produced by the Explosion of Hydrogen and Oxygen. 479 
the galvanometer the discharge was instantaneous, and any 
leak sufficiently serious to affect the reading of the first throw 
would have given rise to a steady current which could not 
have escaped detection. There was, however, greater possi- 
bility of an error in reading the deflexion, especially when 
it consisted of but few divisions; but Hie observations in 
which there was reason to suspect a greater error than 5 per 
cent. have not been given in the tables. 
In comparing the values of g, allowance must be made for 
the fact that in the first six experiments of Table III., and 
in those of which the mean is given after Table I1., shields 
were placed round the tubes G, G’, L, L', outside the cylinder 
and raised to the same potential with the cylinder, with the 
object of collecting a greater charge. The volume of the 
field of force was thus increased from 356 c.c. to 447 c. Cx 
and consequently the values of ¢, had the shields not been 
present, would have been approximately four-fifths of the 
values actually given. But after making every allowance, 
some other cause must be sought for to explain the large 
discrepancies. 
One fact seems to us to stand out very prominently, 
namely, the number of cases in which the electrical effects 
obtained in an experiment made when the contents of the 
cylinder were probably dry, has very largely exceeded the 
effects given by experiments repeated very soon afterwards 
under conditions as far as possible the same. This may mean 
that the presence of moisture impedes the formation of ions 
very seriously. Thus, in the case of the electrometer, the 
first three experiments give values two to three times as great 
as were obtained afterwards. In Tables II. and III. the 
deflexion first obtained is largely in excess of those which 
immediately follow. In Table ILI. the ninth experiment and 
the two which succeed it were made at the same time, and 
the charge measured in the second experiment with the last 
form of apparatus is notably less than that given by the first. 
On the other hand, prolonged attempts at “drying , short of 
actually taking down the : apparatus, failed to give more than 
a fraction of the maximum numbers obtained, and in some 
cases, notably when much air was present, greater values 
were found, when there was probably a considerable amount 
of moisture, than had been anticipated. Still, on the whole, 
we think that the presence of water-vapour checks the 
formation of ions, that when the cylinder and gases are 
perfectly dry the electricity generated per c.c. exploded at a 
pressure of 80 mm. may be as large as ‘(032 microcoulomh, 
but that when much moisture is ‘present the amount may 
