of Bronzes from Machu Picchu, Peru. 571 



considerably thicker at this point than the shank itself. With 

 this possibility in view, a longitudinal section through the 

 center of the entire blade, shank, and handle was cut for metal- 

 lographic examination. The outline of this section is shown 

 in fig. 36. The entire surface of the section was explored 

 under the microscope after the usual preparation. 



The general structure, characteristic of cast metal, is illus- 

 trated by fig. 37, taken from the base of the blade, b, as indi- 

 cated in the sketch. In this micrograph, the greatest possible 

 hetheogeneity was developed by etching rather deeply with 

 ammonia-hydrogen peroxide and then lightly with ferric 

 chloride-hydrochloric acid. The former reagent does not 

 reveal the black cores but attacks the metal rather uniformly 

 as far as the tin-rich boundaries of the crystallites, showing a 

 raised network of rounded units, each possessing a black border 

 which is not a dark etching but is due to inequality of focus 

 at the edges. The large black cores developed by the ferric 

 chloride represent the copper-rich centers of the grains. Under 

 high power, a small quantity of bluish-gray, or slate-colored, 

 constituent can be seen. This is probably cuprous sulphide, 

 which was invariably found in these bronzes and could be 

 brought into harmony with the analytical figures where sulphur 

 was reported in the analyses. (Cf. description of Object Eo. 

 2 for distinction between this constituent and the a -f- h com- 

 plex.) It is always located within the tin-rich network, 

 described above, as are other comparatively insoluble impuri- 

 ties, mechanically mixed foreign material such as tin oxide, 

 of which indications were obtained in some cases, contraction 

 cavities, etc. Rapidly cooled bronzes, which should normally 

 consist of the a constituent alone, usually show traces of the 

 cc -f- $ complex even where the tin-content is lower than the 

 present value, 4.82 per cent. This is true in the case of our 

 own chill-cast specimens ranging from 4 to 8 per cent tin and 

 is due to the comparatively slow rate of diffusion as explained 

 earlier. In the present case, however, no traces of any eutec- 

 toid structure element could be detected by thorough search 

 with the highest power at our disposal ; a one twelfth inch 

 fluorite immersion objective of 1.32 mm. aperture. Such a 

 condition could be produced by slow cooling, but numerous 

 experiments in casting metal at widely different rates of cool- 

 ing have shown beyond question, that the size attained by the 

 primary dendrites when homogenization in this respect is 

 secured by rather uniformly retarding the rate of cooling 



