84 Transactions of the South African Philosophical Society. 
hour-lines of midnight, 6 a.m., noon, and 6 p.m. as the points .of- 
departure and arrival, the total angular deviation of the pencil:.in: 
each interval being entered in its proper column. ‘The results show-' 
ing the mean movement for each day, and each quarter of a day,. 
during each month, are shown in Table 10. If it be distinctly borne 
in mind that an instantaneous reading of wind direction. cannot. be: 
trusted to give information equally trustworthy with that of a mean- 
hourly reading, the agreement between Table 10 and the more formal 
values of Table 9 will be sufficiently satisfactory. It is evident that: 
the angular velocity of the vane is much greater in summer than: in 
winter, the diurnal mean for the year being rather less than one-third 
that of the sun. The remaining two-thirds, due to perturbations—; 
¢é.g., storms—and partly no doubt to instrumental faults, have no: 
appreciable effect upon the velocity of the resultant, although they 
must considerably reduce its length throughout the day. Instances 
in which the vane travels backwards are probably more common: 
after noon than before. They seem to occur either when the vane 
has advanced more rapidly than usual for a few hours previously, or: 
when, upon the dying away of some disturbance deflecting the air 
current, the vane takes the shortest cut to its normal position. That 
there is not, however, any material difference between summer and 
winter in the diurnal frequency-curves of the different wind directions 
is demonstrated by Table 11, in which the numbers are ratios per. 
thousand arranged in quadrants; the months of December and: 
January for three years providing the summer values, June and July 
thé winter. The greatest differences are in Quadrant 4 (including all, 
winds from west, west-north-west, north-west, and north-north-west) 
the summer curve here being appreciably flatter. Any other months: 
give just the same sort of information. Perturbations are seldom of 
long duration, the longest ‘‘ set-in”? wind on record not perhaps 
lasting three days; and even when they occur their effect is con- 
fined almost exclusively to the addition of a practically constant 
number to each hour for the set-in direction. Whence it is that the 
frequency-curves retain their shape unaltered. : 
WInD VELOCITIES. 
Hitherto we have dealt only with the simple directions of the 
wind irrespective of the corresponding velocities. It is usually the 
custom to discuss the two together, ignoring the directions as oh 
separate factor, especially in dealing with the rectangular com- 
ponents, although it is not easy to see why. Table 12 will demon- 
strate that there is some advantage in making a distinction between. 
the two. It is constructed from two years’ observation of -velocity 
