214 Rate of Recombination of Ions in Gases. 
with measurements of ionization by Réntgen rays knows, a 
single experiment can very rarely, if ever, be depended upon 
for an accurate absolute determination of a quantity. In 
‘making my determination of a I made a large number of 
experiments, and the value which I gave in my paper was the 
mean value obtained from about nine different experiments 
which gave fairly consistent results. I think the accuracy of 
my method and the absolute value obtained is pretty well 
confirmed by the close agreement of this value with those 
obtained by Prof. Townsend* and M. Langevin+. Prof. 
Townsend obtained the value 3420 e¢ and M. Langevin the 
value 3200 e, while the result of my experiments gave the 
value 3384 e, where e is the charge on an ion. These three 
values were obtained by entirely different methods, and 
considering the experimental difficulties involved, the agree- 
ment is remarkably good; and I think it leaves little room 
for doubt in regard to these being very near the true value 
of a. 
Again, in regard to the dependence of « upon pressure, it 
appears to me that the method which I adopted for comparing 
a at the different pressuresis a far more definite and reliable 
one than that which Mr. Walker proposes. In my com- 
parison of a at any two pressures, as described on pp. 297— 
298, the measurements were made within a comparatively 
short interval of time, and all the conditions were kept as 
nearly constant as possible; while Mr. Walker’s method 
involves the comparison of entirely independent experiments 
which extended over an interval of about ten days, and in 
which no precautions whatever were taken to keep the con- 
ditions constant from day to day. I cannot see why he 
discards my method in preference to the other. 
In preparing his third table of numbers he has again fallen 
into the error of comparing independent experiments which 
have no particular connexion with one another, without having 
all the necessary data at his disposal. Mr. Walker has appa- 
2 
rently ignored the fact that to compare the values of Gare 
must know the capacity of the apparatus unless it remains 
constant throughout. In the experiments an adjustable con- 
denser was used in parallel with the main apparatus; and as 
it was only my purpose to compare the observations for one 
pair of experiments, such as in A, the capacities were not 
known for this lotof experiments. All that concerned me in 
* Phil. Trans. Roy. Soc. A, p. 157 (1899). 
t These présentée a la Faculté des Sciences, p. 151, Paris, 1902, 
