118 
HAROLD’S DISCUSSIONS. 
This increases its density on that side and diminishes 
it toward the left. Fig. 59 will make this plain. Let 
us suppose that there is a low barometer at T and a 
high at S. On account of the difference of pres¬ 
sure, then, the currents would 
move from S to but on ac¬ 
count of the rotation of the 
earth they would take the 
curved path, S G E. Now 
the crowding effect on the 
right would prevent the cur¬ 
rent from taking this path. 
Its course therefore would 
lie somewhere between S T 
and the curve N E^ as, for 
instance, 8 C. 
Thus the combined effect of the rotary principle 
and of the crowding gives the trades in the northern 
hemisphere a southwest motion, instead of either a 
southern or a circular direction, while the prevailing 
winds north of the horse latitudes continue in an east¬ 
erly or northeasterly direction—that is, they are all 
generally west or southwest winds. 
THE UPPER CURRENTS. 
Now we come to consider the upper currents. 
As shown by the clouds and the smoke from volca¬ 
noes, the air that rises at the equator does not flow 
off directly toward the poles, but turns eastward, so 
that its course is even more oblique than the trade- 
winds below, as shown in Fig. 60. When these up- 
