44 Proceedings of the Euycd Society 
manner as the. former ones, the following results become apparent : 
— (1) The manometric effect is very much greater with, the new than 
with the old apparatus. (2) The curves for any one gas, or mixture 
of gases, do not coincide with one another. Thus corresponding 
ordinates of the two curves for air containing IT per cent, of olefiant 
gas differ by about 4 or 5 per cent., or, as it may be put, one curve 
is 4 or 5 per cent, lower than the other. Of the two curves for air 
with 0*2 per cent, of olefiant gas, one is. 2 to 4 per cent, lower than 
the other. Of the two for air with 0*06 per cent, of the same gas, 
one is from 6 to 10 per cent, lower than the other. Hence the 
apparatus used was incapable of giving perfectly accurate results. 
As we have pointed out above, the reservoir was not sufficiently 
rigid. With a rigid reservoir the curves for any one gas, or gaseous 
mixture, would coincide at any rate in those portions corresponding 
to considerable intervals of time ; but the inevitable difference in 
the initial conditions of the experiments must in general prevent 
entire coincidence in portions corresponding to short intervals of 
time. (3) The dry air curves are with this apparatus the lowest 
on the diagram. To us this result was unexpected ; but Professor 
Tait had foreseen it as a consequence of the change of dimensions of 
the apparatus. (4) As in the preliminary experiments, the curves 
for air containing olefiant gas and for air containing water vapour 
occupy the same region of the diagram. The relative positions of 
the curves may be used to determine the relative absorptive powers 
of the gaseous mixtures. (5) The curves for air containing dif- 
ferent quantities of olefiant gas occupy positions extending over a 
considerable area of the diagram, and they are higher the greater 
the percentage of olefiant gas (from *032 up to IT). By deter- 
mining a sufficient number of such curves a scale of absorptive 
power may be formed, to which the absorption of air containing 
known quantities of vapour may be referred. With our apparatus 
it was impossible to form such a scale accurately for the reasons 
given above. But an approximate result is given by our experi- 
ments. The curve for air with 1 *3 per cent, of water vapour (not 
saturated) has above it the two curves for air with 0’2 per cent, of 
olefiant gas, and below it the two curves for air with 0*06 per cent, 
of olefiant gas. Hence we may conclude that the absorption of air 
containing 1 -3 per cent, of water vapour is between that of air con^ 
