Tentative Model of Cavalry Charging Infantry

This is a preliminary draft to explore the dynamics of the cavalry attack.


A reasonable start is to review some material from Ann Hyland's book.

Ann Hyland: Training the Roman Cavalry

(P 13): "Arrian subdivides cavalry simply into armoured and non-armoured. Today we give them the names heavy and light cavalry. The cataphracts, where men and horses are armoured, obviously belong to the heavy cavalry. . . . The rest of the cavalry is definitely light horse -- missile throwers, javelineers, archers. The javelineers are subdivided into purely missile throwers who never close with an enemy, and the sort who, when their javelins are cast, move forward and fight close in with the spatha."

(P 15): "Horses are in the main timid animals, as are most graziers. Their best means of avoiding trouble is flight. If attacked and unable to flee their second line of defence is to strike with either fore or hind hooves, usually the latter. With a frontal attack teeth may also be used as a defence, particularly if the anima is a stallion. . . . a horse not only assesses the enemy as something that is pursuing or attacking him: his natural timidity, which is present in widely varying amounts, will cause him to shy, or be ready to shy, at a great variety of things. Foremost among these are unusual looking objects, noises closeby, and things that flap and crack."

(P 16): "Any unusual movements of a rider can disturb a horse, particularly anything involving the rider's equipment."

(P 18): "As herd animals horses are easier to train to work in groups than for solo performances. . . . Horses working together are also competitive, or rather territorial, especially if entrires, and work at speed brings out this characteristic. Speed nearly always excites a horse, and when working in a group some individuals can become very hard to control. In some cases they become anti-social to their own kind, lashing out at horses behind them, or bloody-minded with their riders, resenting restraint and employing diverse methods to escape it. These can take the form of fighting the bit with resultant head throwing, and lunging against the bit. This would have made it very hard for a trooper to keep his rein hand still, particularly when also encumbered with a shield. When a horse fights the bit, fluidity of movement is lost and correct pacing extremely hard, if not impossible, to maintain. . . . Traveling straight . . . is one of the most difficult things to teach a horse and rider. . . . Most horses have a tendency to veer to one side or the other."

(P 21): "A 15 hh horse's stride at only an extended canter is approximately 11 ft 6 in to 12 ft." The footnote to this statement reads: "This measurement was taken using Nizzolan, my 15 hh Arab stallion who has a long stride, but not excessively long. He was at a very extended canter and as the near fore hoof touched down on one stride the distance was measured to where it touched down on the next stride, 11 ft 6 in being the actual measurement (the extra 6 in being allowed for a slightly larger horse)." Footnote 14 adds: "In a full gallop stride the measurement per stride would have been increased by 3 or 4 ft according to the length of each horse's stride."

(P 23): ". . . the 'gallop' of the Tactica must be seen as a controlled hand gallop speed where the horse is readily brought under control, and which would not be much faster than an extended canter. ... The shortest distance that American Quarter Horses, the fastest horses on earth over a quarter of a mile, run is over 220 yd. Many horses cover this in under 11.95 seconds. . . . This is close to 40 mph."

(P 33): "Earlier in this chapter I considered the gallop of the Tactica to have been carried out at controlled speed. Xenephon clearly states the horse must be under control at all times or the turns would bring him down."

Quoting passages from Arrian's Ars Tactica as per the Teubner Text of A. G. Roos:

(P 70): Arrian distinguished armoured and unarmoured cavalry. Unarmoured [4.2] "comprise on the one hand spear-bearers, pike-bearers and lancers and on the other skirmishers ... mounted-bowmen or javelin-throwers." [4.3] " The spear-bearers are those who approach the enemy ranks and fight them off with spears or charge and drive them back with pikes like the Alans and Sarmatians, and the skirmishers are those who discharge weapons from a distance, like the Armenians and those of the Parthains who do not carry pikes." He notes that some carry shields and some do not. "The name of skirmishers is given to those who do not come to close quarters but discharge their weapons from a distance; and of these some use throwing-spears and others bows." "Tarentines" -- those who use throwing-spears. [4.6] ... some skirmish only from a distance, keeping far off, or riding round the enemy in a circle, and those are the Tarentines proper. But others first discharge their weapons and then join battle with the enemy, either retaining one of their spears or using a sword (spatha)."

(P 71) [16.8] "It is hard to wheel about with square formations -- [which] are well arranged in as much as those so drawn up are in ranks and files and it is organized for easy charging and withdrawing; and only with this formation do the officers fall in a single body upon the enemy. The best are those that have the double measurement in the length rather than the breadth; for example, if there are ten men drawn up along the front and five deep. For such formations are oblong as regards number, but square in actual shape. For the length of the horse from head to tail fills out the square since the length of the horse is three times a man's width at the shoulders and as when they draw up nine in line along the front, they make the formation three deep. For this too should be borne in mind, that the cavalry drawn up in depth do not afford the same assistance as to infantry in depth, for they do not push on those in front of them, since one horse cannot push against another in the way that infantry push on with their shoulders and flanks, nor when they are contiguous with those drawn up in front do they constitute a single massed weight for the whole body of troops; on the contrary, if they mass and press against each other, they rather cause the horses to panic." [17.1] "An oblong formation in which ... the front is greater than the depth ... is better in contests. ... A single line along the front with no depth is convenient for unsuspected raids ... but for contests it is very disadvantageous." [33.1] Arrian makes it clear that the tactics described are derived from the Celts. His Ars Tactica describes elaborate practice maneuvers. Horsemen gallop in circular patterns throwing their javelins at targets. In the practice formations the turma appears to string itself out as individuals following one behind the other through the loops of the pattern.

(P 71): In 16 and 17 Arrian discusses three basic formations: the square or rectangle, the wedge and the rhombus (diamond). He notes that the wedge and rhombus are better for wheeling [16.8]"For it is hard to wheel about with square formations. Whereas the formation which comes forward to a point, even if it increases in-depth as it goes forward, nevertheless by wheeling in a short space at the very beginning enables the whole formation to extend its line with east. 16.9] Square formations were employed especially by the Persians, by the barbarians in Sicily and by most of the Greeks and those most renowned for horsemanship. [16.10] For this formation more than any other is well-arranged, in as much as those so drawn up are in ranks and files and it is organized for easy charging and withdrawing; and only with this formation do the officers fall in a single body upon the enemy."

Page 76: [43.3] "They then draw their swords and carry them along making a variety of strokes, to the utmost of their skill, either reaching an enemy in flight or killing one who has fallen, or carrying out some action obliquely as they ride alongside."

(P 80): "The pike was also known as a contus. this was a sturdy weapon employed by the heavy cavalry in a massed charge, wielded two-handed, and driven home hard with the full thrust of the body behind it. . . The spatha blade was 34 inches. Josephus mentions three or more light javelins being carried in a quiver."

(P 81): "Vegetius, writing in the latter half of the fourth century AD throws light on to javelins and other weapons. He informs us that five loaded javelins were carried in the concavity of the shield and two other javelins were also carried, the largest 5 1/2 ft long in the stave, with a 9 in head, formerly called a pilum and now 9i.e. in his day) called a spiculum, and a smaller one with a 3 1/2 ft stave and a 5 in head, formerly called a vericulum and in his day , a verutum. This would tie in well with the light rapid fire javelins of 36-9 and 40.8-12 and the lance/javelins of 41.1-4 and 42.1-4."

(P 83): "In the Hannabalic Wars, 218-02 BC, the Numidians employed their javelins in hit and run methods, while the Spanish heavy horse was used in head-on charges[see Hyland Equus, p. 151]. Crassus' debacle at Carrhae in 53 BC had been largely delivered by the Parthian heavy horse cataphracts coupled with the light horse archers. According to Tacitus Samaritans ... employed their national tactic of massed charge with the contus, followed by hand-to-hand fighting with their heavy swords."

(P 111): "At speed, and particularly if doing a linear movement, if two horses did collide there would have been total disarray as those following would not have been able to extricate themselves in time, particularly if in close order."

(P 115): "At one gallop stride a 16 hh horse can cover at least 16 ft (I had a 15.2 hh standard bred trotter who extended trot stride reached eighteen feet)."

(P117): Hyland cites the Strategikon on the size of horses: "Maurice explains a ploy for confusing the enemy as to the size of the army. The cavalry were to be lined up extremely close together so that the whole unit looked much smaller than it really was. This left no room to manoeuvre. Each horse was allotted a space 3 ft wide by 8 ft long. I have measured this out with a horse. Katchina at 14.3 hands hits easily into this space lengthwise, and when he is in fit condition into the width also. Nizzolan, of oriental blood being pure Arabian, and a much leaner type but slightly taller than Katchina at 15 hands fits, with a little room to spare. There is just sufficient room widthwise for the troopers knees, though in Maurice's army they would have been riding stirrup to stirrup. ... A height range of 14.2 to 15.1 hands would be the expected average for chargers of the 6th century."

(P159, footnote 8): "Horses tend to become very excited and sometimes hard to control when galloping in a group. It brings out their competitiveness."

(P166-7): "Why did a horse willingly to into battle time and time again? As stated earlier a horse's prime defence is rapid flight. He is also a herd animal who will run with the majority. The speed and herding instincts can be channeled by man. In cavalry engagements this resulted in a massed charge. Added to the speed was competitiveness, and if stallions were being used a little natural aggression would also have been present. In a close-order engagement speed, noise, and competitiveness created excitement. With this combination the pain of a wound received, unless it was severe enough to disable a horse totally, would have been masked by the excitement of the moment. later when the excitement had abated and the pain was felt the horse would not have connected the action to that delivered it with the pain he felt later. . . .The very nature of a massed charge of heavy horse meant it was delivered and the horses then took time to re-group. If bogged down and hand-to-hand fighting ensued the turmoil would have kept the excitement in the horse at a certain pitch. Any single reciprocal missile that landed on or even in a horse would not necessarily have been attributed by him to far off skirmishers. A continuous retaliatory barrage would have been understood and obviously the horse would have tried to turn tail and run, but by that time, as a skirmisher, his rider would have taken himself and his precious horse out of range."

(P 172): "The modern sports javelin at 8 ft 6 in approximates to the 8 ft Roman hasta, though of course the heads are different."

(P173): "The speed of a foot javelineer prior to the throw was very important. The speed of the horse definitely lent power to the cast. As it leaves the hand the speed of a modern man's javelin weighing 800 grams is between 27 and 30 meters a second. A speed of 33 meters per second is exceptional. Loss of force occurs in flight. As it reaches its highest point, before starting on its downward flight, it is traveling at approximately 9.81 meters per second."

Methods of Attacking

The first thing to consider is just how a mounted man can attack an infantryman who is facing him and not running away.

Frame 1:

This frame shows that it is not possible for a mounted man traveling at speed (the blue arrow indicates motion) to attack directly into a formation of armed infantrymen. Horses will not charge a solid object, especially one that has a lot of sharp objects like spears or pila projecting toward it. Unless the formation breaks up (a different situation that the one being considered in Frame 1) there is no place for the horse to go. The illustration shows a single cavalryman. If he were charging as part of a cavalry formation it would be virtually impossible for the horse to turn or even stop because of the horses on either side and behind.

In this frame, if the horse will not run into the shield wall and it cannot turn aside then it would have to come to a halt. And this would shift all of the advantages to the infantry who could attack the large undefended horse as well as the rider.

Is the solution to imagine that the horse could charge at a slow gallop to the moment of impact when the tip of the spear strikes the soldier and then come to an immediate stop? No. This idea still has difficulties. Even the longest pike, contus, would not project more than a few feet beyond the horse itself. And the horse could not come from even a slow gallop to a complete stop in just two or three feet. Even if it could stop, it would then be well within range of the range of all of the other infantry around it. Neither horse nor rider would survive long. And then there would also be the problem of the horses behind it in the cavalry formation. A sudden stop by the front horse could, and likely would, lead to collisions between the ranks of horses.

Frame 2:

The only way a straight ahead attack at speed is possible is when the cavalryman can ride past the infantryman as the blue arrow indicates. However, if the infantry holds its formation, especially if there are supporting ranks behind (as in Frame 1), then there would not be a lane for the horse. The attack illustrated here could only be possible against an isolated man, against the end man of a formation or against a formation that is so open that horses can easily ride right through it. This is the style of attack motion pictures are so fond of because it looks dramatic. However, in reality, the infantry would be likely to take one of two options. It would panic, break up and start to run away before the cavalry ever got close or it would close ranks and face the cavalry charge. The scenario illustrated in Frame 2, in which a lone or isolated soldier faces the cavalry charge, is most unlikely.

The problems with the straight ahead attack as illustrated in Frames 1 and 2 cannot be solved by giving the rider a longer lance, a contus, because the difficulty is not with the length of the lance but with finding a path for the horse.

Frame 3:

The next three images show different types of oblique attacks. The cavalryman can ride parallel to the infantry front, striking at them as he passes by. The illustration on the left shows a couched lance. With minor adjustments the lance could be lengthened into a contus. On the right the two illustrations show a lance held in the overhead position; first with the attack to the left so that the shield offers some protection. The far right shows the attack from the other side so that the cavalryman has a little long reach with his lance.

It seems unlikely that cavalry could actually attack in any of these three ways for, as the illustrations make obvious, they all leave the horse extremely vulnerable to attack.

Frame 4:

The problems are worse for the cavalryman armed with a sword. He cannot close facing the infantry (the left illustration) because his reach is simply not long enough. And if he tries to fight to the side (the right illustration) he subjects his horse to attack by at least three men.


It is difficult, perhaps almost impossible, for cavalry to successfully come to close combat with a solid mass of infantry if that infantry holds its formation. If the cavalryman comes close enough to use his spear, however long it might be, the infantry can easily attack the horse, which presents a rather large target. A very heavily armored horse might be somewhat safer, but it would certainly not be invulnerable, especially to the penetrating power of the pilum..

A cavalry formation cannot charge directly into a mass of infantrymen for there is no place for the horses to go. Standard battle diagrams like the one at the right in which the cavalry (blue) attacks the flank of an infantry formation (red) break down when one considers the action at the level of the individual soldier who simply cannot attack straight ahead.

Cavalry attacks against infantry should probably be thought of in two general categories.

The first is the massed attack that has the object of creating panic such that the infantry formation actually breaks up before the cavalry make contact. However, if the infantry holds its formation then the cavalry must turn aside before actually making contact. This is a critical consideration.

The second is the skirmishing attack with missiles from a distance.

The following sections will look at these types of attack in some detail.

Flank Attack

What is fairly certain is that cavalry generally did not attack heavy infantry from the front. A steady line of soldiers can stop a cavalry attack if it does not waver. And, once it is halted, the cavalry loses it advantages. Effective cavalry attacks were almost always against the flank or the rear of the infantry. This is sometimes described as attacking the flank and just rolling up the enemy infantry until the army panics and a rout develops. During the rout it is the cavalry's role to pursue and cut down the fleeing soldiers. It is likely that most of the casualties experienced by the losing side happened during this phase of the battle.

The first aspect to be explored is the notion of the flank attack. It sounds fairly straight forward. One's own cavalry defeats the enemy cavalry, chasing them from the field. It then turns against the exposed flank of the enemy infantry formation. This causes panic in the ranks and the formation breaks up. The soldiers run for their lives, the cavalry chases and cuts them down.

When one tries to actually model this, however, there are some difficulties.

The Cavalry

Cavalry came in many varieties over the centuries. For the sake of simplicity the cavalry accompanying the Roman army will be used as a generic model. That cavalry was eventually organized into turmae of 30 men, probably configured in three rows of ten. The ala was comprised of 10 turmae, 300 men. The total cavalry force could vary from as few as 1,000 to as many as 7,000.

The illustration on the right shows the turma and the box used to represent it in subsequent drawings. The ranks are 12 feet apart, the files are 4 feet apart. This is a parade ground formation. In battle, galloping to the attack, the turma would have to spread out more.

Cavalry is best used in wide shallow formations (Arrian: 16.8). It does little good to arrange horses 10 or 20 deep because those at the back have little or no impact on the fighting at the front. Much better is to spread the formation and outflank the enemy.

For the purposes of the model the cavalry will be arranged two turmae deep, that is, 6 ranks of horse. A 1,000 man cavalry force would look something like this.

The Infantry's Flank

The appearance of the flank of the enemy would depend on the type of enemy one is talking about. If the Roman legion is taken as the example then a typical Roman army comprised of four legions would look like this. The front of the army is toward the top.

The right flank of that army would look like this. The army is divided into three lines separated by some distance. In this model the second line is 100 feet behind the first line and the third line is 200 feet behind the second.

The small boxes represent the centuries. Each century, at full strength, has 10 files and 8 ranks for a strength of 80 men. The parade ground formation has the ranks on four foot intervals and the files on three foot intervals.

In battle the second line may well have almost merged with the first line. The third line could be anywhere on the field but late in the battle the third line would very likely be close behind the first two lines.

However, the model will be based on the parade ground formation as illustrated on the left.

The Problem

The problem in conceptualizing the flank attack is the relative sizes of the cavalry force and the flank of the army. When drawn to the same scale and placed next to each other the difficulty becomes immediately clear.

The legion in its parade ground formation is only 366 feet deep from front to back. (In battle the lines might close up making the depth of the legion significantly less.) The 1,020 man cavalry formation is 900 feet wide. When it attacks the flank less than half the cavalry units can actually engage.

The problem is even worse when looked at in detail, as on the right. Of the turmae that overlap the legion only parts of three actually make contact with the infantry. In fact, only 18 horsemen actually engage the infantry.


What should we imagine that the rest of the turmae do? There really is nothing for them to do but to curl around and attack the rear of the army. This is illustrated on the left. Those turmae that are able to engage the infantry are marked with an X inside the boxes.

The turmae without an X have no immediate way to attack. The two turmae attacking between lines one and two might swing around and attack the rear of the front line centuries. The 8 turmae between lines two and three could possibly ride down the corridor between the lines and attack to their right or left as the opportunity arises. However, it could also be a highly dangerous place to be -- caught between two lines of heavy infantry.

But above all, what is clear is that most of the turmae are engaged not in a flank attack but in an attack against the rear of the army.

If this is the nature of the attack, then another diagram may prove more accurate. Instead of the cavalry attacking directly from the side, perhaps it is better to envision it coming in at an angle from the rear. Since most of the attack is against the rear and not the flank, it may have made more sense to direct the impetus against the rear and let a few turmae ride around to attack the flank.

A somewhat more realistic attack model might look more like the illustration on the left. The three blue lines are relatively close together. The enemy infantry is indicated by the single line of red rectangles across top. It has been and may still be actively engaged with the first blue line in hand-to-hand fighting.

The attacking red cavalry is split into a small group of four turmae that attack the right flank and the large group of 30 turmae that attack the back of the blue legion.

This illustration raises several new issues.

The first issue is the terminology "flank attack." As was noted above, this lends itself to vague imagery of cavalry massing against the flank of the army and rolling it up. In fact, however, it is clear that a large number of cavalry cannot attack the narrow flank of the army, be it Roman or any other army. For no ancient army ever routinely arranged itself more than 16 or 20 ranks deep (80 feet) and no large cavalry force would ever attack along a front that narrow a front. So what is called a flank attack might more appropriately be called an attack against the back of the right (or left) flank. And, clearly, the idea of attacking the side and "rolling up" the infantry could not have happened. Virtually all "flank" attacks were actually attacks against the rear of the army not the sides.

The other issue is the rout. This is more difficult. The blue army, about one legion in the illustration above, is surrounded on three sides by the red enemy forces. This blue legion's left flank would be closed by the legion next to it in the line of battle. This legion is not shown in the illustration. The blue legion is thus hemmed in on its left by its own forces, in front by the enemy infantry and in back on on its open right flank by the enemy cavalry.

The men of the blue legion cannot escape; they cannot flee. So if this were the cavalry attack method, there would be no rout among these men at all. Instead, they would be forced to stand and fight where they are. And this often proved a very difficult nut to crack. Once the legionnaires formed a defensive square they were often able to extricate themselves. And, even if they could not do that, they were difficult to kill. And it took a long while. None of this fits the panic and rout that the cavalry attack was supposed to create. It may perhaps be that a different type of cavalry attack must be envisioned.

But first a more in-depth look at the head-on shock attack will be taken.

Shock Attack

By shock attack is meant the attack of heavy cavalry directly into the infantry line. These cavalrymen are armed with a thrusting spear and long sword. They likely wear armor and are protected by a shield. Their horses may or may not be armored (for many centuries the horses were not armored).

The question here is just how is such an attack carried home.

Speed of the Cavalry

The Table at the right shows the speed of cavalry horses as estimated on a previous page, "Cavalry: Improved Model.".

Ann Hyland mentions the "hand gallop." For the purpose of these models the cavalry will be considered to approach at the slowest easy gallop speed, 9 miles per hour, to keep the formation in order. For the final attack the charge is at the full gallop speed of between 15 and 22 miles per hour. The fastest horses are considered to have been reined in to keep more or less in line with the average horses.

The speed of the cavalry changes somewhat for each of the scenarios. Those changes will be outlined in the context of the scenarios themselves.


In the models that follow the infantry will be represented as facing directly toward the cavalry. It would seem that a mass of 1,000 horses galloping down on the infantry formation would be noticed in sufficient time for the officers to turn their men to face the new danger. This could be complicated for men in the first line that are also facing enemy infantry in hand-to-hand fighting. However, since it would usually be the rear of the army being attacked, those centuries should be able to turn right around and face the oncoming cavalry. These men, being part of the reserve troops, would still be armed with their pila. Thus, in these models the attacking cavalry is always seen as attacking men that are facing them and prepared to withstand the charge.

Scenario 1: Shock Attack by Cavalry in Close Order

It is said that well-trained infantry can stand up to a cavalry charge. If the infantry do not break ranks then the horses will not run into the hedge of spears, shields or swords. But what happens then? These models seek to explore the dynamics of that scenario.

The first illustration shows the basic formations. On the left is the century armed with pila. On the right is the turma. At the top is a set of lines that show how far the cavalry can move in one second: 10.3 feet at the trot, 13.2 feet at a slow gallop and between 22 and 32 feet at a full gallop. The turma is positioned about 22 feet from the infantry formation.

Cavalry: The cavalry for this scenario should probably be thought of as heavy cavalry. That is, armored men, perhaps armored horses. The thrusting spear may have been longer than the one drawn for these generic images. The speed of the horses may have been somewhat reduced because of the weight. The spacing for the horses in the illustration is 4 feet between files and about 12 feet between ranks. The distance between the ranks would be 12 feet when walking but would have to become greater when the charge began.

The infantry formation is a standard century of 80 men in 10 files and 8 ranks. In practice the century would be understrength by 10 to 20 percent.

Preliminary Considerations: Even this first illustration raises some questions that must decided before further modeling can be done. The most critical is the spacing of the cavalry files.

As noted above, if the infantry holds its formation then the horses will not charge into a hedge of spears. Therefore, the turma must be able to either turn away or come to a stop before making physical contact with the infantry.

If the turma were to attack in the formation illustrated above then it would be virtually impossible for any of the horses in the front rank to either stop or turn aside. Four feet is not enough lateral room for a horse to turn around at speed. 12 feet between ranks represents a separation of only 1/2 second, not enough time for the following horses to react if the ones in front were to suddenly stop. Therefore, in this formation the turma is not able to either stop or turn aside quickly. It would have to decide well in advance to take either action. The turning radius for the formation is a minimum of 42 feet, as per the illustration at the left. And this would be under the very best of circumstances. In practice, the turma would probably have to begin its turn at least 100 feet from the infantry. This allows a minimal 4 seconds before the time of impact. Four seconds would also be about the minimal time, under ideal circumstances, to bring these 30 horses to a halt.

This would mean that the cavalry commander would have to make a rapid assessment of the tenacity of the infantry being attacked from well over 100 feet away. He would have to decide that they do not look like they are going to give way and then issue the order to halt or turn aside. (A related question is whether or not the infantry would be likely to turn and run when the cavalry is still over 100 feet away. That, unfortunately, is almost solely a matter of psychology and somewhat beyond the scope of this modeling technique.)

Conclusion: Because of the inability of the close order turma to maneuver close to the infantry this particular attack scenario does not appear to be viable.

Scenario 2: Oblique Attack by Cavalry in Close Order

If the cavalry does not charge straight onto the infantry flank but, instead, comes in at an oblique angle then it is much easier to for the formation to turn aside if it has to. In the straight on charge the horses must turn through at least 90 degrees. With the oblique charge a fairly minor change in direction can suffice. This would appear to be feasible with trained cavalry.

Scenario 3: Oblique Attack by Cavalry in Wedge Formation

If the cavalry adopted a wedge formation in which the leader was more clearly visible then the turning maneuver would be even easier to execute.

Scenario 4: Open Order

The problems outlined above arise because the cavalry spacing does not allow sufficient room for horses to turn. This is solved if the formation is more open.

How much space would a horse need to turn around? American rodeo barrel racing provides a fairly clear answer; very little. Quarter horses turn around 24" barrels at speed, completely reversing their direction. The illustration shows one horse turning around a 24" diameter circle. Seven feet between files would be enough to allow one horse to turn inside the other. As for the separation between ranks, a one second separation would give the following horse some time to react to the movements of the lead horse. This translates to about 22 feet. This is close to twice the standard file width of 4 feet. It would have been easier, and provided an extra margin of safety, if the open formation were just twice that of the closed formation; that is, 8 feet between files and 24 feet between ranks.

That open formation would look something like the illustration at the left.

The turma is shown just 22 feet ( about 1 second at a slow gallop) from the infantry flank. At this point the unit could either stop or the horses could turn inside each other to reverse direction.

At the moment depicted in the illustration the cavalry either must halt the attack or fully commit to it. That means that the infantry would be expected to have already broken ranks if it were going to.

It seems obvious that, if the attack were to go forward, that at least four files of cavalry would be able to swing around the back of the century and attack it in the rear.

If the infantry were to hold its position then what can the cavalry do next? Four ranks can swing around the rear and attack there. The other 6 ranks must either stop or reverse direction. In that case the turma could simply ride off and try an attack somewhere else, re-group for another attack against this century or ride around the rear to join in on the attack of the 4 files. A fourth alternative would be for the turma to re-group and then try a different style of attack, a harassing attack (see below) with the goal of disrupting the integrity of the infantry line.

In the meanwhile, the century must face about to meet this threat to its flank and possible also to its rear. This effectively immobilizes it since it cannot move forward while men are facing both to the side and to the back.

Scenario 5: Shock Attack, The Infantry Does Not Hold

If the infantry does not hold its line but, instead, breaks and runs then the cavalry presses its attack home. The particular cavalry envisioned for this model is armed with a thrusting spear and with swords. The next model seeks to explore the dynamics of that successful charge -- just how the cavalry actually engages the infantry when the infantry formation disintegrates.

Frame 1: The first frame shows the moments just before initial contact. In this illustration the century has turned to meet the cavalry. Since the model is for a century in the second or the third line it is possible that all of the files could turn to face the cavalry charge, be it to the side or to the rear. In the cavalry were attacking the first line, the line actively engaged in hand-to-hand combat with enemy infantry, then the scenario would look quite different.

The starting position for the model sequence shows the cavalry charging at a slow gallop so that the horses all stay together. They cover 22 feet per second. In this frame they are 110 feet from the infantry, a distance they will cover in about five seconds. This is the minimum distance for the cavalry officer to decide if the charge should carry forward or turn aside.

Frame 2: The next frame shows the same position but shows the infantry beginning to break as some individuals pull back from the line.

Frame 3: The time is 1 second later. Panic has begun to spread through the ranks. Men turn to run. As the cavalry sees the rout start the lead horses pick up their speed and the lines are not as straight as when they were riding at a slower more controlled pace.

Frame 4: In two seconds the panic has spread through the entire century. Everyone is running now. Men can run at 12 to 15 feet per second, 8 to 10 miles per hour, at least for short periods. Those at the back (left), with no one in their way, are able to run away faster than those on the left who get jammed together as they try to escape. The cavalry is now just 3 seconds from making contact.

Frame 5: One second later. The infantry are in full rout. The cavalry is charging at full speed. The fastest horses cover about 32 feet per second.

Frame 6: Contact is imminent. Because the infantry is running away the initial time of contact has stretched from 5 seconds to 7 or 8 seconds. But the fastest horses will catch the slowest runners in the next frame.

Frame 7: The fastest two riders have caught up with the slowest runners.

This illustration shows a detail of Frame 7.

The speed of the lead horses has to be considered at this point. The two fastest horses have been galloping at 32 feet per second. The soldiers at the back of the infantry are only running about 12 feet per second. If the horses were to continue at their full speed they would overrun the infantry, as shown in the illustration below.

This is the scenario the motion pictures like so well -- mounted men riding through a formation of foot soldiers, slashing them to pieces with their swords. But is this a reasonable image? Would the cavalry wish to ride into the midst of the enemy infantry? Even though they may be frightened and running away, these are dangerous men, trained fighters and well armed. The rider can only effectively strike and defend on one side of the horse. On the other side, the horse is vulnerable to any blows from the infantry. Once the horse and rider enter the formation they are exceptionally vulnerable to a pilum or sword thrust. Even when running for their lives, the soldiers may well strike at a horse if it gets close.

On that basis it seems unlikely that the cavalry would actually ride into the infantry. Rather, they would slow their horses to stay just behind the infantry and do their killing from there. Of course, cavalry could also ride along the flanks and even gallop ahead of the running soldiers to cut them off.

Frame 8: The lead horses have slowed to the pace of the infantry.

Notice that the cavalrymen who strike an infantry soldier do not immediately disengage. The model has them engaged for approximately 5 seconds. This provides some time for the blow to be delivered, for the cavalryman to recover his weapon and ready his horse again, for the infantryman to attempt some evasive move, and time for the infantryman to either defend himself or even attempt an offensive attack. During these five seconds the horse is shown moving around to simulate the way in which an actual encounter might occur. This movement of the horse creates problems for the remaining cavalry coming up behind since they have to channel themselves around these spots of combat.

Frame 9: In this frame 5 riders are attacking runners, the rest of the cavalry flows around them and continues the pursuit.

Frame 10: The pursuit continues. Another runner is speared.

Frame 11:
The first two casualties appear, the red figures. These are the two men first attacked in Frame 7.

Frame 12: As more and more of the cavalry close with the infantry the horses begin to crowd together, sometimes to the point where they cannot effectively attack because of their closeness to each other.

Frame 13: In Frame 13 the timeline in purple at the top has been extended further to the left.. The ticks represent the distance the cavalry can travel in one second.

At this point the attack sequence has been carried far enough to work out some of the mechanics.

The key element is that, as it catches up to them, the cavalry slows down to more or less match the speed of the infantry. After some seconds much of the cavalry force is compressed right behind the running infantry. Once the infantry begins to break apart and create lanes then the cavalry will penetrate the lanes.

The first infantryman was attacked at Frame 7. In the following six seconds there have been six deaths. At that rate it would only take a matter of minutes for the turma to chase down and kill all of the 80 members of the century. More cavalry do not necessarily make the process faster. As individual fights erupt they create bottlenecks. More horses may simply make the congestion worse and not lead to a significant increase in the rate of the killing.


The models suggest several conclusions.

1. The term "flank attack" may be misleading since a large cavalry formation cannot reasonably attack the narrow flank of the army. Virtually all attacks had to be against the rear of the infantry formations.

2. If the infantry did not panic and run away then shock attacks in which the cavalry charged with a thrusting spear had to begin turning aside at around 100 to 150 feet.

3. If the infantry did panic and run the cavalry would pursue them at approximately the speed of the running men since it would be dangerous for any single cavalryman to ride into the midst of the infantry soldiers.

Harassing Attack

The harassing attack would use missile weapons to break down the infantry formation and then pursue the routed army. There were two main types of missiles, the dart or javelin and the arrow. For this model only the dart or javelin will be considered.

These types of tactics were used by many types of cavalry. Most famously, the Numidians.

In her book, Training the Roman Cavalry, Ann Hyland illustrates a number of the formations described in Arrian's Ars Tactica. The illustration below is loosely based on two of her drawings. The maneuvers she describes are training maneuvers in which two teams compete against each other. However, precisely the same maneuvers could be used against any target, especially against infantry formations.

The left formation would allow the 9 riders to circle more or less continuously while they discharged one javelin after another at the enemy line. The formation on the right would let each rider make a single pass across a long front or they could circle back at some point.

It should be noted that the riders attack with their left sides toward the target/enemy. This is to provide themselves with the protection of their shields. The javelin throw is cross-body. Because of the torso twist involved, this cross-body throw to the left generates more distance than a throw to the right.

Hyland (Page 173) has an appendix that discusses the range of the javelin but, oddly enough, does not determine the range of javelins thrown from horseback. Because the trajectory is largely at about a 45 degree angle to the direction of the horse's travel, only half of the momentum of the horse is added to the speed of the javelin. Hyland notes that a javelin speed of 27 to 30 meters per second might be typical for a man throwing from a running start. From horseback the body mechanics are not quite as favorable. At a full gallop the horse only covers around 10 meters per second, half of that would be only 5 meters per second. Compared to the overall speed of 27 to 30 meters per second, the speed of the horse contributes relatively little to the throw.

Hyland does note that the well trained soldier should have been able to achieve some considerable distances by using his entire body in the throw. Nevertheless, it would seem obvious that a throw from a seated position would be less powerful than from a running start.

On the page titled "The Skirmishers" the range of the javelin was estimated at 120' for an average and 150' for a maximum. It seems likely that range of the cavalry javelin would be at least half that but certainly not greater than that -- between 60' and 120'. For the purpose of modeling, an average range of 90' will arbitrarily selected. 90 feet is likely to be within a few yards of the actual distance achieved.

Arrian distinguishes between skirmishers who attack only from a distance and those who draw their swords and attack close-up once their javelin supply is exhausted. For the purposes of the models, the close-up attack can be viewed as somewhat similar to the shock attack examined above. This model will focus solely on the missile attack that seeks to break up the infantry formation.

The model for the skirmishing attack is much simpler to work out. The cavalrymen ride to within missile range and, using one configuration or another, throw javelins into the massed infantry. Any number of formations could be used, including the two shown above. The illustration below shows three centuries being attacked by a turma. The turma stays 60 feet from the century, just outside pilum range but well within the 90 foot range of their own javelins.

In the model as illustrated above, the cavalry ride around a loop approximately 300 feet in circumference. At a "hand gallop" of about 13 feet per second it takes about 25 seconds for each man to the complete loop. At that speed it would take about 75 seconds for every man in the 30 man turma to make one circuit. Allowing for normal delays it would be reasonable to allow 90 seconds, 1 1/2 minutes, for the cycle. If each soldier has 8 javelins and they throw two javelins on each pass then the turma's attack would last for about 5 minutes at which time it would have discharged a total of 240 javelins at the three centuries.

However, there would be almost infinite variations possible. The turma could turn in a tighter circle so that it attacks only one century. It could, alternately ride along an extended front and not make a full circle at all. The turma could also close to a much closer range if it were not particularly concerned that the soldiers might launch a pilum volley at them.

It is easy to imagine that two turmae could work together to attack both the side and the flank of the centuries as in the illustration below. In this illustration the turmae are closer to the infantry, 40 instead of 60 feet. The turma on the right is in the formation Arrian describes, a circle. Each of the ten man lines will take its turn circling until its supply of javelins is depleted. Then the next line will take its place.

In the section above on the Shock Attack it was postulated that pure flank attacks were unlikely because of the width of the cavalry formation compared to the depth of the infantry formation. For skirmishing attacks the situation changes because the turma may adopt a formation similar to those illustrated above, derived from the exercises Arrian describes.

Would an attack of this nature break down a determined infantry formation? It seems unlikely. The effectiveness of the hand-thrown javelin against a heavily armored soldier protected by a large shield could not have been all that great. Perhaps the psychological impact of having enemy horse on both the flanks and the rear would be sufficient to cause panic.

The Numidians were noted as being superior in chasing down enemy infantry that had already been routed. If their tactics were similar to those illustrated in the model, then that would be understandable. The skirmishing tactics may not have been effective at breaking an intact determined and resolute infantry formation but they may have been quite effective against infantry already broken or on the verge of being broken.


The formations illustrated above seem too contrived, too tightly organized, too formal for use in actual battle. Nevertheless, skirmishers are likely to have used some type of formation that allowed each man to circle his target and throw his javelin. The standard Indian attack against the circle of soldiers or covered wagons seen in almost every old western movie is a good example. Even very loosely organized cavalry almost instinctively adopts some type of circling motion to bring the attack home. The illustrations, even if considered to be somewhat unrealistic, do at least have the merit of being based on some ancient source material. And it is reasonable to believe that the training exercises Arrian describes were intended to have some relevance to actual combat.

General Conclusions

The purpose of this page was to explore the mechanics of the cavalry attack against infantry.

Heavy cavalry could and did attack infantry formation and could break them up. Skirmishing cavalry, such as the Numidians, were best at pursuit but could also be effective at breaking up infantry -- though this may have been more through the psychology of the attack from the rear than from the actual attack itself.

It is difficult to model the actual cavalry attack itself.

Based on some basic principles it would seem that cavalry cannot attack directly into an infantry formation. If the infantry does not panic then the attacking cavalry formation has to start to turn aside almost 100 feet away, less if the attack is oblique or in a wedge formation. When cavalry pursues infantry that is running away it probably chases at about the same speed the infantry is running. The skirmish attack likely involved some form of circling maneuvers, but probably not as tightly organized as the training exercises Arrian describes. The primary goal of all cavalry attacks was to create panic among the infantry and cause them to break formation and run away.

Because ancient armies were relatively shallow and cavalry formations were wide it probably did not make sense to think of a large body of cavalrymen attacking the flank, side, of an army. Rather, the cavalry would almost have to focus its attack on the rear of the army, with any attack against the exposed side being only an incidental aspect of the general attack.

Psychological Considerations

Napoleon is quoted as saying that

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.