(17) 
A NOTE ON THE POSSIBILITY OF LONO-EANaE WEATHER 
FOEECASTS. 
By J. R. Sutton, F.R.S.S.A. 
Long-range weather forecasts — that is to say, forecasts which profess to 
foretell the character of coming seasons or periods — may be grouped into 
three main classes : 
(1) The random guesses of the weather prophets, mostly published in 
cheap almanacs and newspapers. With a little ingenuity these may be 
made to show a specious accuracy. For instance : suppose it to be known 
beforehand that, on an average, there have been ten wet days in January ; 
then all that is necessary is to predict ten wet and twenty-one dry days in 
coming Januarys. The wet and dry days may be assigned to any dates at 
random, and of these, in the long run, four hits will be scored to three 
misses. Again, if it be known that there have been, on an average, two wet 
days in June, the same sort of random guessing will give us, in the long run, 
thirteen hits to two misses — all of which will look very clever to the 
uninitiated, especially when an occasional lucky hit with a very heavy 
rainfall happens to be made. 
(2) Forecasts based on cycles — which always break down sooner or later 
when put to the test of prediction. 
(3) Forecasts based on recognised physical or statistical principles. 
These are much less ambitious than the other two, and, generally speaking, 
are limited in scope. Two of these, of South African interest, may be men- 
tioned : One, by Claxton, showing that " winter droughts at Durban have 
invariably been followed by summer droughts in Mauritius the other, by 
Mossman, suggesting a relationship between the temperatures of the Ant- 
arctic during August and September and those of Kimberley during the 
following October to December.f Each of these depends on sound physical 
principles. The first is based on the easterly drift of barometric depressions 
* T. F. Claxton, "Note on the Connection between the Rainfall at Durban and 
Mauritius," 'Trans. S.A. Phil. Soc./ 1907. A similar sort of correlation probably 
exists between the rainfall of the Eiver Plate delta and that of south-west Cape 
Colony. 
t R. C. Mossman, " Southern Hemisphere Seasonal Correlations/' ' Symons's Met. 
Mag-./ 1913. 
2 
