428 
MR. F. OSMOND AND PROFESSOR ROBERTS-AUSTEN 
fissures, the thickness equals 1 g to 5 g, or a mean thickness of 2'5 /a. Generally speak¬ 
ing the condition of these joints appears to be closely related to the mechanical 
properties of the alloy (table, p. 418.). 
Conclusions .—We do not contest in any way, as our previous publications abund¬ 
antly prove, the importance of the part which may be played in the mechanical 
properties of the alloys by the residues which remain liquid after the main mass of 
the alloy has solidified, the alloys being tested either at the ordinary temperature or 
when heated. But, in order that it may be possible for such cements to intervene 
and affect the mechanical properties of alloys, the cements must at least have a real 
existence. Nothing indicates that they do exist in ten out of twelve of our alloys, but 
we would not even express ourselves too positively on this point, for some new method 
of etching may reveal new facts. The impurities which are sought for may happen to 
concentrate themselves beyond the particular region which has been sectioned. These 
are, however, for the present gratuitous suppositions. Polishing only indicates the 
presence of cement in two cases. The little secondary crystals which we have already 
described might readily be mistaken for cements of definite or indefinite composition 
if they were found only in certain specimens and then in such proportions as could be 
accepted. But we meet with them everywhere and in all cases their appearance is 
constant in forms and dimensions, and moreover we see them collect into crystallites 
which pervade the whole mass. These crystals are therefore usually and indubitably 
due to the crystallization of gold itself, although the alloying substances sometimes 
(indium and probably potassium) join up the crystals in question. For the same 
reason the dark lines of the joints, traced as furrows by the etching, are very rarely 
the empty tracks of cement which has been dissolved away by aqua regia ; their 
formation, which it is easy to follow in all its phases, directly connects them with 
secondary crystallization. We are led to the belief that in the case of ten of our alloys 
of gold with about 0'2 per cent, of various impurities, solidification of the whole mass, 
although prolonged and less rapid than in the case of pure gold, has been directly 
accomplished without interruption, and that the foreign bodies have remained partly 
or wholly as solidified solutions, the impurities being dissociated into their atoms 
in both solid and liquid. We can at least say, without going beyond the actual 
evidence before us, that the dissemination of the foreign bodies eludes the power 
of the methods of investigation which we have employed. Under these conditions 
it is difficult to invoke, as explaining the mechanical properties of the alloy, the 
intervention of hypothetical cements with relatively low fusing points. 
The absolute dimensions of the grains or crystallites cannot, as we have already 
seen, account for the mechanical properties. The micrographic examination of the 
Gopper-zinc alloys (brasses) has recently led M. Chakpy* to a similar conclusion, and 
it is now certain that the large size of the grains does not, in itself, constitute a 
# Loc. cit. He also shows that ordinary brasses may he effectively annealed at 500°. 
