438 Scientific Proceedings, Royal Dublin Society. 
of ground air (rich in carbonic anhydride) from the soil, owing to 
the lowering of temperature. 
At sea no such influence can be exerted, but an absorption of 
atmospheric carbonic anhydride may occur at the surface of the 
water, owing to lowering of temperature, thus reversing the land 
effect. This question has received scarcely any attention. 
(3) The effects of Atmospheric Precipitates, and especially of 
snow, which appears to increase the amount of atmospheric carbonic 
anhydride. No reasonable theory has been proposed to account 
tor this curious phenomenon, andit would be interesting to ascertain 
whether it occurs at sea as well as on land, and the same remark 
would apply to fog and rain, both of which appear to affect the 
amount also. 
Other supposed causes of variation are worth studying, such as 
the effects of the seasons, direction and force of the winds, the pre- 
vailing type of weather, height above the Harth’s surface, &c., 
but those which we think most interesting are such as a scien- 
tific mission would be under peculiarly favourable conditions, to 
investigate, and especially the proposed Antarctic expeditions— 
English and German—whose work will be performed largely at 
sea. 
It is the investigation of the carbonic anhydride over the ocean 
which, in our opinion, presents so many problems of interest, and 
which, up to the present time, has received so little attention. 
In the memoir we have referred to, we have also described a 
method for determining atmospheric carbonic anhydride—a 
modification of Pettenkofer’s process—by means of which results 
of great accuracy may be obtained. Thus in the final set of test 
experiments with artificial mixtures of purified air and carbonic 
anhydride, a mean error (in six determinations) of about 1 per 
cent. of the gas was found corresponding with some 4 parts per 
million of the air taken. 
For certain purposes—and especially for employment by a 
scientific expedition—it seemed to us, however, that a different 
process is required, in which the operations at the place of obser- 
 yation should be simple, and of such a nature as to permit the 
actual determinations to be made later when the resources of a 
properly ejuipped laboratory are available. The method employed 
