Analysis of alloy composition is an obvious avenue for research, but as Jet there is not a large database of analyses (see Rovira 1994 for a review of gold). As the opportunities increase to have analysis done on pre-Hispanic metalwork, the database will become large enough to allow statistically valid comparison and grouping of objects now scattered throughout international collections. However, without comparability in the analytical approach, there are likely to be some problems which are illustrated by the Muisca tunjos discussed below.

Photograph 1: Muisca tunjos (height of largest, 11 cm). The numbering correlates the objects to the analyses in tables 1 and 2, and figure 2. [photo A. Milton, British Museum].

Muisca tunjos are miniature representations of artefacts, animals or (most commonly) human beings, singly or in groups. A few are made from pure copper or gold, but the overwhelming majority are cast in gold alloys. They were made as votive offerings by the Muisca of the altiplano of Cundinamarca and Boyacá, central Colombia, in the period between the 6th and 16th centuries AD, and are usually found (sometimes several together, in ceramic offertory vessels) at isolated landmarks such as lakes, peaks, and caves (Plazas, 1975; | Lleras-Pérez,1997). They were made by the lost-wax method of casting; | the model was cut from wax sheet les s than 1mm thick and applied 'wires' of wax provided the detail as well as giving the sheet strength and doubling as runners and risers for the casting (photographs 1 and 2).

The metal surfaces are conspicuously rough and unfinished, presumably because of the votive nature of these pieces. Ornaments made by the Muisca for wear, such as the necklace pendants in photograph 3, are also cast but they have a much smoother surface.

Photograph 2: Magnified detail of tunjo 6 showing the modelling of the face and cast structure. (width 1 cm), The black patches are residues of the casting mould [photo S. La Niece, British Museum].

Several of these tunjos have black mould material still adhering to them and no. 8 in photograph 1 has the casting sprue, where the metal was poured in at the top of the mould, still attached to the foot (these figures were cast head downwards). It is not at all uncommon to find tunjos where the casting sprues and feeders for the molten metal have not been trimmed off (see Bray, 1982, nos. 59 and 60). Many also have major casting faults, for example the holes in numbers 1 and 2 in photograph l.