108 Proceedings of the Royal Society of Edinburgh. [Sess. 
that it becomes “ too cold ” by piling np under the stratosphere and readjust- 
ment of the several layers within the stratosphere, so that pressure on the 
sample which causes the bulging is reduced, while that over the surrounding 
regions is increased.* Radiation is left out of account — whether rightly or 
wrongly, it is not possible at this stage to say. 
The motion of the critical layer is on the average from west to east, but 
not invariably so, and apparently the temperature-relations which have 
been described are not dependent upon wind direction. Other phenomena, 
so far as they have been observed, seem to indicate a similar symmetry, 
but there is no sufficient evidence for supposing that the phenomena are 
necessarily centred locally. In fact, according to the distribution of isobars 
at 4 kilometres computed by Teisserenc de Bort (Lemma II.), the average 
motion does not differ much from a circulation round the pole which, once 
set up, might be persistent with little change if it was everywhere 
adjusted to the barometric gradient. The actual motion, however, certainly 
does change, and is, in fact, constantly changing. 
Let us consider the conditions of Teisserenc de Bort’s average isobars 
and the forces which are available to produce the perturbations of a 
supposed original circumpolar circulation indicated thereby. I have 
already remarked that, for such a circulation as that represented by 
Teisserenc de Bort, the isobars for 4 kilometres may fairly be accepted as 
applicable at 7 kilometres also, because the changes of pressure-difference 
between 4 kilometres and 7 kilometres are in ordinary circumstances very 
slight. 
Taking the average map for January, it will be noticed that the 
isobars at 4 kilometres are clearly not circles round the pole. If they 
were so, a steady circulation would be a natural conclusion. It has been 
already postulated in Lemma II. that they are in reality indented ovals or 
approximate figures-of -eight with the lobes over the Asiatic and American 
continents and the inward bends over the two oceans. I purpose consider- 
ing first the effect of convection as a possible cause of the deviation from 
the circular shape. The shape which we have to explain is exactly 
opposite of that which is often shown on synchronous charts of the 
distribution of pressure at the surface of the Northern hemisphere in 
winter, and which has “ highs ” over the continents and “ lows ” over the 
oceans. I remark in the first place that, to derive the figure-of-eight shape 
from the circular shape, one cannot rely simply upon the nutation of a 
west-to-east circulation round the pole ; one must superpose either a pair of 
* See a note on the Perturbations of the Stratosphere in Publication 202 of the 
Meteorological Office. 
