20 PRKSIDENTIAL ADDRESS SECTION A. 



that the mean path of cyclones is over the equatorial regions 

 where, in fact, they do not occur. Statistics are to be applied 

 to reliable documents, which, needless to say, is very often 

 neglected. They are to be applied to a large number of observa- 

 tions if the results are to represent a physical fact, and not 

 merely a meaningless arithmetical result. But, above everything, 

 one must have, in applying the statistical method, an open mind — 

 that is, one must try to find only what statistics really give, not 

 what one wants to get out of them. If properly applied, this 

 method can either give the law we look for, or show us in what 

 direction we have to look for it, for the underlying philosophical 

 principle is that an effect which reciu's with a certain constancy 

 is not due to mere chance, but to some cause which produces it. 



The first fact on which weather predictions may be based is 

 that, in general, pressure systems move from west to east. This 

 is especially true for the cyclones and also for the anticyclones 

 in North America and South Africa. For the European anti- 

 cyclones the question is not so clear. There are exceptions, 

 however. Cyclones are known to move westwards, others are 

 stationary. The next consideration is their rate of displacement. 

 Here we have a less satisfactory state of things. Anticyclones, 

 as a rule, move slowly and hang over the same regions. To the 

 cyclones we can assign an average rate of displacement deduced 

 from a large number of observations. But here the averages are 

 deduced from velocities ranging from zero to a hundred miles or 

 more. 



Another important consideration is that of the tracks 

 followed by the various pressure systems. Here we are on still 

 less certain ground. In North America the cyclones and anti- 

 cyclones follow one another fairly regularly, and move also with 

 a fair regularity along certain mean tracks, which change, how- 

 ever, with the seaons. The same may be said of the tropical 

 cyclones in the West Indies, Mauritius, and the Far East. In 

 Europe, however, it has l)een practically impossible to find a 

 system of tracks that can be of any great use in weather forecast- 

 ing. Koppen, Van Bebber, Rykatchew, who especially investi- 

 gated this point, find from five to ten mean tracks over Europe. 



The solution of the pressure transformation has been tried 

 in another way also. From the study of the daily charts a series 

 of weather types has been deduced, and their formation and 

 transformation carefully studied. The possible influence of large 

 areas of heat, rain or snow, on neighbouring pressure systems 

 has been patiently investigated, and rules for forecasting trans- 

 formations or displacements in these systems deduced therefrom. 



When, by some such consideration, the forecaster has decided 

 upon the probable transformation of the existing system of 

 pressure, he has to pass to the transformation of the existing 

 weather. Here, besides the general correspondence of pressure 



