Byzantine Lamellar Armour: Conjectural Reconstruction of a Find
from the Great Palace in Istanbul, Based on Early Medieval Parallels

Peter Beatson - NVG Mikligard

The ruins of the Great Palace of Constantinople were explored in excavations carried out in 1935-38 under the direction of Prof. J.H. Baxter of the University of St. Andrews, Scotland, which revealed the now-famous 5th century floor mosaics displayed in situ in the Istanbul Mosaic Museum. Among the small finds were a quantity of weapons and armour, including parts of one or more lamellar armours. The armour was described in a disappointingly cursory fashion1 :

‘Find No. 292. Found in Ec VI 3 in the debris covering the paved street. A quantity (more than 200 pieces) of fragments of strip armour. Large numbers had been fused together by fire, and in the process of cleaning a coin of Manuel I (A.D. 1143-80) was found sticking to a group of fragments.

The normal fragments were found in six regular sizes. They varied in width from 3 to 6 cm.; where the whole piece was found, the length was approximately twice the width. All were pierced with holes for attachment; the regular arrangement was three along one end, one on the other, and two along each side. On more than half the examples of each size a flange along the centre was beaten out from back to front.’

A photograph2 accompanying the report shows 3 or 4 pieces comprised of corroded and fragmentary lamellae (Fig. 1 A-D). All seem of the ‘flanged’ (or better, ‘ridged’) variety, the original widths can be estimated as 3.0, >=2.8, 3.3, and 4.2 cm, but none appear to have the whole length preserved. On only one (fragment A) are holes preserved (diam. ~2.5 mm), presumably a pair of side holes. Although not stated in the report, the ends (one, or both?) of the lamellae appear rounded.

Figure 1. Fragmentary lamellae from the Great Palace, Istanbul. Late 12th century 3. Sketch after photograph in Martiny et al., pl. 58.7.

In Figure 2, a possible reconstruction of a lamella is presented. As in the description, the lamella is twice as long as it is wide. The two holes in each side are placed as indicated by the extant holes in fragment ‘A’. One end of ‘A’ is sufficently intact to allow for only a single hole close to the edge, near the centre line. If adjacent lamellae overlap as much as suggested by fragment ‘D’ (see below), the placement of the remaining three holes ‘along the end’ is restricted. To be functional they must be placed so that they would match corresponding holes in the neighbouring lamellae when they overlap, OR lie in a part that doesn’t overlap, i.e. on the centre ridge.

Figure 2. Reconstruction of single lamella, and a row showing overlapping edges butting against the centre ridge, as in fragment ‘D’ (see Fig. 1). The shading indicates the surviving part of fragment ‘A’.


In fragment D a few overlapping rows of lamellae can be seen. Allowing that the lamellae are twice as long as they are wide, the overlap is considerable, perhaps indicating the lacing system allowed ‘telescoping’ together of the rows for flexibility, as was observed in a lamellar corslet excavated at Visby, Gotland4 (Sweden), and some recent Tibetan examples5. In fragment D it appears that the edge of each lamella butts against the centre ridge of its neighbour in the row, as shown in Figure 2.
In the ‘standard’ lamellar lacing technique6 , a lace running through the matching pair or pairs of side holes would be used to fasten single lamellae together into rows, while a separate lacing system running from a hole in the centre or lower part of lamellae in one row to a pair in the top of the next would be used to unite them in an upwards-overlapping manner (Fig. 3). However, the pattern of the Great Palace lamella is not suitable for such a system, mainly due to the lack of a pair of holes at the top. A suitable alternative is to be found in a Merovingian period Frankish armour excavated at Krefeld-Gellep, Germany7 (Fig. 4).

Figure 3. The ‘standard’ method of fastening lamellar. a. one lamella; b. outside view; c. lengthwise section; d. inside view. Though from a recent Tibetan lamellar cuirass in the collection of the State Ethnographical Museum, Stockholm, numerous medieval finds of lamellae have the same pattern of holes, and are supposed to have been fastened similarly. From Thordeman, 1939, Fig. 238.

Figure 4. Lacing pattern (partially exploded) of lamellar armour from grave 2589, Krefeld-Gellup, 6th century AD. After Pirling, 1986.

The lamellae are bound in rows by a leather strip folded over the rounded upper and lower edges, and held with a double lacing passing through the single holes. The ridge prevents excessive sideways shifting. Loops of thonging passing through the two pairs of side holes serve to attach each row to the ones above and below. A single hole could well serve in place of the lower pair of the Krefeld-Gellup system, as is frequently done in the ‘normal’ system shown in Fig. 3. The result is shown in Fig. 5.

Figure 5. Proposed lacing system for the Great Palace lamellar, exploded.


An exact parallel for the pattern of holes proposed for the Great Palace lamellae has not been published, though some are close, eg. Fig. 6-2, -3. Lamellae with centre ridges or bosses have been found elsewhere, eg. Fig. 6-1, -3. As well as stiffening the lamellae, the ridge presumably kept neighbouring plates from shifting and cutting their laces.
The ‘banded’ appearance8 of lamellar armour in Byzantine art may also be considered. The horizontal lines separating rows could be leather strips folded over the upper edges, and the vertical double lines occasionally appearing to separate the plates could be a misinterpretation of the centre ridges of the lamellae.

Figure 6. Other finds resembling the reconstructed Great Palace lamellae.

Locations: 1. ‘Börg’ (the fortress), Birka, Sweden. 2. Grave 121, Homokmégy-Halom, Hungary. 3. Doneshkoe gorodishe, Russia. Dating: 1. 10th cent. 2. late 7th-8th cent. 3. 950-1250AD. Sources: 1. Thordeman 1939; 2. Garam et al. 1975; 3. Kirpitchnikov 1971. All to same scale.


Numerous questions remain. From what piece or pieces of armour did the find originate? Do the six different-sized plates mean that several separate armours were deposited together, or did they serve different functions in the same panoply, for example smaller plates coming from sleeves, and larger ones9 from the thorax? Why were ‘more than half’ the plates ridged, while the rest were flat? What other hole patterns existed? Not all the plates were of the ’normal’ type described, what other shapes were there?
Until further information10 on this fascinating find can be uncovered, the reconstruction presented remains highly conjectural. It does, however concord with the appearance of ‘banded’ lamellar which, has been noted to occur frequently in Byzantine artistic representations11 , and it can also be demonstrated to have precedents in other, more completely described archaeological material.
A prototype (under construction) should demonstrate the practicality (or otherwise) of this lacing system, I urge others to experiment with the design and forward their comments.


Arena, M.S. and L. Paroli. Museo dell’Alto Medioevo Roma (guide). Rome: Instituto Polygrafico e Zecca dello Stato 1993.

Dawson, T. (1992). Banded lamellar- a solution. Varangian Voice iss. 23 (July 1992), p.16.

Franceschini, E.B.R. (1995). Winter in the Great Palace: the persistence of pagan festivals in Christian Byzantium. Byzantinische Forschungen 21, p.117-132.

Garam, É., I. Kovrig, J. Gy. Szabó and Gy. Török. Avar Finds in the Hungarian National Museum. Budapest: Akadémiai Kiadó 1975.

Kirpitchnikov, A.N. Drevnirusskoe Oruzhie 3: Dospech, Kompleks Boev’ich Sredstv IX-XIII v.v. (Arkheologiia SSSR). Leningrad: Nauka 1971. (trans - Old Russian Arms 3: Armour, Armaments and their Usage, 9-13th centuries).

Martiny, G., G. Brett and R.B.K. Stevenson. The Great Palace of the Byzantine Emperors: Being a First Report on the Excavations Carried Out in Istanbul on Behalf of the Walker Trust (The University of St. Andrews) 1935-38. London: Oxford University Press 1947.

Pirling, R. Römer und Franken am Niederrhein. Katalog-Handbuch des Landschafts-museum Burg Linn in Krefeld. Mainz am Rhein: Phillip von Zabern 1986.

Robinson, H.R. Oriental Armour. London: Herbert Jenkins 1967.

Roth, H. Kunst der Völkerwanderungzeit (Propyläen Kunstgeschichte, supplementary series: vol. 4). Berlin: Propyläen Verlag 1979.

Thordeman, B. Armour From the Battle of Wisby 1361 (Vol. 1, Text). Stockholm: Kungl. Vitterhets Historie och Antikvitets Akademien 1939.

Appendix - Some finds of Avar-type lamellar in and near Europe

The Krefeld-Gellup cuirass is not an isolated find, lamellar seems to have been briefly adopted by neighbouring Germanic tribes during the period of Avar ascendance. All four burials listed below contained hundreds of plates, probably a complete armour.
The Byzantines also keenly appreciated and copied Avarian war technology, as Maurice’s Strategicon clearly shows. As the only European power where lamellar remained continuously in service in the medieval age, the Byzantines may well have preserved the lacing system introduced by the Avars.

1. Kertch (Crimea), Ukraine. Avarian, later 5th century (post 457AD). Ref. Thordeman, 1939. Held: State Historical Museum, Moscow. Note - possibly a mixture of lacing systems.

2. Krefeld-Gellup, Germany. Grave 2589, Frankish, 6th century. Ref. Pirling, 1986. Held: Landschaftsmuseums Burg-Linn, Krefeld Germany.

3. Castel Trosino, Italy. Tomb 119, Lombardic, early 7th century. Refs. Arena and Paroli, 1993; Thordeman, 1939. Held: Museo dell’Alto Medioevo, Rome. Note - reconstruction drawing of entire panoply displayed at museum.

4. Niederstotzingen, Germany. ?Grave 12b/c, Alamannic, 7th century. Refs. Pirling, 1986 (includes reconstruction drawing of cuirass with skirt); Roth, 1979. Held: Württembergisches Landesmuseum, Stuttgart.


1. Martiny et al., 1947, p.99.

2. op. cit., pl. 58.7.

3. Franceschini (1995) conjectures that the burnt layer in which the armour was found may be a trace of the catastrophic fires set during the seige and capture of Constantinople in 1204.

4. Armour 25 from the Visby mass war grave of 1361. Thordeman, 1939, p.218-219, fig.199.

5. op. cit., 245-248, figs. 236-238.

6. see footnotes 3 and 4. Variants of this system are frequently illustrated in popular works, and are well known to medieval reenactors.

7. Pirling, 1986. This form of lamellar was probably introduced to Europe by the Avars: see Robinson 1967, fig. 29, and Thordeman 1939, fig. 232 nos.11-15.

8. Examples: The ‘Joshua’ fresco, Hosios Lukas 10th cent. (near Delphi, Greece), steatite icon of St. Theodore Stratelates, 11th cent. (Vatican Museum 982).

9. It is perhaps relevant that in the Visby armour a few crudely-made lamellae were of double or greater width, representing replacements made during later repairs. All the lamellae were, however, the same length. Thordeman 1939, p. 403.

10. A recent visit to St. Andrews University failed to uncover the notebooks of the excavation, however some archived materials, including personal correspondance, as well as the original figures for the Report (Martiny et al. 1947) are held in the papers of Sir David Russell, Prof. J.H. Baxter and the Walker Trust. It is to be hoped that the artefacts themselves are still extant, the excavation permit required that they remained in Turkey.

11. Dawson, 1992.


Webbed by Peter Beatson, 1999-2000.
All rights to materials herein belong to the original owners.