THE INCIDENCE OF OYSTER SHELL IN THE SPRINGHEAD POOL
LITTLEDEAN WATER SHRINE
(c) 10th September 2019 D.M.Macer-Wright
Occasional Paper No. 2 with Update added as a footnote 5th April 2021 regarding the evidence for quenching.
Three single oyster shells were found during the 1984 excavation season of the Littledean Roman Temple. These came from a deposit identified as a ritual structured closing fill of the rectangular pool of the Romano British Water Shrine in advance of re development as a major nymphaeum, probably undertaken at the end of the 1st century or early in the 2nd. This deposit was composed of many artefacts spanning a cultural time frame from Neolithic to Roman, bringing together a diverse range of objects of ritual significance. No other complete oyster shells were found from the area of the Roman remains.
The late Professor Barri Jones took numerous clay samples from a section cut through the pool from east to west in August 1985. These samples were stored at Manchester University until 2018 when they were returned to Littledean. The samples remained in their original bags and had not been analysed. The ‘brick hard’ clay samples had a number of small bits of stone in them and many had the smoothed impressions of the 1985 excavation spade cuts. One of the larger samples had a remarkable and astonishing small yellow/green sandstone Celtic head stained with red ochre, which was totally encased within the clay, A CELTIC RED OCHRE STAINED HEAD Littledean Water Shrine Occasional Papers No 1. It is the less visually obvious contents of the clay samples however which provide the interest for this particular paper.
Robin Holley wet sieved several of the clay samples collected by Professor Jones from the “pool section” and produced some very significant findings. Central to his findings is the incidence of oyster shell fragments and dissolved oyster shells which range from c. 10mm x 5mm down to dust. Given the virtual absence of oyster shells from dry contexts around the site, this material is anomalous. The deposit of oyster shell remains in the springhead pool deviates from the normal deposits of oyster shells on Roman and prehistoric sites. Winder is the authority for oysters in the Roman period in Britain. It is clear from the Elm Farm site at Heybridge, Essex – Oysters and Other Marine Shells by Jessica M. Winder that there are no parallels to be drawn for Littledean,
The other material found by Robin in the samples was residue from late iron age metal working. This was suspected as the 1984/85 east/west and north/south pool sections consistently showed iron staining with fragments of rusted iron nodules within the clay and sand sediment sections. Robin isolated small fragments of both oxidised and reduced burnt clay, small pieces of smithing slag, flakes of metal and small grains and nodules of magnetic hammer scale throughout the samples, evidence consistent with quenching red hot metal. Investigation of the samples before preparation for sieving indicated the shell and metal working fragments were spread through the clay and sand sediments of the pool, forming an homogeneous mixture. Three clay samples with a total weight of 5.17 Kg were analysed by Robin. All three produced significant quantities of oyster shell fragments, small pieces of iron/slag or clinker, iron fragments and single small pieces of bronze and bronze slag.
Whilst it is tempting to suggest some ritual explanation for the oyster shell, the evidence for contemporaneity of shell and metal working debris suggests otherwise. A possible explanation which can be put forward as an hypothesis is that the crushed and powdered shell combined with burnt clay, slag and metal flakes, and including the clay and sand deposits, derives from metal working processes. The fact that the metal working residues were deposited in water indicates the primary mechanism for these deposits was quenching.
Direct technologies leading to small flakes of iron and hammer scale would include forge welding/quenching/tempering. In this process smithing techniques have involved and indeed some artisan smithies still do, the mixing of silica sand with calcium carbonate (lime or crushed sea shells) to make a paste which was used as a cleaning flux to coat the metal surfaces to be welded. This would certainly explain the use of oyster shell in this context in order to create lime from the calcium rich shells. The combination of lime and sand produces a flux which drives off the dirt and impurities allowing a sound and greater success for a fault free weld. Chandler Dickinson, an American historical swords maker, has ably demonstrated the practical issues of to flux or not to flux and the benefits of using sand are fundamentally clear. Significantly there are also high levels of silica sand in the Littledean samples, which sand derives from the local quartz conglomerate sandstone and Old Red Sandstone.
Thus those very ingredients locked up in the clay and sand deposits could indeed derive from the use of a flux paste comprising silica sand, oyster shell powder and red clay. For these materials to end up in the pool demands a process whereby blade hardening was being employed to produce superior weaponry. H.Foll is a particular authority on Celtic swords; how they were made, analysis of original artefacts and a thorough knowledge of the classical literature H. Föll (Iron, Steel and Swords script). He demonstrates how cross sections of La Tene and other Celtic period swords have shown that forge welding by ‘piling’ (the building up of layers) was carried out by Celtic smiths since circa 500BC. The evidence for quenching at an early date is not evident however and it would seem this was a development in Celtic metallurgy which came relatively late. That is not to say the technology was unknown more that it was maybe considered unnecessary, until late in the Iron Age when an advantage from finer weaponry became an asset.
The other process which avoided the need for forge welding was differential hardening. The Japanese sword makers in particular used a methodology which included sand derived from grinding sandstones together to release the very fine micaceous sand, to use in the clay paste which was coated around the blades before heating in the forge. This allowed the sword edges to be extra hard and the backs to flex giving the distinctive shape of the Samurai swords. The clay paste could have charcoal powder or soda ash (plant ash) mixed in it. The clay wrapped swords were hardened by quenching followed by heating to temper the blades.
Given the apparent contemporaneity of the metal working evidence and oyster shell in the samples, along with clay and sand, one can postulate that the homogeneous samples contain all the constituent parts that one should expect from forge welding when using the technique of piling and also from differential hardening with release of clay into the water. That is to say in the first the forge welding uses flux whilst in the second differential hardening uses clay. The by/waste products are likely to be hammer scale, metal flakes, slag pieces, reduced and oxidised burnt clay fragments, quartz sand, fine clay sediment and material remains high in calcium, which in this case is oyster shell. All these are present in the clay samples taken from the pool section in 1985.
Amongst the critical artefacts which support any argument for high end weaponry is the significant class of objects known as hones or whetstones. These objects are quite numerous at Littledean as objects placed in the backfills to the late neolithic stone holes or pits after standing stones had been removed. It can be demonstrated that standing stones which surrounded the pool were universally removed during the metal working phase and up to the Romano British period. Amongst the backfills were pieces of iron slag, broken hones/whetstones, grinders and sandstone surfaces with the tell tale signs of grinding wear. Another pit which may have held a wooden tree trunk for an anvil base, had a small skull shaped piece of iron slag in it, placed beside another ochre stained small sandstone head, both in association with a 16 x 16 x 4cm broken Old Red Sandstone slab from a grinding platform, which may have been part of a post pad. The importance of this head is twofold; firstly it was deposited with iron slag, placing it in the late Iron Age metal working phase and secondly the ochre stain places it with the head referred to earlier from the clay deposits. A third small sandstone ochre stained head from pool 4a in an IA context confirms these heads probably all belong to the Iron Age period and one can conclude from the contexts that they all belong to the metal working phase.
Last but not least three of fourteen soil samples (21%) from contexts associated with the springhead pool produced single fragments of bronze, which combined with a good piece of slag and IA pot from the subsidiary water feed to the IA phase of the springhead pool, provide evidence for bronze finishing, eg sword hilts.
The evidence allows for a reasoned conclusion that the incidence of oyster shell throughout the sediments of the springhead pool provides a compelling case for both quenching and differential hardening of iron artefact during the late Iron Age.
FOOTNOTE ADDED 5th April 2021
Following completion of archaeological work and analysis of the evidence a significant Quenching Phase can now be proposed. The 1st broadly circular or horseshoe shaped shrine dated to the 1st century AD, and probably contemporary with a period spanning either side of the Roman Conquest of 43 AD is identified as a ritual quenching pool. The pool was enclosed by a fence with a stone entrance portal on the east side comprising a major Bronze Age standing stone on the south side and possibly a repositioned Bronze Age stone on the north. Apart from a narrow band between the pool edge and the fence the whole interior was taken up by the pool. The shrine was roughly 3.5m in diameter and the pool 3m in diameter. Thus the rim at c.0.25m was barely wide enough to walk around. The dating concurs with a period of quenching at or towards the end of the Iron Age in Western Britain. The sedimentation fills described in the above paper are all contemporary with this 1st phase structure.