82 
Proceedings of the Royal Society of Edinburgh. [Sess. 
heavier air is not far to seek, it is not so easily identified on the scale of a 
meteorological chart. 
Convection in the atmosphere may also be due to the variation in 
the gravitational acceleration due to the motion of the air with reference 
to the earth. 
The gravitational acceleration depends partly on the statical attraction 
of the earth’s mass and partly on the centrifugal action due to rotation. 
The ordinary values of the constant of gravitation assume the rotation to 
be that of the solid earth, and the acceleration of gravity upon air moving 
over the earth’s surface is consequently different from that for calm air. 
Hence the air which forms part of a westerly wind is specifically lighter 
than air at the same temperature and pressure which is calm ; and, on the 
other h^tnd, air which forms part of an easterly wind is specifically heavier. 
These variations in what, contrary to the usual convention, may rightly be 
called the “ specific gravity of the air ” have not yet been generally taken 
into account in meteorological practice, but they are of real significance, 
and are the subject of certain classical papers by von Helmholtz and 
Brouillin on the circulation of the atmosphere. 
4. The Law of the Limit of Convection. 
Convection in the atmosphere is limited to that portion of it in which 
there exists a sensible fall of temperature with height. 
This portion, which comprises about three-fourths of the atmosphere, is 
called the troposphere , and is a layer of air about 10 kilometres thick 
surrounding the whole earth. It is surrounded by an outer spheroid of 
air comprising the remaining fourth part of the atmosphere, which is 
called the stratosphere , in which there is no sensible fall of temperature 
with height. The boundary between these two layers is not at a fixed 
height ; it is apparently a flexible, and therefore deformable, surface, but it 
is not penetrable by air. 
The height of the boundary differs in different latitudes, being highest 
over the equator and getting gradually lower towards the poles ; it differs 
also in different localities, being higher over an area of high pressure than 
over one of low pressure. The local differences are due to deformations of 
the boundary by the accumulation or withdrawal of air from underneath. 
At any place the boundary oscillates about a mean position which should 
be regarded as the height of the boundary of the stratosphere for the 
place. There is no physical reason why the boundary of the stratosphere 
should not be penetrated. All that is required to produce that effect is an 
accumulation of air warm enough to cause upward convection. All that 
