Recombination of Ions in Gases. 213 
as proposed, is correct, the values of a, whether obtained from 
(3) or from the relation between N; and N, must be the same 
and quite independent of the intensity of the rays, provided 
that the intensity is the same for any one determination of N 
and Q;” and, he might have added, of N; and N. I admit 
that the determination of 2 ought not to be affected by the 
intensity of the rays so long as the intensity remains constant 
during the measurement of ali the quantities involved; but 
suppose the intensity of the rays when determining N; is 
different from the intensity when determining N, then the 
variation certainly would affect the result. Formule (4) and 
(5), which Mr. Walker used in determining an and aq, involve 
the quantity ) which depends upon ¢, which in turn depends 
upon N;. Now suppose that at any point in the experiments 
necessary for one of these curves, which involved a lon 
series of readings, the intensity of the rays should slightly 
increase and maintain this increased intensity for the re- 
mainder of the curve, then the curve would become slightly 
too flat, and the value of ¢ obtained from it for any given 
value of N; might be, especially at the lower pressures, 
considerably too large. The value of A thus obtained would 
be smaller than the true value of X corresponding to N. This 
would of course cause a much greater error in formula (5), 
since X appears in the second power instead of the first. This 
is quite apparent in the values obtained by Mr. Walker, for 
the values of ae are all still farther from the true value than 
those for an. 
These two methods of determining « also involve the 
absolute measurement of a very short timet with considerable 
accuracy, to do which elaborate precautions are necessary. In 
my experiments I do not claim to have attained this necessary 
accuracy in the measurement of the time intervals. The 
relative value was of more importance than the absolute 
value. 
Besides this, since these methods of determining ay and 
aq involve N,, it means the measurement of much smaller 
effects than in the case of Nand Q. One of the great diffi- 
culties of the experiment was to obtain an apparatus to give 
large effects for measurement so that the results would be 
more reliable. The results from the formula which I used 
involving Q and N are, since N and Q are large, much more 
to be relied upon than those which depend upon the deter- 
mination of smaller quantities. 
Besides these facts, Mr. Walker has made his determinations 
of a by all three formule in every case from single isolated 
experiments. As any one who has ever had any experience 
