1913-14.] Principia Atmospherica. 109 
anticyclonic systems, elongated north or south, over the oceans, or a pair 
of cyclonic systems over the continents, of which we can at present only 
determine the southern portions ; or we might arrive at the actual shapes 
by adjustments of both kinds. If we assumed positions for the original 
circular isobars, it would be a simple matter to give numerical values of 
the superposed anticyclones or cyclones. But the circumpolar circular 
isobars are hypothetical, and, at the present stage, the numerical work 
indicated would be unremunerative. Let us assume, however, such an 
initial circumpolar system, and consider the physical forces which would 
disturb its motion. 
The only force immediately at hand is that of gravity, due indirectly 
to the cooling of the surface air on the land and frozen sea in the arctic 
night operating in accordance with Law 3, the law of convection. This 
may produce a real effect of some magnitude on land-slopes. It is not, 
I think, necessarily effective over level surfaces, because there is no slope 
down which the cooled air can flow. 
I have always hesitated about the common explanation of the trade- 
winds and other well-known phenomena based upon the reverse process of 
surface-heating. Surface-heating and surface-cooling necessarily produce 
a certain amount of expansion and contraction, but not necessarily any 
continuous convection current. Convection requires the juxtaposition of 
warm air and cold air, and, if the region is big enough, the result of 
surface-heating may easily give rise to a heated volume of air surrounded 
by isobars and air-currents that prevent any continuous process of general 
convection. Local convection there would be, but that need only extend 
high enough up to take up the day’s heat. All the main air-currents of 
the globe have pressure-distributions to guide them. They cannot usefully 
be called convection currents. 
So, if we had, say, a million square miles of level ice round the pole, 
I cannot see that the cooling of that area need produce any considerable 
effect upon the distribution of pressure ; but if the cooling takes place on 
slopes, we at once get the force of gravity to help, and one can no more 
suppose the downward flow of the air to be stopped than the flow of a 
river to be permanently arrested. Hence there must be in winter a 
continual flow of air off the great land-areas of the Northern hemisphere 
if they have any slope. The air-fall off Greenland, for example, must be 
enormous. Every description by explorers in the Antarctic seems to 
support the suggestion of a great cold-air cascade from the Antarctic 
continent. How much flows, and where it flows to, I cannot say ; ulti- 
mately it must find its way to warmer latitudes by some route or other ; 
