while many gold objects are cast, worked from sheet or have simple chased, repoussé or dot-punched decoration (Shimada and Griffin 1994, Bray 1978b, Plazas and Falchetti 1994), there are examples of more complex decorative craftsmanship which leave characteristic toolmarks on jewellery. It would be interesting to compare such toolmarks with the collections of tools in museums (eg. Carcedo Muro in this publication). Two objects with interesting toolmarks are discussed and illustrated.

The earspools are made as two similar items from sheet gold (0.25 mm thick, 88.0%Au, 9.3%Ag, 2.7% Cu) formed into complex three dimensional shapes, Fig. l7. Each earspool has a tubular sleeve (which allows the two to connect together in the photograph) but when used they would probably have been worn as a pair, each secured with a wooden plug fitted into the end of the tube at the back of the ears. The earspools are 30mm in diameter and 40mm long when the two are fitted together. The main earspool discs each have a very tall central dome (10mm high) probably made by hammering anta or into a former. Details of the curved form of one of the earspools and the three characteristic types of  toolmark on the gold can be seen in the oblique magnified SEM image, Fig.18.

There is a simple circular line formed by a single narrow flat edged chisel that has been tapped anta the sheet and moved sequentially after each mark thereby creating a continuous line around the centre ring.

One of the main features of the design is a ring of sixteen triangular punch marks around each disco The marks are all of similar shape and size, confirming that they were made by the same punch. The disc would have been backed by a resilient material that allowed the punch to make the deep impressions without distorting the surrounding sheet. The punch marks are evenly spaced and are of consistent depth but some have just pierced the sheet along sharp edges. Both of the earspools have identical marks confirming that they are a related pair. Such characteristically shaped toolmarks can be used as a good indicator for associated workshop origins of assemblages of goldwork.

Figure 17: Coclé earspool, 30mm diameter, showing the punched decoration on the disc and the over-lapping tubes between the two halves. [photo: A. Milton, British Museum].

A third type of toolmark is seen as a regular indented pattern punched anta the raised inner rim. The marks show that a thin flat edged tool was gently tapped anta the pre-formed rim making indents that give a beaded effect. SEM examination shows slight miss-registering of the lines in a number of places indicating double punching of these lines.

The earspool halves are each joined to a backing sheet at the outer rim which is folded over a backing sheet making a mechanical joint. The backing sheets each have a long central tube which would pass through the ear lobe to be fastened at the back with a wooden plug. (Because the robes are of slightly different diameters the earspools are shown pushed together).

Figure 18: SEM oblique view of the Coclé earspool showing detail of the three types of punch marks. [photo: N.D. Meeks, British Museum].

The general form of the nose ring and its silver-rich composition suggests that it might come from the central coastal region of Ecuador, post 1000AD-conquest (Warwick Bray personal communication). The nose ring is 22mm diameter and made from thick, hammered sheet (45.6%Au, 45.9%Ag, 8.5%Cu) forming a hollow tube 7mm diameter with a seam running around the inner circumference (Fig.19).

Figure 19: Penannular ring from Ecuador, diameter 22mm. [Photo: A. Milton, British Museum].

The ends of the tubular ring have a gap of about 2mm. Making such a tight, hollow ring is itself a difficult task. A decorative bi-convex sheet sleeve, 8.5mm wide, is wrapped tightly around the tubular ring and the ends just overlap, (Fig.20a). This averlap joint was examined in the SEM and there appears to be copper-rich solder at this joint which has not properly fused and has subsequently parted. The subtle bi-convex curvature of the sleeve again shows considerable dexterity by the goldsmith. The sleeve has geometrical decoration mainly of grooved lines, and with some circular dots which arrear to be punched. In the SEM the decorative grooves arrear slightly concave and clearly show fine toolmarks at right-angles to the direction of the grooves Figs.20b. The toolmarks were made by gently tapping a sharp tool with a hammer to carve out the metal. The grooves are only O.3-0.4mm wide and skilfully applied.

a	b
Figure 20a and 20b: SEM detail of the toolmarks on the penannular ring from Ecuador, (a) general view of the decorated sleeve on the ring, (b) detail of the chiselled toolmarks. [photo: N.D. Meeks, British Museum].

There are many small metal goldsmiths tools, polished stone anvils and hammers in the collections of South American museums, ego the Museo del Oro, Bogota (Plazas and Falchetti 1994), and the Museo Nacional de Antropologia y Arqueologia, Lima, (Carcedo Muro in this publication). It would be possible to take impressions of tool edges with silicone rubber and examine these in the SEM to determine the type of mark the tool would make and thereby identify for what purpose the tools were used (Sax and Meeks 1995). Similarly the study of toolmarks on goldwork in general can be made using silicone impressions which removes the restriction of only studying objects small enough to be examined directly in the SEM.

The detailed metallographic and analytical examination of examples of Central and South American goldwork from the British Museum using scanning electro n microscopy and microanalysis has allowed the collections to be categorised within the existing technological framework of south American gold and has also shown several points of interest. It appears that the combination of copper-rich alloys plus depletion gilding involving heating, quenching in pickle and possible boiling, results in the age-ordering over archaeological time to produce binary phase structures. By comparison, the copper-rich alloy fish hooks of similar composition, which are not depletion gilded, had both cored (cast) and homogenised (worked) microstructures without segregated phases. This supports the suggestion that the processes involved with depletion gilding are responsible for the ultimate appearance of the segregated two-phase structures. The binary phase alloys also show reversal of normal corrosion, in that the gold-rich phase corrodes anodically while the copper-rich phase is cathodic and passive. Thick depletion gilding with its associated diffusion zones and porosity indicate repeat cycles of heating and oxidation followed by pickling to produce the desired thick golden surfaces.

The metallography of Cauca river fish hooks shows that two methods of wire manufacture were used, hammering to fully form the wire and casting (in a similar fashion to false filigree work) followed by minimal working to finally shape the hooks. Analysis of the fish hooks and the associated alluvial gold grains shows that the hooks were made from both the natural alluvial argentiferous gold and from ternary alloys made by adding copper to the alluvial gold. The range of copper content suggests that the actual composition mar not have been important for this type of object, although selection of particular alloy compositions could have helped, particularly during casting, if the metalsmiths had chosen to do so.

Workmanship of a fine quality is seen on the earspool and penannular ring which show a variety of identifiable toolmarks on the goldwork, and this is an area of study that could be extended. There is scope for the examination and recording of toolmarks either directly in the SEM or by taking silicone rubber moulds of the toolmarks and examining these in the SEM to build up full technological documentation of the South American goldsmiths crafts on such objects.

Many of the South American objects at the British Museum have been cleaned before acquisition so there is no record of the treatment used. It is therefore not possible for us to know exactly how the present golden surfaces compare to the original manufactured surfaces. The green copper corrosion products commonly found covering the gilt surfaces of tumbaga alloy objects from burial (Lechtman 1984, Shimada and Griffin 1994) indicates loss of copper from the copper-rich alloy. Scott (1983a) discusses the problems associated with corroded tumbaga alloys and the care required during treatment to preserve the technological information within the objects.

By comparing sections of gilt surfaces with published microstructures, typically those of Heather Lechtman and David Scott, it is believed that structures found within 1-2 microns of the surfaces of the examined objects are essentially characteristic of those created during manufacture, and that any changes due to superficial cleaning or polishing are probably insignificant and occur only on the outermost surfaces.

Whether corrosion during burial could create the characteristic porous microstructure of a depletion gilt surface on a copper-rich object where none existed originally needed to be considered. Microanalysis of the uncleaned surfaces of a range of fish hook alloys compared to cleaned areas of underlying metal did show that surface enrichment due to burial had occurred within a few microns of the surface which altered the surface composition by up to a few percent. But there were no obvious porous structures that are characteristic of deliberate depletion gilding. This is because the reaction kinetics are very different for depletion gilding which involves fast, high temperature oxidation and pickling compared to slow, long-term, ambient temperature loss of copper during burial. It is this characteristic porosity that sets depletion gilding apart from natural surface enrichment over time.