitia Centuriata to have won election to the consulship. Livy even claims that he was the sole choice of the Assembly at the election, which strongly attests to his popularity, and that he actually presided over the vote to appoint his colleague. His success also makes it certain that he possessed considerable support amongst his fellow senators and, given that the presiding magistrate could do much to influence the outcome of an election, it is unlikely that he and his fellow consul were hostile to each other. There is no good reason to accept Livy's depiction of Aemilius Paullus as an adherent of Fabius' strategy of avoiding battle. It is not even certain that Fabius himself believed this to be the right way to proceed in the spring of 216, and that he continued this strategy in the years after Cannae does not necessarily mean that he advocated it before this disaster.3
Even if Fabius still advocated his policy of delay, then the plans of the Senate for the campaigning season of 216 make it clear that he was in a minority, and that they expected a direct confrontation with Hannibal. Fabius' six-month dictatorship had given the State time to recover from the defeats at Trebia and Trasimene. His army of four legions was relatively well-trained and had won some minor successes, even if part of it had also been defeated under Minucius. To this force they added four new and less experienced legions, so that around 80,000 infantry and 6,000 cavalry would face Hannibal's 40,000 foot and 10,000 horse. A continuation of Fabius' strategy did not require such a large force and concentrating so many soldiers and mounts greatly increased the problems of supply. Morale was good both in the population as a whole and amongst the soldiers, who were eager for battle. Many senators and their sons were serving either as tribunes, in the cavalry or on the commanders' staffs. One of the tribunes was the former Master of Horse, Minucius. The allied soldiers in particular were desperate to fight and avenge the devastation Hannibal's progress had wrought on Italian fields. Polybius claims that the Senate told Paullus to seek battle and attributes a speech to him in which he explained to the soldiers the reasons for Hannibal's recent victories and assured them that the enemy would not be able to stand up to the combined might of both consular armies. In Polybius' version, Varro and Paullus did not disagree over whether or not to fight a battle, but instead over when and where to do so.4
At the opening of the campaigning season in 216, Hannibal was still at his winter base at Gerunium in Apulia, watched warily by the army commanded by Geminus and Regulus. Both of these men had their power extended as proconsuls for this year, but it is not clear whether Regulus remained with the army for the coming campaign. Livy claims that he gained permission to return to Roman on the grounds of age and infirmity, and certainly Polybius' statement that he was killed is untrue, since he became censor in 214. Once the crops had ripened enough to make foraging possible, Hannibal led his army south. The proconsuls followed him at a safe distance and sent repeated messages to the Senate asking for instructions, explaining that they could not close with the enemy without being forced to fight a battle. This was plausible enough, since the country the armies were marching through was open and fairly flat. Hannibal pressed on and captured the ruined stronghold of Cannae, which was still in use as a Roman supply depot.5
It is not certain when the new consuls arrived and the two parts of the army were joined together. Polybius implies that this was not until after Hannibal had taken Cannae, less than a week before the battle, whereas Livy tells of their arrival before he had left Gerunium. Polybius' account is probably to be preferred, not only because of the implausibility of much of Livy's account of Hannibal's withdrawal, but also because the difficulties of feeding such a large army make it unlikely that the Roman force remained concentrated for such a long period. Further uncertainty exists over the precise location of the battlefield, chiefly with regard to which side of the River Aufidius it was fought on. We do not know what the course of the river was in the third century, but it is clear that this differed from its present-day line. Although some authorities have placed the battle to the north - the term is used loosely, the river actually running from south-west to north-east - on what is normally referred to as the left bank, it is an easier reading of our best sources to locate the fighting on the south, or right bank, assuming that the river's course originally lay further away from the hillock of Cannae itself. Such a positioning makes the movements of both armies more intelligible and will be followed here.6
The Romans advanced carefully in pursuit of Hannibal, having learned from Flaminius' failure to scout properly before Trasimene. They seem to have come along the coastal plain, perhaps to avoid having to pass any suitable ambush positions. On the day they came into sight of the enemy army, they stopped and camped 6 miles away. The Romans were to the east on the flat and open plain which rims down - there is a very slight gradient -to the sea. The consuls held command of the army on alternate days, the normal practice on the rare occasions both consuls operated together, but one which Fabius had refused to adopt with Minucius in the previous year. Paullus is supposed to have advised against advancing directly on the enemy in this country which favoured Hannibal's more numerous cavalry, but the next day was Varro's turn to command and he decided to press on. As the Romans marched across the plain, the column was attacked by Hannibal's cavalry and light infantry and thrown into some disorder before the Romans formed up part of their forces to drive the enemy back. The Roman velites and cavalry fought with formed maniples of legionaries in close support, giving them an edge over the opposition. Skirmishing lasted till nightfall, and it is doubtful that the Romans had covered many miles by the time they made camp. The next day Paullus continued the advance and brought the army up to camp by the river bank, only a couple of miles from Hannibal's position.
Polybius claims that Paullus still did not like the ground, but felt that the armies were now too close together for the Romans to disengage safely. It may be that Polybius was simply trying to shift the blame for the defeat onto Varro, but the difficulty of withdrawing in the face of the enemy was genuine. The Romans would be vulnerable as they withdrew across the open plain and anyway such a retreat would have had a demoralizing effect on the soldiers. It should never be forgotten that the legions at this time were not composed of the highly disciplined, professional soldiers of later years, but were still a volunteer militia of citizens who looked forward to returning to civilian life as soon as the campaign was over and the threat to their State ended. At the moment the army was enthusiastic, confident in its own numbers and encouraged by the promises of its leaders and the victories in the skirmishes over the winter. If their commanders appeared to lack a belief in victory and decided to flee from the enemy who was freely devastating Roman and allied fields, then the army's spirit would begin to drop. Apart from the risk of lowering morale, there was another pressing reason for the army to seek an early battle. Feeding so many men was an immensely difficult and never-ending problem, made far worse by the loss of the supplies at Cannae. If the campaign were prolonged, then the two consuls would be forced to divide their forces in order to keep men and horses fed. Paullus divided his army and sent the smaller portion across the Aufidius into a separate camp with the express intention of protecting the foraging parties which were to be sent out on this side of the river.7
Hannibal faced similar problems, made far worse by his lack of any immediate source of supplies beyond what his soldiers could forage or capture. Livy claims that immediately prior to the battle the situation had got so bad that many of his mercenaries, and particularly the Spanish contingents, were contemplating desertion. Hannibal is even supposed to have considered making a run for Gaul with his senior officers and cavalry, although his desperation may simply be another attempt to emphasize the wisdom of Fabius' strategy if only it had been followed. The capture of Cannae was only a temporary solution to his army's requirements and with the Romans so close he could no longer risk spreading his army out into detachments to forage. Both sides thus needed to seek battle in the immediate future if supply problems were not to put pressure on them to retreat or disperse, both of which would put them in danger. Yet the size of the Roman army was daunting and one of Hannibal's officers, a certain Gisgo, is supposed to have commented on their superiority. The general is said to have looked solemn and then quipped that whilst there may be a lot of Romans over there, there is not one called Gisgo, prompting a burst of laughter, perhaps forced, nervous, sycophantic, or a mixture of all three, from his assembled staff.8
Several days followed during which the armies stared at each other and skirmished in the usual way. Though both sides were eager to fight a battle, neither wanted to provoke one until they felt ready. By this time Hannibal had moved on from the hilltop citadel of Cannae itself and crossed the river, camping on the same side as the larger Roman camp. The most likely location for his camp is the high ground on which the modern village of San Ferdinando di Puglia now lies. On the next day, 31 July, the Carthaginian army was ordered to prepare for battle, cleaning their armour and sharpening the blades of their weapons. On 1 August Hannibal's army marched out to deploy on the open plain in front of the ridge. This was Paullus' day of command, but his only move was to deploy strong covering forces in front of both of the Roman camps. Hannibal seems to have been content with giving his soldiers this demonstration of Roman timidity. Numidian light horsemen crossed the river and rode up to harass the slaves gathering water for the smaller camp. Paullus remained on the defensive and Hannibal made no further moves to force a battle.9
The Roman commander's reluctance to fight in the open plain is understandable given Hannibal's marked superiority in cavalry, but Polybius tells us that his soldiers resented his passive behaviour, a natural mixture of enthusiasm and nervousness making them long to get the anticipated battle over with. Varro is supposed to have been similarly roused by the sight of
Numidian cavalry riding up to the Roman camp, and when he took over command of the army on the next day, decided to give battle. However, he did not plan to do so under the same circumstances which his colleague had just refused, but across the river on the narrower plain north of Cannae itself. Livy claims that he issued the orders without even bothering to consult Paullus, but this is extremely unlikely and not claimed by Polybius. It may be that Paullus did not believe it wise to fight, but this would then make his willingness to close with the enemy in the first place rather strange, and it is distinctly probable that he agreed with Varro's decision. One scholar has even suggested the ingenious and attractive theory that the battle was actually fought on Paullus' day of command, but this is impossible to prove and it is safer to stick with our sources.10
Early on the morning of 2 August, Varro had the red vexillum, the square Roman flag carried by the consul's bodyguard, displayed outside his tent in the traditional signal for battle. It is probable that orders had been issued to the tribunes during the night, giving them time to prepare their men, for just after dawn the army began to march out of the larger camp and cross the river. Joining the troops from the smaller camp, the Romans formed a single line of battle with its right flank resting on the river. This vital position was held by the Roman cavalry who should have been 2,400 strong if all eight legions had their full complement of horsemen. The left flank, resting against the hill of Cannae itself, was held by the Latin and allied cavalry who composed the remainder of the army's 6,000 horse, and were thus probably around 3,600 strong. The alae were normally supposed to supply three times as many horsemen as the legions, but the lower proportion at Cannae may have been a result of the heavy losses suffered by Centenius' force the year before. However, it is equally possible that some of the legions were below strength in horsemen, in which case the allied contingent would have been larger.
The centre of the army was composed of its strongest component, the heavy infantry of the legions and alae. There were perhaps 55,000 heavy infantry supported by 15,000 velites, allowing for the contingents left out of the battle for various reasons. They were in the usual triplex acies, but with one major difference, for Polybius tells us that the maniples were placed closer together than usual, each one's depth 'many times' wider than its frontage. We do not know the precise dimensions of this formation and estimates have varied from a total depth for the three lines of between fifty and seventy ranks, giving a frontage for the centre of perhaps half a mile to a mile, with perhaps each maniple deploying five men abreast. There were several reasons for adopting this formation. The first was simply one of space, for the flat land between the hills and the river was narrow and would not have permitted all of the legions and alae to deploy in their normal, shallower formation, but given that the Romans had chosen to fight in this position they clearly did not believe this to be a major problem. The deeper, narrower formation allowed both the individual maniples and the army as a whole to move more quickly whilst the ranks kept their dressing, for the wider a formation is, the quicker it will fall into disorder as it marches across even the flattest ground. Although some of the Roman soldiers had been in service since 218, and a good number had experience from the previous year, more than half the army consisted of recent recruits whose standard of drill cannot have been high. Moreover, the entire army had had little or no time to train together, and none of the officers had experience of leading or serving in such an unprecedentedly large force. The formation adopted was simple enough to work with such material and was able to create tremendous forward pressure. Visually it was intimidating to any enemy in its path, whilst the Roman soldiers enjoyed the security of being surrounded by so many of their comrades. The deep formation would also make it harder for any of the soldiers to flee. The men in the front ranks would be unable to escape until the men behind them had given way and these were removed from the immediate risks and stress of combat. Once the Roman mass had begun its lumbering advance it would be difficult to stop. At the very least it ought to have far greater staying power than Hannibal's less numerous infantry. The price was a loss of flexibility, for the reduction in the gaps between the maniples made it virtually impossible for these to change formation or wheel to face another direction.
Varro placed himself at the head of the Latin cavalry, whilst Paullus commanded the Roman horse and the proconsul Servilius Geminus led the infantry centre. Paullus' position with the prestigious Roman cavalry has been used as evidence by those who believe that the battle was fought on his day of command, but it is actually uncertain whether there was a normal position for the supreme commander of an army. In the few battles where both consuls were present they do not appear to have placed themselves in any particular precedence, since traditionally they were not expected to fight together.11 The entire army must have occupied a frontage of between 1 and 2 miles and it is distinctly probable that it was angled back from the river, facing roughly south-west, to allow it to fit into a plain which was no more than 1/3 miles wide. The Roman plan was simple and based upon their experience in earlier battles. At Ticinus and Trebia the Roman cavalry had been outnumbered and outfought, allowing the enemy cavalry to outflank the entire army. Yet at Trebia the Roman infantry had broken through the enemy centre, whilst even in the disorganized fighting at Trasimene they had put up a strong resistance and the vanguard had actually smashed through the Punic line. The strengthened Roman centre should be able to repeat these successes and crush the Carthaginian centre. All that was required of the cavalry wings was for them to protect the flanks of the infantry long enough for them to win the battle in the centre. Terrain prevented the cavalry wings from being outflanked by their numerically superior enemy and the Roman tactics here were to be purely defensive, designed to hold their ground for as long as possible. It was probably for this reason that the two consuls held command of these critical positions, their presence intended to inspire the cavalry to stand against their more numerous foes. With his infantry beaten and scattered, Hannibal's army would be permanently defeated, even if his cavalry were eventually successful in their combat with their Roman counterparts. Varro's plan was not subtle, and nothing illustrates Polybius' earlier comment on the Romans' reliance on brute force better than their tactics at Cannae, but it might easily have worked and anything much more sophisticated would have been impossible with the army under his command. Hannibal had been brought to battle on ground of the Romans' choosing where they hoped to negate his cavalry superiority and could be sure that no ambush lay behind their lines. He no longer had the elephants which had panicked earlier Roman armies and now he would be crushed by the numbers and courage of Rome's greatest strength, her sturdy citizenry of farmer-soldiers.12
There was no guarantee that Hannibal would accept battle in the narrow plain. Paullus left 10,000 men to guard the larger camp which remained on the same side of the river as the enemy. It is not certain whether these were a whole unit, perhaps a legion with itsala, or detachments from several units. There is no good reason to believe that the entire triarii were given this task, since, contrary to some claims, this was not their normal role. Paullus is said to have ordered them to attack the Punic camp if Hannibal took the bait and crossed the river to fight. If this is true, then this was a bold plan, but characteristically Roman, and the capture of Hannibal's camp and baggage would make sure that the enemy had no chance of reforming his army to continue the struggle. At the very least it might compel the enemy to weaken their force by leaving a detachment to protect the camp while the rest of the army fought the main battle. In fact, the Punic commander quickly decided to accept the challenge to battle and does not seem to have made any special provision for the defence of his base.13
Hannibal responded quickly to the sight of the Romans crossing the river, which may suggest that his army was at least partially prepared to move out in any case. His slingers and javelin-men were sent across the river as a covering force to allow the remainder of his army to move out and deploy. The main body forded the Aufidius in two places, which suggests that they were formed into two columns. The army then wheeled into line facing the Romans, its left flank resting on the river. The 10,000 cavalry were divided between the wings, but in this case Hannibal put all of his Numidians opposite the Latin Horse and concentrated his close order cavalry, who rode with saddles and bridles, on the left. It is unclear just how many of each nationality were present, but at least 4,000 of the cavalry were Gallic and several thousand Spanish, so it is probable that the Punic left wing significantly outnumbered the Romans facing them. The Numidians may have been roughly equal in numbers to the Latins, but certainty is impossible. Hannibal had 40,000 infantry, but this total included the light infantry. He had had 8,000 of these at Trebia and it is doubtful whether his recruitment of Celts had added substantially to this total, for skirmishing was not common in Gallic warfare. This left 32,000 close order foot, of whom the majority were Celts, perhaps as many as 20,000, for he had received no more drafts of Libyans or Spanish. Possibly there were 8-10,000 Libyan infantry and around 4,000 Spanish.
Together the Spanish and Gallic infantry formed the army's centre, deployed in alternate companies. Polybius uses the word sperae, one of the terms he uses to mean 'maniple', and it is likely that he uses it to mean units of a few hundred men, although there was probably no standard size. This interspersion of companies from two distinct ethnic groups suggests that Gauls were now fully absorbed into Hannibal's army, so that there was no need to place them in larger, tribal contingents. The Libyans were split into two halves, each roughly the strength of a Roman legion, and placed on the wings, formed into deep columns. Although this is not clear from our source, it is probable that they were actually behind the edges of the line of Spanish and Celts, out of sight of the Romans. They may well have composed the second of Hannibal's columns, crossing the river upstream of the main force, concealed from enemy gaze. Once the army had reached its positions, Hannibal led forward the centre companies of his infantry, causing the whole line to bulge towards the enemy, the units echeloned back on either side of the new, narrow front. The general himself, with his brother Mago, was with the Gauls and Spanish, while Hasdrubal led the heavy cavalry, and Livy says that Maharbal controlled the Numidians. Hannibal had rightly guessed that the main Roman effort was to be made in the centre and had adjusted his deployment and issued orders accordingly. His plan was to use the enemy's own strength against him, but it is easier to describe how this was achieved than it was to anticipate the action.14
It must have taken hours for both the armies to reach their positions and deploy into battle, the tribunes scurrying about to join the two Roman armies together and jostle the men into place. When they were ready, over 125,000 men and 16,000 horses were gathered in an area no larger than 5 or 6 square miles, whilst more soldiers and tens of thousands of slaves, servants and camp followers looked on from the three camps. The noise of their movements drowned out the constant chirruping of the cicadas which fills the air on summer days in this plain. So many feet and hoofs threw up clouds of dust which swirled in the strong gusts of the hot Volturnus wind which blows from the south-east. The dust was another irritant to men beginning to swelter in their heavy armour under the hot glare of the sun. On the Roman side the army presented a fairly uniform appearance, although we must remember that these were citizen soldiers and there is no good reason to believe that they wore tunics of the same colour or that shields were painted with unit insignia. Our sources were most struck by the diverse dress of the enemy army. On the one hand were the Libyans, dressed in Roman helmets and armour, and with oval scuta, then the Gauls stripped to the waist (since this is probably what Polybius means by 'naked'), and the Spanish in their white tunics with purple borders, to which we might add the unarmoured Numidians with their distinctive hairstyles and riding their small, shaggy horses. It is uncertain how accurate this picture is. The Spanish had left home two years before and one may wonder how many still wore their native garb and had not replaced it with whatever was available locally or could be made in camp. However, they probably had retained their native weapons, the Spanish carrying their short-stabbing swords, and perhaps a few of the curved blades similar to the Greek kopis, whilst the Gauls had their long slashing blades.
For a while the armies stared at each other, whilst their light infantry skirmished between the lines. Neither side seems to have gained much advantage in this combat and eventually the skirmishers pulled back behind the main lines of their infantry. Hannibal's light troops may then have moved to support his cavalry on the wings as they did at Trebia, for early in the fighting Paullus was hit in the face by a slingstone cast by one of the Balearic slingers. The Roman velites seem to have pulled back through the small intervals left in the line of hastati. However, the first close combat occurred when Hasdrubal led his Spanish and Gallic cavalry against the Roman horse. A vicious melee developed, the sources once again stressing that it was unlike most cavalry combats, consisting not of charges and pursuits, but of a standing fight. Again we hear of men dismounting to fight on foot. Eventually the ferocity of Spanish and Gallic horsemen proved too much and the Romans were killed or put to flight. It is unclear how much advantage Hasdrubal had from his numbers, for the confined space between the infantry centres and the river may have prevented him from bringing them to bear. It may simply be that the Roman cavalry had got as used to being beaten by the Punic horsemen as the latter had to winning. In hand-to-hand combat confidence was often of greater importance than numbers or equipment. The Romans fled, but many found that their escape was cut off by the river and were slaughtered by their exultant opponents. The combat had been fierce, but according to Livy had not lasted very long, although it is always difficult to know what to make of such vague allusions to time. Before it was finished the heavy infantry had met in the centre.15
Hannibal's men do not seem to have advanced further once he had formed his convex line pointing towards the enemy, so it was probably the Romans who marched forward, eager as they were to decide the combat before their cavalry were beaten. The cacophony of noise can only have been appalling as the Romans cheered, blew their trumpets and clashed their weapons against their shields, the Celtic and Spanish warriors answering them with their own war cries as each side tried to terrify the other into submission. As they came closer, the Roman line checked and began to hurl their pila, the enemy replying with showers of their own javelins. Despite their numbers the Romans did not throw many more missiles than their opponents, for the men in the rear ranks even of the maniples of hastati could not do so without severe risk of hitting their own front ranks. Soon the Romans, encouraged by their officers and the men behind, surged forward into contact. The combat fell into the usual pattern, with brief flurries of savage hand-to-hand fighting, after which the exhausted participants pulled back a few yards to draw breath, taunting and lobbing missiles at their enemies, until they regained the confidence and energy to renew the fight. Livy speaks of the Romans 'for a long time repeatedly pushing forward', before they began to win ground against strong opposition.16 The Gauls were renowned for their ferocity in the early stages of battle, but supposed to weary quickly and lose heart if they did not seem to be winning. At Cannae, as at Telamon, they confounded the literary cliche of the fickle and easily tired barbarian, and put up a long and sturdy resistance. There were several reasons why they were able to do so. In numbers they were roughly equal to the Roman hastati and, since they occupied roughly the same frontage, their companies were formed in similar depth. The stiffening of experienced Spanish infantry may also have helped to steady the Gauls, and they were inspired by the presence of Hannibal and Mago, who rode around, close behind the righting line, yelling encouragement to their warriors. Pride probably had a lot to do with it, for both the Gauls and Spanish were products of warrior societies which prized military glory above all else. At Cannae these men had been specially chosen as the first to meet the enemy, even being advanced ahead of the main army where all could witness their valour, in a gesture not unlike that of the Gaesatae at Telamon, running naked ahead of the whole army, challenging the enemy and daring them to come on.
Only slowly did the Romans force the Celts and Spaniards back, and at first they did so step by step, still facing forward. The bulge in Hannibal's line was flattened, and still the Romans pushed on, till in the centre they drove the enemy back further, so that now the line was concave instead of convex. More of the front lines were in contact and the fighting general, but the main effort was still in the centre where the two sides had first met and where the Romans were winning. Roman officers, including the many tribunes, the proconsul, and Paullus himself who had ridden to the centre after the defeat of his cavalry, urged the legionaries on, led them in charges, and fed in maniples from the reserve lines to support the hastati, desperate to keep the forward momentum going and exploit this success. Gradually the Roman infantry lost their neat formation, as the narrow gaps between the maniples vanished and the units merged into one great crowd. The intervals between the three lines had probably also been reduced by the deep formation of the individual maniples. There was always a tendency for very large mass formations to lose order and degenerate into a mob of men pushing forward, (as Napoleon's army was to discover when the declining quality of its infantry led to the use of gigantic formations at Wagram, Albuera and Waterloo). Yet the forward pressure created by the densely packed mass of Roman infantry was inexorable and, eventually, the Gauls and Spanish began to break. In the centre they at last gave way, and the Romans surged forward, victory in sight. It was probably now that the Gauls suffered a good proportion of the many casualties, as those who did not run quickly enough or were slowed by their wounds were hacked down by the elated legionaries. The Roman mass burst through the centre of the enemy army and, in the rear, the Roman commanders urged more men on to support them.
On either side of the victorious Roman infantry were the columns of the Libyan infantry. We do not know whether Hannibal had given their commanders instructions to begin to move when the enemy reached a certain point, or whether he now sent orders by courier for them to do so. Calmly, the columns turned to face inwards, and although there has been considerable debate over precisely how this manoeuvre was performed, this need not concern us, since so much depends on the details of the formation they started the battle in, concerning which we have no precise information. Then, ranks neady dressed, they advanced to take the mass of Roman infantry from both sides. The disorganization amongst the Romans was appalling and no one was able to assemble a coherent fighting line to face these new threats. The maniples were hopelessly confused and the men turned as individuals and small groups to confront the advancing Libyans. Most of the Romans were weary from the fighting, since even those not actually in the front ranks had endured the stress of close combat, and now they faced men who were well formed and fresh. It is even possible that they did not immediately realize that these new troops were enemies, for the African soldiers were dressed in Roman equipment and in battle men often become disorientated and lose their sense of direction. All forward movement in the Roman centre ceased, the two bodies of Africans compressing the mass of soldiers like a vice. In the lull, the Gauls and Spanish who had broken began to rally and return to the fight.17
Varro must have watched the early stages of the battle with some satisfaction, as his infantry started to achieve the breakthrough which was to smash the enemy army. His own command was faced only by skirmishing Numidians, who never risked a charge and fled whenever the Latins advanced towards them. His men suffered a steady drain of casualties, but there was no reason for him to advance and drive the enemy back, since as long as he remained in place and protected the flank of the infantry the legions could perform their task and win the battle. It is doubtful that the consul could have seen the defeat of the Roman cavalry on the opposite wing, but even if Varro had knowledge of it, there was nothing that he could have done to prevent it. Various rumours circulated in the aftermath of the battle to explain the Roman defeat, and one of these was that a party of Numidians had pretended to surrender, only to produce swords which they had concealed on their persons and attack their captors in the rear, but Polybius does not mention this and it is most likely untrue.18
On the opposite flank, Hasdrubal had allowed his cavalry to pursue the fleeing Roman horse for a short distance along the river, but soon rallied them. It was always difficult to reform cavalrymen once they had begun to scatter in pursuit of a helpless enemy and it is a tribute to Hasdrubal's ability and the discipline of his men that he so quickly re-established order. The narrowness of the plain probably helped to keep the pursuers together and they were prevented from charging off too far into the distance by the smaller Roman camp, only a mile or so along its bank. The Spanish and Gallic cavalry then moved round behind the Roman army and prepared to charge into the rear of the Latin horse. Without waiting to receive them, Varro and his men fled in panic as soon as they perceived the threat. Such routs were not uncommon when a force was unexpectedly confronted by a new threat, but the ground may have added to the Latin horsemen's nervousness, for if they stayed to fight they would have been trapped between the Numidians, Hasdrubal's men, their own infantry and the steep slopes of the high ground around Cannae. Once again the commander of the Punic left wing displayed admirable control over his Celtic and Spanish warriors, halting them and leaving the pursuit of Varro's troopers entirely in the hands of the Numidians. The fact that his men had not actually made contact with the enemy probably made it easier to keep them in order. Hasdrubal wheeled them round and began a series of charges against the rear of the Roman infantry. Thetriarii may no longer have been a clearly distinct line, having been absorbed into the general mass, and anyway there were probably few senior officers in the rear of the army to organize resistance as most will have made their way forward to control the critical fighting against the Punic infantry. There was no question of a line of spearmen being able to turn around and ward off the approaching cavalry. In some places a dense group of men presented a wall of spear points to deter the oncoming horsemen, in which case they were bombarded with thrown javelins, but elsewhere the Punic horsemen were able to charge home into the panicked and disordered men.19
The Roman foot were now almost completely surrounded. Such was their disorder that they could make little use of their numbers, which were still greater than the enemy's. In the milling mass of men there were no formed reserves to be sent forward to reinforce a combat. Everywhere they were steadily driven back, pressing the crowd more closely together and adding to the confusion. Still the Romans fought on, although admittedly for many of them flight was impossible. This phase of the battle is passed over briefly by our sources, and often by modern commentators as well, since it is not a story of tactical brilliance, but of prolonged butchery. It must have taken hours for the Carthaginians to massacre their enemy. The pauses between the brief minutes of furious hand-to-hand combat doubtless grew longer as the Punic soldiers had to overcome their exhaustion before renewing the killing. For hours they pressed on, their shields and the chests of their horses stained red with blood, the edges of their swords blunted by so much killing. Hannibal lost 4,000 Gauls, 1,500 Spaniards and Libyans and 200 cavalry, a total of about 11.5 per cent of his entire army, still more if these figures only included the dead and need to be increased to include wounded. This was a staggeringly high loss for a victorious army in the ancient world and a testament to the long and ghastly struggle fought to destroy the surrounded Roman host.
Our sources give various figures for the Roman casualties. The normally reliable Polybius is obviously confused at this point, because his figures for their losses produce a total higher than the one he gives for the entire army at the beginning of the battle. Livy says that 45,500 Roman and allied infantry and 2,700 cavalrymen were killed and in this case his version seems more plausible. Some 3,000 foot and 1,500 horse were captured immediately, but to these we must add the roughly 17,000 men who surrendered in both the Roman camps by the next day, since only a small proportion of the fugitives who had fled to these were willing or able to fight their way to safety. The losses amongst senior officers had been especially bad. Paullus was killed, allegedly after refusing the offer of his horse from the tribune Cnaeus Lentulus, who had found the wounded consul sitting on a rock in the midst of a mob of fugitives. Geminus was dead, as were Minucius Rufus, both of the consuls' quaestors, and twenty-nine out of forty-eight military tribunes. In addition Livy says that eighty odier senators, or men due to be enrolled in the body at the next census, had also fallen.20
These figures need to be put into perspective. On 1 July AD 1916, the British army began its offensive on the Somme, suffering an appalling 60,000 casualties on this first day. It was a disaster which still haunts the national psyche, much as Cannae was to remain a powerful image to the Romans for the remainder of their history. In the popular mind the losses to the mostly volunteer army is often equated with 60,000 dead, but in fact out of the total of 61,816, there were 8,170 killed, 35,888 wounded and 17,758 listed as missing, 10,705 of whom were later found to have been killed. The French suffered even higher losses on the first day of Nivelle's offensive the following year. In each case these casualties were spread along a front many miles in length - on the Somme the British Expeditionary Force attacked along a 16-mile front.21 At Cannae, over 50,000 coipses lay heaped up in a few square miles of open plain. Livy's description of the appalling sights on the battlefield on the next day may owe much to his imagination, but does convey something of the horror. He speaks of ‘so many thousands of Romans, infantry and cavalry mingled', bloodstained men rising from amidst the slain only to be cut down by the Punic soldiers, others unable to walk, begging to be put out of their misery, some who had scraped holes in the ground to bury their heads and smother themselves; and he tells the story of a Numidian, pulled alive from underneath the body of a Roman soldier who, in his death throes, had bitten into the man's nose and ears.22
Polybius commented that the battle proved that it was better to fight a battle with half as many infantry as the enemy, but with a great superiority in cavalry than to fight with roughly equal numbers of both, but it must be emphasized that it was only through Hannibal's tactical skill that the victory had been possible.23 The Carthaginian general had exploited the diversity of his multiracial army to defeat the homogenous forces of his opponents. Thus his Numidians had kept the Latin cavalry occupied, whilst his heavier horse routed the Romans, and in the infantry centre his wild but ill-disciplined and poorly armoured tribesmen had engaged the enemy in a hard struggle, before they finally gave way and the Romans were lured forward in pursuit, exposing their flanks to the Libyan foot in reserve. It is probably a mistake to assume that the Gauls and Spanish were exposed in this way because they were expendable in comparison to his trained African phalanx. Only the Libyan infantry had the training necessary to wait quietly in reserve and then manoeuvre to trap the enemy. However, the eventual scale of the Punic victory should not conceal the many phases where the complex plan might have collapsed. The Spanish and Gallic cavalry might not have been able to defeat the Romans as quickly as they did, nor was it certain that Hasdrubal would be able to restrain them from pursuing first the Romans and then later the Latin cavalry. The warriors in the centre might not have held out for as long as they did in the face of the tremendous Roman pressure. If they had broken quickly, then the advancing legionaries may still have been in good enough order to face the massively outnumbered Libyans. Hannibal's decision to stay with his centre emphasizes the importance of this. He had had to rely on Hasdrubal's skill to keep his heavy cavalry well under control. Luck had favoured Hannibal, as it has most successful commanders.