42 SECTIONAL ADDRESSES. 
others, a fact which we should expect from the general properties of the 
space lattice. It is not explained, however, why these etching pits should 
appear at first separate from one another, the intervening portions of the 
surface being unattacked. Minute particles of some impurity, causing 
local electrolytic differences, suggest themselves as a possible cause, but 
it is unlikely that they would be so evenly scattered in, for instance, a 
quartz crystal as to produce the regular distribution which is often 
observed. Minute inequalities of level, which may be of a periodic 
character, are more probable, and this suggestion is strengthened by the 
observation that a polished face of rock salt dissolves evenly in water, 
whilst a natural cleavage face shows etching pits. 
Lastly, another cause of want of homogeneity in solids is the presence 
of portions which have been deformed beyond their elastic limit. Such 
deformation alters the electrolytic potential of a metal, so that a couple 
is set up between the deformed and undeformed portions, even bringing 
about action in otherwise remarkably inactive iron of high purity used by 
Lambert in his experiments on corrosion. A true theory of corrosion will 
have to account for the formation of etching figures in apparently homo- 
geneous substances. 
It is now possible, when pursuing the study of solids, to eliminate one of 
the disturbing factors, the intercrystalline boundary, by making experi- 
ments with specimens composed of a single crystal. There are several 
ways of preparing single metallic crystals of such a size as to allow of the 
determination of their physical and mechanical properties. Carpenter 
and Elam have strained sheets of pure aluminium in tension, producing 
a small permanent elongation, and this sheet, after suitable annealing, 
shows such a remarkable increase of size of its crystal grains that frequently 
one occupies the whole specimen. Czochralski’s method is to dip a silica 
point into slightly undercooled molten metal, and then to raise it by clock- 
work at a rate which just keeps pace with the growth of the crystal, thus 
obtaming a thin cylindrical specimen. Davey has prepared large single 
crystals of copper by allowing the molten metal contained in a tube to 
freeze slowly from one end, whilst tungsten filaments of great length have 
been prepared by suitable thermal treatment during and after drawing. 
All these specimens have been studied, their great ductility being a 
characteristic feature. Even so brittle a metal as zinc has an extraordinary 
ductility in single crystals. The mechanism of deformation has been 
examined in detail by means of X-rays, aluminium having been studied by 
Taylor and Elam, zinc and tin by Polanyi and his colleagues, and tungsten 
_by Goucher. There is now a large body of evidence as to the directions of 
slip in a crystal during deformation, and this knowledge is essential to any 
understanding of the nature of cohesion, with which the chemical properties 
are no doubt closely connected. 
We may now turn to the subject of chemical reactions which take place 
in the interior of a solid, either originating at the surface or from nuclei 
which make a spontaneous appearance in the course of cooling below the 
melting point. A chemical change which has begun at some point in or at 
the surface of a homogeneous crystalline mass cannot advance unless the 
atoms are able in some way to change their places. Gross movements, 
represented in gases and liquids by convection currents, are out of the 
question, but the slower process of diffusion, by which atoms or molecules 
