Electric Origin of Rigidity and Consequences. 429 
But M’/ is proportional to e*s?/K, and in the metals we have 
found s=(m/p)?. If, then, we take Kv(m/p)*, that is Kva, 
to be a constant 8%, we eliminate K from (17) thus 
y= GRbT?/47OavI? 2 ee (18) 
As I the inertia of an electron has been assumed constant, 
and L/y is the electrical specific resistance at @, this can be 
written 
(=) wt®/@~ y=constant ye ae Ee ote 
y\p by P 3 
This is the relation discovered by W. Williams (Phil. Mag. 
[6] ui. 1902). It must be noted that several metals like 
Ag, Cu, Tl, and Pb, which have ? replaced by a simple 
multiple of @?, show no appearance of the effect of the 
simple multiplier in the formula of Williams, while some 
metals are exceptions to this rule. It is possible that for 
these two classes of exceptional metals I cannot be assumed 
to be constant, but that the ratio of the simple multiplier of 
8? to I is constant or a simple number. On account of the 
importance of (18a) I shall reproduce the data gathered by 
Williams with the addition of a few, to show how nearly the 
left side of (18a) is the same for most of the metals. That 
equation makes resistance proportional toabsolute temperature, 
which is the well-known approximate truth pointed out by 
Clausius. Our comparison will be restricted to the case 
where 0=273, unless it is otherwise stated. The resistance 
1/y is given in ohms for a cm.’ For v the highest known 
valency is given. 
Na. K. Cu. Ag. Au. Mg. Ca, 
> :..... 51 84 17 5 20 41 15 
10M /p ... 237 454 71 102 101 139 254 
BG 5325. 720 830 170 194 147 270 (279)* 
Es Be a: 369 33D 1333 1175 1310 1023 853? 
1 a Oe 1 1 2 1 3 2 2 
He 22, 25 ol 13 37 52 
Zn. Cd. Hg. Al. in, TT. Sn. 
Myo, 58 7d 210 29 84 176 105 
10M/p ... 91 129 141 106 158 172 163 
BOO. C222. 298 316 (778)* ° 222 417 302 230 
We Fearon 76 593 254 923 449 561 5038 
Aan 2 2 2 3 3 5) 4 
39 28 34 14 14 31 9 
* Calculated by empirical formula vIM* =-044. 
