DISCHARGE OF NEGATIVE ELECTRICITY FROM HOT PLATINUM. 
267 
oxygen dissolves in the platinum at low pressures, while at high pressures—since the 
leak becomes nearly independent of the pressure—there is an indication of chemical 
combination. 
It will be seen that these results are very analogous to the results obtained with 
the negative leak in hydrogen. This increases proportionally to a power of the 
pressure at first, but in the course of time a compound is formed, and the leak is then 
nearly independent of the pressure. The difference between the two cases is that the 
positive leak always gets to a definite value at any pressure if enough time is allowed. 
This means that the compound formed can easily dissociate. The negative leak at 
first, before the stable compound is formed, exhibits hysteresis effects, that is, it lags 
behind changes in the pressure. Richardson finds precisely the same thing with the 
positive leak, and it is almost certainly due in each case to the gas in the wire taking 
time to get into equilibrium with that outside. 
The negative leak is often large on first heating a wire, and falls off with the time. 
The same thing applies to the positive leak, and in each case it is almost certainly due 
to the escape of gas absorbed while the wire was cold. The variation of both leaks 
with the temperature is represented by the formula x = A6x~ Q/20 . 
I think, therefore, that the negative leak in hydrogen bears the same relation to 
the hydrogen that the positive leak in oxygen does to the oxygen. If one of them is 
due to the presence of the gas in the surface layer of the wire, then the other is also. 
The precise manner in which the presence of hydrogen in the surface layer alters the 
negative leak will be considered in Section 9. 
8. The Theory of the Negative Leah. 
Professor 0. W. Richardson (‘ Phil. Trans.,’ A, 343, 1903) has given a very simple 
and elegant theory of the negative leak from hot bodies. According to this theory 
the metal contains electrons which move about freely inside the metal and have a 
velocity distribution like the molecules of a gas. Those electrons which collide with 
the surface with a normal velocity greater than a certain value escape. This theory 
leads to the formula x — A 0 i e~ Q/29 , and according to it A is proportional to the number 
of free electrons per cubic centimetre in the metal. According to this theory A ought 
to be nearly independent of the pressure and nature of the gas present, because the 
electrical conductivity of the metal is very little affected, even by hydrogen. 
In my previous paper I showed that hydrogen changed the value of A by a very 
large factor, and therefore I suggested that Richardson’s theory required modification 
in order to enable it to account for the facts. 
It is shown in Section 1 of the present paper that if the true values of Q and A are 
denoted by R and D, and if we suppose that R and D may be functions of the 
temperature 6 as well as of the pressure, then the variations of Q and A can be 
explained either by supposing that 1) is constant and R a function of 6 and p, or that 
2 M 2 
