Barometric Variability at Kimberley and Elseiuliere. ill 
January, 1901, and -069 inch in January, 1905. And these values are 
shown in Tables 6 and 7 respectively. 
Table 8 gives the deviation of the annual mean variability from the 
total mean. For example, the total mean variability at Cape Town, as 
shown in Table 3, is *108 inch, and the mean for 1905 is -100 inch. Hence 
the deviation for 1905 is — -008 inch. 
The individual annual mean values of the monthly maxima of varia- 
bility are given in Table 9. For example, the maximum variability for 
each month of 1905 at Cape Town was : — 
Inch. Inch. 
Jan -216 July -333 
Feb -255 Aug -315 
Mar -212 Sept -542 
April -251 Oct -386 
May -324: Nov -254 
June -362 Dec -248 
The average of these separate maxima is "308 inch, which is the value 
placed in the Table. 
It is fashionable nowadays to compute the coefficients of correlation 
for all manner of presumably mutually dependent things. Calling the 
coefficient of correlation r, and the "probable error" +B, we have 
the following results for sundry pairs of the stations shown in 
Table 8 :— 
r 
E 
Durban and East London 
+-690 
•075 
Durban and Kimberley 
+-525 
•112 
--184 
•188 
+-359 
•123 
+-132 
•183 
+-480 
•130 
+-165 
•199 
These results may, in some of the cases, be more curious than 
valuable. It is certainly remarkable that the coefficient for Durban 
and Adelaide should be so much greater than, and so much at variance 
with, the coefficient for Durban and Perth. And equally remarkable that 
the coefficient for Cape Town and Kimberley should be so small. This 
is a matter, however, that must stand over for the present, awaiting 
further elucidation later on. Particulars of barometric variability from a 
number of South American coast stations would probably help us. 
However important barometric variability may be as a factor of 
weather, there is no present evidence forthcoming to show that it is 
