509 
of Edinburgh , Session 1868 - 69 . 
the volume of the metal. He found that a wire of palladium 100 
millimetres in length became 10T5 millimetres when saturated with 
hydrogen. Now, if a uniform hydrogen atmosphere surrounds a 
symmetrical piece of palladium, there is no reason why it should 
penetrate with a greater rapidity one surface rather than any 
other. But if the absorption is not uniform on all the surfaces, 
from want of uniformity in the hydrogen atmosphere, the surface 
absorbing must produce a flexure of the plate from the expansion 
of the metal. If a thin plate of this rigid metal can be so arranged 
as to induce absorption on one side rather than the other, then, as 
a necessary consequence, the plate will become convex on the side 
where the greatest relative absorption is taking place ; and as the 
saturation approaches uniformity, the convexity should disappear, 
the plate regaining its original form if the elasticity of the metal 
is not changed during the action. A plate of palladium, when 
it functions as the negative pole during the electrolysis of water, 
is subjected during the course of the action to the supposed 
non-uniformity of the gaseous atmosphere — if the surface of the 
plate is parallel to a similar platinum electrode. The amount of 
chemical action effected by the unit surface of the electrode, and 
its distribution, depends upon the position of the plate relatively 
to the lines of force emanating from the poles. Now, of the two 
surfaces of the palladium plate, the surface next to the positive 
pole has the greatest concentration of the lines of force, and there- 
fore the greatest evolution of hydrogen takes place on this surface, 
so that the quantity of alloy formed in the first instance is in 
excess of the amount produced on the other side. The plate, from 
the great expansion on the outer side, becomes convex, until the 
progression of the action from the other side is able to compensate 
this flexure by a corresponding expansion, thereby bringing the 
plate to its original form. But I have already explained how this 
backward motion goes beyond the original position, producing a 
flexure in the opposite direction, only in some cases returning to 
its normal shape. This, in all probability, arises from the com- 
bined effect of expansion and compression. Graham has shown 
that this tenacity of the alloy is considerably diminished as com- 
pared with that of the palladium, so that in the first flexure of the 
plate the expansion has to compress the rigid palladium, whereas, 
