80 SCIENCE PROGRESS. 



kept for some months. This may be an extreme case ot 

 the weakness inherent in a conglomerate alloy. 



A study of the proportions and temperatures at which 

 the metals when melted in pairs form homogeneous liquids, 

 ought therefore to be the first step in a scientific study of 

 alloys. Matthiessen was aware of the importance of this 

 subject, and made experiments on the miscibility of pairs 

 of metals ; but the later experiments of Alder Wright are 

 more accurate and are the only ones we shall quote. 



The behaviour of partially miscible liquids to one 

 another has, however, been much more perfectly studied by 

 Alexejew 1 and Konovalow. 2 These experimenters did not 

 examine metallic solutions, but it is almost certain that the 

 general conclusions they arrived at are applicable to alloys. 

 The facts, so far as they affect our subject, can be very 

 briefly stated. If a liquid B is gradually added to another 

 liquid A and the mixture is sufficiently stirred, we shall at 

 first get a homogeneous liquid, which may for convenience 

 be called a solution of B in A. Continued addition of B 

 will make this solution stronger, but it will remain homo- 

 geneous until A is saturated with B. A further addition of 

 B, or the removal of some A by evaporation or freezing, 

 will now cause the appearance of a second layer of liquid 

 not miscible with the first. This new liquid may appro- 

 priately be called a saturated solution of A in B. A further 

 addition of B will not alter the composition of these two 

 layers, but will cause the second layer to increase at the 

 expense of the first, and if we add enough of B the first 

 layer will disappear and we shall have only the saturated 

 solution of A in B, while with a still further addition of B 

 we shall obtain an unsaturated solution of A in B. In 

 other words, there is an infinite number of different pro- 

 portions in which A and B may be shaken up together so 

 as to give two layers of liquid, and in all cases the com- 

 position of these two layers will be independent of the 

 proportion of A and B employed. These two layers of 



1 Wied. Annalen, 28, 305, 1886. 



2 Ibid., 14, 219, 1881. 



