B.—CHEMISTRY. 45 
_ nickel or a fused mass of albite and anorthite, begins to solidify, the 
- composition of the crystals has to adjust itself continuously in order to 
maintain equilibrium with the changing liquid phase, as was shown by 
Roozeboom in his classical work on solid solutions. Such an adjustment 
is only possible by means of diffusion, and when cooling is sufficiently 
slow, the adjustment does in fact keep pace with the change in the liquid, 
but with more rapid cooling the interior of each crystal differs in com- 
position from its outer layers, there being a concentration gradient from 
the centre to the boundary. This condition produces the ‘ cored ’ crystals 
which are familiar to every metallurgist, and the ‘ zoned’ crystals of the 
mineralogist. In most alloys this want of homogeneity disappears after 
a sufficiently long period of heating at some temperature below that of 
which the first drops of liquid are formed, but alloys of bismuth and 
antimony fail to become uniform even after weeks of annealing, whilst 
the felspars and similar minerals have never been persuaded to lose their 
zoned structure by any methods known in the laboratory. 
Bruni has shown and Vegard has confirmed the observation by the 
X-ray method, that true interdiffusion occurs between potassium and 
sodium chlorides when mixed and heated in the solid state. Electrolytic 
transport is observed in the solid halides of silver and in mixtures of silver 
and copper sulphides, but the modern view of the structure of such sub- 
stances represents them as built up of ions rather than of neutral atoms, 
and this must be taken into account in any interpretation of the facts. 
The apparent absence of diffusion in minerals which have once solidified, 
even when given geological periods of time, is a serious difficulty in the 
way of any general theory of diffusion. Such examples of the passage of 
alkali metals through quartz and other silicious minerals under the 
influence of a difference of electric potential are probably not instances 
of true diffusion at all, but merely of the passage of traces of impurities 
through a mass which is not completely impervious. We have always 
to bear in mind that crystals, whether of natural origin or prepared in 
the laboratory, are rarely perfect, and may contain cavities and capillary 
passages through which matter may pass without disturbing the crystalline 
lattice. This idea of the imperfection of crystals has found an interesting 
application in the work of A. A. Griffith on the rupture of solids and of such 
_ semi-solid substances as glass and fused silica. The tensile strength of 
metals and of these substances is far smaller than would be expected from 
calculations of the theoretical cohesion of the materials. Griffith supposes 
that actual solids and glasses contain innumerable fine cracks, which 
reduce the strength. By special means he has been able to prepare rods 
of glass and silica in an unstable state, in which their strength and 
elasticity are enormously greater than in their normal condition. It has 
even been suggested that means may be found for bringing our ordinary 
metals and structural materials into a similar condition, which would 
enable them to withstand loads several times greater than those which 
are normally possible, although the prospect of a sudden return to the 
: stable condition with its accompanying weakness may alarm the engineer. 
7. However, the use of materials in an unstable condition is already 
_ familiar to metallurgists. Hardened steel is an instance. At high tem- 
peratures the structure of most of our steels is homogeneous, the carbon 
being in solid solution in the iron, which is then in the y-condition. As 
