Soon after dawn on 2 August the troops in the larger Roman camp formed columns in the main roads between the tent lines and in the open space (intervallum) behind the ramparts. Each maniple, legion and ala assumed a position in the column corresponding with its place in the battle-line. A camp normally had at least four gateways and each column marched out from a different gate. Varro led the army out of the camp and across the river. There was one ford between the two Roman camps, but there may have been other crossing places to the east which were also employed. On the right bank of the Aufidius, the army was joined by the troops from the smaller camp and together they deployed into battle order.

All our sources emphasize that the decision to deploy the army and offer battle was taken by Varro alone. This was right and proper, since it was his day to exercise command, but Livy goes so far as to claim that he issued the orders without consulting, or even informing, Paullus. In this version, Varro simply raised the red vexillum standard - the square flag which marked the commander’s position during a battle - outside his tent, the traditional symbol to tell the soldiers to prepare for battle. He then formed his own legions into column and led them out. Paullus, seeing all this happen, felt obliged to follow with his own troops. None of this makes any sense. Varro held supreme command of the entire army for the day and it is absurd to suggest that he failed to issue orders to one important section of it or to inform his colleague. It should also be noted that the process of preparing the soldiers for battle and parading them preparatory to moving out was long and complex. Close supervision was required on the part of all the army's officers, and especially the military tribunes of the legions and praefecti of the alae, to ensure that this was carried out as smoothly and quickly as possible, checking that the columns used for deployment were formed in the correct order and that when the army finally was able to move out it went by the proper route to the right place. The process must have taken hours for any army, and was made especially difficult by the size and mixed levels of experience and drill of the soldiers at Cannae. It is impossible to imagine that Paullus was unaware of all the activity in camp until Varro had begun to lead his forces out.1

Aemilius Paullus cannot have been ignorant of his colleague's intention to offer battle on 2 August. The day before he had refused to meet Hannibal's challenge, keeping most of his soldiers inside the camps. This does appear to give a clear indication that Paullus genuinely believed that it was unwise to fight. Polybius claims that he felt supply problems would force Hannibal to move his camp within two days if there was not a battle. If the Carthaginian army was to disengage and withdraw, then this would encourage the Romans and perhaps grant them an advantage in any future encounter. The smaller Roman camp had been expressly set up to place pressure on the enemy's foragers. There was perhaps another reason why Paullus might have been less reluctant to fight a battle on the next day. As far as we can tell, on 1 August the Carthaginian army formed up between their own camp and the larger Roman camp. Livy tells us explicitly, and Polybius appears to imply, that the Numidians had to cross the Aufidius in order to attack the smaller Roman camp which seems to confirm that the main Punic army had been formed up on the left bank of the river. Both of these sources make it clear that Varro led the troops from the larger Roman camp across the river to the same side as the smaller camp and deployed the combined army into battle order in front of it. He was not then accepting battle under precisely the same conditions that his colleague had declined the previous day. The Romans had deliberately chosen different ground, despite this involving moving the larger part of their forces to the new position. Varro could not even be sure that Hannibal would accept a battle in this new location, but may have felt that simply offering to fight would help to encourage his soldiers after the humiliation of declining battle and seeing one of their camps attacked. Perhaps Paullus still believed that it was unwise to fight, even in the alternative position. This is impossible to know, but it is worth remembering that the ability of the Numidians to dominate the eastern bank right up to the outposts immediately outside the smaller camp called into question the Romans' ability to deny Hannibal's army provisions.2

Whether or not Paullus agreed with his judgement, Varro had the right to make the decision and had not simply reversed his colleague's choice. Fighting on ground of your own choice was one of the skills of the good commander portrayed by Hellenistic military theory, and this was something which both Sempronius Longus at Trebia and Flaminius at Trasimene had failed to do. Before moving on to consider in detail the choice of battlefield and how the terrain affected the subsequent battle, it is worth mentioning a theory which claims that in fact Paullus rather than Varro was in command on 2 August and committed the army to battle. This is an attractive idea, allowing us to claim that we have seen through the propaganda in our sources, but is based on highly tenuous assumptions. It is much better to follow the literary tradition and accept that Varro was in command, although he may have acted with the approval of his fellow consul.3

Panorama showing the view from Cannae today. The cluster of white buildings in the distance (centre) is San Ferdinando, the probable site of Hannibal's camp. The current line of the River Aufidius (Ofanto) is marked by the green line of low trees and bushes in the middle distance. The battle was fought on the level plain in the centre.


It is rare for the precise location of any battle fought in the classical world to be known with certainty. Literary style was important to ancient historians and too much topographic information was likely to overburden any narrative. As a result, even accounts written by senior officers who were present at a battle mention few geographical features. The campaigns in Italy from 218 to 216 BC are described in some detail by our sources and it is usually possible to determine the general area in which a battle occurred. It is then a question of attempting to relate the snippets of information provided in their accounts of the battle to the terrain in this area today in order to locate the battlefield. Most useful are those major geographical features which are unlikely to have changed in the last twenty-three centuries. Usually there are several sites in the right area which could conform to the ancient sources. The probable size of the opposing armies, their tactical systems and the objectives of each side in the campaign itself provide the context in which we must try to judge on which of these sites the battle is most likely to have occurred. This is an uncertain process, inevitably relying on many guesses and impressions, and as a result it is unsurprising that there is a broad range of opinions concerning the site of most battles. Cannae is no exception to this.

The location of the town of Cannae itself on the line of hills south of the River Aufidius (Ofanto) is one of the few certainties, even if the Roman remains there in fact date to a later period. We know from our sources that the Roman army had constructed two camps. The larger camp lay on the side of the river from which the Roman army had approached Cannae and faced Hannibal's second encampment. The smaller camp was on the other side of the river, about a mile (1.6km) from the main position, and even further from the Punic camp. The battle was fought on the same side of the river as the smaller camp. Polybius tells us that the Roman line was formed with its right flank resting on the river and correspondingly that the Punic left was also anchored on the Aufidius. The Roman line faced south, the Carthaginians north, so that neither side suffered the disadvantage of fighting with the sun in their eyes. Other sources repeat a Roman tradition that their soldiers were hampered by a strong wind blowing towards them. Although doubtless exaggerated, the prevailing wind in this area is the hot Voltumus, which blows in very strong gusts from the south-west, again suggesting that the Romans faced south or south-west. The Greek historian seems to have believed that the River Aufidius flowed from south to north. In fact it runs more south-west to northeast, but meanders considerably. The general direction of the river’s flow cannot have been any different in the third century BC, but its actual course is far less clear. During the last century the line of the Ofanto has changed on numerous occasions, sometimes even varying from year to year. The contours of the shallow valley through which it flows impose a northernmost and southernmost restriction, but in places this would still permit a variation of several kilometres. As a result, one of our few fixed points on the battlefield proves to be far less certain than we might like.4

So much has been written about Cannae that it is unsurprising that the site of the battle has proved one source of major controversy. In many cases arguments have focused on minor details, but other disputes have been much deeper. The most fundamental question is on which side of the river the battle was fought, since this determines our understanding of the orientation of the battlefield. In recent years very few scholars have argued for a location on the left, or western bank, but in the past a number of eminent scholars have advocated this view, notably Hans Delbruck and Konrad Lehmann. If the battle was fought on the left bank, then the Romans - given that their right flank rested on the river - must have formed up facing towards the sea. There are several problems with this. In the first place it is very difficult to see how Polybius can have believed that the Roman line was facing south, even with his misunderstanding of the river's orientation. Secondly, for the Romans to have been camped to the west and Hannibal to the east, nearer the sea would suggest a very different build-up to the battle to the most obvious reading of our sources. It is hard to see how the armies could have ended up in this position if the Romans had followed Hannibal to Cannae from the area around Gerunium. Advocates of this view sought explanation in the problematic chronology for the campaign mentioned in the last chapter. If Hannibal left winter quarters at Gerunium when the harvest became available in early June and the battle was not fought until 2 August, then there was plenty of time - nearly two months - for more manoeuvring than is described by our sources. It was possible for Hannibal to have crossed the Aufidius and raided more widely in Apulia, before turning back north, or north-east, to seize Cannae. During this expedition he was followed by the Roman army which kept to the high ground and avoided contact. Only when the consuls had joined did the Romans choose to close the distance and a battle occur. Adherents to this cause suggest that the consuls deliberately chose a narrow battlefield west of the town of Cannae and north of the river, believing that this would protect their flanks from the superior Punic cavalry. They add that the fact that Roman fugitives from the battle gathered at Canusium would make far more sense if their army had deployed to the west of the Carthaginians.5

Previous page: The actual site of the battlefield of Cannae is not known with absolute certainty and must be deduced from the fragments of information provided in our sources and examination of the topography of the area. For the reasons given in the text, this assumes that the river ran further north in 216 and that the battle was fought in the position marked.

Whilst it is just possible that our sources skimmed over more than a month of operations and that the battle occurred in this way, this does seem unlikely. It is also difficult to see what useful purpose Hannibal would have served by marauding about southern Apulia. His aim in this campaign was to bring the Roman army to battle and to destroy it. The reluctance of the proconsuls throughout the winter to risk serious engagement and their care to follow him at a safe distance when he finally left Gerunium must have made it clear that there was little chance of joining battle until the new consuls arrived. With the supplies captured at Cannae, and the ease of foraging in the surrounding area, there was really no need for a mobile campaign further south. Hannibal had already demonstrated the Romans' inability to present his going wherever he wished. The ability of the Roman legions at Trebia to retreat to Placentia even though the enemy army lay in between suggests that we should be cautious about making too much of the routes taken by fugitives

In addition the terrain to the north of the Aufidius fits rather better our sources’ description of the Roman march towards Cannae across open country than any possible route from the south-west. On balance, it is far more likely that the battle was fought on the right, or southern, bank of the Aufidius and that the Romans had their backs to the sea and faced roughly south-east. In all of the chapters of this book I have assumed that this was the case and that therefore the larger Roman camp was on the left bank and the smaller camp on the right.

In recent years the River Ofanto has tended to run fairly close to the line of hills on which sits Cannae itself. Many scholars have assumed that there could not possibly have been sufficient space in the plain between the two to accommodate the armies, especially the huge Roman force. J. Kromayer, who assisted by G. Veith, produced in the early twentieth century what is still the classic study of the battlefields of the ancient world, therefore placed the battle to the east of Cannae itself, on the broad plain which slopes very gently down to the sea. This area is certainly wide enough to permit the deployment of nearly 140,000 men. In this interpretation the smaller Roman camp was little more than 4km from the sea and the Roman battle line a short distance in front of it. The left wing of Hannibal's army would have been fairly close to the edge of the hill of Cannae itself. His camp is unlikely to have been as far to the east as the high ground around San Ferdinando di Puglia, but would have been somewhere on the open plain north of Cannae. There could have been no intrinsic value to such a position apart from the pressure it applied on the enemy by its proximity to their camp.6

However, this battlefield is no less open and suitable for cavalry than the one north of the river where Paullus refused to fight on 1 August. It is difficult to understand why Varro would have made the effort of shifting the bulk of his forces across the river to the right bank to fight on virtually identical terrain. Only the Roman right flank resting on the river would have been secure, since the left had no terrain feature to anchor itself upon. In addition the slope, although gentle, would have placed the Romans at a slight, but not insignificant disadvantage. If the Roman army had decided to fight in this position on 2 August then the criticisms levelled at Varro would seem to be valid, for this area offered far more advantages to Hannibal's cavalry than it did to the Romans. Just about the only reason why the consul might have chosen to move to such a position was an urge to challenge the enemy's dominance of this bank and the area around the Roman camp which had been threatened by the Numidians the day before. The problems were not solely confined to one side, for wherever Hannibal’s camp lay his columns would have had to march through an awkward little defile around the hill of Cannae in order to reach the battlefield. This would have made his deployment a more difficult and time consuming process, and a good commander tried to ensure that his troops entered battle as fresh as possible.

Both of these interpretations assumed that the Aufidius flowed on virtually the same line in 216 BC as it did in the early twentieth century. As we have seen, it is in fact perfectly possible that it followed a very different course. Assuming that the Aufidius actually lay much nearer to the northernmost limit, Peter Connolly suggested that the battle was in fact fought north of the line of hills around Cannae on a plain about 2km (c. 1.3 miles) in width. In order to fit such a large Roman force, whose probable frontage he calculated at around 3km (c. 2 miles), into this area, Connolly argued that the Romans deployed at an angle, so that in fact they were close to facing south as Polybius described. This looks a little awkward on his maps, but it should be noted that his estimate for the Roman frontage is too high as we shall see in the next section. It is much easier to fit the Roman army into this position than at first appears. The advantages of this position are obvious. The Romans were able to anchor one flank on the river and the other on the high ground near Cannae, making it impossible for the Carthaginian horse to envelop their line as they had at Ticinus and Trebia. It would make far more sense for Varro to have chosen to offer battle here, rather than simply crossing the river further east to fight in a plain very little different from the one outside the main camp.7

If you find an error please notify us in the comments. Thank you!