Two distinct alloy groups were found within the 44 fish hooks analysed. Sixteen hooks were gold/silver alloys without any copper (Ag range 7.2% 16.7%) and, significantly, were similar in composition to the alluvial grains. 28 hooks were of ternary alloy with a wide range of copper compositions between 51.9% and 0.7%, but the silver content of these (equivalent Ag/Au composition range 7.4% - 15.5%) is also similar to the copper-free hooks and to the alluvial grains (see Fig.16 where the orange dots of the gold/silver alloy hooks overlap the green dots of the alluvial grains). One fish hook has much higher silver (27%) and no copper, but is still within the range of natural alluvial gold (Rehren and Temme 1994).
The copper contents of the fish hooks fall on a line that can be projected back from the alluvial gold composition towards the copper apex of the ternary phase diagram, with fairly constant silver/gold ratios. This is similar to the results of Rovira (1994 Figs. 14,15,18, pp 335-337) showing that all of these fish hooks were made from alluvial argentiferous gold including those with alloyed copper. Similar trends are noted by La Niece (in this publication) with the composition of tunjos figures, and also for other south American objects (Rehren and Temme 1994).
The range of compositions found for the fish hooks suggests that there was no attempt at producing an alloy with a consistent copper composition and this point is also noted by Rovira (1994). It could be argued that for some of the copper-containing alloys the melting point is particularly low or the freezing range of the alloy is long and therefore these features would have helped casting properties of the alloys. Even the presence of small amounts of copper significantly reduce the melting point of gold (Fig. 3). The variable composition mar reflect on the abundance or scarcity of extracted or traded metal at any particular time. Economics of metal use mar have played some part, as winning copper from the ore was a complex process (Shimada 1994, Merkel et. al. 1994) with relatively fewer sources of copper than gold. The alluvial sources of gold were more widespread and various practical methods were devised to exploit the sources from numerous streams and rivers (Plazas and Falchetti 1978, Bray 1978b). The very low copper content of some of the fish hooks suggest that scrap alloy mar have been incorporated with the alluvial gold, or some of the local alluvial gold had a small copper content as reported by Scott (1991a), Rheren and Temme (1994) and Rovira (1994) from some South American contexts.
Figure 16: Analyses from table 3 of the Cauca river fish hooks (orange) and alluvial grains (green) plotted on the phase diagram for the ternary alloys of gold-copper-silver. Curved lines show the liquidus contours relating to composition. [Phase diagram, Metals Handbook, 8th edition, Vol. 8, Lyman 1973].
The appearance of two types of fish hook alloy (those of ternary alloys containing copper and those of alluvial argentiferous gold) and two types of wire manufacture (cast wire and hammered wire) mar reflect on the goldsmith's preferred individual practice rather than there being a 'standard' workshop procedure for fish hook manufacture. Although some fish hooks show an association of wire type with composition, ie. copper-rich/cast and argentiferous gold/hammered, this is not always the case.
Osmium/iridium/ruthenium inclusions in the fish hook wire.
Several of the fish hooks had silver/grey coloured inclusions of osmiuml iridium/ruthenium in their surfaces. They were found in both the argentiferous gold wire and the ternary gold wire. These platinum group element (PGE) inclusions are found in alluvial gold both in South America (Scott and Bray 1980, 1994) and in many other sources in the world (Ogden 1977). They can have a wide composition range, often within a single cluster of grains (Meeks & Tite 1980) and those examined here are no exception. Examples are given in table 3e. Two inclusions have low platinum and are in hooks from both the Cauca river and the Tumaco/Buenaventura region. However, the composition of these inclusions is unlike those of the platinum placers of the Pacific coastal area where alluvial platinum is encountered (Scott and Bray 1980, 1994, ZeilI979).
The inclusions are extremely hard and could cause difficulties in hammering gold into thin sheet or wire with the danger of damaging the tools, particularly the finely polished stone hammers and anvils if precautions were not taken. They are not removed from the gold during melting as they are refractory alloys with little solubility in gold (Craddock 1995). Plazas and Falchetti (1978) suggest 'bitter gold' was probably refined by cementation to remove metallic impurities before use (although this would also remove silver and copper). Craddock (1995 & personal communication) suggests that gravity segregation by melting the gold in deep crucibles might have been more effective for removing Os/Ir/Ru inclusions than cementation. Inclusions have not been noticed in other objects examined so far in this project but they were reasonably obvious in the fish hook wire, and caused splitting and weakening of the gold wire in one hook where it was bent to form the loop.