Shift of Trade Winds — WENTWORTH 
87 
ticns, of which northeast is one. The total of 
wind in each sector for each month is tabulated 
both by miles and by hours, and is summarized 
in the same way for each calendar year. 
This study is based on the number of hours 
shown in the two sectors east and northeast, 
since these two sectors, or one fourth of the 
whole wind rose, have ar counted, over the 40- 
year period, for an average of about 81 per 
cent of the total time. It is evident that there 
is a strong central tendency in these two sec- 
tors, and the shift of trade-wind direction in 
degrees has been deduced by comparing the 
changing proportions carried in the two sectors. 
For example, in the period of 2 or 3 years 
around 1908, the percentage of northeast wind 
by hours was about 56, while that of east wind 
was only about 24. On the other hand, by about 
1935 and for 3 or 4 years before and after, the 
percentage of northeast wind was only about 
22, and that of east wind had increased to 57. 
The annual percentages are shown graphically 
in Figure 2. 
The annual data are somewhat ragged, sug- 
gesting that a number of minor factors are in- 
volved, but the general shift is nevertheless 
clearly shown. In order to smooth out the 
lesser variations and combine the concentration 
of wind in the two sectors, the algebraic dif- 
ference m percer "ages (east-northeast) for each 
year has been tabulated and then a 5 -year mov- 
ing average computed. This average has been 
plotted in Figure 1 with the point of the arrow 
placed at the middle year of the 5 -year period. 
LU 
2 
LU 
o 
O 
in 
UJ 
o 
O 
CD 
UJ 
o 
O 
UJ 
o 
O 
00 
Fig. 1. Diagram showing changing wind directions at Honolulu from 1907 to 1945. Each arrow repre- 
sents the mean position for 5 years, recorded at the middle year, in terms of the excess of east wind over 
northeast wind, positive to right, negative to left. The attitudes of the arrows give approximate azimuths, 
the vertical being N 67.5 E. 
East 
