The Battle of Pharsalus:
The Two Armies

The blue square on the right represents the location of Pompey's camp, the green square on the bottom right is Caesar's camp. The two diagonal boxes are where Morgan would locate the actual battle itself.

Without critiquing Morgan's arguments, for now it will work to simply accept these locations. However, the symbols for camps and armies should be refined. Pompey's camp should be between 33% and 100% larger than Caesar's, depending upon which set of numbers for manpower one uses. Correspondingly, the battle lines need to reflect the actual composition of the armies and cavalry formations.

The size of the two armies

The Romans had a rule of thumb that determined the appropriate camp size based on the number of men in the army. Therefore, to establish the size of the camps the first step is to settle on numbers for the two armies. The starting point is Caesar who gives these numbers

According to Caesar


45,000 men, 110 cohorts @ 409 men
3,000 Archers
1,200 Slingers
7,000 cavalry
800 Volunteers
57,000 Total from Caesar's details


22,000 men, 80 cohorts @ 275 men
1,000 cavalry
1,000 Archers and Slingers (not mentioned but he had to have some -- number estimated)

According to Keppie


36,000 men, 88 cohorts @ 409 men
12,000 horse, auxiliary, others
48,000 Total

According to Delbrück:


40,000 men, 88 cohorts @ 450 men
3,000 cavalry
??? archers, slingers, others


30,000 men, 80 cohorts @ 375 men
2,000 cavalry
2,000 antesignani
??? archers, slingers & others

As noted above, writing in 1900, Delbrück seriously doubts Caesar's numbers. Of more recent authors, the only one I have found who questions Caesar's numbers is Keppie.

One must decide. Given Caesar's known penchant for exaggerating the size of his opponents armies and the apparently sound arguments in favor of the 22 cohorts being assigned to garrison duty it seems reasonable to accept 88 cohorts as Pompey's strength and reduce his numbers accordingly. The difference between Keppie's 36,000 and Delbrück's 40,000 will not significantly affect the model. Keppie simply scaled down the army by counting off the 22 cohorts. Delbrück adjusted the total number. Delbrück's number is closer to Caesar's and, in that way, is a less radical departure from what most accept. It will be used for the model. The difference between 7,000 and 3,000 cavalry is significant. Delbrück is less persuasive in his discussion of Pompey's cavalry; the main argument seems to be that it was difficult to recruit that large a cavalry force. Keppie and everyone else seem prepared to accept the 7,000 number. Somewhat reluctantly bowing to the majority opinion, I decided to accept the 7,000 number for the model, even while believing that Delbrück might be right and that it is too high. For the same reason I stayed with the 22,000 infantry for Caesar, but will follow Delbrück by including the 2,000 antesignani to fight with the cavalry. Slingers and archers should probably be added as a part of Caesar's army but I cannot find anyone who puts numbers to them. In 3.46 Caesar makes reference to slingers but does not refer to having archers. The model will not include them.

The revised army sizes to be used in the model are as follows:

For the Model


40,000 men, 88 cohorts @ 450 men (Delbrück)
7,000 cavalry (compromise number)
3,000 archers (Caesar)
1,200 slingers (Caesar)


22,000 men, 80 cohorts @ 275 men (Caesar)
1,000 cavalry (Caesar)
2,000 antesignani (Delbrück)

The camps

Alan Richardson has an article, "The Order of Battle in the Roman Army: Evidence from Marching Camps," Oxford Journal of Archaeology 20(2) 171-185 2001, which gives a Roman rule-of-thumb for calculating the dimensions of a camp based on the army's manpower. Using his formulas Caesar's camp would have enclosed about 33 acres (133 decares) and Pompey's about 82 acres (331 decares). The map on the right shows the camps in their proper sizes.

The camps were about 30 stades (3.94 miles, 6.34 kilometers) apart. The circle on the map is 30 stades in diameter.

The infantry


Pompey's infantry model has 88 cohorts, 40,000 men, arranged in 3 lines, with each line being 10 ranks deep. The 88 cohorts will be considered to be organized into 8 full legions and one partial. Given the mixed nature of Pompey's troops, it is possible that not all of his forces were organized around the legion, but since there is no good information about any alternative organizational structure the legion formation will be used. Each "legion" is arranged in the standard triplex acies formation of 4-3-3. There are 8 legions (80 cohorts) in that formation. The remaining "legion" is arranged 3-3-2. The cohorts average 450 men arranged in 10 ranks and 45 files. Alternate centuries have 7 files and 8 files.


Caesar's infantry has 80 cohorts, 22,000 men, 3 lines. Pompey's front was 35 cohorts wide (8 legions x 4 cohorts in the first line plus 3 first line cohorts from the extra 8 cohorts). The 35 cohorts of 45 files each had 1,575 files. The model will suppose that Caesar at least approximated the same frontage.

There is no mention in this battle of Pompey trying to simply outflank Caesar with his main army. If the two armies had arranged themselves with an equal number of ranks Caesar's army would have been just a little over half as wide in its frontage. Pompey could have easily enveloped Caesar. That this was not a tactic would seem to indicate that Caesar somehow matched Pompey's width on the battlefield.

What would have kept Pompey and his generals from trying to overlap Caesar's infantry, even given that Pompey had already decided to keep his ranks 10 deep? It would seem that Pompey did not expect to be able to outflank Caesar. He counted on his cavalry to do that job. This could indicate that he anticipated that Caesar would match his line width -- possibly because that was the normal thing that was done, or possibly because the two armies had lined up opposite each other on several previous days and each had seen the other's dispositions. For whatever reason, it seems that Pompey anticipated that the two lines would be equal and he developed the cavalry strategy. However, even at the last minute Pompey might have decided upon envelopment by infantry if he has seen Caesar's line to be significantly shorter. If, for example, it were only half as long as his own. I think it most likely that, when Pompey looked across the battlefield, he saw Caesar's line as roughly the same length as his own, as he had expected to, and therefore continued with his cavalry flanking tactics. At least, that is the basis I will use for the model.

Caesar's 80 cohorts divide nicely into 8 "legions" even though that is an artificial construct (the 8th and 9th legions, for example, were so depleted that they were more like one). The 8 "legions" in triplex acies formation would have 32 cohorts on the front line. If each cohort had 48 files (6 centuries of 8 files each) then the total would be 1,536 files against Pompey's 1575. This would certainly qualify as "roughly the same length."

If Caesar's cohorts were 48 files wide then, having an average of 275 men each, they would have been only 5.7 ranks deep. Within the cohort, each century would be 8 files wide, 4 centuries would be 6 ranks deep and the other 2 only 5 ranks deep; this totals 272 men to the cohort. The three extra men will not be accounted for in the model.

The illustration on the left shows the two sizes of centuries for Pompey (blue) and Caesar (red). The narrow box at the bottom of each century rectangle shows the possible location of the optio, aeneator and signifer as was discussed in one of the earlier sections. They are represented as standing at the back of the rectangle, somewhat behind the main body.

The drawing on the right shows the six centuries arranged into a cohort. The darker blue and red rectangles are the symbols that will be used in the army models. The small rectangles enclosing the optio group is not included in the cohort box because it would give the whole cohort the appearance of being much deeper than it really was.

Below the two cohorts for Pompey and Caesar are the standard six cohorts used in the earlier battle scenarios. Those were built from centuries of 8 ranks by 10 files. The different sizes for centuries and cohorts from one model to the next is not a defect. Most certainly the sizes did differ considerably from army to army and even from battle to battle. The example here merely serves to highlight this.

Putting these cohort symbols together makes the legion. On the left a 4 legion army of Pompey's cohorts (blue), Caesar's cohorts (red) and below them, for comparison, the 4 legion army used in the battlefield scenarios previously.

In the earlier scenario the models used individual centuries as the smallest units. For Pharsalus, with its much larger numbers, the smallest units shown will be the cohort.

Caesar's cohorts and legions have a broader front than Pompey's, as is necessary if Caesar is to match Pompey's front with a smaller force.

The complete armies for Pompey and Caesar at Pharsalus are below. The two armies are shown in the same starting position as has been used in the previous Frame 1 drawings. That is, they are positioned 600 feet apart.

It needs to be noted that the armies continue to be shown as if they were made of regular legions when, in fact, this was not the case. Pompey appears to have had cohorts in groups that were not exactly legion size. They may have been organized into ad hoc legions, or not. The texts are silent about this. Caesar certainly had some irregularities in his formation. On his left the 8th and 9th were so under strength that they amounted to a single legions. There is no description of just how they might have been arranged -- did each keep its 10 cohorts but just have them smaller, did each re-configure its remaining men into fewer cohorts of more or less normal strength. It would be pure guesswork to try to model the real situation and would introduce yet more unknowns into the models. I chose to use standard legion formations for the given numbers of cohorts and not to try to depicts odd formations about which we know virtually nothing.

These armies can then be placed onto the map of Pharsalus as below. The infantry armies are not shown right on the creek bank since all that would be necessary would be to stay close enough to the creek to prevent any sizable force from making a flanking attack. The armies are shown about 500 yards from the creek as it was drawn on the Greek map. At this point the creek bed is about 85 yards wide. Even though the creek itself may have been dry, the banks are said to have been up to 6 meters high, easily preventing any troop movements. There are about 2,400 meters, a mile and a half, between the flanks of the infantry and the nearest hill; it is within this space that the cavalry will maneuver.

The cavalry

Caesar says Pompey had 7,000 cavalry and lists 7 national units and their strengths. Those 7 are shown below. The small boxes are the 30 man "turmae" units used previously for modeling.

This accounts for only 2,800 men, leaving 4,200 unaccounted for.

With no additional information about the cavalry, the model will simply double the 7 units (5,600 men) and then add a 600, a 500 and a 300 man unit (1,200) to reach the total 7,000 man force. Caesar had 1,000 cavalry, shown in two 500 man units (red) in the drawing below.

As much as possible the individual groups of cavalry will be held together during the modeling process. If cavalry came in national units of 200 to 600 men strong then it would seem most likely that these groups would fight together rather than allow themselves to be split up.

The skirmishers

Source Notes

Gabriel & Metz

From Sumer to Rome: p 69 ff. "The most common use of the bow in ancient armies was as a weapon of indirect fire in support of infantry on the battlefield or in siege operations. Large groups of archers were positioned behind the main bodies of the contesting armies and fired in salvo to rain down a hail of arrows upon the enemy. archers were rarely placed at the flanks of the formations, a position that would have made them vulnerable to attack by cavalry or light infantry. Placing archers in the front to act as a screen would have made them terribly vulnerable to arrow attack from behind the enemy formation. The need to operate the bow with both hands precluded the use of the shields for protection.

"the massed infantry formations of ancient armies formed up approximately 100 yards apart and were 50 yards deep. Archers were positioned behind the infantry, a position that gave them a cone of fire against the enemy infantry formation of approximately 100 to 125 yards. As long at the enemy remained in place, archery fire could be brought to bear upon their formations. However, once the armies began to close with one another in a relatively short time, the angle between the archers and the enemy target formation became so acute as to make firing in salvo very difficult."

... A composite bow can fire a 553 gain arrow 250 yards at a 30 to 35 degree angle in about 5.8 seconds into a crosswind of less than 8 miles per hour. at that range, as our experiments demonstrated, an experienced archer can place 100 percent of his arrows within a 50 to 20 yard target box. Moreover, at that range area accuracy is easily achieved, and the archer can range his fire back and forth or from side to side over the entire target area with little difficulty. However, at 300 yards, the degree of accuracy drops off considerably to where only 50 percent of the arrows can be expected to fall within the target area. Ranging becomes almost impossible. As the target moves closer, from 200 to 100 yards, accuracy remains at 50 percent. p72 "An arrow fired from 250 yards passed through the formation of 40 degrees so that it had only one chance of striking 2 soldiers in the same horizontal plane. The changes of hitting the target at all were approximately 1 in 5 or 22 percent. p73"although archery fire may not have been effective against massed infantry, it was very effective against chariots and cavalry"


The Roman Army at War: p 188"On the open battlefield it seems that archers, who could fire over the heads of ranks in front, did not perform well in formed bodies, and were unable to defend their own frontage (i.e. they could not stop an enemy charge on their position by their own fire). Against cavalry, who could charge quickly and thus reduce the number of arrows fired at them, this is not surprising, even though horsemen made better targets than infantry. . . . It was a common practice to station light infantry and archers in support of cavalry, as will be discussed in a later section, but invariably once the enemy had routed this cavalry, the supporting infantrymen were quickly cut down. . . . p 189 "In once instance men ... the archers formed the fourth rank of a formation with three ranks of infantry in front and three of cavalry behind them. On other occasions, the archers and slingers were protected by earthworks and supported by artillery. On each occasion ... the attacks were stopped ... by the weight of missiles" ... However these archers would not have fired any more arrows than a cohort in the open so "In each case the difference seems to have been that the archers did not run away when the enemy came close." p190 "Archers normally appear supporting other troops and were often stationed behind them, firing over their heads. As such, they were especially useful supporting heavy infantry against cavalry, as we shall see in the section n dealing with this. If they were stationed ahead of the main line, it was normally as skirmishers, with other slingers and javelin men, who might have weakened the enemy, but could not close with him and would have withdrawn before any charge." p231 "Skirmishing troops normally fought in a fairly loose and open order as a protection from enemy fire. This made them exceptionally vulnerable to cavalry, since very few individuals would have stood against charging horsemen. As a result skirmishers charged by cavalry and, unable to escape or shelter behind other troops, suffered extremely heavy losses and even annihilation. "

P 243) "Caesar claimed that the German light infantry moved clutching the manes of the horses, being swift enough to keep pace with them. Does this imply that the light infantry also charged and fought interspersed within a cavalry unit's formation? It is difficult to believe that men could have kept up with galloping horses. If, as was commonly the case, one of the two sides in a cavalry charge fled before contact, then the infantry would not have served any useful purpose, being too slow to pursue. . . A cavalry unit with foot soldiers interspersed between its files would itself have had great difficulty manoeuvring properly, or in adopting any of the skirmishing formations of the type described by Arrian. It seems more likely that these mixtures of cavalry and infantry were in fact lines in which whole units of infantry were placed alternately to whole units of cavalry. The infantry must have been formed in fairly dense formations, since otherwise they were exceptionally vulnerable to enemy cavalry. … When friendly cavalry fled from an enemy charge, a dense knot of infantry could have provided shelter form pursuit. The pursuers were in a fairly loose formation, and vulnerable to a volley of arrows, pila, or javelins from the infantry, whilst probably not in good enough order to charge and break these. Cavalry combats were whirling affairs, with each side chasing and being chased by the other, backwards and forwards for quite long periods of time. Infantry block could lend stability in such a combat, being hard to defeat and static, able, unlike cavalry, to actually hold ground. Their fire support, particularly if archers, was very valuable. Only once the enemy had driven off the friendly horse, were his cavalry able to concentrate on the infantry… a legionary force, well protected by armour and shields, could have held out for a long time. Poorly protected javelin-men and unshielded archers might quickly have …broken."


The Face of Battle: P 90: In his chapter on the Battle of Agincourt in the section titled The Battle Keegan describes how the English archers used stakes to fortify their position against the cavalry. Although this has no direct parallel in Roman warfare it does serve to underline the vulnerability of unsupported archers unless protected by something (stakes, heavy infantry) that could withstand a cavalry charge. In that same section he estimates the range of the English longbows at 250 to 300 yards, with 300 being "a tremendous carry."


Three types of skirmishers are a part of the model. Pompey fielded both archers and slingers, Caesar had some variety of light infantry associated with the cavalry. Skirmishers require more space than heavy infantry for several reasons. Their weapons may require more space for safety reasons. Slingers are an excellent example of this. Also, skirmishers, at least when supporting cavalry as all of these were, must be highly mobile. Moving quickly in a tight formation would be difficult, the skirmishers needed more space around them.

The first image shows three archers in different poses and two light infantry running. The box around each man is 4 feet wide by 6 feet deep.

The light infantry is a challenge to model. There are only a few references to these troops and I have found no descriptions of their armament or method of fighting. They appear to have been different from either the archers or slingers so might well have been armed with a javelin. The model shows them with a shield and javelin. If the view was closer the extra javelins that they hold in their shield-hands could be seen. Until a better description of how they were armed or how they fought, this will have to do for a model.

The next drawing shows slingers in three poses. At the bottom the circles give a rough idea of the clearance needed for them to use their slings. The men in each row are 8 feet apart and each row is 8 feet from the next. the men are shown as offset only because this seems like a reasonable configuration. Archers could be arrayed in deep ranks firing volleys up over those in front. Slingers might need more room and a better line of sight.

It should be noted that in these illustrations there is no evidence to support the distances or the arrangement of the men. The drawings stand on their own merits as either reasonable representations, or not.

Using these distances larger units can be constructed. This drawing shows a small unit of light infantry (left) and archers (right). The units have 80 men, based on the nominal strength of the century.

These formations are purely speculative. In some respects they appear odd. One might expect a unit of skirmishers to be wider than it is deep. And that may well have been the case. But in favor of the above configuration is that narrow deep formations are much easier to maneuver than wide shallow ones. And skirmishers were expected to maneuver quickly.

The slinger formation is wide and shallow. Given the way a sling is used it hardly seems feasible for them to be arranged in a deep formation.

The proposed unit for slingers is only 4 ranks deep and each row is staggered to give more vision and more clearance to the men.

Six of these units make a cohort sized unit. The archer and light infantry cohort is on the left, the slinger cohort, arranged in two lines, is on the right. Each cohort has a nominal strength of 480 men.


Pompey had 3,000 archers (6 1/4 cohorts) and 1,200 slingers (2 1/2 cohorts), Caesar had 2,000 light infantry (4 1/6 cohorts). Those units are shown here. From top to bottom they are: Pompey's archers, his slingers, and Caesar's light infantry (red).

In comparison the infantry cohorts these will appear large in the models but that is only because the men are so spread out.

The armies

With the building blocks in place it is now possible to construct the two armies. The cavalry are placed on the flank, the skirmishers are placed on the flank but behind the cavalry.

This drawing shows Pompey's army in blue and Caesar's in red. The cavalry units are on the right. Behind Pompey's cavalry is a line of archer cohorts and then two lines of slinger cohorts. Caesar's cavalry appears as two barely distinguishable rows, behind them are his cohorts of light infantry. At this point there has been no particular effort to arrange the cohorts of cavalry or skirmishers.

The complete armies can now be fitted onto the map of Pharsalus.

Next Page: The Battle of Pharsalus: Deployment of the Armies

Site Map

© 2003, Gary Brueggeman. All rights reserved world wide. No part of this work may be reproduced in part or whole, in any form or by any means, without permission from the author.