Those familiar with this period and its accoutrements will likely be fairly familiar with the form, and feel, of an early Anglo-Saxon sword, perhaps based on handling carefully made replicas. The validity of any observations made by handling such swords, however, will always be limited by the accuracy of those replicas, ranging from reenactment blunts (the safety adaptations of which add up to completely change the mass and handling of the blade, (1) ) to reproductions of particular archaeological finds made by expert historic bladesmiths, who often work from observations made by personally examining finds to collect dimensions. Such smiths have a working sense for the dimensions and properties of early Anglo-Saxon swords, but for many years, publications on the subject have been light on technical details: until only a couple of decades ago it was typical for archaeological reports to fail entirely to document the width and length of blades, or to confuse length of the blade with length of the object as a whole. As archaeological iron continues to degrade, post excavation, and can never be wholly stabilised (Welton, 2016) the urgency of recording these seemingly arcane details for future study is increasingly recognised.
A cursory review of major recent publications, however, reveals that an important detail for characterising these blades – critical for modelling their original weight and handling characteristics – is routinely not being recorded or reported. This has led us to ask a troubling question; do we really know what thickness early Anglo-Saxon blades had? Could generations of reproductions, and the conclusions drawn from them, be completely wrong?
This discussion and the small piece of research reported here was prompted by an enquiry from historic craftsperson Tom Betts who, working on an early Anglo-Saxon sword and approaching the daunting task of the grind, asked me what weight (for its length) or maximal thickness he should aim for. This question gave me pause, for it is not one I had hitherto considered.
Does blade thickness matter?
Doubling any one of these dimensions would double the volume, and therefore mass of the sword. To say nothing of the distribution of this mass (and the moment applied to the wrist – the way, in practice, we experience a sword’s ‘heaviness’ or ‘balance’) it is obvious that thickness is therefore important, and that the overall mass of the sword is highly sensitive to it.
The formulae for calculating the area of such a shape are quite complicated, requiring, first, the calculation of the theoretical circles (of radius r) and the distance between their centres (a). For added clarity we have replaced the terms length and height (of the ellipse), with width and thickness (of the sword).
It is beyond the scope of this article to explore this geometry in any more detail (for more see here (link) ) but it is sufficient for our purposes to merely observe that when solved, there is a 1:1 relationship between thickness (T) and area (A), such that doubling thickness doubles area. For a solid with this cross-section (eg. sword blade), with all other variables kept constant, doubling thickness therefore doubles volume, and therefore mass.
Width, W = 2 * √( r * T - T² / 4)
thickness, T = 2r - √ (4 r² - W²)
radius of circles, r = ( W² + T² ) / (4h)
Distance of circle centres, a = 2r - h
Perimiter, p = 4r * arccos [ 1 - T / ( 2 * r ) ]
Area of the ellipse = rp/2 - W * ( r - T/2 )
It should therefore be clear that blade-thickness matters; it is absolutely understandable that any smith wanting to make a replica early Anglo-Saxon sword would need this measurement, and be frustrated that data on this appears to be unavailable.
Why has blade-thickness been overlooked?
Measurements of individual rusted blades are certainly worth taking with a huge pinch of salt, and in some cases where poor preservation or massive amounts of concretions are obvious, might be entirely meaningless. However, as it can be argued that the sources of error affecting these measurements (vis. their representation of original blade thickness) can work to inflate, or deflate them depending on conditions, I would argue that there is potential for meta-analysis of thickness measurements of a large enough sample of sword remains from different contexts to yield reasonable and reliable ranges and averages for early Anglo-Saxon sword thickness in general. The potential for such information to be useful in the future, and its potential irretrievable loss due to post-excavation decay, I would argue, at least justifies this stat being routinely recorded out of an abundance of caution, even if it might seem silly at the time.
Thickness data directly from blades
In the report on the Snape (Suffolk) cemetery, discussion of the sword from g47, with substantial organic remains of the scabbard, nevertheless did mention the thickness of the extant remains as a whole – 17mm – though due to the organics this has no bearing on the thickness of the blade (Filmer-Sankey & Pestell 2001). Likewise, reporting the cemetery at Pilgrims Way, Wrotham, Kent, Stoodley et. al. (2015) documented the extant thickness of the three swords found. The first is described as having “very little organic material preserved” and a reported thickness of 4mm (Grave 7003/ON 42) which we can take to be approximately representative of the blade’s true thickness, whereas the other swords had far more substantial scabbard remains preserved, and consequently had extant thickness reported as 10mm in both cases (Grave 7010/ON 57, and G7067/ ON 223).
The Northumbrian sword discovered in a shallow inhumation burial at Acklam Wold, North Yorkshire in 1980 is reported as “5.8cm wide by 0.3cm thick at the hilt end” (Ager & Gilmour, 1988). Reporting the sword found during excavation of the small cemetery at Wigber Low in the Derbyshire Peaks, Collis et. al. (1983) documented its blade length and width, together with the length, width, and thickness of the tang (7mm) but not the thickness of the blade. This unfortunately cannot be taken as a proxy for thickness of the blade-root, as the swaging of tangs often made them thicker than the blade-core - a fact borne out by hilt-plate data (see later).
Of particular interest are the descriptions of swords from the Dover Buckland Cemetery, Kent (1994 excavations; Parfitt et. al. 2012) where seven swords were recovered. The report (p49) states that:
‘maximum surviving thickness [for the Dover Buckland Swords] is around 3-4mm (but this compares with thicknesses of 4-5 early Anglo-Saxon swords housed in the British Museum with surviving original surfaces eg. From Waterbeach Cambridgeshire, the Thames and London, and may possibly reflect some corrosion post-deposition).' (Parfitt et. al. 2012)
- The Grave 264 sword (SN 1019) section (cut from 330mm below the blade shoulder) had a 4mm max thickness.
- The Grave 265B sword (SN 1020) section (cut from 280mm below blade shoulder) had a 4mm max thickness.
- The Grave 346 sword (SN 1021) section (cut from 270mm below the blade shoulder) had a 4mm maximum thickness.
- The Grave 375 sword (SN 1022) section (cut from 330mm below the blade shoulder) had a 3.5mm max thickness.
- The Grave 437 sword (SN 1023) section (cut from 560mm below the blade shoulder) had a 3mm max thickness.
It is now is possible to say more of the dimensions of the 6th century sword from the 'Marlow Warlord' burial in Buckinghamshire excavated in 2020. Although the interrim report (Bunker & Thomas 2022) documented the length (87cm), blade length (75cm) and maximum width (6cm) - each in the fairly typical range for early AS swords- it did not report thickness. Thickness measurements, however, were collected at various points along the blade post-conservation (Matt Bunker, personal comm. 2023) which give us not only a clearer idea of this blade's maximal cross-section but also how it changed along the blade's length. Six measurements were taken from the area of the blade immediately below the guard, both on and 10mm either side of the centre-line, and averaged 8.39mm. At 60cm down the blade the same array of measurements averaged 5.99mm. The difference of over 2mm in thickness (a reduction of 29% thickness across 80% of the blade length) is indicative of modest distal taper. These thickness measurements are substantially greater than other swords discussed here, but as the blade was in relatively good condition cannot be easily dismissed as an artefact of expanded corrosion.
Another sword from another 'sentinel burial' at Brimble Hill near Swindon, Wiltshire - excavated in 2000 (accompanied by a shield-boss dateable to the late 6th or early 7th century) but only recently analysed and in poorer condition, had a total length of 89.3cm, blade length 78.0cm, was 48mm wide at the shoulders tapering to 40mm wide at 68cm (87%) down the blade. In this case thickness measurements were collected at 10mm intervals along the centreline of the blade and were consistently 5mm (Matt Bunker, personal comm).
The thickness data from these few reports are summarised below.
Inferring sword cross-section from Staffordshire Hoard hilt-plates
The blades of the swords were not included in the Hoard, but rather, the gold and silver fittings which had once adorned their hilts, and had been often quite brutally removed prior to deposition. (A detailed discussion of the various sorts of hilt fittings, form, function and combinations can be found here: link) Although many fittings were found badly mangled, being of precious metals they are effectively immune to corrosion and were found chemically intact. Because these fittings were precisely crafted to be tight-fitting, we can use them to learn a great deal about the blades they once adorned, of which nothing at all survives.
We have already seen how two unusual fittings – both gold adornments of seaxes – bore slots for their blades which could be measured and cross-referenced with our database of grave seax finds, to allow us to reconstruct the entire form of their missing blades (Thompson & Thompson 2014). Likewise, the lower-most hilt-plate of a sword bears a slot, relatively precisely fitting the root of its blade. Measuring this slot can therefore give a (maximal) estimate for the width and thickness of the blade it once adorned. The Staffordshire Hoard garnet-adorned seax assemblage validates this approach, for its slot was, strictly, teardrop-shaped to enclose the blade shoulders but not the blade-root, which instead butted up against the gold of the lower fitting leaving a measurable dent – the direct footprint of the blade’s cross-section. The extremely close correspondence between the slot and this dent, in the thickness axis, demonstrates how precisely these pieces were fitted to their blades, and therefore justifies the use of gold and silver fittings’ slots as proxies for the cross-sections of the blades themselves.
34 gold LL hilt-plates*, and 5 silver LL hilt-plates were identified, thus, in theory, representing the cross-section of 39 sword blades. One other fragment was identified as certainly a lower-guard’s hilt-plate but of unknown position; a further 12 fragments were identified as likely from lower-guard hilt-plates but of unknown position. (* Nb. this likely includes the seax hilt-plate)
This is, however, complicated by severe damage to the hilt-plates (particularly the softer gold ones) mostly attributable to their brutal dismantling prior to deposition. Fern et. al. (2019) reported that sword-blade width at lower guard could be determined from the blade slots of 8 hilt-plates, ranging from 49 t0 62.5mm, with four between 52 and 57mm wide. (We previously discussed the aparrent bimodal distribution of sword widths suggesting two parallel distinct subtypes of Anglo-Saxon swords here). Measurements for the minor-axis of the blade slot, representing blade thickness, were not reported.
Our own visual survey of the Staffordshire Hoard images catalogue identified only five LL hilt-plates surviving sufficiently intact and un-deformed to provide a reliable full blade cross-section, and another two where the tips of the hilt-plate were damaged such that width could only be roughly estimated, but with the middle section undistorted allowing a reliable thickness measurement to be obtained.
Tangentially, it is interesting to observe that all upper hilt-plates had rectangular tang-slots which were considerably thicker than typical for blade-slots in LL hilt-plates, including in the case of Cat244 where both hilt-plates remained attached together. This makes sense considering tangs were swaged rather than made by stock-removal (compressing the end of the blade-core’s width down into a narrower spike, it must necessarily grow in length and thickness) but this has practical implications for the hilt assembly, in that the LL hilt-plate’s slot would be too thin to slot over the tang, so would have to be slid up from the blade tip during assembly. By extension this confirms to us that, at no point along a finished blade could its thickness or width have exceeded that of the blade slot; the blade must have tapered in profile and thickness or at least remained constant. The blade-slot thus gives us both a close approximation of the actual cross-section of the blade at its root, and the whole blade’s cross-sectional maximum.
- Cat 244 The best-preserved lower hilt-plate assemblage (and matched to upper hilt-plate assemblage cat 243). Thin gold, with lower-lower of tray type. Lower-guard top plate features legacymark of a beaded wire grip-collar. LL. Hilt-plate 86mm long. Lentoid lot 58mm wide. 5mm thick. Accompanying lower-guard top hilt-plate, tang slot 33 x 8mm.
- Cat 264 Gold. Tray-type. Unusual blade-facing slot flange. Plain, integral ‘dummy rivet head’ bosses. LL. Hilt-plate 91mm long. Lentoid slot 62mm wide. 5mm thick.
- Cat 292 Gold. Tray type. Part loop of filigree wire surround of boss rivet extant on one side; legacy mark on the other. End distorted. LL hilt-plate. Original length would have been approx. 85mm long. Lentoid slot. 61mm wide. 4.5mm thick.
- Cat 331 Gilded silver. Tray-type LL. Hilt-plate. Multiple drill holes in each tip; one containing a bent rivet shank. Hilt-plate 87mm long. Lentoid slot 54mm wide. 3.5mm thick.
- Cat. 371 Gilded silver. Tray-type LL hilt-plate. Single drill hole in each tip. Circular legacy mark of now missing boss/head. Hilt-plate 90mm long. Slot 56mm wide. 5mm thick.
- Cat 290* Gold. Tray type. Both ends bent round, but the middle region relatively undistorted; beaded wire surrounds for rivet heads in the tips. Hilt-plate was likely around 7cm long with the bladewith blade slot likely 4-5cm long (estimates). Blade slot 4.5mm thick.
- Cat 337* Gold. Tray type / slot flange. Very small fragment of LL hilt plate with only the full orbit of the slot preserved. Original size of hilt-plate unknown, likely around 7cm. Slot 47mm wide, 3mm thick.
Slot width varied from 45 to 62mm in this sample; average (mean) slot width was 55mm (median=56, sd=6.5, n=7). Slot thickness varied from 3 to 5mm, with an average (mean) width of 4.4mm (median=4.5, mode=5, sd=0.8, n=7).
There is a hint of a possible positive correlation between thickness and width (figure 2), as we might expect for natural variation in the output of bladesmiths nevertheless maintaining rough proportions. However, owing both to this apparent trend being weak (Adjusted R-squared: 0.2) and the unfortunately excessively small sample-size this trend is not statistically significant (p=0.18). This means that where no correlation existed between width and thickness, we would still expect to see a distribution like this in 18 of 100 cases purely by chance.
We can therefore now state that we have a highly reliable (unaffected by corrosion) albeit not necessarily representative measure for the thickness of early Anglo-Saxon swords. Strictly this sample comes from a single context, and the particular pieces may represent gear from a single army, so might not be representative of early Anglo-Saxon swords as a whole. It is interesting to note, however, that these measurements are broadly in agreement with the small sample of reported blade thicknesses (see section “Thickness Data directly from Blades” and Table 1) which ranged from 3 to 4mm.
Direct measurement and perhaps scanning and digital reconstruction of mangled hilt-plates from the Hoard may offer the potential of increasing our sample-size for Staffordshire Hoard sword cross-sections. However, the addition of measurements from extant hilt-plates of other early Anglo-Saxon swords in museum collections might allow these averages and ranges to be more representative of swords in early Anglo-Saxon lowland Britain as a whole.
We therefore suggest that where specific data is lacking, early Anglo-Saxon swords should be modelled or reconstructed with a thickness falling within this range, and particularly, should not exceed 6mm maximal thickness.
Stoodley, N., Mepham, A.B.P. and Watson, J., 2015. An Anglo-Saxon Cemetery on Pilgrim’s Way, near Wrotham, Kent.
A. Bayliss, J. Hines, K. Høilund Nielsen, G. McCormac & C. Scull. 2017. Anglo-Saxon graves and grave goods of the 6th and 7th centuries AD: a chronological framework. (SMA Monographs 33). Society for Medieval Archaeology.
Ager, B.M. and Gilmour, B., 1988. A Pattern-Welded Anglo-Saxon Sword from Acklam Wold, North Yorkshire. Yorkshire Archaeological Journal, 60, pp.13-23.
Blackmore, L., Blair, I., Hirst, S. and Scull, C., 2019. The Prittlewell Princely Burial: Excavations at Priory Crescent, Southend-on-Sea, Essex, 2003.
Brunning, S. 2019. The sword in Early Medieval Northern Europe: Experience, Identity, Representation. The Boydell Press. Woodbridge.
Bunker M., Thomas G. 2022. The Marlow Warlord: An Early Medieval Sentinel Burial of the Middle Thames. University of Reading.
Bunker, M. 2023. Personal communication with the author regarding measurements of swords from Marlow and Brimble Hill. 08/07/2023.
Collis, J. and Ager, B., 1983. Wigber Low, Derbyshire: a bronze age and Anglian burial site in the White Peak. Department of Prehistory and Archaeology, University of Sheffield.
Evison, V.I. and Annable, F.K., 1994. An Anglo-Saxon Cemetery at Great Chesterford, Essex
Filmer-Sankey, W. and Pestell, T., 2001. Snape Anglo-Saxon cemetery: excavations and surveys 1824-1992 (No. 95). East Anglian Archaeology.
Gilmour, B.J., 1991. Developments in iron smithing and decorative welding techniques found in Anglo-Saxon swords and related edged weapons. University of London, University College London (United Kingdom).
Härke, H., 1989. Early Saxon weapon burials: frequencies, distributions and weapon combinations. Weapons and Warfare in Anglo-Saxon England, pp.49-61.
Härke, H.G., 2004. Swords, warrior graves and Anglo-Saxon warfare. Current Archaeology.
Lang, J. and Ager, B., 1989. Swords of the Anglo-Saxon and Viking periods in the British Museum: a radiographic study. Weapons and Warfare in Anglo-Saxon England, pp.85-122.
Leahy, K., 2011. Anglo-Saxon Crafts. The Oxford Handbook of Anglo-Saxon Archaeology, pp.440-59.
Meadows, I., 2019. The Pioneer Burial: A high-status Anglian warrior burial from Wollaston Northamptonshire. The Pioneer Burial, pp.1-82.
Mortimer, P. and Bunker, M. eds., 2019. The Sword in Anglo-Saxon England: from the 5th to 7th century. Anglo-Saxon Books.
Parfitt, K. and Anderson, T., 2012. Buckland Anglo-Saxon Cemetery, Dover, Excavations, 1994. Canterbury Archaeological Trust Limited.
Thiele, A., Hošek, J., Kucypera, P. and Dévényi, L., 2015. The Role of Pattern‐Welding in Historical Swords—M echanical Testing of Materials Used in Their Manufacture. Archaeometry, 57(4), pp.720-739.
Thompson A & Thompson Æ. 2014. “Wyrmfang” – Staffordshire Hoard (c7th) narrow-seax. [Online] [URL= https://www.thegns.org/thegns-of-mercia-reconstructions-blog/wyrmfang-staffordshire-hoard-c7th-narrow-seax] Accessed: 06/07/2023
Welton, A.J., 2016. Encounters with iron: an archaeometallurgical reassessment of early Anglo-Saxon spearheads and knives. Archaeological Journal, 173(2), pp.206-24
West, S.E., Crowfoot, E., Härke, H. and Start, M.Y., 1988. The Anglo-Saxon Cemetery at Westgarth Gardens, Bury St. Edmunds, Suffolk: Catalogue (No. 38). Suffolk County Planning Department.
Wickham H (2016). ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. ISBN 978-3-319-24277-4, https://ggplot2.tidyverse.org.
R Core Team (2023). _R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.
All analyses were performed using R Statistical Software (v4.3.1)