80 Transactions of the South African Philosophical Socvety. 
revealing the prevailing directions of the wind.. Such a conclusion 
can only be drawn when the tabular numbers from which they are 
derived show also a distinct and overwhelming majority for one 
direction over each of the others. Should there be, say, two 
directions, each containing preponderating numbers, then the 
resultant will lie between them, and may (almost certainly will, in 
the case of valley winds, or of land- and sea-breezes) lie along a 
direction from which relatively few of the winds come. Referring 
again to Table 1, it will be seen that although certain directions do 
prevail at the expense of the others month by month in a moderate 
degree, yet any preponderance for the whole year is comparatively 
insignificant. Month by month, then, we may take it as a rule that 
the prevailing direction is approximately that of the resultant, and 
such that (see Fig. 1) it blows from the denomination of the month 
to the origin of the figure with a frequency and proximity propor- 
tional to the space described, whereas for the year it is not so. 
Table 3 gives Table 1 in a condensed form, the wind numbers 
being arranged in quadrants. It both smooths and emphasises the. 
monthly variations of Table 1. Apparently, however, Table 4, in 
which the numbers of Table 1 are arranged in periods of four months 
—a, condensation suggested by Fig. 1—is more advantageous. From 
this we conclude that the prevailing directions are north-north- 
westerly from January to April, easterly from May to August, . 
south-westerly from September to December. Fig. 2 is a graphical — 
representation of this feature, and Table 5 the components. It 
appears to be extremely likely that if sufficient observations can be 
taken it will be found that the wind in the course of the year backs 
with uniform angular velocity clockwise round the compass. 
Of the 25,898 hours of wind analysed throughout the three years 
the final resultant contains the small components of only 50 hours to 
the north, and 100 hours to the west, being exceeded in magnitude 
by those of every separate month with the exception of July. The 
resultant of each of the three years, also, is much greater than that 
of the final resultant of the three together; moreover, neither bears 
the least resemblance to either of the others. They are :— 
March—Dec., 1896, and | N. E. 
Jan, Hepealego) wiaenene ( +255°4—1273°3 
TBST DSRS nah ice Cnt ame —500°3— 244°6 
UB9S tps beens eae + 995:24- 1417-7 
AMS ai in bu. sis atei be eee + 168— 33:4 
* January and February, 1899, are added, to complete a year, to the ten 
months March—December, 1896, and so as to avoid breaking up the full years 
1897, 1898. 
