As the sun rose on Tuesday, the second day in August, the scarlet tunic signifying battle could be seen displayed above the tent of Terentius Varro, whose turn it was to command the Cannae army. Polybius says Varro’s men were eager for the fight, were at a near fever pitch of anticipation from the waiting. Orders would have been distributed to the tribunes in the night. The tribunes then would have assembled the men and cavalry in time to march out of camp just after dawn, cross the river, and join the troops in the smaller encampment on the right bank. All were now present, with the exception of ten thousand (probably a legion and an ala) left to guard the main camp and stage an assault on the Punic encampment during the battle. It is likely that the men guarding the main camp were the bulk of those fated to survive the day and become the living ghosts of Cannae. Those less fortunate, around seventy-six thousand men, would move into the customary formation—velites out front; triplex acies, compacted in the middle; and cavalry on the flanks—all awaiting the Carthaginians and destiny.
But exactly where were they? The short answer is that we will never know the precise site of the battle for sure; but that said, the issue has stirred up enough controversy over the years to make it worth considering. Geographically, there are basically two reference points—the location of Cannae itself and the River Aufidus, now called Ofanto. There are also two reliable historical artifacts: we know from Polybius that the battle was fought on the same side of the river as the smaller Roman camp, and the Roman line faced roughly south, with its right flank anchored by the river. It also makes sense that the Romans would have wanted their left flank resting against the highlands on which Cannae was perched, the idea being to make it impossible for Hannibal’s cavalry to sweep around either side to envelop them. The problem is that the distance between the bluffs and the modern Ofanto is far too narrow to accommodate anything like the size of the Roman army, no matter how compacted.
This led a number of respected scholars to propose that the battle was actually fought on the left side of the river, or on a broad plain to the east of Cannae. But the problem with the first view is that it clearly contradicts Polybius, who seldom made this sort of mistake; the drawback to the second is that the flat area to the east is easily wide enough to give Hannibal’s cavalry complete freedom, which raises the question of why Varro would have bothered crossing the river to fight there. Yet all of these interpretations assume that the course of today’s Ofanto is identical to that of the Aufidus, likely a bad bet, given the passage of twenty-two hundred years. This assumption is questioned by modern historians Peter Connolly and Adrian Goldsworthy. Their ingenious alternative is that the ancient river ran considerably to the north as it passed Cannae, leaving flats of about 1.3 miles, wide enough to fit the Roman order of battle as it was assembled that day. This hypothesis remains open to conjecture, but this alternative location seems to be the most plausible for what would shortly become the most prolific killing ground in the history of Western warfare.
If this was indeed the point of deployment for the Romans, it must have inspired great confidence. The inexperienced citizen and allied cavalry, stationed at the extreme ends of the line, right and left respectively, had been relieved of any offensive responsibilities; the cavalry had simply to guard the flanks while the infantry did its work.
Similarly, the numerous but qualitatively inferior velites that were spread out in front of the army had no particular mission once the force was deployed, and they could conveniently retreat between the maniples if pressed.
Meanwhile, the reinforced triplex acies seemed unstoppable, and if anything slowed it down, it was at least impenetrable. It must have appeared to Varro and Paullus that they had finally positioned their forces in a way that even Hannibal could not bend to his own advantage.
Now it was his turn to do just that. Hannibal apparently sent the Balearic slingers and Numidian foot soldiers across the river at about the same time the Romans crossed, but the mission does not seem to have been to interfere with enemy deployments so much as to set up a perimeter behind which the Punic cavalry and heavy infantry could line up. When this was done and he was certain the Romans were ready to fight, Hannibal followed. We can conjecture that the cavalry crossed the river first to reinforce the screeners. Next the Gallic and Spanish infantry joined them, lining up in the center, followed by the two bodies of Libyans, who took their place on either side but remained in columns to form the backward block letter C. By this time the horsemen would have split apart and moved to the flanks, the Spanish and Gallic heavy cavalry on the left facing their Roman equivalents, and the Numidians on the right matched against the allied mounted elements.
As orderly and purposeful as these pre-battle rituals seem in print, the real thing must have provided, even before the fighting started, plenty of distractions and cause for disorientation. At this point the field must have been a jumble of cacophony—horns blaring, drums pounding, swords beating on shields, shouts and war cries reverberating back and forth, to and fro—all the sounds that men muster as they steel themselves to face death and intimidate those they hope will be their victims. Also, more than 125,000 men and in excess of 15,000 horses tramping about in a confined area must have kicked up huge quantities of dust, and it appears that Hannibal’s familiarity with the environment now dovetailed with his desire that the Romans not accurately perceive the true nature of his infantry formation. He had apparently observed earlier that a southeasterly wind, the Vulturnus, gusted with increasing force during the morning, and could be counted on to whip up the dust and blow it into the Romans’ faces, a vexation apparently confirmed by a fragment from the near-contemporary poet Ennius.
Finally, and probably most important, this was August in southern Italy; we can count on it having been hot, and it was bound to get hotter as the day progressed. Most of the Roman heavy infantry and at least the rearmed Libyans would be burdened with between fifty and eighty pounds of arms and armor as they fought for their lives throughout the day. If Trebia had been orchestrated by hypothermia, Cannae was destined to be an inferno where untold thousands were likely to be felled by heat exhaustion, and access to drinking water may well have allowed the Carthaginians to persist in their butchery during the last and most murderous phases of battle.
So it was as they began to fight. The ancient sources agree that the light troops were first to engage, and from Polybius we hear they were evenly matched, neither side gaining an advantage before withdrawing, as was customary, behind their respective lines of heavy infantry. If it had been otherwise, presumably we would have heard more from other sources. So it seems the velites’ numerical preponderance had won the Romans at least a standoff—an auspicious start, considering the multiple drubbings they had taken on earlier occasions. Still, the Punic auxiliaries may have inflicted a very significant casualty at the outset. Livy (22.49.1) reports that the consul Aemilius Paullus, who was with the Roman citizen cavalry, was dealt a severe head wound from a slinger just as the battle commenced—an injury bad enough to leave him unable to ride a horse and bad enough to force his bodyguard to dismount in order to protect him. Polybius does not mention this incident, but it is still suggestive, considering what was about to happen.
The first decisive Punic move came when the Spanish and Gallic cavalry under Hasdrubal—leader of the service corps and destined to perform brilliantly on this day—charged headlong into the opposing horse on the Romans’ right flank. With their adversaries anchored on the river and outnumbering the Romans by more than two to one (around sixty-five hundred to twenty-four hundred), there were none of the normal wheeling maneuvers. Instead the Carthaginians seemed intent on going right through the Romans.
The combat that ensued, Polybius (3.115.3) tells us, “was truly barbaric.” In large part this was because it was mostly on foot. Roman cavalrymen had a decided proclivity toward fighting on the ground, and many of these troopers must have been inexperienced and new to their mounts. But they also may have chosen to dismount because of Paullus’s wound. Plutarch maintains that when the consul was forced from his horse and his attendants got off theirs to protect him, the cavalry assumed a general order had been given and so also dismounted—a development that supposedly caused Hannibal to comment: “This is more to my wish than if they had been handed over to me in fetters.” While it is unlikely the Punic commander actually observed the cavalry getting off their horses, the act nonetheless proved fatal to most of the outnumbered Romans, who were basically annihilated. This is also where many of the Roman senators and others of the equestrian class would have gathered to fight, and ended up making their last stand. It is not clear if Paullus died here—he and his staff may have escaped to join the rest of the army—but Livy’s version makes it seem that this was his end. So at a point when the battle had barely begun, it seems logical that the republic had already been dealt a grievous blow. And it would only get worse. Rather than chase down the last of the survivors who managed to get back on their horses and flee, Hasdrubal reeled in his men from the pursuit, then rested and re-formed them to inflict further mayhem on another part of the battlefield.
Meanwhile, the heavy infantry engagement had begun and had already taken shape, literally, in an unexpected way. As the line of Gallic and Spanish infantry advanced (one source estimates the formation was roughly 840 men wide and 26 deep46), it bowed outward to form a crescent. While some maintain this was natural for a line of men moving forward, others believe it was a last-minute decision on Hannibal’s part. Whichever it was, this convex formation had an immediate and beneficial effect for the Carthaginian side. For as the Roman hastati charged and reached pila-throwing distance, the shape of the line left only a narrow group in the Punic center vulnerable to this potentially devastating missile barrage, and may have led many legionaries to waste their shots while still out of range.
The same thing would have happened as the sides closed for sword-play. Initially at least, the Roman manipular order and their own training would have more or less automatically kept their line straight, and so only the center group of Spaniards and Gauls would have been engaged. The key to Punic success turned on the interior line retreating slowly and in a controlled manner. This was why Hannibal and his brother Mago (presumably joined by other officers and Celtic nobles) stationed themselves here, immediately behind the front, to better manage the action and encourage these most critical of fighters. And the initial geometry of battle served exactly its purpose by committing only a relatively few combatants, and by keeping the huge Roman force at bay until the legionaries in the middle managed to push those at the Carthaginian center back.
As we imagine clusters of bare-chested Gauls flailing their broadswords, interspersed with Spaniards fighting from a crouch behind their shields, all seeking to fend off the surging Romans—themselves bashing forward with their scuta, seeking an opening for their gladii—we should not forget that this sort of combat, essentially a series of individual duels, was both physically and emotionally exhausting. It could not be sustained for more than a few minutes. Once the Punic line failed to collapse immediately, these spasms of violence had to be followed by rest periods when both sides drew back to catch their collective breath for a few minutes. War cries and insults might have been hurled back and forth, followed by pila and other projectiles picked up or passed forward, and then close combat would have been reinitiated. Over time the lulls would have grown longer and the mêlées shorter.
This interrupted rhythm of violence also was to the Carthaginians’ advantage, allowing them to regroup, regenerate, and fall steadily backward in relatively good order. Seeing this, the Romans naturally pushed ahead with increased confidence and growing excitement, focused on their objective of breaking through at the center as quickly as possible. As this happened, the retreating Punic line began to assume an increasingly concave shape, and a critical juncture was reached. Polybius (3.115.6) reports that the Gallic and Spanish infantry in the middle were forced into such a rapid retreat that the Punic line started to break up. As the Roman tide surged forward, it cast caution and training aside and followed the line of least resistance, crowding inward toward the center. The intervals separating the three lines of the triplex acies, and the spacing between maniples, disappeared, and its general organization started to disintegrate. Collectively, the legionaries thought they could see victory just ahead, but it was a mirage; instead, as-yet-unnoticed defeat stared them down from either side in the form of two serried blocks of Libyan heavy infantry, the jaws to the Carthaginian trap.
The moment of Hannibal’s killer epiphany had arrived. The order went out, and man by man the Africans on both the left and right sides pivoted inward, dressed their ranks, and in unison fell upon the Roman flanks, most likely the location of the least-experienced citizen and allied troops. There was little the Romans could do in response besides turning as individuals to face the threat; as units, their formations were too compressed and disorganized to maneuver effectively. They were reduced to a crowd of loners trying to fight off a coordinated engine of destruction. Meanwhile, the emotional shock waves rippled inward, spreading paralysis throughout the Roman ranks and halting the forward momentum of the entire army. Their fate was all but sealed.
That took place in another quarter. Terentius Varro, the overall Roman commander, was with the allied cavalry numbering around thirty-six hundred on the left wing, doing not a lot in the face of a roughly equal number of Numidian horse under either Hannibal’s nephew Hanno (Polybius 3.114) or the resourceful Maharbal (Livy, 22.46.7). The Numidians skirted and swarmed their adversaries as best they could, but were probably thwarted by the Cannae bluffs anchoring the Romans’ position creating the kind of standoff the Romans had wanted on their flanks, a standoff likely satisfactory to Varro. It was around this time, Livy maintains, that the Numidians supposedly staged their fake surrender. But even if this did occur, the Roman consul, who would have had little idea of what had taken place on his opposite flank, was shortly in for an even more unpleasant surprise.
Hasdrubal, fresh from obliterating the Roman horse on the other side of the combat zone, led his reconstituted force of Gallic and Spanish heavy cavalry across the battlefield behind the deployed lines of legionaries, and was soon bearing down on the allied horse with a force that was nearly twice the size. But even before the Carthaginians could bring the charge home, their intended victims evaporated in a panicked stampede, apparently sweeping Varro and his attendants along in their wake. The Numidians, devastating in pursuit, took out after them, killing or capturing all but three hundred of the allies, though Varro escaped to nearby Venusia with seventy of his bodyguards. With Paullus very possibly killed on the right, and Varro removed from the left, Hasdrubal may have already shorn the quadruple consular army of both its consuls.
Yet the focus of his appetite for destruction remained unerring. Rather than rising to the bait of the chase, yet again the Punic commander re-formed his cavalry and instead headed toward the rear of the Roman infantry, quite apparently intent on closing their last avenue of escape. Here, Polybius tells us (3.116.8), Hasdrubal delivered multiple charges at different points, seemingly with devastating effect. On first glance this seems puzzling, since the triarii in the rear should have been well equipped to turn and resist, lining up on one knee with their shields resting against their shoulders and their long spears protruding to form a barrier that horses would not charge. But instead of a solid wall of triarii, it is far more likely the Roman rear was cluttered with a soft mass of up to twenty thousand velites, who had withdrawn behind the maniples shortly after the battle had begun. Most of them were probably adolescents, were very lightly protected, and were lacking room to throw their javelins, and with no avenue of escape, they were virtually the perfect prey for heavy cavalry. Terrified by the horses and the slaughter of their comrades, they would have recoiled inward, exposing their backs and hamstrings to spear thrusts and sword slashes while they pressed desperately against an ever more compacted and undifferentiated human mass.
The same crowding into helplessness must have been taking place on the flanks, as the Libyans on either side continued to press home their attack, an ever-tightening human vise. Meanwhile, the nearly routed Gauls and Spaniards, no longer pursued by the Romans at the forward edge of battle, would have been given time to regroup and turn the tide in their favor. A terrible dynamic was taking place. Assailed from all sides, beyond the control of its officers, with those on the outer edges having no place to go but inward, the Roman army, by pressing itself into paralysis, was becoming if not the instrument of its own destruction then at least complicit in the process.
Somewhere between sixty-five thousand and seventy thousand Romans and allies—depending upon how many had already fallen—were now surrounded. Tactically the battle was over, but the killing had just begun. There was no alternative. The army was still too large and full of fight to be taken prisoner; besides, with its leadership immobilized in the press, it had no real means of surrender. The only choice was its effective extermination, a task the Carthaginians accomplished through systematic butchery almost until the sun set on this terrible day.
Not only does the process beggar description, but exploring the details of the massacre might seem to serve little purpose beyond pandering to some bloodlust with a kind of pornography of violence. Yet war is truly terrible, and to turn our eyes away from its results is in itself an act of cowardice. Hannibal’s great victory, his tactical masterpiece celebrated through the ages, produced, in the end, little more than corpses. But this is probably better illustrated by recounting the subsequent course of events than by moralizing over it now. Nevertheless, there is a more prosaic but still historically valid reason for prying into the details of this exercise in mass homicide. As one source put it, “What remains unclear is how encircled troops, with nowhere to run, could be slaughtered in such a one-sided fashion.” We live in an age when killing is cheap, virtually automated; that was far from the case at Cannae. Other than those who succumbed to the heat, each of the men who died had to be individually punctured, slashed, or battered into oblivion. One modern source estimates that in order for the necessary killing to be accomplished in the eight hours that Appian (Han. 25) estimated the fighting lasted, over one hundred men had to be dispatched every minute. Yet even this astonishing figure underestimates the swiftness and profusion of the slaughter, since the estimate assumes that the killing took place at a regular rate throughout the day and not in a great spasm toward the end, as actually happened. In essence, so many victims, so little time, and that doesn’t even attempt to reflect on the ruthlessness and horror of it all.
Nevertheless, logic tells us that the liquidation of the Roman army at Cannae, if it is possible to reconstruct at all, must have been a matter of mechanics and motivation. We can start here. Among the ancient sources, only Polybius (3.116.10–11) has left us something approaching a plausible description of what must have been the most horrific several hours in all of Western military history: “The Romans, as long as they could turn and present a face on every side to the enemy, held out, but as the outer ranks continued to fall, and the rest were gradually huddled in and surrounded, they finally were all killed where they stood.” In other words, they were finished off from the outside in, peeled like an onion. This makes sense, at least basically, but there were likely to have been other lethal dynamics at work.
Hannibal’s skirmishers—the Numidian foot soldiers and Balearic slingers—having earlier withdrawn to safety behind his main line, must have been intact and available. It’s hard to imagine that he wouldn’t have had them let loose a hail of javelins, stones, and even expended pila onto the stationary mass, a deadly barrage that could hardly have missed in such a target-rich environment, nor avoid inflicting serious injury on Romans who were either too crowded or too exhausted to raise their heavy shields for protection.
Meanwhile, the infantry of Libyans, Gauls, and Spaniards would have continued with their grim work at the circumference. One modern source in an otherwise believable reconstruction of the carnage describes victims “dispatched with frenzied blows, usually to the head.” This seems to miss the mindset implied in the quantity and rapidity of the butchery. Hannibal’s soldiers were practiced killers; very likely most had adopted the cooler, utilitarian approach of the predator, having drawn on our emotional heritage as hunters of the most prodigious and ruthless sort. Moreover, they would have known how to kill quickly and efficiently. If the victim’s back were turned, then a spear or sword thrust to the kidneys would have been so painful as to instantly paralyze, and would have killed within seconds through massive internal bleeding. Or if the victim were facing forward, an equivalent stab to the lower abdomen would have produced the same results almost as fast. Yet such a death stroke—or even more so, multiple death strokes rapidly delivered—implies a certain acceptance, or at least passivity, on the part of the recipient.