Electric Origin of Rigidity and Consequences. 431 
this formula, since Cly and Cl supply everything to make 
Cl;, and the ) of Au unites with a Z to form a doublet, and 
leaves Au to act as monovalent with one free 
C] c] &. Itis possible, then, that in metallic gold 
bb the atom has two doublets. Similarly. with 
Ba Aland In. These three triad metals fall into 
Au line with the others if their atoms contain one 
5 or two doublets instead of the three assigned 
2 to them in the table. In the case of Sn the 
C tetrad valency of the atom may be due to two 
= and two 9 electrons. This would give the 
atom of metallic tin two doublets and would bring Sn into a 
less exceptional position. One fact is very suggestive of 
truth in our speculations about the valencies of Au, Al, and 
Sn, namely, that it is so difficult to determine their ionic 
velocities that even yet values are not given. The interesting 
chemical inquiries suggested cannot be followed up here. 
We have shown good cause why the four smallest values in 
the last row of our table should not be held to be at variance 
with the principles at the basis of (18 a), but should lead to 
a means of estimating the actual number of doublets in an 
atom of uncombined metal more refined than that of treating 
it as the same as the highest valency in all cases.’ These 
considerations throw doubt on the value for Tl] and Pb. 
The case of Cu is also not quite clear. But in considering 
the table as a whole, it must be remembered that 107/ry ranges 
from 15 for Ag to 1080 for Bi, that v ranges from 1 to 8, 
M/p from 6:7 to 40°4, 10%) from 91 to 830, and T from 234 
to 2350, JT entering also as a square. The approach to 
constancy in these circumstances is such as to show a sub- 
stantial soundness in the train of thought leading to (18 a). 
In the metallic conduction of heat, we shall at present ignore 
that part of it which is identical with the molecular process of 
heat-conduction, as it takes place in electric insulators, and 
investigate only the chief part effected by the electric doublets. 
In electric conduction we regarded the electric gyrostats as 
forced to vibrate in a field of electric force X. But in a mass 
of metal free from electric force and at a uniform tempera- 
ture each doublet is thrown into oscillations of a certain mean 
frequency because of atomic motion. These produce con- 
junctions favourable for the breaking of doublets and the 
forming of new ones. ‘This process causes a movement of 
electrons through the mass of metal which is very similar to 
that of the molecules in a perfect gas. If there is variation 
of temperature in a metal, the motion of the electrons is the 
cause of a flow of heat from the hot parts to the cold. Riecke, 
— 
