66 
Proceedings of the Royal Society of Edinburgh. [Sess. 
observer on the earth it will, therefore, appear as a whole to revolve rela- 
tively to the earth in an opposite direction to that of the earth’s rotation, 
i.e. to circulate anticyclonically. Conversely, the contracting air over C, as 
it yields to the centripetal pressure, will commence to circulate cyclonically. 
It should be observed that the energy of the winds is, therefore, not derived 
exclusively from the solar heat, but in part from the earth’s rotation. In 
every wind system part of the terrestrial kinetic energy of rotation is being 
converted into heat. 
In general, however, the motion of the air will not be uniform. If the 
existing currents are readily deviated into the required type of circulation, 
their inertia will intensify the anticyclonic or cyclonic effect. If, however, 
they are opposed to the motion induced by the thermal changes (i.e. if the 
area A was originally a region of barometric depression or B a region of 
high pressure), the pressure which every horizontal wind exerts to the 
right in the northern hemisphere (or to the left in the southern hemisphere) 
will cause the rapid extinction (or even prevent the formation) of the 
barometric gradients due to the temperature changes, and the result will be 
merely a diminution in intensity of the depression originally existing over 
A, or the anticyclone over C. 
The circulatory air movements tend (in virtue of the principle of side 
pressure quoted above) to maintain the existing inequality of pressure and 
delay the entrance of air into the region over C, or its exit from the region 
over A. They would not of themselves suffice to prevent or reverse the 
effect were it not for the inequalities of density, which are not instan- 
taneously destroyed by the process of mingling. These inequalities of 
density give rise to a further set of effects arising from the principle that 
in any fluid in motion denser portions tend to move towards regions of 
higher pressure, and rarer portions towards regions of lower pressure. As 
this generalisation of the principle of the centrifuge is apparently new, 
I attach a formal proof. 
Let A, L, M (fig. 3) represent three consecutive portions of a fluid in 
motion, which pass the point P with the same velocity. A, L, M are taken 
as spherical portions, all of the same diameter but of different densities — A 
being of average density, L of less than average density, and M of more 
than average density. They are supposed to be so near that the pressure 
conditions do not alter materially between the times of their successively 
reaching P. As A passes P it is deflected from its undisturbed course by 
the pressure of the surrounding fluid and follows path PQ, which we may 
call the “ average path.” L, reaching P with the same velocity, and being 
subject to the same pressure-forces (since it is a sphere equal in volume to A), 
