Coin pellet mould and crucible fragments from Old Sleaford, Lincolnshire

Transcription

1 ANCIENT MONUMENTS LABORATORY REPORT SERIES Coin pellet mould and crucible fragments from Old Sleaford, Lincolnshire Keith Robbins and Justine Bayley Introduction Old Sleaford lies at the east end of the Ancaster Gap which cuts through the limestone ridge of Lincolnshire. An earlier Iron Age, then Belgic, settlement developed into a Roman town south of the crossing of the river Slea by the fen-edge road. The long sequence of occupation included two main deposits of late Iron Age mint debris. The smaller deposit (Heyworth and Wilthew 1987) lay within the medieval churchyard on an old ground surface disturbed by the building of an outsize Romano-British corn-drier while the larger group had accumulated in the fills of a criss-cross of ditches. There were no related structures, and only two apparently associated Iron Age coins silver half-denomination uninscribed. The material seems to be contemporary with the distinctive regional pottery decorated with curvilinear designs incorporating notched rouletting and stamps which was associated with wheel-thrown Belgic wares (Elsdon 1975). The samples described here are part of the larger deposit, found in 1963 on Site H. The assemblage appears to be one of the largest yet recovered from any Celtic mint in Europe. The total collection of mint debris consists of c.6000 fragments, 95% of which are identifiable as coin pellet mould fragments. The remaining 320 pieces are probably fragments of crucibles. Coin pellet moulds are a characteristic find from major Celtic sites of the immediate pre-roman Iron Age. The moulds are in the form of shallow reduced fired clay trays with approximately round holes or indentations sunk in them in parallel rows arranged in chessboard fashion. The holes are set fairly close together and without any runnel or connection between them. The majority of the indentations appear to have horizontal bottoms with others rounded to a degree, the holes taper slightly to the bottom. They have been found all over Europe as well as in Britain (Tournaire et al 1982). The use of the term `coin pellet moulds' has been disputed by some authors; Casey (1983) argues that there is no conclusive evidence for their use in the production of pellets specifically for coin production but Collis (1985) supports the use implied by the conventional name. The term `coin mould' is or could be misleading as the coins themselves were not cast in the moulds; thus `coin pellet mould' is a more appropriate term. The mostly likely explanation for their use appears to be in the production of pellets of metal, blanks from which coins could be struck. It has been suggested that small lumps or granules of metal of a predetermined total weight were placed 1

2 cold in the holes, which were then melted by intense heat directed at them from on top, employing a charcoal fire and blowing air over each depression in succession (Tylecote 1962). The use of precious metals such as gold and silver in the moulds meant that the molten metal would not have wetted the surface of the moulds and so would have formed small pellets with convex surfaces. The final operations were to hammer such pellets roughly into flans, and to strike these between two slightly dished dies, producing coins. It is thought different sizes of holes were for preparing blanks of different weights for coins of different denominations. A major programme of X-radiography of all the coin pellet mould fragments was undertaken in the Ancient Monuments Laboratory shortly after their discovery and before they were washed. This identified one fragment that had a pellet remaining in the mould, though hidden by adhering soil. The mal-formed pellet had a concavo-convex form as the molten metal had wetted the mould fabric and formed a meniscus rather than solidifying with the normal convex upper surface. It was removed and examined metallographically by R.F. Tylecote and was found to be a copper-silver alloy with approximately 65-70% silver. This result was included in an interim report on the moulds from Old Sleaford (Jones et al 1976). Recently the mounted section has been re-examined and analysed in a scanning electron microscope (SEM) by Kerstin Eckstein. The sample was deeply corroded but a small, relatively unaltered zone in the core of the pellet had a normalised composition of 64.2% silver and 35.8% copper which confirms the earlier estimate (Figs 1 and 2). Examination and analysis The coin pellet moulds sent for analysis (AML No ) were selected by Sheila Elsdon as representative of the range of sizes found on the site. A number of crucible fragments (AML No ) were also examined and analysed by energy dispersive X-ray fluorescence (XRF) in an attempt to identify traces of metal on their surfaces. The ultimate aim was to determine what metal was melted in the coin pellet moulds and crucibles. All the comments below refer only to the small sample of the finds which were analysed. There are a number of factors that have to be taken into consideration when interpreting the results of XRF analyses such as these. Elements such as silver and tin give far weaker signals than copper, lead or zinc and so are more difficult to detect when present at low levels. In addition the proportions of the metals present depends not only on their original concentrations in the metal melt but also on their chemical nature. Elements like silver which are relatively unreactive are only physically bound to the ceramic and so are under-represented. The concentrations of lead and particularly zinc are enhanced as they can act as glass-forming elements and so are chemically bound into the crucible slag. Zinc has a very high vapour pressure and tends to diffuse into crucible walls so it is well represented when the crucible is analysed, even if it was only present in minor amounts in the metal melt (Bayley 1989). Tables 1 and 2 show the metals detected by XRF on the mould and crucible fragments respectively. 2

3 Coin Pellet Moulds The dimensions of the analysed fragments are given in Table 1. The total number of coin pellet mould fragments analysed was 22, of which six had hole diameters of 11-14mm at the mouth (top), the majority of the samples had a diameter of 6-9mm, and five fragments had smaller diameters, of 4-5mm. The depth to which the holes, large or small, are sunk is variable between 4-10mm. Clifford (1961) states the depth is usually rather more than half the thickness of the slab for coin pellet moulds at the site of Bagendon. This is the case for the slabs with small holes at Old Sleaford but it does not appear to apply to those with larger holes, whose depth is considerably less than half the thickness of the slab. As only a small sample of the coin pellet moulds have been examined in this present study, no firm conclusions can be drawn about their shape. Only one fragment exhibits a right angle corner, so the complete size and shape can only be surmised. The right angle fragment contains the greatest number of indentations, 19 in three rows each with five holes and one row with four holes. Morphological study of the bulk of the finds will allow more detailed information to be recorded. Initially the samples were all X-rayed. Close examination of the radiographs indicated that three fragments had possible traces metal surviving (CM 1136, CM 1084 and CM 2345). These appeared to be droplets of metal that had failed to coalesce into a pellet. These three samples, together with the remaining fragments, were analysed by XRF. In all the mould samples zinc was present and in half copper was identified, although sometimes only in trace amounts. Both copper and zinc are likely to have been present as impurities in the metal being melted. They are more readily oxidised than precious metals and their oxides would have wetted the surface of the mould and therefore are more easily detected as they would have become chemically bound into the clay. Gold was sought but not detected on any of the coin pellet moulds. Both sample CM 1084 and sample CM 2345 were found to contain silver as well as copper and zinc. The majority of coins of this period were made of silver alloys and it is therefore likely that this was the main metal used in the production of the pellets. This was confirmed by analysing a metal deposit removed from sample CM 1136, employing a scanning electron microscope (SEM); the results indicated the sample was a copper-silver alloy. Close examination of the SEM micrograph of the sample (Fig 3) showed the metal fragment had consisted of two phases, a silver-rich phase and a copper-rich phase, though the latter had corroded away leaving voids. As a result of the loss of copper the original silver:copper ratio cannot be determined. The remaining metal was found to be 94% silver, so the original silver content of the sample must have been less than this. The copper-rich phase was minor but significant, so the composition was probably similar to that of the complete pellet mentioned above. The SEM micrograph displays an interesting crystal structure, produced as a result of the cooling conditions which affected the metal. Crucible Fragments The initial investigation of the fragments involved a visual examination, this had the aim of highlighting any samples which appeared to have metal residues on their surfaces for later analysis. The remaining fragments were grouped according 3

4 to their surface appearance and morphology (see Table 2 for list of classes). This classification was useful in dealing with the large sample, however the groups show some overlaps as many fragments could be classified within a number of groups. The size of the fragments in the sample ranged from approximately 10mm to 80mm, with the majority falling somewhere in between. Due to the small size of the fragments any attempt to determine the shape of the vessels from which they came was very difficult. However some general remarks are possible. Of the total of 222, 67 were identified as rim fragments, and 25 of these had visible vitrification on the rim surface. During the later Iron Age the typical crucible in lowland Britain was triangular in plan with three pouring lips formed in the rim of a hand-made bowl (Bayley 1989). Bayley (ibid) suggests crucibles were heated from above during this period as the most intense vitrification is always around the rim. The 25 vitrified rim fragments may be seen as examples of this. The majority of the rims are not vitrified so one must question if they have undergone the same heating process but somehow avoided vitrification, or if they were heated in different ways. Six of the rim fragments appear to be pouring lips, two of which have visible vitrification along their edge. XRF analysis indicated the presence of copper, zinc and a trace of tin. The fragment thickness varied from around 5mm to 20mm, with the majority approximately 10mm, however around 10% only had one original surface surviving. Fragments from each of the groups, divided by surface appearance, were examined under a low-power microscope to provide information on their fabric and its degree of vitrification. The fabrics were initially classified as either clay without any mineral inclusions, clay with chalk inclusions, or with quartz inclusions. However all three main fabric groups overlap with each other; for example, a fragment may be mainly void of inclusions but have a small number of chalk and quartz inclusions. The examination indicated there was no relationship between sample fabric and surface colouration. A common feature amongst the fragments was the appearance of voids, a result of organic temper being burnt out. Sample CR 3018 is not a crucible but a fragment of hearth lining. Those samples which were grouped as having possible visible metal residues (a total of 34) were initially analysed. However this is not the total number of fragments in the sample with metal residues, as other fragments undoubtedly also have metal residues on their surfaces. The results (see Table 2) indicate the presence of copper and zinc on all but one or two fragments, while silver, tin and lead were recorded on approximately half the fragments. The results indicate two important points, the first being that silver, tin and lead were identified only on the inside surfaces of the fragments, with two exceptions for lead. The second point is that silver, tin and lead were only identified by XRF when copper and lead were present in greater than trace amounts. This suggests that silver, tin and lead may be present in amounts which XRF cannot detect on the surfaces of fragments with only trace amounts of copper and zinc. The XRF results can thus be interpreted to suggest all crucibles were used for similar alloys, the variation in the elements detected being due to the levels at which they were present on the fragments. The crucible sherds displayed several different surface colourations, which 4

5 provided a means of grouping similar fragments. In addition to the crucible sherds with visible residues referred to above, samples from each of the other groups were analysed by XRF to determine if any metal traces were present (see Table 2). Fragments with red and pink surface colouration contain copper and zinc. Samples with yellow surface colouration also contained copper and zinc but always in only trace amounts; in three of the four samples lead was also present in trace amounts. Of the four white fragments analysed, three produced traces of copper and zinc while the fourth had relatively high traces of both together with a trace amount of silver. The pattern was repeated for fragments with a black glazed surface; one fragment which had a relatively high amount of copper also showed the presence of silver together with a trace of lead. This supports the point made above, that silver, tin and/or lead may be present, though not in detectable amounts, where traces of copper and zinc are identified. On crucible fragment CR 2925 a metal droplet approximately 10mm in diameter was noted (Fig 4). The localised vitrification around it suggests a small quantity of metal may have been melted using a blowpipe to create localised heating rather than by heating the whole vessel. Energy dispersive analysis in an SEM indicated the metal consisted of silver and copper. The amounts varied throughout the body of the metal, the silver-rich areas contained around 70-78% silver and 10% copper, while in other areas 56-70% silver and 30% copper were detected. In both areas two phases can be identified, a copper rich phase with around 70% copper and 10% silver, and silver rich phase with 80-90% silver and around 10% copper (Fig 5). A small number of inclusions were identified containing approximately 65% copper. This droplet, like the prill from coin pellet mould CM 1010, was somewhat corroded so the lower silver contents are likely to be representative of its original composition; ie, similar to the analysed prill. Interpretation and Conclusions Crucible Fragments The lack of any substantial morphology for the fragments, together with their varied surface appearance may initially be seen as evidence for a range of forms and uses. However there is no positive evidence for a variety of forms; all sherds could derive from a bowl form with pouring lip(s) in the rim. The variability of surface colour and degree of vitrification noted can be paralleled on a single crucible as the temperature and redox conditions in the hearth were never uniform. All crucibles are reduced fired, as metals must be fired under reduced conditions to prevent them from being oxidised and lost into a massive crucible slag. However during the later Iron Age the typical crucible in lowland Britain was heated from above and this resulted in the outside of the base sometimes being oxidised fired. The differing oxidation and reduction conditions, together with the chemical differences in the vitrified deposits, may be responsible for the varied surface colour and appearance. Likewise the degree of vitrification over a crucible is never uniform, a non-vitrified fragment may well originate from the same crucible as a heavily vitrified piece. The results of the XRF analysis of the 5

6 fragments indicate that some if not all provide evidence for non-ferrous metal working. The elements silver, tin, lead, copper and zinc were all detected in some samples. Bearing in mind the problems in interpreting XRF data outlined above, it is likely that the major (and perhaps only) alloy being worked was debased silver. SEM analysis of the metal droplet showed it has a silver-rich composition, probably typical of the metals being melted. Coin Pellet Moulds The analytical evidence from the moulds examined agrees with the results reported previously from Old Sleaford (Jones et al 1976, Heyworth and Wilthew 1987), that alloys of silver and copper similar to those melted in the crucibles were being used for the production of pellets of metal for coinage. There is however nothing to suggest that molten metal was poured into the coin pellet moulds. References Bayley, J. (1989) Non-metallic evidence for metalworking, in Y. Maniatis (ed) Archaeometry: Proceedings of the 25th International Symposium, Casey, J. (1983) Review of `Coinage and Society in Britain and Gaul: some current problems', edited by B. Cunliffe, Britannia 14, Clifford, E.M. (1961) Bagendon: a Belgic oppidum. Collis, J. (1985) Iron Age `coin moulds'. Britannia 15, Elsdon, S.M. (1975) Stamp and roulette decorated pottery of the La Tene period in Eastern England (BAR 10). Heyworth, M. and Wilthew, P. (1987) Coin pellet moulds from Old Place, Sleaford, Lincolnshire. AML Report 220/87. Jones, M.U., Kent, J.P.C., Musty, J. and Biek, L. (1976) `Celtic coin-moulds from Old Sleaford, Lincolnshire', Antiquaries Journal 56, Tournaire, J., Buchsenschutz, O., Henderson, J. and Collis, J. (1982) Iron Age coin moulds from France. Proceedings of the Prehistoric Society 48, Tylecote, R.F. (1962) The method of use of early Iron-Age coin moulds. Numismatic Chronicle (7th ser) 2,

7 Table 1: Analysed coin pellet mould fragments fragment hole fragment hole elements detected by XRF number diameter thickness depth copper zinc silver Small CM mm 6-8mm 6mm * * CM mm 8-9mm 5-6mm * CM mm 11-12mm 4-5mm t* * CM mm 7-9mm 4-5mm * * CM 179 5mm 9-12mm 5mm * CM 1249??? * * CM 243??? * * Medium CM mm 15-17mm 6mm * * * CM mm 14-16mm 7mm * CM mm 7-8mm 6-7mm * * CM mm 9-12mm 4-6mm t* * CM 111 8mm 12-16mm 6-7mm * CM mm 11-14mm 9-10mm t* * CM mm 21-23mm 7-8mm * * CM 123 9mm 16-18mm 8mm t* * CM mm 13-15mm 5-6mm * Large CM mm 16-19mm 6mm * CM mm 12-15mm 8mm * CM mm 16-18mm Filled * * CM mm 16-18mm 7mm t* * CM mm 16-18mm 6mm * CM mm 11-14mm 7-9mm * Key: * = element detected, t = trace amount only 7

8 Table 2: XRF results for crucible fragments fragment face of fragment elements identified by XRF number analysed copper zinc silver tin lead CR 2710 inside * * outside * CR 1187 inside * * * * * CR 1034 inside * * * *? * CR 529 inside * * * t* * CR 1301 * CM 1287 unknown t* * CR 1210 inside t* t* outside * * t* CR 1909 unknown * * t* CR 2576 inside: metal spot t* * inside * * outside * * * CR 2438 unknown * * CR 1621 inside * * * outside * * CR 1441 inside * t* CR 1347 inside h* * * * t* CR 1351 inside t* t* CR 1038 inside * * * t* * CR 770 inside h* * h* * * outside * CR 411 inside h* * * * CR 286 inside t* t* CR 2451 inside t* CR 2711 inside * * * h* * CR 1880 inside h* h* CR 2310 inside? * * CR 2121 inside * * t*?* CU 2064 inside * * h* * CR 2960 inside t* t* CR 2958 inside * * outside OS 851 inside t* t* outside, black * * outside, white * * CR 2979 inside h* * * 8

9 fragment face of fragment elements identified by XRF number analysed copper zinc silver tin lead CR 2744 inside t* t* outside *?* * CR 3059 inside: metal spot * * * * outside * * CR 1196 inside * * * * * CM 3076 inside t* CM 2993 inside t* t* CR 3017 inside h* * h* Samples with red colouration on the surface CR 2063 inside t* t* CR 3023 outside * * CR 3095 * Samples with pink colouration on the surface CR 2892 inside * * outside * * CR 2047 inside, deposit h* * inside t* t* CR 2962 inside t* * * Samples with yellow colouration on the surface CR 2164 inside t* t* t* t* CR 2980 inside t* t*? 1873 inside t* t* t* t* CR 2934 inside t* t* CR 1353 inside t* t* t* diff t* fabric? Samples with white colouration on the surface CR 2687 CR 2649 inside * * t* CR 2573 unknown t* t* CR 2738 unknown t* 9

10 fragment face of fragment elements identified by XRF number analysed copper zinc silver tin lead Samples with black glazed surface CR 567 CR 1391 CR 1977 outside h* t* * t* CR 2940 Samples with chalk in their fabric CR 2569 inside outside CR 2996 inside t* * CR 2034 inside t* * CR 801 inside t* Key: * = element detected, t = trace amount only, h = relatively high signal 10

11 Summary. Excavations at Old Sleaford, Lincolshire in 1963 discovered over 4000 fragments of late Iron Age coin pellet moulds and 320 fragments which were thought to be from crucibles. A sample of both groups was examined and analysed. The artefact assemblage is interpreted as evidence of a mint on the site. The analytical evidence supports this view, the results suggest that the coin pellet moulds were used in the production of silver alloy flans (blanks) for coin production. Analysis of the crucible fragments by XRF identified copper, zinc, silver, tin and lead on the surface of a number of them, indicating at least some of them were used for melting alloys that contained silver. This report complements AML Report 220/87 which describes comparable material from later excavations on an adjacent site. 11

12 Figure 1: Section of pellet from coin pellet mould CM Note the two phase dendritic structure and the porosity and deep corrosion of the metal. The upper surface of the pellet is on the right side. Scale bar 1mm. Figure 2: Microphoto of a central area of pellet from CM 1010, showing the two phase structure. Scale bar 30μ.

13 Figure 3: Microphoto of a metal droplet from coin pellet mould CM 1136, showing porosity and the two phase structure. Scale bar 50μ. Figure 4: The metal droplet on crucible fragment CR Figure 5: Microphoto of a section through the metal droplet from CR 2925, showing the dendritic two phase structure. Scale bar 200μ.