508 Proceedings of the Royal Society 
By this apparatus the small deflection of the palladium plate is 
greatly magnified, and the direction of motion well defined. Sup- 
pose the palladium plate, A in the figure, is connected with the 
negative pole of the battery, the glass index, after a short time, 
begins to move from left to right on the plane of the diagram, to 
the extent of 8 or 10 centimetres on the scale. As the saturation 
goes on, the index begins to move backwards from its first position, 
going towards the left even to a greater extent than its first de- 
flection towards the right. Continuing the action, it again returns 
to near its original position. The power of being able to return to 
the position it had at starting seems to depend on the position of 
the plate, as regards the distribution of tensile strain, produced by 
rolling ; at least, after repeated use the plate lost the power of re- 
turning, after having passed towards the left. If the plate, after 
saturation, is connected with the positive pole of the battery, the 
first effect on the index is to move quickly towards the left, then 
to return to where it was — this double motion taking place before 
any gas makes its appearance on the palladium. If the platinum 
electrode B is placed on the opposite side of A, and the saturation 
of the plate repeated, the index goes through the same series of 
positions, but the direction of motion is reversed. The direction of 
motion depends, therefore, on the relative positions of the electrodes, 
but is constant for the same position. This is easily shown by 
allowing the index to commence its motion, say from left to right; 
then, by moving the positive electrode to the other side of the 
palladium plate, the motion immediately commences in the oppo- 
site direction, although the saturation was far from being complete. 
The motion of the index, when oxygen is thrown on the hydro- 
genised plate, depends also on the position of the electrodes. The 
index has also a motion at right angles to the plane of the scale, 
the resultant motion being compounded of the separate flexures of 
the plate. Many other devices could be used to show the motion, 
such as a plate bent into the form of a cylinder with a narrow channel 
left between the two edges, which should shut and open alternately, 
according as the positive electrode is without or within the 
cylinder. 
Graham has shown that the formation of the alloy of liydro- 
genium and palladium is attended with an enormous increase in 
