Metallographic examination of four 7th–8th century long-blade weapons from Želovce (Slovakia)

The article presents the metallographic examination of three sabres and one double-edged sword coming from the 7–8 century Slavic-Avar site of Želovce (Slovakia). All four weapons had been subjected to metallography as early as 1975, but the results were not published in sufficient detail. With this article, written in honour of Radomír Pleiner, the authors wish to repay this debt. The blades are compared with other (metallographically examined) weapons from Želovce, and the manufacturing methods of early medieval production of sabres are discussed.


Introduction
In 1975, Radomír Pleiner metallographically examined one sword and three sabres from the 7 th -8 th century Slavic-Avar cemetery of Želovce (south Slovakia). The metallography of the sabres was published in a preliminary fashion in 1979, but no photographs or drawings were included (Pleiner 1979). Later, in 1989, R. Pleiner published simplified drawings of the metallographic samples, but no details of the metallography were provided (Pleiner 1989). In fact, R. Pleiner never published these weapons and the results of their investigation in full detail. In 1991, Ľ. Mihok published the results of the metallographic examination of several long-blade weapons from Želovce , including the same 'Carolingian' sword that had been previously investigated by R. Pleiner. The opportunity was taken to check the metallographic samples stored in the Institute of Archaeology in Prague and to present details of the metallographic examinations.
was recently published by G. Csiky (2015), who also introduced a specific classification system for these objects. Following Csiky's classification, edged weapons (corresponding to category of artefacts 'E') coming from Avaric sites are categorized according to both the form and cross-section of blades: double-edged swords (E.I), single-edged swords (E.II), sabres (E.III) and seaxes (E.IV ;Csiky 2015, 152). Observed through this grouping, the main trend in 7 th -8 th century Avar weaponry is the gradual decrease of sabre blade curvature, the relatively permanent presence of single-edged swords and the increasing appearance of crossguards. Weapons with a single-edged and curved blade, suitable for cutting, can be considered sabres. Sabre blades could also be provided with a false edge (or 'elman') running some distance (up to 260 mm) from the point upward (i.e. the lower third or quarter of the blade was double-edged). There are several theories on the function of the doubleedged parts; they offered more options to cut and facilitate thrusting, 'since a lenticular cross-section has several advantages over a triangular cross-section' (Csiky 2015, 195). This particular feature is seen on Želovce sabres as well; the length of their false edges lies in a very wide range going from 0 to 205 mm.
Some sabres (such as those from the Želovce-cemetery graves nos. 167, 335, 442, 818) have a cross-or star-shaped crossguard, formed by two arms and two langets in the middle. The function of the langets was to fit the crossguard tightly to both the wooden grip and the scabbard's throat, thus protecting the blade from atmospheric factors and preventing accidental unsheathing when riding a horse.

The archaeological background of the Želovce blades
The archaeological site of Želovce (excavated between 1963 and 1968 under the direction of Z. Čilinská 1973) was situated on the Slavic-Avar border, and today lies roughly 2.5 km northeast of the village of Želovce (Veľký Krtíš district, south Slovakia). It was a large burial ground (with 870 graves) and represents one of the most important Slovak sites of the 7 th and 8 th centuries. It consisted of two parts (as a structural analysis revealed) most likely belonging to two individual settlements (Čilinská -Wolska 1979, 154). A total of 628 graves were excavated in the first (earlier) part of the burial ground; burials were made there from the 630s. In the second (later) part of the burial ground, 242 graves were excavated, and it was in use from the second half of the 7 th century. The latest burials at the Želovce site date to the second half of the 8 th century (Čilinská 1992, 30-36). The Želovce burial ground features an unusually high number of weapons (17 sabres, one 'Carolingian' sword, one single-edged sword-palash, and one seax), which is unmatched by other contemporary burial grounds in the Carpathian Basin.

)
The presence of welding seams indicates that the blade was welded from three strips. One side was more carburised than the other. Before welding, the strips were somewhat homogenised by extensive forging at rather low temperatures. Even the most carburised places barely reach the quality of medium-hard steel, hence the material was not suitable for quenching, which was not applied. This weapon is of mediocre quality.

Metallographic description:
Two samples ('A' and 'B') were collected for metallography from the middle part of the blade. Part of sample 'B' was also used for chemical analysis.
from Dabas/Gyón-Paphegy, one from Csolnok, three swords (grave nos. 78, 97, 149) 2 other swords (as stray finds) and 2 spearheads (grave no. 129 and stray find) from Környe, 3 spearheads (grave nos. 221, 238, 321) and one sabre from Košice-Šebastovce, eight edged weapons (grave Nos. 78,124,126,235,311,335,442,818) from Želovce, and other spearheads and swords from Budakalász and Szegvár-Oromdűlő (Csiky 2015, 293-294: nos. 7-11). 2 However, the Környe burial belongs to the Early Avar period and thus does not chronologically correspond to the Želovce cemetery. The surface layers of the metallographic samples are corroded. In sample 'B', the slag inclusions are chained in roughly three bands running lengthwise from the edge towards the back. Considering their number, corresponding to level 3-2 on the Jernkontoret scale, the metal is of mediocre purity. Sample 'A' also contains slag inclusions chained into several bands, but inclusions individually scattered throughout the sample prevail. The number of inclusions corresponds to levels 3 and 4 on the Jernkontoret scale.
Macroscopic etching with Oberhoffer's and Heyn's reagents revealed that the blade profile consists of several strips. In sample 'B' a strip (separated from one side by the weld) of fine-grained (ASTM 12) ferritic-pearlitic spheroidised structure runs through the centre of the sample. This structure slowly changes into ferrite (ASTM 11-12) with traces of pearlite and is separated by a weld from another strip of ferritic-pearlitic structure with a grain size of ASTM 12. Specimen 'A' shows a striped microstructure similar to that of specimen 'B', i.e. a continuity of the strips and two distinct welds. A strip of ferritic structure with traces of pearlite (grain size ASTM 11-12) is observed on the left side of the sample. This structure is beyond a weld changed into a fine-grained ferritic-pearlitic structure. Pearlite with a ferritic network and needles appears in places. This structure slowly changes into ferrite with traces of pearlite and tertiary cementite on the boundaries of grains of ASTM 8-9 size. This structure is separated by a weld from a strip of ferritic and ferritic-pearlitic structure. The blade has a very fine ferritic-pearlitic structureglobules of cementite can be seen within ferritic grains.
2) Sabre from grave 30 (specimen no. 302; fig. 3) The blade was forge-welded from two, or (more likely) from three strips of material, of which each (mainly the central one) was unilaterally strongly carburised in advance. A significant diffusion of carbon to the adjacent ferritic areas occurred in the course of subsequent forging. The cutting edge was slightly hardened, perhaps by cooling it down in a stream of cold air or in warm oil, etc.

Metallographic description:
A specimen for both metallographic and chemical analysis was cut off crosswise from the middle part of the blade. The sample is covered by a thin corrosion layer. Non-metallic inclusions are chained in several bands running lengthwise from the edge towards the back. The number of inclusions corresponds on average to levels 2 and 3 on the Jernkontoret scale.
Etching revealed a highly heterogeneous structure. There is a narrow strip of ferrite with traces of pearlite at the back of the blade. The cutting edge shows a fine-grained sorbitic structure, which, towards the centre of the blade, changes in places to pearlite with a ferritic network, and further into a strip located in the lengthwise axis of the sample with a structure of ferrite with some pearlite (grain size corresponds to ASTM 9-10). At the right side of the sample there is a fine-grained ferritic-pearlitic structure that changes into ferrite (of ASTM-8 grain size) with traces of pearlite.
3) Sabre from grave 44 (specimen no. 396; fig. 4) The blade was made of soft carbon steel. Considering the distribution of slag inclusions, it is likely that the billet was welded from at least two strips of such steel, of which one contained irregularly distributed zones richer in carbon. The cutting edge remained relatively soft and was not (and could not be) quench hardened; on the contrary, it seems that intensive final forging took place at rather low temperatures. The blade was a simple product.

Metallographic description:
Two samples ('A' and 'B') were collected for metallography from both the cutting edge and back of the blade. Part of sample B was also used for chemical analysis.
Half of the sample 'B' has ferritic structure with traces of pearlite; the grain size corresponds to the ASTM 8-10. A narrow strip of ferritic-pearlitic structure is on the right side of the sample. The left side of the sample shows a very fine-grained (ASTM 10-11) ferritic-pearlitic structure that includes zones of pearlite with ferritic network and needles. A pearlitic structure appears on one side at the back of the sample. A predominant part of sample 'A' contains a ferritic structure (of ASTM 9-1 grain size) permeated entirely by spheroidised cementite. The middle portion of the blade shows an oblique strip of pearlite structure with ferritic network and needles, which gradually changes over to the structure of ferrite and pearlite (traces of pearlite in places). The cutting edge contains ferrite with a small amount of pearlite.
While the metallography of the sabres was published at least in a fragmentary manner, an examination of the 'Carolingian' sword (undoubtedly from grave no. 124) was never published. One of the possible explanations is that Pleiner lacked information he needed for the proper description and identification of the weapon. 3 His original interpretation of the sword production describes a blade made of a single piece of iron subjected to secondary carburisation (either the edges of the iron semi-product were carburised prior to the final forming to the shape of the blade, or the cutting-edge(s) of the nearly finished blade were carburised). However, this interpretation appeared slightly inaccurate when the sample was checked. Therefore, a re-assessment and re-examination recently conducted by the authors is presented here. 4) 'Carolingian' sword (grave 124; specimen no. 299; fig. 5) The blade consists of steel cutting edges welded onto a middle portion consisting of surface panels and a core of iron. The overall character of the revealed part of the surface panel corresponds to a layered-and-twisted composite, which, however, cannot be considered a standard pattern-welded composite. The reason lies in the very low differences in the local contents of phosphorus. The blade was not quench hardened (at least not in the whole volume) in the place of sampling. The very low visibility of the welding lines, along with the extensive carbon diffusion beyond the welds, indicate that during the forging cycles the blade was exposed to high temperatures for a long time. The find seems to be a failed attempt to produce a more glamorous pattern-welded weapon.
Metallographic description: Two samples were cut out roughly halfway down the blade, each from one side. While one sample had a well-preserved metallic core, the other was entirely corroded.
Metal purity varies considerably in both the central and cutting-edge portions of the blade (from level 2 to level 5 on the Jernkontoret scale).
When etched with Nital, the metallographic structure revealed three basic areas. Area I shows a pearlitic structure with a hardness of 220±27 HV0.2 (maximum hardness was 251 HV0.2 on the cutting edge, decreasing to HV0.2 towards the central portion). Area II is a mixture of pearlite and ferrite; the carbon content gradually decreases from circa 0.7 to 0.25 percent towards the central portion of the blade; the hardness reaches 133±8 HV0.2. Ferrite is present in the form of polyhedral grains network as well as lamellas penetrating the pearlitic grains. Area III is ferrite with some pearlite in places (max. c. 0.2% C). Hardness is 116±19 HV0.2, grain size c. ASTM 7-8.
Etching with Oberhoffer's reagent gradually revealed traces of welding lines separating the cutting edge (A), the core (B) and one surface panel (C). Sharply bounded areas appeared (C*) within this panel (C). Their hardness (133±6 HV0.2) is somewhat higher than the hardness (121±7 HV0.2) and apparently also the phosphorus content of the surrounding matrix.   5. Sword, Želovce (Slovakia), grave 124: a -the examined weapon; b -schematic drawings and photos of the blade samples (from the left: unetched state; after etching with Nital; after etching with Oberhoffer's reagent; layout of areas described; distribution of the microstructures and of the main welds across the sample; hardness distribution chart); c -the other sample, which was entirely corroded; d -the expected method of assembling individual rods to make the blade, if double-edged (cutting-edge welding onto a middle portion with pattern-welded surface panels); e -pearlitic structure in the Area I (etched with Nital). Photo and drawings after Z. Čilinská (1973, 199, Taf. XXII: 16) (a) and by J. Hošek (b-e). Obr. 5. Meč, Želovce, hrob 124: a -analyzovaná zbraň; b -schematické nákresy a fotografie vzorku čepele (zleva: neleptaný stav, po naleptání nitalem, po naleptání Oberhofferovým roztokem, rozložení popisovaných strukturních oblastí, vyznačení hlavních svarů a zachycených strukturních oblastí, graf průběhu tvrdosti); c -druhý odebraný vzorek, který byl zcela prokorodován; d -předpokládaný způsob sestavení jednotlivých prutů při výrobě čepele, je-li dvoubřitá (břity navařeny na střední část nesoucí povrchové damaskové panely); e -perlitická struktura v oblasti I (leptáno nitalem). Foto a kresby podle Z. Čilinské (1973,199, Taf. XXII: 16) (a) a J. Hoška (b-e). The archive of metallographic specimens also contains sample nos. 684-687, which were detached from the Želovce weapons, but we do not know from which weapons in particular these samples were taken. Although R. Pleiner and his assistant Mrs. B. Novotná started with the metallographic examinations, they never completed them. The folders with semi-finished metallographic reports (each of which is labelled only as 'sword -Želovce') also contained a letter from Ľ. Mihok. The letter (now preserved in fragments) 4 explains that Ľ. Mihok sent photographs of metallographic specimens to R. Pleiner in 1990, which he himself examined and also published a year later (see . Therefore, we can only assume that the samples, which R. Pleiner had at his disposal, were provided by Ľ. Mihok. In that case, the samples should come from sabre nos. 78, 335, 442 and 818. Because these samples are of low importance for the research of the production techniques of Avar sabres, their description is not included. Ľ. Mihok and his colleagues also examined several weapons from the Želovce cemetery : four sabres (from grave nos. 78, 335, 442 and 818), one single-edged sword-palash (from grave no. 167), one seax (grave no. 311), one 'Carolingian' sword (grave no. 124), and one dagger (no. 235). The obtained results (except for the 'Carolingian' sword) are summarised in fig. 7. The blades of sabre nos. 78, 335, 442, 818 and the blade of dagger no. 235 were made by edge-to-back layering. Blade 818 is a steel-iron-steel sandwich, while blades 335 and 442 are a steel-and-iron sandwich. The blade of sabre no. 78 was welded from two plates of steel, and it seems that the cutting edge was also subjected to secondary carburisation. Sword-palash no. 167 was most likely made of a single piece of heterogeneous iron. Mihok himself believed that the palash has a side-to-side banded blade, but he did not provide any acceptable evidence for this hypothesis. The same situation concerns seax no. 311, the blade of which consists of heterogeneous iron and which is described as a side-to-side banded item. Sword no. 124 has, according to Ľ. Mihok, a single-edged blade made from an iron back and welded-on cutting edge of steel. No traces of quenching were detected in the examined weapons.
Another sabre examined metallographically by Ľ. Mihok and his colleagues (Mihok -Pribulová -Mačala 1995) came from the 7 th -8 th century Slavic-Avar cemetery of Košice-Šebastovce (Slovakia). The sabre is presumably the one from grave no. 161 dating to the first half of the 7 th century (for more details see Krivák 2017, 69). According to Mihok, the blade of this weapon was made as a sandwich (a variant of the edge-to-back layering) consisting of a mildly carburised central plate to which a somewhat less carburised plate was laterally welded from each side. In addition, it seems that the cutting edge of the blade was secondarily carburised (but no more than 0.3% of carbon was observed). Two samples were taken from the blade crosswise, one near the hilt, the other near the point, but none shows traces of quenching.
Three single-edged long-blade weapons dating from the 7 th to 8 th century were also examined by M. Mehofer (2006); all three pieces (two sabres and one single-edge sword) come from the Avar burial ground of Zillingtal (Austria). Possible manufacturing techniques used to produce the weapons are described as follows (Mehofer 2006). The 8 th century sabre from grave B 23 was made from a single piece of iron (with some steel in places), and no traces of hardening were observed (samples were taken half-way down the blade). The sabre from grave D 338, dating to the second half of the 7 th century, was apparently made from a single piece of metal irregularly varying between iron and steel. While samples taken half-way down the blade show no traces of hardening, samples taken close to the point indicate rapid cooling or even quenching (although the carbon content was insufficient for the formation of typical quenched microstructures). The blade of the single-edged sword from grave D 3, dating to the second quarter of the 7 th century, was made of at least two lamellas, as suggested by a corroded weld running from the back downwards. The carbon content of the blade is presumably low and its distribution uneven.
Certain comparisons can also be made with 7 th to 9 th -century sabres from Volga-Bulgarian cemeteries, which were examined and recently published by Yu. Semykin (2015). Two sabres come from the Novinkovskij kurgan burial ground /Новинковский курганный могильник/ (Russia) dating from the late 7 th to the mid-8 th century. The blade of the first one (no. 45) was made of heterogeneous material, mostly iron, but in places also steel. The other (no. 29) was presumably made of a similar material, but it seems that cutting edges were also secondarily carburized and subjected to quenching. While samples from sabre no. 29 were taken approximately half-way down the blade, the sample from sabre no. 45 was taken close to the hilt and just from the cutting edge. One sabre (no. 35) comes from Boľshe-Tiganskij I burial ground /Больше-Тиганский I грунтовый могильник/ (Russia) dating from the mid-8 th to the mid-9 th century. The blade of this weapon was forged from a piled billet (combining iron and steel). The sample was taken approximately halfway down the blade and just from the cutting edge. Five sabres come from the Boľshe-Tiganskij burial ground /Больше-Тиганский грунтовый могильник/ (Russia) dating from the mid-8 th to the 9 th century. The blade of sabre No. 21 was presumably made from a layer of steel welded to one side on a piece of iron (Semykin 2015, 139). 6 A sample for metallography was taken from a cutting edge in the upper part of the blade (closer to the hilt), and no traces of quenching were detected. The four other sabres (nos. 22 to 25) were made of steel with visible welding lines running from cutting-edge to back. Sabre nos. 22, 23 and 24 were sampled near the point, the hilt, and half-way down the blade, respectively. All three blades were subjected to quenching limited to the cutting edges. Three samples detached from blade no. 25 indicate that the blade was quenched along the entire length and in the whole volume. One sabre (no. 1), dating from the mid-8 th to the mid-9 th century, comes from the 'burial ground at the 116 th kilometre' /могильник у 116 километра/ (Russia). The blade shows a layer of quench-hardened steel welded to one side of a piece of heterogeneous material varying between iron and steel. The sample for metallography was taken from a cutting edge in the upper part of the blade (closer to the hilt).
Considering the small number of the 7 th -8 th century sabres metallographically examined, it seems pointless to strive for a detailed comparison of the sabres from Želovce with those from other contemporary sites. We can just conclude that they did not differ significantly in any technological aspect. However, we can clarify the current knowledge of the techniques used in the manufacture of sabres. Their blades were in general forged from billets prepared by edge-to-back layering; 7 some of them consist entirely of steel, others entirely of iron or heterogeneous material varying between iron and steel. Various forms of sandwiching or cutting-edge welding-in seem to be a standard for making blades, deliberately combining iron and steel. Perhaps carburising cutting edges could also be practiced (as suggested by some of the examinations). No blades with cutting-edges of steel welded onto an iron back (corresponding to side-to-side banding) were encountered. No patternwelding appears on sabres. In fact, the constructions of the sabre-blades correspond well to the constructions used in manufacturing knives in the given cultures (see e.g. Semykin 2015, 110;Mihok -Pribulová -Mačala 1995). The blades of 7 th -8 th century sabres were seldom quench-hardened in their whole volume (i.e. across the whole cross section and along their whole length).
Comparing the 'Carolingian' sword from Želovce (grave 124) with other contemporary swords is rather difficult. First, it is not clear whether the blade is single-or double-edged. Čilinská (1973, 57) described the sword as a single-edged sword, and Mihok was convinced that the results of his examination confirmed this (Mihok -Pribulová -Mačala 1995;Mihok -Holý -Čilinská 1993). However, in 1975 the weapon was already heavily corroded and 'preserved in poor condition'. Pleiner cut out two samples from the blade, one from each side, but only one sample contained a metallic core (see fig. 6). In addition, newly obtained metallographic results suggest instead that it was a double-edged pattern-welded blade (or rather an attempt to produce a pattern-welded blade). If so, the sword would be a common 7 th -8 th century 'spatha' in terms of the construction used, but an unusually poor piece in terms of the low quality of the pattern-welded composites.

Conclusion
The sabres from Želovce, examined in 1975 by Radomír Pleiner, show features similar to those of other contemporary sabres in terms of the manufacturing techniques employed. Deliberate combinations of iron and steel, encountered in 7 th to 9 th century sabre-blades, were typically based on edge-to-back layering, and most of the examined blades show no traces of quenching. The 'Carolingian' sword from Želovce (grave 124) has a welded and most likely double-edged blade deliberately combining iron and steel. The cutting edges were welded-on and the middle portion was provided with surface panels of layered and apparently twisted material resembling pattern-welded composites. If the sword is indeed a 'pattern-welded' specimen, like the majority of the 7 th -8 th century swords, the patternwelding was virtually invisible.