22 Report S.A.A. Advancement of Science. 



causation into a morning and evening tendency of the wind (on 

 a long average of observation), to draw towards the time of 

 sunrise and sunset, to compensate the overflow from off the 

 heated hemisphere which takes place aloft in a contrary 

 direction." 



I have quoted these passages in full, because they stand for 

 what, 40 years ago, was regarded as proved. Two assumptions — 

 which, however, were not then thought to be assumptions — underlay 

 the theory : one, that the air expanded by the heat of the sun flows 

 upwards and outwards from the light hemisphere to the dark ; 

 the other, that what is called the "pressure of dry air" can be 

 obtained by subtracting the vapour pressure, as determined by the 

 hygrometer, from the total barometric pressure. 



Now the diurnal variation of wind-direction is not quite in 

 accordance with Sir John Herschel's idea. Taking Kimberley as an 

 example, since we have here no very decisive prevailing wind- 

 direction, we ought to be in a very good position to detect any drawing 

 of the wind if it exist. But what really happens is this : Though 

 there is no decisive prevailing direction -when the whole year is 

 considered, there is some tendency to a seasonal prevailing direction, 

 which, in the course of the year, backs completely round the compass. 

 Thus in the spring the prevailing direction is S.W., in the summer 

 nearly N., and in the winter about S.E. But superimposed on this 

 is a strong diurnal rotational veering of the vane evident at all 

 seasons. (Fig. 5.) In this diurnal scheme the normal direction is 

 N.E. at sunrise, N.W. at noon, S.W. at sunset, and S.E. at 

 midnight. Now, all over the world this same diurnal rotation may 

 be wdth more or less trouble separated from the prevailing directions. 

 It has been demonstrated for several stations in India. I have 

 shewn that it is quite obvious, even at East London, where the 

 prevailing directions (with high velocities) are almost exclusively N.E. 

 or S.W. — i.e., up or down the coast. (Fig. 6.) If we separate this 

 rotational veering into its mechanical components at right-angles N. 

 and E., we find that the E. component varies directly with the 

 temperature of the air, and therefore indicates some such relation to 

 the prevailing direction, as the diurnal temperature variation does to 

 the annual. Also, if we combine velocity and direction, there is some 

 sort of obscure agreem.ent between the E. component wind movement 

 and the barometric phases, which is chiefly noticeable in the second 

 of the periodic terms in the sine series representing each. More than 

 this cannot be said. Now this rotational veering of the vane implies, 

 as F. Chambers has pointed out, not a movement inwards of the air 

 to replace that which is supposed to have flown off above, but an 

 outward movement, that is, not a convection current, but an anti- 

 convection current. Also the diurnal velocity of the wind is not greatest 

 at 4 p.m., when the barometer is lowest, as it perhaps might be if 

 there were a partial vacuum to fill, but is greatest some hours earlier 

 — earlier, indeed, in some months than the epochs of maximum 

 air-temperature. It cannot be said, therefore, that the supposition 



