A.—_MATHEMATICAL AND PHYSICAL SCIENCES 37 
only 0:2. The immediate effect of this is that numerical values of the 
winter oscillation give us a means of predicting three months in advance, 
at any rate approximately, the summer values of the oscillation and 
therefore of the pressure, temperature and rainfall associated with them. 
In Fig. 9 are the relationships of the values of the pressure, temperature 
and rainfall of December to February, with the numbers indicating the 
fluctuation of the previous June to August. These express relationships 
which have held for about fifty years, and show that we have arrived, 
not at a mathematical figment, but at a physical reality of commercial 
value. 
These methods of prediction can be improved on by study of the 
relationships of individual areas. For example, the coefficient of 0°64 
of rainfall of north-east Australia with the oscillation of the previous 
winter becomes 0-79, when we base it on previous pressure at Honolulu, 
1900 ‘Oo ‘25 ‘30 
2 vaca ANN PS 
STEN neers oe 
ae ee ARE 
Fic. 10.—N.E. Australian rainfall, October to April. 
Port Darwin and South America ; a comparison of the actual rainfall with 
that given by the formula isshownin Fig.10. Similarly, theo-56 of South 
Africa becomeso:72. Butacertain amount of the improvement effected in 
this way by selecting the biggest factors is bound to be fictitious, even 
when there appear to be adequate independent reasons for thinking that 
the relationships are real ; and, if this precaution is ignored, the more 
promising the formula, as indicated by the closeness of its apparent 
relationship, the greater is the likelihood of disappointment. 
It must be admitted that a real control of 0-7 by previous conditions 
is about as good as is now available for forecasting, and the difference 
between the actual and the forecasted amounts will still be considerable ; 
so predictions can only be issued with restraint if public confidence is 
to be won. The natural consequence is silence, except when the 
indications are markedly favourable or unfavourable: in a race with 
thirty starters a conspicuously good horse may, without undue risk, be 
backed to come in within the foremost six, and we may feel confident 
that a thoroughly bad animal will be in the last six; but it would be 
unwise to hazard much on the likelihood that a commonplace individual 
will finish among the central six. It may at first sight seem a confession 
