180 
bility has been already suggested by Rutherford ("Radio- 
activity," p. 33). All electron, which has just been ejected 
from an atom by a passing a jjarticle, does not go far before 
encountering a neighbouring atom- Tlie encounter, perhaps, 
results in a temporary attachment, for we know that ion- 
clusters are formed in this way. In any case, it is probable 
that the electron loses much of its velocity of projection. Now, 
it is still under the attraction of the atom from which it has 
come. Supposing this atom to have only lost one electron, the 
strength of the electric force which it exerts at the distance 
of the mean free path is e(iual to e/r- = 3 x 10-^710-^°= 3 E.S.U,, 
or 900 volts per cm. This is large compared to the usual im- 
pressed electric forces of experiment. It is by no means im- 
probable, therefore, that the electron may finally slip back 
into its old place- Such a possibility is not considered in the 
equations as usually formed. For all writers begin their 
arguments by the statement : — "Let p be the number of posi- 
tive ions in a cubic centimetre, and n the number of nega- 
tive-'' In doing so they tacitly assume that the relations of 
any one ion to all others of opposite sign are of the same 
character. But if a pair of newly-formed ions ran a special 
chance of recombination until they got away from each 
other, then the relations of either of these two to the other 
would be quite different from its relation to all other ions. 
Let us, then, for the moment suppose that there is a spe- 
cial form of recombination, which we may call 'initial," as 
distinguished from general recombination, and let us con- 
sider the nature of its effects, in order that we may find 
means of testing the correctness of the supposition. 
Now, it is clear that the effects of initial recombination 
do not depend upon the shape of the ionisation chamber, and 
this at once differentiates them from those of general recom- 
bination. For the special or initial recombination concerns 
only the ion and its parent atom. But general recombination 
depends on the chance of an ion meeting others of the oj^po- 
site sign, which chance depends on the number in a c.cm-, and 
this, again, on the shape of the chamber. If, for example, 
a particles cross a chamber 3 mm. wide, and a sufficient 
potential gradient is applied, most of the ions will be car- 
ried to the electrodes. If the width of the chamber is in- 
creased to 6 mm. the magnitude of the stream of ions is 
doubled, each positive meets twice as many negatives as be- 
fore, and therefore the chance that any one ion enters into 
recombination is twice as p^reat. Suppose the saturation cur- 
rent for a 3 mm. chamber were 100, using any arbitrary sys- 
tem of units, and the actual current for a moderate potential 
were 90, then for the 6 mm. chamber, under an equal poten- 
tial eradient, the current would be 160, not 180: the satura- 
