Roman Cavalry

Left. The steppe saddle. Note the lack of padding, detachable flaps to protect the leg, and the woven girth strap directly attached to the robust frame.

Right. A wooden framed four-horned saddle with Carlisle-style saddle cover with side flaps. The front of each saddle is towards the top of the picture.

The typical Roman horse of around 14 hands, unshod, strong, with a broad back, at the canter. A strung recurve bow is carried in a case. The rider is holding a contos in both hands, using weight distribution and legs to control the horse.

The kontos, in this case a 4m long slender lance in two halves joined by an iron collar. An iron ferrule is attached to the butt. The horse barding is in part based on drawings of the destroyed Column of Theodosius copied by Franco Giovanni Battista il Semolei in the 16th century, now in the musée du Louvre. Note the manica, rigid scale cuirass, and leather thigh guards – a speculative assemblage.

Horse archery at the canter against a static folkon formation at close range. The rider’s body is held within a subarmalis and a cuirass of locked scale. The horse is balanced with all four hooves off the ground. Note the bucket of plumbata carried from the front right of the saddle.

Recreating Roman Cavalry

This article is intended to serve as a simple introduction to recreating the equipment and skills of the Roman cavalryman. Many groups contact Comitatus regarding Roman cavalry equipment, and I hope this serves as a starting point for others wanting to recreate a cavalry impression. I also hope this sets out in part the standards we expect from aspiring riders.

Over the past few years we have recreated Roman cavalry from the 1st to the 6th centuries. But it is in the 4th century that the “perfect cavalryman” first makes an appearance. A trooper able to use the bow and kontos, while controlling the horse with his balance and legs. The article concentrates on my experience of 4th century riding.

Saddles and Tack
Peter Connolly used the surviving evidence in the form of leather covers, their stitching, stretch and wear marks, as well as metal horn plates, to produce a working Romano-Celtic saddle. He produced a design based upon a solid wooden four-horned frame. The size of the horns are in part dictated by the surviving copper alloy horn plates, possibly acting as stiffeners. Some plates are of a surprising thickness perhaps suggesting they are for protection. However these protectors or stiffeners do not give an absolute indication of the angle of horns. This can be derived from sculptural evidence. Surviving pieces of harness fitting also give clues to the nature of harness and how the saddle was attached to the horse.1

Various other attempts have been made to reproduce four-horned saddles using alternatives to the solid wooden frame, in part perhaps to justify simpler and cheaper reconstructions. Flexible padded saddles as used by Junkelmann for 1st - 2nd century cavalry simulations, without a wooden frame can produce similar wear and stretch marks as found on surviving saddle covers.2 The metal horn stiffener can be attached to the padded horns internally or externally, or not used at all.3 The fact that individual names have been found scratched or punched on to the stiffeners has been used as evidence that they were used externally.

Finds presented at the Carlisle Millennium Project conference in 2004 were found during excavations on the Castle Green between 1998 and 2001.4 Two saddle covers were illustrated which both showed stretch marks where they had been pressed down over a wooden frame. The covers were very worn and had both been patched many times. Overall the stitch pattern used on each cover was the same as has been found on other sites, but these covers retained trapezoidal flaps of leather, about half as deep as they were long, with the widest edges lowest when on the horse. They demonstrate that rather than just being sewn up under the saddle as originally believed, leather covers could be secured over the horns and wooden frame of the saddle. These saddle covers simply hung down the sides of the horse, even having a substantial fringed curtain of leather hanging from the lower edge. These seem to be covers from riding saddles rather than pack saddles, protecting the rider’s legs against the girth and the edge of the wooden frame. The girth strap, rather than being stitched into the saddle cover as initially thought, could be attached directly to the saddle frame giving greater stability. A piece of wood was exhibited which exactly conformed to the curved piece of the saddle frame that crossed the withers in the Connolly reconstruction.5 Such a design does pose questions about how much padding was used in the saddle, and how it was held in place.

Many people have ridden in reconstructed Roman saddles, recreating the cavalry of the 1st and 2nd centuries. Junkelmann included a reconstructed late Roman horseman in “Die Reiter Roms” 1992.6 I wanted to systematically recreate the skills and equipment of the late Roman cavalryman. This meant that I had to learn to ride. The process I went through would in part mirror that of an infantryman, inducted into the cavalry. I was used to the weapons and equipment, but not to horses. However Roman recruits had the option of very severe bits and less humane methods of horse control. It took me a year of intensive practice to become a convincing Roman cavalryman. The lack of stirrups probably simplified the learning process, and to gain experience I rode a variety of horses, and several different period saddles. Spurs were not needed on all horses, and this suggests that not all Roman cavalrymen would wear them.

The purpose of the saddle is to lift the weight of the rider from the horse’s spine. Both the solid framed and padded styles of four-horned saddle can meet this basic requirement. However I initially believed the solid wooden frame of a four-horned saddle was inflexible, and potentially painful for a horse’s back. Each saddletree would only be able to be used on one shape of horse, and even a saddle made to fit a specific horse would cease to fit if the horse lost condition on campaign. This would result in pressure sores, calloused and thickened skin. But experience has altered my view. A solid Roman saddle with a wooden frame can be made to fit most horses, with the addition of good padding in the form of a saddlecloth or furs. The same was true of the solid wooden framed military saddles of the 19th century, when cavalrymen were taught how to fold their saddle cloths to fit their horse and saddle, especially on campaign7 .

My saddle horns lack the copper alloy horn stiffeners, and at different times I have broken both the rear horns. However the design of the saddle could allow the leather cover to be lifted off, or the stitching to be cut open, and the horn replaced in less than two hours. The copper alloy stiffeners certainly seem important to re-enforce the horns and make them stronger. The rear horns are particularly important in bracing the rider against powerful thrusts, and when riding uphill. This may explain why the rear copper-alloy stiffeners can stretch completely across the rear of the saddle, giving optimum re-enforcement. My saddle is a light 4.8kg design, easily carried and stored, with very little padding for the riders comfort. The second saddle I used was a little larger, at 5.2 kg, while Connolly’s initial reconstruction was 6.8kg. The variance partly being due to size, the copper alloy stiffeners or lack of them, and the amount of stuffing in the saddle.

Padded versions of these saddles made without a wooden frame often have a metal bar towards the front of the saddle for stability. Reconstructions are generally very heavy at 11-12kg, and larger than examples based on a wooden frame. The weight of the rider forces the seat of the saddle downward and the horns lock around the rider’s legs. While this gives a very secure seat, the rider will find it difficult to get out of the saddle if the horse falls so some movement in the saddle is to be preferred. The wooden frame seems by far the more usable of the two designs and the Carlisle finds certainly seem to prove its validity. It is a good design, but it is time consuming to produce and the horns are intrinsic weaknesses. Initially the saddle feels as if the rear horns do not offer sufficient support. Indeed, the angle that the saddle sits on the horse is very important. If the rear of the saddle is not high enough, the rider’s full weight is constantly hammering on the two rear horns. However riders soon get used to trusting the rear horns, and using their legs to grip the front horns. In a short time the rider becomes confident enough to lean well out of the saddle, instinctively riding with the bent legs and downward pointing toes familiar from Roman monuments. However long periods of riding can be very hard on the rider’s legs, and serious cramp can result in having to be lifted out of the saddle.

The girth and other tack can vary between reconstructions. A split girth holds the saddle in place more securely. The saddles do not fit as securely as modern saddles, and breast and breaching straps help hold the saddle in place. A surcingle, a simple strap around both horse and saddle, can be used to fasten the saddle more securely if needed. Reconstructions of tack from the 1st and 2nd centuries are generally highly decorated with copper alloy fittings, often tinned or silvered, based on archaeological finds.8 Few such fittings date from the 4th or 5th centuries. However throughout the Roman period there was large scale use of amulets on horse tack made from the bases of shed antlers. The denticulated edge is no more than the natural coronet of the burr, channelled and perforated by the presence of blood vessels in the velvet during growth. One or more holes drilled in the disc allowed for suspension from the harness. The most common design is the phallus, perhaps to ward off the evil eye, and the use of antler may suggest that it had some special talismanic significance.9 Triplet straps hanging from the front and rear of the saddle are very useful for securing equipment, and may have helped secure the leather cover to the wooden frame.

The horse is directed by weight distribution, leg pressure, verbal commands and primarily the bit in the horse’s mouth held by the reins and bridle. Every horse needs different degrees of direction. Romans used either the snaffle bit of Celtic origin not unlike a modern bit, or the potentially severe curb bit. The Romans could also use the hackamore to increase leverage on the horse’s jaw.10 Various metal examples have been discovered, yet many more could have been made of leather or even dried grass.

A simple hackamore would have no bit, and the 1st century tombstone found in 2005 in Lancaster seems to show a bitless bridle.11 This system is useful for young horses, or those with sensitive mouths, but is generally not associated with Romans. Today metal hackamores could be covered with sheepskin for the horse’s comfort, and it is possible that some Roman hackamores would have been covered also. The rider has to learn to neck rein, using one hand to control the horse by exerting pressure on the horse’s neck with the reins, or even at times his shield.

IIn the 4th century the steppe saddle was introduced into the west by the Huns and their allies.12 It is a simple and strong design. It is possible that some saddles were built with a one piece wooden tree, but reconstructions are made of no more than four pieces of wood joined and shaped to transfer the weight of the rider to the horse’s sides. In time this saddle would develop into the medieval saddle and the modern Portuguese and Spanish saddles. The proportions of the pommel and cantle can only be deduced from surviving metal decoration. The earliest such fittings from Europe are a set of early 5th century curved and triangular-shaped gold sheet mounts from Mundolsheim, Alsace.13 These suggest a very high-fronted saddle, used to display wealth and status. Lower status riders could have used lower fronted saddles, for which rare, small and functional fittings have been found from later dates14 .

The steppe saddle does not need integral padding and can be left as just bare wood, weighing 6.4kg. It sits on several layers of wool or fur to protect the horse. It does not need breast or breaching straps, although they may be of use over long distances and rough terrain. Coming from a four-horned saddle, the Roman rider is initially concerned about sliding out of the “side door”. They try and hook their legs under the front cantle to secure themselves in the seat, as they would hook their legs under the front horns of the four-horned saddle. But the steppe saddle is not designed for this and the position soon becomes very uncomfortable. Instead the rider must use a straight leg and a very deep seat when cornering. Such a position is relatively easy on the riders legs and can be maintained for long periods of time. But this saddle is a design that naturally benefits from the invention of the stirrup. The rider can continue to ride with a straight leg, but with the addition of long stirrups giving greater stability. They can raise themselves in the saddle to cushion the effect of the movement of the horse whilst engaged in horse archery, important in a steppe culture. And perhaps most useful of all, the stirrup allows the horse to be easily mounted.

The issue of just how Romans mounted their horses is unresolved. Contemporary books mention mounting from either side of the horse.15 Fences and infantry are both good mounting blocks, and in armour it is just possible to mount while stationary with the assistance of a spear. Rope attached to the spear and used to carry the weapon over the shoulder can make a simple mounting step. A strong loop of rope over the front horns can also make a useful “step” for mounting, but there is no evidence of such devices.

The Horse
Much has been written on the size of the Roman horse.16 Ann Hyland has considered the various ancient breeds potentially available to the Roman cavalry, although she shows a bias towards the Arab.17 To generalise, there is a consensus that Roman horses in the west were around 13 to14 hands, with some as tall as 15 hands.18 Roman monuments certainly show us a small stocky animal. Recent work on bone evidence suggests that actual military horses were what we would call ponies, robust specimens of 13.2 hands, with small regional differences.19 Certainly it is safe to assume that strong animals were needed to carry armoured riders over considerable distances, and speed would have been a secondary consideration. Junkelmann used Camargue horses for his reconstructions, while native British breeds similar to the Fell or Dartmoor pony would be best adapted to the British climate.20 Size is crucial in determining speed of manoeuvre, and the effectiveness of Roman cavalry.

The rider needs to trust his mount implicitly. Not only does the rider lack stirrups, but for some manoeuvres he will not be using any reins. The control comes from weight distribution, verbal commands and leg pressure. After riding several horses including polo ponies and thoroughbreds, I settled on an Irish Cob of just over 14 hands to act the part of a Roman military horse. This breed has a reliable temperament, can easily carry a man in armour, and has strong hooves. The gelding was unshod, while some Roman horses may have been. It took time for the horse to learn how to respond to neck reining. The horse also has to become used to the Roman saddle, as well as the rider’s armour and equipment. Riders have to find a way of carrying their shield, bow, arrows, lance and sometimes javelins, either hanging from the saddle or themselves. As the late 6th century Strategikon states, riders must be able to hang their lance from their shoulder while drawing the bow and placing an arrow on the string.21 They must then be able to replace the bow and ready the lance. Put simply riders must learn to look like Roman soldiers, comfortable with their kit and weaponry.

The Roman-style cob showed a typical speed of 10kph at the walk, 15-20kph for the trot, 40kph for the canter, and 56kph at the gallop. Trotting without stirrups is uncomfortable for any length of time, and accuracy with missile weapons is difficult, so generally manoeuvres were carried out at the canter. Such a speed makes the rider and horse a difficult target, while allowing them time to accurately use their weapons.

Cavalry Weapons: the bow and the contos
In the late antique period two cavalry weapons stand out. The perfect horseman must master both the contos and the bow.

Riding carrying a large heavy shield is difficult. Carvings, such as the Arch of Galerius, suggest that the shield could be carried on the left arm using a series of straps. Many modern interpreters have followed this method, going further by padding out the interior face of the shield, so it rests more comfortably on the left leg. However there are no archaeological finds to back up this method. Practice shows a large oval shield can be carried using the horizontal handle with the boss protecting the left hand. The horse can be turned to the left using the reins in the left hand, while the shield can be pressed on to the horse’s neck to help it turn to the right. The shield can be rested on the cavalryman’s left foot when not actively in use. However Maurice describes the ideal cavalryman as carrying the kontos,22 the 4m long lance, and bow.23 These weapons require the use of both hands. Maurice believed you could not draw a bow effectively while carrying a shield.24 However I have experimented with a small shield, 450mm in diameter, strapped to the left fore arm. The small shield allows the kontos to be held in both hands, and allows for the use of bow even when using a western release.25 The shield boss is retained to allow the shield to be carried in a more conventional manner. Procopius mentions the use of small shields strapped to the upper left arm.26 Agathias mentions cavalry serving under Narses at Casilinum using shields, spear, bows and arrows.27 The perfect horseman must master both the kontos and the bow.

The kontos is a 3-4m long lance, designed to outreach cavalry opponents. It can also pick out infantrymen from static formations, keeping the rider beyond reach of spears and swords. The horseman holds the weapon with both hands, giving great power to the thrust, controlling the horse with his legs. The kontos can be used either in a low guard across the horses neck, or in a high position to strike downwards. Such a long weapon is liable to flex if the diameter of the shaft is too narrow. Accuracy is easier at the walk, the canter or the gallop, while the trot can emphasise the amount of flexibility in the shaft.

Comitatus riders carry the spatha from a waist belt, secured by a scabbard slide. Often the bow case is tied to the same belt, or is attached to another waist belt. The bow case is secured with two ties, just in case one breaks. When carrying the bow in the bow case it is liable to bounce around hitting the leg. Carrying the spatha over the case helps secure it. The case can be pulled out from behind the spatha, and twisted around so the bow points forward to ease its withdraw. When replacing the bow the case is twisted so the top laths point towards the rear, and the case is secured once more behind the spatha.

But archery is the hardest skill to learn, and it is possible that many second line units never used bows, or perhaps used them stationary. The right-handed rider should be able to loose the arrow forwards to the right or left of the horse’s head. This means he must pull his right shoulder backwards and stretch forward with his left arm. However draw length may be limited, especially while wearing rigid scale, in turn limiting the power of the shot. Shooting to the rear, the famous Parthian shot, in many ways seems easier. The left arm is extended over the horse’s rear and while the right holds the string to the ear. The draw length is maximised, but care must be taken to ensure that the left shoulder does not lie too far within the bow. If it does the string may hit the left shoulder, robbing the shot of its power. Loosing the arrow to the left when doing the Cantabrian circle is a basic skill.

The rider starts with the arrow held under tension on the bow with the left hand. He rides unusually with the reins in the right hand. The rider approaches the target at approximately 40kph and puts the horse into a right turn. When the horse is balanced he drops the reins, relaxes, looses the arrow, finds the reins, and turns to the right to come back to the starting point. Riders recreating Mongolian horse archery with stirrups, shoot to the right side of the bow, using an eastern release, and are able to shoot and “reload” on the move. They carry spare arrows in their left hand resting against the stave of the bow. After the release they can use their right hand to pull down a new arrow on to the string, to the right side of the stave. The way eastern arrows are fletched means there is no right or wrong way to place the arrow on the string. Riders can stand in their stirrups to minimise the rise and fall as the horse travels over the ground. Roman riders at the canter without stirrups, shooting in a western style, using western fletchings, take longer to knock a new arrow to the string. I have always shot using a western style and find it almost impossible to adapt to the Mongolian release.

Convincingly reconstructing the working uniform and equipment of a Roman soldier is a difficult task. It cannot be simply fancy dress, but be a practical answer to living and working in the outdoors. Reconstructing the tack of the Roman horse is a greater challenge. While being authentic in style it must also fit the horse without discomfort, and stand up to continual wear and tear. Making the equipment of a Roman soldier work on the back of a horse is the culmination of both tasks. It takes time to bring everything together and make it work. Finally, learning the skills of the Roman cavalryman takes the challenge to another level. But the rewards are worth it.


1 Connolly (1987) 7-28 and Bishop (1988) 105-108
2 Junkelmann (1992) 34-74
3 Bishop (1988) 104
4 Winterbottom (2006) 7-11
5 Winterbottom S. (2006) 7
6 Junkelmann (1992) 154-155
7 Congdon J.A. (1864) 101-103
8 Bishop and Coulston (2006) 120-123
9 MacGregor (1985) 107
10 Junkelmann (1992) 31
11 For illustrations see Bull (2007) 15 and 28. However the illustration reconstructing the cavalryman shows a bridle with a bit.
12 Evans (2004) 23-26
13 Schnitzler 1997 86
14 Evans (2004) 23
15 Vegetius, Epit. I.18
16 The English-speaking world measures the height of horses in hands, measured at the highest point of an animal's withers where the neck meets the back, chosen as a stable point of the anatomy, unlike the head or neck, which move up and down. One hand is 4 inches (10 cm). Intermediate heights are defined by hands and inches, rounding to the lower measurement in hands, followed by a decimal point and the number of additional inches between 1 and 3. Thus a horse described as 14 hands is 140cm.
17 Hyland (1990) 11-29
18 Dixon and Southern (1992) 163-173
19 Johnstone (2004)
20 Junkelmann (1990) 44-48
21 Maurice, Strat. 1.1
22 Bishop and Coulston (2006) 130
23 Maurice, Strat. 1.1
24 Maurice, Strat. 2.8
25 Coulston (1985) 281 states this is impossible.
26 Procopius 1.1.12
27 Agathias, 2.8.8

Bishop M.C. and Coulston J.C.N. (1993) Roman Military Equipment (2nd edition Oxbow 2006).

Connolly P. (1987) “The Roman Saddle” Third Roman Military Equipment Seminar, BAR International Studies 336 (1987).

Dixon K.R. and Southern P. (1992) The Roman Cavalry (B.T. Batsford 1992).

Hyland A. (1993) Training the Roman Cavalry from Arrian’s Ars Tactica. (Alan Sutton 1993).

Hyland A. (1990) Equus, The Horse in the Roman World (Batsford 1990).

James S. (2004) Excavations at Dura-Europos 1928-1937 Final report VII The Arms and Armour and Other Military Equipment (British Museum Press 2004).

Johnstone C.L. 2004 “A Biometric Study of Equids in the Roman World” (Ph.D. thesis, Uni. of York 2004).

Junkelmann M. (1990, 1991, 1992) “Die Reiter Roms” Volumes I, II and III. (Philipp Von Zabern.1990, 1991, 1992).

MacGregor A. (1985) Bone Antler Ivory and Horn (Croom Helm Ltd. 1985).

Maurice Strategikon (Translated by George T. Dennis 1984).

Procopius “History of the Wars”. Loeb Classical Library.

Vegetius: Epitome of Military Science. (transl. N.P. Milner 1993.

Coulston J.C. (1985) "Roman Archery Equipment" in Bishop M.C. The Production and Distribution of Roman Military Equipment. Proceedings of the Second Roman Military Equipment Seminar, BAR International Series 275, Oxford, 220-366.