110 Proceedings of the Boyal Society of Edinburgh. [Sess. 
but these air-flows must be a real cause of alteration in the distribution of 
pressure, and it is to the land-slopes which are losing heat that we may 
trace an indubitable influence, and therefore a disturbance of the uniformity 
of circulation. Apart from compensation, a flow-off of 1 metre thickness 
of air would mean a reduction of pressure by 0*1 millibar.* 
Similar phenomena must of course happen locally, and they are well 
known in mountainous regions, though we can hardly expect the smaller 
local examples to show much effect in the distribution of pressure over 
the globe. 
But we may assume that cold land-slopes in winter are the cause of a 
constant abstraction of air from the lowest layers of the atmosphere in 
those regions. The cold air flows away by gravity, and since the surface 
pressure is apparently still maintained, the efforts to redress the loss of air 
have to be carried out in the upper atmosphere and in accordance with its 
laws ; consequently we should expect to find a cyclonic circulation in the 
level in which the replacement is taking place. The cyclonic circulation 
may operate to prevent the pressure being made up overhead, but it cannot 
prevent the cold air from flowing downhill unless the reduction of pressure 
is enough to reduce the density by as much as the low temperature 
increases it, and this is a difficult task, for near sea-level it takes more 
than 3 millibars loss of pressure to make up for a single degree loss of 
temperature. 
Hence we may suppose that the constant drainage of the land-areas 
would result in the superposition of a cyclonic distribution at high level 
over them, and the continental lobes of Teisserenc de Bort’s isobars for the 
upper air may well be due to this cause. 
But the .cause is obviously a very variable one, depending upon the 
distribution of cloud and other circumstances. Statistically, its effect upon 
the circulation of the upper air is to exaggerate the pressure gradient for 
westerly winds over the temperate zones of the continents, and to diminish 
the gradient northward. Thereby we introduce into the circulation local 
accentuation of current, which must be disposed of by some dynamical 
process. 
* The facts which are here represented are sometimes taken as indicating the formation 
of anticyclones over the Arctic and Antarctic land-areas. When those areas are represented 
by plateaus 10,000 or 15,000 feet in height, the surface anticyclone may become merely a 
hypothetical construction supposed to occupy the space which is really occupied by land 
and not by air at all. To a considerable extent the great Asiatic and American anticyclones 
depend upon the reduction of observations to sea-level under conditions which can have no 
real existence. The mountain slope might possibly operate, in the maintenance of a 
cyclonic circulation in the upper air, much like the hole in the bottom of a basin, and the 
actual land-surface at the high level might therefore be a region of cyclonic circulation. 
