This article was originally published in Byzantine and Modern Greek Studies number 22, 1998, pp. 38-50. The numbers in square brackets are the page numbers of the original publication. Please cite the B.M.G.S. edition, not this version. The WWW is no substitute for scholarship!
 The material aspects of the Byzantine army is a field which has always been the poor relation of scholarship on its organisation and logistics, and publications of the last decade have unfortunately confused the issue as much as elucidated it. Byzantium had a rich tradition of military literature, in unbroken continuity with the already sophisticated practice of the western empire of Rome. Manuals from late Antiquity to the tenth century provide considerable detail of the equipment a Byzantine soldier should ideally have, and in doing so show in the armed forces of the empire a pragmatic willingness to absorb useful equipment, as much as effective tactics, from its neighbours and enemies. The quality of its equipment must also have been a factor in the remarkable success of the army and navy in preserving the empire as much as they did against so many foes for a thousand years. In view of this, the relationship between the ideals of the manuals and the reality is an important issue, one which demands a laborious search for evidence beyond literary sources. Economic conditions impinged on more than the amount of man-power to be mobilised. They also influencing the quantity and even the very type of equipment that could be supplied to the troops. We shall look here at three items of armour which were essential elements of middle period panoply, the kremasmata, the kabadion and the klibanion with the aim of establishing their nature more precisely.
The major work of the last decade is Taxiarchis Kolias' Byzantinische Waffen. Although based upon a compendious reading of Byzantine literature, its conclusions have been undermined by a deficiency of comparative material and insufficient grasp of the practicalities of the subject. Unfortunately the position of Byzantinische Waffen as the largest and most accessible publication in the field have led to its  problematical propositions being disseminated and perpetuated by later authors. One difficulty in the volume is the nature of the kremasmata and kabadion and their relation to each other. It must be accepted that kremasmata is a padded and quilted skirt hanging below a soldier's cuirass to protect his legs. An arrangement of this type is unmistakably illustrated in many detailed pictorial sources of the middle period.(1) At one point Kolias proposes that the kabadion is the same thing as kremasmata.(2) His basis for this conclusion are is both are made of the same materials. This is nothing like sufficient evidence. Anvils and fish-hooks are both made of iron, but few fish are caught with anvils. Later he decides that it is a variety of garment, when it is recommended as the sole protection for archers.(3) Contemporaneous with this source, De Cerimoniis also mentions kabadia, describing them as 'the ethnic habit' of certain nationalities visiting Constantinople.(4)
To find the kabadion appearing prominently again in Byzantine sources we must move onto the fourteenth century. In the treatise on the court offices of Pseudo-Kodinos it is stipulated as the primary garment of most courtiers. Here it comes in diverse colours, brocades and bearing opulent decoration,(5) quite unlike the 'coarse silk or cotton as thick as can be stitched together' and furnished with zabai, (interpreted as strips of mail(6)) as the description of kremasmata goes, but certainly in accord with it being a garment.(7)
Reiske's commentary on De Cerimoniis points us towards a cognate  term in Arabic, transliterated in various sources 'aba,(8) kabâ and qabâ, which is often mentioned in the compendious Arabic literature on textiles and dress.(9) It is to an Arabic source that we turn to find the precise nature of the kabadion. In the ninth century al-Aghani explained that 'when a man tears his qamis (shirt or tunic) from the opening below the neck to the foot it becomes a qabâ'.(10) Hence the kabadion/kabâ is a coat opening down the centre front, traceable to the well illustrated Persian garment mentioned in Herodotos by the earlier Hellenised name kandys(11), perhaps with kabadhV as an intermediate form.(12) Garments of this type were already ubiquitous throughout the Mediterranean world by the fifth century,(13) and several fine complete, or largely complete, examples dated to the tenth century were found in the Buyid cemetery at Rayy.(14) The description in the  Praecepta of knee length kabadia as 'short'(15) is a clear indication that something close to the floor sweeping garments familiar from the fourteenth century portraits of Theodore Metochitês(16) and Grand Admiral Apokaukos(17) and others(18) were already established in civilian dress by the middle of the tenth century. The length of Byzantine men's garments was a element in Luitprand of Cremona's diatribe against them in the wake of his visit of 968-9.(19)
Eric McGeer's translation of the Praecepta Militaria and a portion of the Taktika of Nikephoros Ouranos, necessarily depended heavily on Kolias,' with consequent unfortunate lapses in the quality of his translation. In the light of a better understanding of kabadion the passage of the Praecepta III.8:
eiV de taV zwnaV autwn foreitwsan oi toxotai kabadia proV to skepesqai meroV ti twn ippwn autwn, fulattesqai de autouV apo ths zwsewV kai katw.
should be read thus:
This is still more subtly informative about the nature of these kabadia. They must not be split up the back,(20) for if they were, the skirts would simply fall either side of the horse and give it little protection, as can be seen in Fatimid ivories in Berlin,(21) and commonly in the surcoats of European knights.(22) It is clear this variety of kabadion was not similar to the Persian and Lombard forms already mentioned,  which are constructed as a quite straight tube with little or no extra volume in the skirt, but made in the manner of an eighth to ninth century example found at Mochtchevaya-Balka in the Caucasus.(23) This has a panelled skirt attached by a horizontal seam at the waist, allowing for greater fullness and the skirt is also split on either side.(24) With such an arrangement the two front panels would hang over the rider's legs while the back panel would lay along the horse's spine.
The klibanion demands some careful attention. It has largely gone unquestioned that the short, usually sleeveless cuirass shown in so many pictorial sources is the whole story, however both the texts of the tenth century manuals and some lesser known pictures of a little later tell another story.
The manuals do not define the form of the klibanion explicitly. Several mentions make it clear that this is because the term had several meanings.
At II.3, and III.8 the author of the Praecepta writes of skirmishers and archers equipped only with klibania and stresses at the first point the lightness and mobility of such armour. This is echoed by Ouranos. In contrast, at III. 4 of the Praecepta Phokas stipulates that the klibania of the kataphraktoi should have sleeves and kremasmata, a specification again repeated by Ouranos. In this contrast we have the clarification of the dream of Achmet mentioned by Kolias. Klibanion might refer to as little as the breast and back, but could also mean a full harness consisting of breast and back, shoulder guards, sleeves and skirt. The fewer of these pieces it had, the greater the danger to the we arer, as Achmet observed.(25)
 Beyond these uses, klibanion referred to anything made of lamellar, such as horse armour. (Phokas III.5, Ouranos 60.5) This brings me to the matter of construction. In 1988 I embarked upon a project to construct a functioning metal klibanion. Drawing on the limited material published to that date(26) and some pioneering work by colleagues, Steven and Martin Baker on the reconstruction of Near Eastern and Central Asian style leather lamellar,(27) I had gone ahead with the assumption that the Byzantines used the same sort of assembly methods as other peoples of the Near East and Central Asia. By 1991 I had produced an effective solid-laced thorax closely approximating the overall style of the pictorial sources after much trial and error. A problem I encountered rapidly was that in the athletic activity of combat the rows of plates would separate within minutes. In solid-laced lamellar each row of plates is tied directly to those above and below to form a virtually rigid cuirass, one which, contrary to Haldon, is very much more rigid than common scale armour.(28) The flexing of my torso pushed the rows past each other and, with the metal to metal contact where the laces passed through, the plates sheared the laces like scissors. I returned to an observation that I had passed over up to that point. Byzantine lamellar, and some Islamic, is depicted with narrow bands spacing the rows. I surmised that this may have been a strip of leather laced between the rows to cushion them and eliminate the scissoring effect, and it did prove so in practice.(29)
Nonetheless, I was not completely satisfied with this result. Through the process of construction, re-construction and use, a germ of doubt about the overall form of the klibanion as depicted in the more commonly reproduced pictorial sources had grown into full blown  scepticism, essentially on account of how little it protected. The skirts and sleeves sometimes shown resembled pteruges, an antiquated and rather ineffective armour, too much to be immediately plausible. The explanation given by the tenth century literary sources dispels this scepticism well enough, yet the searching out of less well known pictures and re-assessing familiar ones prompted by that scepticism yielded valuable observations.
The patternings on Byzantine depictions of lamellar, when clear, show striking differences from other depictions and reconstructions from archaeological finds. Across the corpus of pictorial sources there are also remarkable consistencies which led me to conjecture that these differences were not just artistic distortions. The construction of lamellar as commonly understood (I shall use the term 'generic' henceforth.) has each plate overlapping the adjacent plate and laced firmly to it. If the tops of the plates are rounded, (as they rarely are in generic lamellars) this results in a pattern where the visible vertical edges of the plates do not align with the lowest points of the row of arcs at the tops. In all Byzantine pictures I have seen to date this is not so; the entirety of every plate is visible and there is no overlap. This is viable if the banding is not merely a narrow strip at the top of the plates as I previously conjectured, but backs the whole row of plates which are fixed to it side by side. Such a construction method produces a fabric which is much more horizontally flexible than the generic method and easier in the labour of execution. Other benefits are a saving of fifteen to twenty percent in materials, and concomitantly weight; which is a significant consideration in the case of metal lamellar, and less exacting requirements for hole placement. Further ease of manufacture was afforded by the practice of riveting the plates to the backing rather than lacing them on. It is indicated by the plates showing a round dot or circle at the top in place of the short stripe that represents a lace. This development appears to have been introduced in the eleventh century and becomes the method most commonly illustrated in the twelfth century.(30) In addition, I concluded  that the assembly of these rows was not solid, but of the hanging form. In hanging lamellar the rows are suspended from a loose lace that allows them to move up and down considerably. The most common genus of eleventh and twelfth century lamellar looks like this:
Lamellars of this type are depicted with from two to five suspension laces, although in practice even numbers of laces would work best, and since the bottom holes are likely to have been used also to tie rows together on the inside, any more than two exposed laces is rather extravagant redundancy. Hanging lamellar plates often have a vertical overlap of approximately half, which means that any weapon has to penetrate two spaced layers of armour. The great benefits of spaced armour against artillery was re-discovered in the twentieth century and applied to tanks. My own tests confirm that the same benefits apply to medieval use. My reconstructed klibania have proven to be completely resistant to thrown and thrust spears, to swords and even proof against arrows.(31) By comparison mail is proof against none of these attacks, unless they are light or glancing, and scale armour is little better. Kolias proposes the superiority of the lorikion  over the klibanion and its use by officers solely on the basis of their fewer numbers in the ship complements specified in De Cerimoniis.(32) He overlooks the fact that the author states that the light form of lorikion are for the use of men who do not partake of hand to hand combat; siphon crews, helmsmen and lookouts.(33) This is significant. Even if the unattributed 'common' lorikia are for officers, it does not imply their superiority. On the contrary there could be advantages in supplying officers with an inferior armour. The Byzantine army was not an ancient or European war-band. The job of its officers was to command, not to lead. Inferior armour might discourage officers from heroic excesses. The imperviousness of the klibanion explains the anecdote recounted by Anna Comnena when Alexios took two charges from Frankish cavalry and was merely pushed partly off, then back onto his horse without sustaining any injury.(34) Backed by a thick kabadion, and mail in the case of someone of status, and covered by an epilorikion, an iron lamellar corselet would be almost impenetrable.
Other developments occur in the eleventh and twelfth centuries in addition to riveting. Kremasmata continue to be shown, yet the surface patterning on them looks less and less like quilting and becomes increasingly suggestive of some form of lamellar. On a spectacular fresco of Saint Nestor in the Church of Saint Nikolas at Kastoria, these hints at another reality become explicit.(35) The sleeves and skirt of the saint's klibanion are made of an identical lamellar to the chest, but hung upside down. Long lamellar corselets are well known in the Levant from the early middle ages, however generic styles are consistently homogenous: the lamellar overlaps the same way from neck to knee.(36) Inverted lamellar limb pieces have major advantages.  Blows striking those areas are invariably travelling downwards. In the case of cuts to the thighs the angle is steep and the swing has had a longer arc to develop momentum. Such cuts strike the edges of the tops of the rows in generic lamellar facilitating penetration, while inverted lamellar sheds them, as scale also would.(37) Inverted lamellar sleeves also appear on a fresco in the Serpent Church at Göreme.(38) It appears to be a distinctively Romano-Byzantine practice albeit of uncertain prevalence, for the only unquestionable precursor to this picture I have found to date is a thigh-guard of inverted leather lamellar of the third or fourth century found at Dura Europos.(39) Possibly it was re-introduced in the late tenth or the eleventh century, having fallen out of use in late antiquity as lamellar itself appears to have done. The absence of any mention of lamellar in Prokopios and the Strategikon attributed to Maurice is striking, and it is by no means certain that the author of the anonymous treatise on strategy is writing of anything more solid than mail.(40) In a period when the resources of the empire had been, and were still being, stretched by war on several fronts the disappearance of lamellar in favour of mail is hardly surprising, its effectiveness notwithstanding. It occurred for the same reason it recurred in later Byzantium; the expense and complexity of its manufacture. Mail (and scale) is made of small, identical components. The origins of wire drawing are lost in the mists of time. Wire was wound onto a bar then cut into rings. The most exacting part of the operation was to overlap, flatten and punch the ends of each ring to allow it to be riveted shut; a necessity due to the weakness of wrought iron and the fineness of the wire. Mail making was a cottage industry which was not uncommonly practised by women  in medieval Europe and the same may have been true in Byzantium.(41) Producing scales is only a little more complicated, and the assembly of corselets of mail or scale is a simple, repetitive operation which requires little or no modification of the components to produce a comprehensive garment. On the other hand, to make lamellar one must forge out the metal, trim the plates to shape whatever the material and punch or drill them with at least seven, and often more, holes. Individual plates must then be assembled into horizontal rows and those rows linked vertically to make the klibanion. Since lamellar plates are larger than scales,(42) and the finished fabric is less flexible, the neck and arm openings of the chest piece must be made with specially shaped plates in order to achieve a harness which gives the best possible protection while allowing the necessary freedom of movement and some degree of comfort.
The Nestor icon, previously discussed, shows other points of detail which are thought provoking. The backing strip does not show beyond the tops of the plates, which are themselves not rounded at the top. The first characteristic would be a small saving of materials, while the second would be a large saving of labour which otherwise produces a purely cosmetic effect. We should not be surprised to find such short cuts employed in the borderlands. The skirt on Nestor's klibanion has no split in the front and so is that of an infantryman. The earlier manuals hardly ever mention the equipment of the infantry, (the various authors all seem to regard them as of little importance) and when they do nothing as solid as lamellar is hinted at. Several conclusions might be drawn from the icon. One is that the artist wanted to make the saint accessible by showing him as a common soldier, which is  a prevalent theme in Byzantine iconography, and so deviated from his high standard of realism. More plausible is the conclusion that, at least by the twelfth century, there were infantry more heavily armoured than in the tenth century, a possibility consistent with the suggestion of that era being a time of economic improvement.(43) A third possibility, which I am in the process of testing in practice, is that with the skirts being separate from the chest piece, it may possible to build them is such a way that the same pieces can be assembled with the splits front and back for cavalry use, or at the sides for infantry. This possibility would have benefits for both equipment supply and troop deployment.
One fine illustration provides a valuable contrast to the paradigm set out above. It is the triumphal portrait of Basil II in the Psalter in Saint Mark's Library, Venice. The Emperor wears a klibanion constructed thus:
This armour is unique in that it is the only unmistakable depiction of a solid-laced example I have found to date. Two factors indicate its solidity. One is the brick-work pattern, the plates offset horizontally from one row to another. The other is the lack of any laces crossing the boundaries of the rows. The horizontal lace is the one that initially fixes the plates to the backing band and the other lace passes through the top holes of the lower row of plates and the lower holes of the  upper row to tie the whole rigidly together. This armour could be used; as I have written earlier, my first cuirass was solid-laced, however its rigidity does make such an armour notably awkward for any activity which requires flexibility. Yet being an imperial parade armour made for symbolic display rather than practicality and protection, such was not a great consideration. Rather it was made for opulence, and necessarily to resemble the classicising artistic convention that we can see in so many works of the tenth century.(44) Consider the neo-classical excesses of similar European Renaissance armours.
A combination of casting a wider net for evidence and a practical approach to interpreting and testing that evidence can yield a more accurate picture of Byzantine military equipment. A picture which matches a sophistication of equipment to the sophistication of strategic and tactical theory which is found in the literary sources.Timothy Dawson
A very much more comprehensive treatment of middle Byzantine arms and armour entitled 'Suntagma Hoplon' is forthcoming from Boydell and Brewer in A Companion to medieval arms and Armour ed. David Nicolle. Available in July 2002.