82 Transactions of the South African Philosophical Society. 
irregularities in the totals at the foot of the columns, each number 
has been reduced to proportional parts of a thousand in Table 8. 
Here we elicit the remarkable fact, to be further touched upon when 
the velocities come to be considered, that the highest percentages at 
the times of maximum frequency occur in the case of those winds 
whose maxima are at, or near, noon. Indeed only these run into 
three figures. It is partly a result of this that the lowest percentages 
at the times of minimum frequency occur at midday to those winds 
whose maxima are at or near midnight. There is, however, a con- 
siderable contrast between the curves for the two classes : the former 
having a long-drawn-out minimum with a rapid rise to and fall from 
maximum ; the latter an equally long-drawn-out maximum with a 
rapid fall to and rise from minimum. 
The hourly components of wind-frequency are shown in Table 9. 
They are calculated from Table 7 in the same way as Table 2 is 
calculated from Table 1. Their interpretation may be best illustrated 
by an example: For the hour ending at noon the components in 
Table 9 are: N = +344:2; KE = —378:1. This means that for 
that hour during three years the northerly components exceeded 
the southerly by 344:2 hours, and the westerly exceeded the easterly 
by 378-1 hours. The results are shown graphically in Figs. 3 and 4. 
In Fig. 3 the heavy line is the north component curve ; the other 
is the east component curve. The portions of both curves are 
conventionally positive above the zero line (upon which the hours 
are marked), and negative below. Thus for all portions of the curves 
above the line the directions prevailing are northerly and easterly, 
but southerly and westerly for all portions below. We see, according 
to the figure, that the east component reaches its greatest positive 
and negative excursion at the respective hours ending 6 a.m, and 
3 p.m.—say at mean times 5.30 a.m. and 2.30 p.m.—these hours, as 
it happens, being the times of the mean minimum and maximum 
temperatures of the day. In other words, the vane is swinging from 
east to west with a rising temperature, and vice versd. The dotted 
line in the figure is the diurnal temperature variation verted. That 
is, it represents the consecutive hourly values of M—t where M is 
the mean daily temperature, and t the temperature at any hour. By 
suitably choosing the scales of hours and degrees, the curves can be 
made to look very much alike. That the east and west swing of the 
vane is a direct effect of the temperature is scarcely to be doubted. 
The barometric curve (see Table 12 and Fig. 7) follows the east 
component curve by some two hours so far as the portion included 
in the daylight hours is. concerned, but the night maximum of the 
former has no corresponding feature on the other. The north com- 
