102 
SIR B. C. BRODIE ON THE ACTION OF ELECTRICITY ON OASES. 
I. 
Unabsorbed gas. 
II. 
T. 
III. 
C. 
IV. 
B=® 
T 
V. 
Ozone per cent. 
22*6 
2-06 
7-63 
3-7 
81-3 
32-73 
3-01 
10-63 
3-53 
85-5 
15-92 
1-46 
5-48 
3*75 
79-8 
31-42 
3-0 
10-78 
3-59 
83-5 
28-14 
2-55 
9-76 
3-82 
78-3 
11-03 
1-13 
4-32 
3-82 
78-3 
16-57 
1-58 
6-19 
3-91 
76-6 
9*43 
0-97 
3-46 
3-57 
84-1 
10-51 
1-09 
3-88 
3-56 
84-3 
The maximum result attained in these experiments is that in which 85 ‘5 per cent. 
c 
of the matter of oxygen is converted into ozone, in which case the value of the ratio ^ 
is 3 - 53. The value of this ratio, 3 - 5, corresponds to a gas in which 85*7 per cent, of 
that matter has been thus converted, and which is constituted of ozone and oxygen in the 
proportion of four units of the former to one of the latter gas, thus 4| 3 +i 2 . The numbers, 
it is to be observed, given in the last column are undoubtedly somewhat too low, from 
the circumstance, already mentioned, that the unabsorbed gas contains a minute quantity 
of hydrogen*, and also possibly from the removal of traces of iodine from the absorption- 
bulb by the rapid current of carbonic acid, so that the proportion of ozone found is 
certainly not less than is there indicated. 
These experiments, taken in connexion with those described in my previous investi- 
gation, leave no room for doubt as to the true nature of the unit of ozone, the compo- 
sition of which must henceforth be regarded as established on evidence hardly less 
conclusive than that on which our knowledge rests of the composition of the unit of 
water. Nevertheless it is not to be anticipated that this result will be equally clear to 
all minds. As in the analogous case of the dual nature of oxygen, it is not a point to 
be demonstrated by any single experiment, but is a conclusion derived from various 
trains of reasoning, each of which has to be separately mastered and appreciated. 
In interpreting the experiments on which this conclusion is based, I have made frequent 
use of the well-known method of the science of probabilities, the method of least squares, 
which (so far as I am aware) has not hitherto found any serious application in chemistry ; 
but in such cases as the present, where the experiments are complicated and where an 
exact result can only be obtained by the successful performance of numerous operations, 
each of which is liable to error, this method is of essential service. We cannot, indeed, 
get rid of accidental errors from our experiments’; but instead we eliminate from 
our conclusion the result of those errors considered individually. It has, moreover, the 
very great advantage of enabling us to estimate the numerical result of our experiments 
at neither more nor less than its true value ; and removing that value from the uncer- 
tainties incidental to the appraisement of individuals, assigns to it its true position 
according to an external standard. 
* Conf. note, p. 92. 
