of Bronzes from Machu Picchu, Peru. 563 



produced no change in the original casting structure. In a 

 later communication (22), the same author shows a set of 

 photo-micrographs which represent stages in the homogenization 

 of an alloy of the same copper-content at 800°. Homogeniza- 

 tion was likewise complete in this case after an annealing period 

 of three hours. The condition of inhomogeneity in cast alloys of 

 the solid solution type is due to incomplete diffusion during the 

 solidification interval and throughout the subsequent period 

 of cooling. It is, therefore, relieved, in accordance with the 

 laws of diffusion, by exposure for an adequate period of time 

 at an effective temperature. 



Fick, in 1855 (23), gave a mathematical analysis of diffu- 

 sional phenomena based upon Fourier's theory of thermal 

 conductivity and fundamentally related to the principle that 

 the rate at which a dissolved substance diffuses into the solvent 

 is proportional to the difference in concentration from point 

 to point. According to Nernst's theory of diffusion (24), 

 osmotic pressure is the driving force which causes the move- 

 ment of diffusion from regions of high concentration towards 

 those of low concentration. Roberts- Austen (25) showed that 

 fluid metals diffuse in one another according to Fick's law. 

 He also measured the diffusivity of several solid metals in one 

 another at a number of temperatures. Recent progress in the 

 study of constitutional relationships has enabled us to clearly 

 specify those metallic readjustments which are essentially 

 diffusional in nature and not complicated by other molecular 

 changes due to solution, precipitation, chemical reaction, etc. 

 The homogenization of a solid solution is distinctly a process 

 of this sort which, in all probability, proceeds according to 

 Fick's law and is very likely due to osmotic pressure between 

 regions of unlike concentration. 



It is thus clear that we must consider primarily the variables 

 of time, temperature and concentration in dealing with this 

 subject. Any experimental result is not clearly defined unless 

 it includes proper specification of all three factors. Accord- 

 ingly, we cannot make broad use of data similar to that 

 mentioned above (Portevin). 



The rapid increase in the rate of diffusion with rise of tem- 

 perature is a general condition of particular significance in the 

 annealing of such alloys as german silver and cupro-nickel 

 which require very high temperatures for rapid effects. It is 

 well known that ordinary bronzes, brasses, etc., when abnor- 

 mally inhomogeneous and, therefore, unstable do not spontane- 



