Despite biased historians and few primary sources, the numbers in the ancients’ battle tales are surprisingly reliable.
In 256 the navies of the Carthaginian and Roman republics confronted each other within sight of Cape Ecnomus on the southern coast of Sicily. The Romans were escorting a fleet of transports carrying an army intended to invade North Africa and take the BC, war to the enemy homeland. Knowing this threat was bound to provoke a response from the Carthaginians, the Romans had reinforced the crews of their warships with large numbers of picked legionaries serving as marines.
Altogether, Rome had stationed 330 war galleys in formation to protect the vulnerable transport vessels. The warships were almost all quinqueremes or “fives,” the standard line-of-battle ship in this era. There were three banks of oars on these vessels. The ships were named for the team of five rowers who together wielded each set of three stacked oars, two rowers manning each of the top oars, one the bottom. At full strength, a quinquereme was crewed by 280 rowers and about 20 deck crew.
The Romans had increased the number of marines on each vessel to 120 men. Altogether there were said to be some 140,000 sailors and marines in the Roman fleet, not including the passengers and crews of the transports. Facing them was a slightly larger Carthaginian fleet of 350 warships crewed by 150,000 men.
These numbers are staggering. If they are accurate, then in terms of the number of combatants, Ecnomus was one of the biggest naval battles in history—perhaps the largest ever. That’s surprising, given that it occurred in the third century BC and was fought by two nations that would be considered small by modern standards. Carthage controlled a stretch of the North African coast, as well as parts of Spain, Sicily, and most of the other islands in the western Mediterranean. The Roman Republic controlled the Italian Peninsula south of the River Po, but the intervention in Sicily, which had provoked the war with Carthage, was its first overseas adventure.
Even more strikingly, Rome had only begun to turn itself into a naval power just a few years before. How it managed this and shattered the power of Carthage, with its long-standing maritime tradition, is one of the most striking revolutions in the history of warfare. Ecnomus proved an overwhelming Roman victory. For the loss of only 24 ships, the Romans sank 30 Carthaginian vessels and captured another 64.
Ecnomus was the largest naval battle of the First Punic War, and the entire conflict, which marked the beginning of Rome’s domination of the Mediterranean world, was eventually decided at sea. The Romans managed to fight in circumstances that negated their enemies’ initial superiority in seamanship, and won all but one of the major battles. In some two decades of fighting, the Carthaginians had lost 500 warships. Roman losses were markedly heavier at around 700 galleys, the overwhelming majority of them lost in storms rather than through enemy action. In the end, the Roman Republic proved more willing to keep on building and manning fleets than its opponent.
There is no reason to doubt the basic course of events described by our sources, and certainly not the overall outcome. The figures are all supplied by the Greek historian Polybius, a man with military experience who is generally considered one of the most sober and reliable historians of the classical period.
He did, however, write his account more than a century after the Battle of Ecnomus was fought. Is it reasonable to accept the spectacularly large figures he gives for the fleets in this and other battles? More broadly, are the numbers of combatants and casualties suggested by Polybius and other ancient chroniclers credible estimates or fantastical fabrications?
Numbers present a real problem when studying the ancient world’s wars. The study of military history must always rest on the bedrock of facts. At best, we want to know with some precision the size and composition of the forces involved in an engagement, their deployment and how it related to battlefield topography, information available to commanders, the orders they issued, and the myriad details of timings and movements. (One reason military history is sometimes unpopular in the wider world of historical studies is that some very fashionable approaches to research do not always cope well with facts and practicalities.) Some of these details cannot be established even for comparatively recent and well-documented battles.
When it comes to ancient warfare, historians face far more severe problems. Most battlefields cannot be located with precision. Accounts of battles given by surviving sources, even on the fairly rare occasions when there were eyewitnesses, never tell us all we would like. Nor is it always straightforward to gauge the reliability of what they do tell us. Accounts may have been deliberately distorted by bias or shaped to fit literary conventions of the day. Alternatively, the author may not have had access to accurate information.
Transmission is another problem. In many ways, this is the hardest of all to judge. Far less than 1 percent of the literature produced by the Greeks and Romans has survived until the modern day. In no case do we have an original manuscript of, for example, Thucydides, Polybius, or Caesar’s Commentaries. Before the invention of the printing press, every work had to be copied by hand. The earliest texts of all significant ancient sources are at best medieval. Dozens, perhaps even hundreds, of copies separate the texts we have from the original document written or dictated by the author. This raises the very real possibility of errors creeping in over the centuries. Even today proofreaders have difficulty checking numbers. If anything, Roman numerals were even more susceptible to accidental errors by copyists.
On occasion it is easy to dismiss the numbers provided by sources. When Herodotus claims that Xerxes invaded Greece in 480 BC with an army numbering some 1.7 million men, we can safely reject this as wild exaggeration since this would have been impossible logistically. No doubt the Persian army was exceptionally large, but even 1 million is a blatantly implausible figure.
Similarly, when we read that King Radagasius raided Italy with 400,000 Gothic warriors in AD 405, we should be immediately skeptical. A Roman army numbering 30,000 men at most—probably fewer, in fact—routed them. The figure is given by a more recent and often unreliable historian, and is also far larger than figures given for similar tribal armies in more reliable histories. Like Xerxes’ 1.7 million men, such a wild figure flattered the victors by making the Gothic warriors’ defeat more spectacular.
At a very basic level, large numbers are simply more dramatic. The same tendency exists in modern media, in which headlines tout the largest estimate available, numbers are rounded upward, and casualties are converted to fatalities. The practice of exaggerating the size of an enemy army and the casualties inflicted on it helps to excuse defeats and enhance victories. It therefore becomes difficult to know enemy numbers with precision. The opposition obviously does not volunteer this sort of information.
In any case, it is unlikely that a leader like King Radagasius would have known precisely how many warriors were in his army. These were not regular soldiers, but individuals who followed their own chieftains and were free to join or leave an army almost at will. At best, the chieftains likely had only a rough idea of their armies. It was, therefore, even more difficult for their enemies to gain knowledge of precise numbers.
This is part of the essential nature of irregular forces in any period, especially when they are formed for a brief campaign. In the American Revolution, it was common for military leaders to underestimate the number of militiamen under their command. This was partly deliberate, to emphasize the role of the Continentals just in case political pressure developed to cut the already meager resources given to the regular army. Yet there was also the basic problem of counting militiamen who might have arrived only hours before a battle.
What are we to make of Julius Caesar’s statement that the migrating group consisting of the Helvetii and other allied tribes, which he encountered and defeated in Southern Gaul in 58 BC, consisted of 368,000 people? He claims that this figure came from captured documents written in Greek characters, recording a census held by the Helvetii during the planning for their migration. By the end of the campaign, 6,000 migrants were enslaved, 32,000 belonging to one tribe were permitted to settle, and some 110,000 were sent back to their homes. This leaves a shortfall of 220,000, and Caesar allowed his readers to see this as the number of enemies his legions killed.
We have only Caesar’s word for this. There is nothing implausible about the existence of written records: although the Gauls were not a literate society, a few examples of their written records exist. They used Celtic words written with either Greek or Latin characters. Many tribes minted coins bearing the names of their kings. In addition, Caesar never claimed that all of the migrants were warriors.
To set against this, Caesar was writing propaganda intended to impress a Roman audience. Politics and war were inseparably linked in Rome, and military glory was readily transferred into political prominence. A triumph to commemorate a victorious war was one of the highest honors a Roman senator could win.
Almost as important as winning a triumph was making clear that it was more glorious than any triumph celebrated in the past. The Roman aristocracy competed not just with their contemporaries but also with their ancestors. It was a culture of superlatives in which it was vital to do things first, or at least do them bigger and better than any predecessor. As a result, there was a long tradition of quantifying victory in quite remarkable, and at times perhaps spurious, detail. Each enemy defeated was listed by name, along with the number of towns and villages captured; plunder gathered; and enemies who surrendered, were enslaved, or killed. A law was passed stipulating that 5,000 enemies had to have been killed in battle before a general became eligible to be awarded a triumph. It is doubtful that the count was carried out rigorously, or in any way independently.
Just because there were good reasons for Caesar to exaggerate does not necessarily mean that he did so. This brings us once again to the question of whether the figures he supplies are plausible. Certainly, the numbers throughout Commentaries on the Gallic War are internally consistent. Tribal armies always appear on a similar scale whenever figures are mentioned; they also fit with those provided for Gallic, Germanic, or British tribal armies in other sources. This may mean only that Caesar was careful to be consistent in his exaggeration.
The eminent military historian, Hans Delbrück, was convinced that Caesar and all other Greek and Roman authors invariably inflated the size of the armies fielded by “barbarian” armies. Some of Delbrück’s reasoning is highly suspect: He consistently asserts, but offers no proof, that civilization made men soft. Therefore, a given number of hardy barbarians would always defeat an equal number of Greeks or Romans. Consequently, the latter must have enjoyed a significant numerical advantage whenever they defeated a barbarian army. The idea seemed to owe more to the celebration of the tribal “ancestors” of recently unified Germany than anything else.
More significant are Delbrück’s appeals to more practical concerns—his Sachkritik, as he termed it, or factual criticism. Based on contemporary estimates of the population of various regions in the Greek and Roman period, he maintains that tribal people simply could not have fielded such large armies as claimed by historical sources. Unfortunately this is a field in which precision is impossible, and great caution ought to be employed when estimating numbers of people.
We have no reliable figures for the overall population of the Roman world at any period. Even less is known about the people who lived outside the empire. Archaeological surveys have located many more settlements in the tribal areas than were known at the end of the nineteenth century and the start of the twentieth. Where excavation has occurred, these settlements often have been found to be larger and more permanent than expected. We still lack sufficient data to make even confident guesses at overall population figures. Having said that, there is now enough evidence to suggest that the figures in our literary sources are certainly not altogether impossible. In some cases, they may even be quite likely.
More specific factors also can be used to judge plausibility. Delbrück pointed out that 368,000 Helvetii simply would have been incapable of moving and supplying themselves as a single massive wagon convoy. This is certainly true, but at no point does Caesar actually state that this is what the migrants did. On the whole, it is easier to read Caesar’s account as suggesting that many smaller groups moved separately in the same general direction, clustering only as they waited to cross a river or met any other obstacle. It is rarely easy to decide what is inherently most plausible in situations like this, since we lack enough other information to permit a reasonable guess at real numbers.
The situation is at least a little clearer when we deal with the more formally organized armies of the Greeks and Romans. All city-states took considerable care in maintaining lists of their male citizens. Therefore, even their temporary militias of hoplites were carefully numbered and recorded. Precise, trustworthy figures for Greek armies are also available.
The Romans carried out a full census every five years, recording the property of male citizens. A Roman’s property dictated his political rights and his military obligations to the republic. The names of men enrolled in a legion were listed in the census books. Polybius supplies a detailed list of the number of Roman citizens and allies eligible for military service in 225 BC, just seven years before Hannibal would invade Italy in the Second Punic War. The total is extremely large, amounting to more than 770,000.
Obviously, no substantial state could ever mobilize more than a proportion of its available manpower, but this is still a huge total. That said, the number does fit with the few surviving census figures from the same period. Further confirmation comes from the large number of legions that Rome raised during the struggle with Hannibal.
Permanent, professional armies tend to generate documentation, and the Roman army of the imperial period was no exception. A few such documents have survived, written variously on papyrus, wooden writing tablets, and, in the case of some desert outposts, on large pottery fragments. A soldier’s military career was recorded in detail, from the moment he was recruited until his eventual death or discharge. In the barracks were daily duty rosters and notes of postings and leave, as well as details of equipment issued, damaged, and replaced.
The army tracked its cavalry horses with similar care. Each cavalry unit had its nominal role, updated regularly by one of the clerks in the headquarters—perhaps even on a daily basis. There were also briefer documents, not listing names but simply the totals of each rank, on the strength of the unit on that day, and noting whether these were on detachment or away from the unit for any other reason, if they were in hospital or otherwise unfit for duty, and finally providing a total for the effective strength of the unit.
The following text was found at the fort of Vindolanda near Hadrian’s Wall in Britain, and dates probably to somewhere between AD 92–97 (the question marks denote indecipherable figures):
The commander of a Roman army unit could expect to know with some precision the total number of men under his command. This was obviously easier when in garrison than during a campaign. Even then there is evidence for small detachments being scattered over a wide area within and outside the province. At least some of this information would have been forwarded up the chain of command. The slow pace of communication meant there would have been some delay passing the information, but the governor would have had access to relatively recent figures for the entire army under his command.
The emperor in Rome also kept track of his army, and was the only person with the authority to order a unit moved from one province to another. The imperial bureaucracy at Rome tracked the careers of the army’s middle-ranking officers, the centurions, on an individual basis. There were about 4,000 army centurions in the second century. Some centurions served in more than a dozen legions; many were posted in turn to provinces on opposite sides of the empire. The central administration at Rome controlled this system. It also may have received information about each ordinary soldier.
More than half the army consisted of noncitizens serving with auxiliary units. When honorably discharged after 25 years of service, these men received Roman citizenship, and a record of this was kept at Rome as part of the census.
So when Caesar notes the size of his army in the Commentaries, we need a good reason to doubt his accuracy. Such information also seems to have been publicly available, and we can have almost as much confidence when dealing with numbers in the accounts of Roman historians such as Tacitus.
Obviously, generals were usually not concerned with absolute precision. As with most regular armies, Roman units were of a standard theoretical size and organization. This was a great asset when planning and, in particular, arranging to supply an army. Evidence suggests that planners used round numbers for the amounts of food and other supplies each type of unit required. Under normal circumstances, Roman commanders knew what size task a legion or a cohort could handle, and they gave their orders accordingly.
Sometimes it was necessary to adapt the reality of the army’s strength to something closer to conventional size. At the Battle of Pharsalus in 48 BC, Caesar combined the significantly understrength Eighth and Ninth Legions under one command, and allocated them a sector of the line that normally would have been given to a single legion.
As with armies in all periods of history, the units of the Roman army only rarely operated with their exact theoretical complement of men. A legion was supposed to have 5,000 men, but Caesar mentions legions operating with 4,000, 3,500, or 2,250 legionaries at various stages in his campaigns. He also tells of a legion reduced to just under 1,000 men that nevertheless was an effective unit.
Several of the auxiliary cohorts recorded on surviving documents do not appear to conform to any of the standard orders of battle believed to have existed. Such irregularity will not surprise anyone familiar with the military, although it does sometimes disturb scholars who have come to see the Roman army’s structure as a puzzle to be reconstructed.
The Romans did not consider organization an end in itself but a means of helping the army operate effectively. Variation and a degree of flexibility were quite natural.
It appears, however, that the same mentality of some current historians sometimes influenced our ancient sources. Knowing that a legion was supposed to muster 5,000 men, and that an army consisted of four legions, sources were quite likely to record its total strength as 20,000 men. In reality, the number of effectives in a battle was most likely substantially lower, and perhaps very occasionally higher. Neat round numbers, especially when they coincide with convenient multiples of individual unit sizes, should be regarded with caution. Yet at the very least, they probably give an idea of the theoretical size of the army. Casualty figures were kept as part of the detailed Roman army record-keeping, especially when it had become a permanent regular army. This was also true of the citizen armies of Greek city-states. Some of the monuments listing the Athenian dead by name have survived. Alexander the Great had statues erected of the 25 Companion cavalrymen who fell at the Battle of Granicus in 334 BC.
Fragments of a large Roman cenotaph from the late first century have been excavated at Adamklissi, Romania. Once it stood next to grand stone trophies and the huge, drum-shaped Tropaeum Triani, decorated with carved scenes showing Roman soldiers defeating barbarian enemies. It looks as if the army of the province used the site regularly for ceremonies. Only a few names survive, but the list includes the fallen soldier’s rank and place of origin.
The most senior man was a prefect whose name is lost. Interestingly, his city of origin is listed as Pompeii, but his affiliation at the time was to Naples. Clearly he was born before the eruption of Vesuvius in AD 79 destroyed the city, and he was forced to move his place of residence elsewhere. The monument also records some of the first soldiers from the British Isles to serve with the Roman army.
Calculations based on the original size of the sides of the monument suggest that there was room for some 3,500 names, but only a few score survive. We do not know whether the Adamklissi monument commemorated the dead from a single battle or those who fell in the course of a war. If the monument honors the latter, were they men from the entire army, or simply those from the units of the province that built and maintained the monument?
In conflicts fought between civilized states, there was a reasonable chance of getting good information about enemy numbers and casual- ties. Although none have survived, we know that there were accounts of the Punic Wars written from the Carthaginian perspective.
Deliberate distortion is still possible, especially when it comes to estimating enemy losses. This was easy to do when that enemy was a barbarian nation, since no one was likely to have any evidence to the contrary. It was much harder for a general to conceal his own losses, although we know of at least one Roman commander who tried—and failed.
On the whole, there is an encouraging consistency in the numbers given for losses in many sources. The losing side almost invariably incurred disproportionately heavy losses. Most casualties seem to have been inflicted during the pursuit, after one side had broken from close combat. By contrast, the victors’ losses were rarely heavier than 5 percent of the total force engaged and often were considerably lower. On rare occasions, in especially hard-fought battles, the proportion might rise, but the winner’s casualties were higher than 10 percent only exceptionally.
Usually the sources do not give separate totals for dead and wounded, which often means we cannot know whether the figure given includes all casualties or only fatalities. Even so, there are some interesting patterns. Julius Caesar’s centurions seem to have been far more likely to be killed than the rank-and-file legionaries. This certainly fits with their aggressive style of leadership.
Such percentages make it very unlikely that the Adamklissi monument commemorated the dead from a Roman victory, since 3,500 casualties would imply a massive victorious army. The monument may have memorialized the losses from a serious defeat, although in this case it would not suggest that the army was especially big, which would tell us something about the scale of fighting in many frontier campaigns.
The Battle of Cannae in 216 BC was Hannibal’s greatest victory and one of the most complete tactical victories in military history. Polybius provides the clearest account of the fighting and his is by far the earliest surviving source, but he is slightly confused in the numbers he gives for the Roman losses. According to him, the Romans fielded 6,000 cavalry and 80,000 infantry, which makes his total of their losses—89,000—an impossibility. Perhaps the author himself simply made a mistake, although this may be a case where errors crept in as the text was copied over the centuries.
Livy wrote a century after Polybius but mentions several different figures for both the starting size of the Roman army and its losses. In the end, he opted for 48,200 dead and some 19,000 prisoners. This seems to have become the established view as other authors give the rounded-up figure of 50,000 dead.
Interestingly, Polybius put Hannibal’s losses at some 11.4 percent of his army, and Livy puts them higher, at about 15 percent. These were exceptionally heavy losses for a victorious army, but would seem to fit with the descriptions of the battle. Most of the Roman dead were not killed as they ran but were surrounded and trapped by the Carthaginians. They had to be killed one by one, and it is entirely convincing that this proved costly for the victors.
Whenever we consider numbers in ancient accounts, we need to consider them in the context of all the other information about the campaign and of warfare in the period in general. Ultimately, we have no choice but to judge the numbers on their plausibility within this context. When different sources provide alternate figures for the same battle, then each source’s overall reliability, closeness to the events, and any bias must be considered as we decide which tradition to follow. In the end we may choose to doubt or even reject the figures, but if we do, there are no better figures to replace those that were rejected. However plausible they might seem, modern estimates will never be more than guesses.
Then how should we deal with the figures Polybius gave for the fleets and losses at the Battle of Ecnomus? At a basic level, it is typically easier to establish the size of a navy than the size of an army, simply because ships are easier to count than individual soldiers. However, a number of eminent scholars have tried to reduce the numbers Polybius gave the rival fleets simply because they seem too large to be plausible.
This downward revision relies on a very literal reading of his account of the First Punic War, and it assumes that the construction of every ship made by either side is specifically mentioned. This reading also suggests reduction on each side by about 100 vessels. Why this makes the size of the fleets more “plausible” is unclear.
The Roman shipbuilding program, as described by Polybius, was certainly spectacular. In 261 BC, the Roman senate decided to construct a fleet of 20 light triremes or “threes,” and 100 quinqueremes. The Romans had never before built a quinquereme but had managed to capture a Carthaginian “five” that had run aground. They took it to pieces and copied the design.
The Romans often boasted of their willingness and their ability to copy the military technology and tactics of their enemies. Recent archaeological findings strongly support this story.
The wreckage of a small Carthaginian warship was found off Massala in Sicily. It may have been sunk in the final battle of the First Punic War. What is most interesting is the evidence that its construction seems to have followed a clear, standard design. Analysis revealed traces of paint that marked where the workmen had to cut the timbers. There were also Punic characters on the keel showing where the ribs were to be fitted. Since the shell of the hull was put together before the ribs were added, these markings would have been invisible to workmen standing inside, so the same sequence of letters was repeated along one of the strakes.
In another case, a word of instruction was painted upside down in relation to the finished boat but it would have appeared the right way up to a workman during construction.
All this demonstrates that Carthaginian warships were built to a template, making possible a pre–Henry Ford version of mass production. No one would have suggested that the ancients were capable of thinking in this way before this discovery, but one should always be cautious about underestimating human ingenuity.
Many of the techniques involved were not specific to shipbuilding, and the Romans had plenty of carpenters and other craftsmen. Building a Punic warship was not the same as building a dreadnought or even a warship from the age of sail. It is more than likely that early Roman vessels were less than perfect in their construction, but the builders would have learned quickly. Given resources and determination, both of which the Roman Republic possessed in abundance, the rapid construction of large fleets was entirely possible.
All of this certainly makes Polybius’ numbers for Ecnomus within the range of plausibility. He, or his sources, may have been mistaken. He may have counted all ships as quinqueremes when significant numbers of them were smaller. Even if the numbers are accurate, some warships may have been undermanned. Therefore, the total figures for participants could be reduced somewhat.
In considering events more than 22 centuries past, one cannot afford to be too dogmatic. Uncertainty about so many details is an inevitable part of studying the ancient world, especially its military history. Yet in the end, there is a better case for accepting, rather than rejecting, the numbers in many ancient sources.
Originally published in the Autumn 2008 issue of Military History Quarterly. To subscribe, click here.