ee te i 
B.—CHEMISTRY. 43 
ean make their way through the solid, must be possible. Only by assuming 
the reality of diffusion in solids can one explain the changes brought about 
in metallic alloys by heating and cooling, or the structure of minerals in 
igneous rocks. No very refined observations are necessary to establish the 
fact of diffusion, although quantitative measurements in this field are 
difficult. A large steel forging which has cooled slowly shows, when etched, 
triangular markings which recall the Widmanstatten figures seen in 
meteorites and are, in fact, of similar origin. From a solid solution in which 
carbon is unformly distributed throughout the crystals of iron, almost 
pure iron has separated in these characteristic bands, leaving the carbon 
concentrated in the remaining material which fills the meshes. For such 
a structure to be produced, some of the atoms of iron or carbon, or both, 
must have travelled through the crystalline steel over distances of the 
order of a millimetre in the course of some hours. Experiment shows that 
diffusion in solids, whilst naturally a slow process in comparison with diffu- 
sion in liquids, proceeds at quite measurable rates, the distribution of the 
invading atoms at different distances from their place of entrance following 
the familiar law, so that a coefficient of diffusion may be calculated from 
analytical results or from microscopical observations. The classical example 
of such measurements, and for many years the only one, is the study of the 
diffusion of gold in solid lead, undertaken by Roberts-Austen in 1896. 
It was then shown, and the figures have since been confirmed by a very 
accurate series of determinations by Van Orstrand and Dewey, that gold 
diffuses into solid lead at 200° at a rate which is 1/420 of that at which it 
diffuses into liquid lead at 550°. This is not the best pair of metals which 
could have been chosen, as lead and gold form compounds with one another, 
so that something more than mere physical diffusion is involved, but the 
choice was an obvious one, on account of the delicacy of the analytical 
methods of determining the distribution of gold in successive layers. Even 
at 100° the diffusion has a measurable value. A much simpler example 
is that of silver and gold, two metals which resemble one another closely 
in chemical character and in atomic volume, so that diffusion causes less 
change of properties than in any pair of less closely similar metals. The 
experimental results prove, as might have been anticipated, that diffusion 
is a much slower process when there is so little difference in chemical 
character. The value of the coefficient of diffusion varies with the condition 
of the experiment, a solid solution which contains much of the diffusing 
element offering a far greater resistance to diffusion than does the pure 
solvent metal. The same is true of other pairs of metals, and of the 
diffusion of carbon into iron, a process of the highest technical importance. 
When the two kinds of atoms are closely alike, the tendency to diffuse must 
be small, but it is certainly not zero. By making use of an ingenious device, 
Hevesy has been able to determine the coefficient of self-diffusion of 
liquid and solid lead. Two isotopes should not differ appreciably in their 
rates of diffusion, so that when the radioactive isotope thorium B is allowed 
to diffuse in ordinary lead the experiment is equivalent to selecting a 
certain number of lead atoms and attaching labels to them by which they 
may be identified in the course of their journey. In this way he found that 
the diffusion in liquid lead near to the melting point was of the order of 
that of salt in water, but that in the solid state it was very small. Further 
experiments, using a thin foil, proved that at 2° below the melting point 
