514 E. HUNTIKGTOX SOLAR HYPOTHESIS OF CLIMATIC CHANGES 



which can not possibly be as high as 1, nor can they possibly fall as low 

 as zero. If the data used in the computations are free from perceptible 

 error, a coefficient approaching 0.50 indicates a high degree of connection. 

 The coefficients may have either a plus or a minus sign. The sign is posi- 

 tive when the maxima of one phenomenon are connected with the maxima 

 of the other, while it is negative if the maxima of one are connected with 

 the minima of the other. 



In the case of the two curves of figure 9 the correlation coefficients 

 have been computed in various ways, as shown in Table 6. The table 

 shows the correlation coefficient between the number of storms in any 

 given year and the sun-spot numbers of various years or combinations of 

 years. For instance, the first line, ^'^Storminess of a given year and sun- 

 spots of the third year previous, -[-0.074,^^ indicates that when the storms 

 of a given year are compared with the number of sun-spots during the 

 third year before that time, the correlation coefficient is only +0.074. 

 So small a coefficient is unimportant. It indicates that there is no par- 

 ticular relation between the storms, say of 1914, and the sun-spots three 

 years previously — that is, during 1911. The next line shows that when 

 the storms are compared with the sun-spots of two years before — that is. 

 when the storms of 1914, for example, are compared with the sun-spots 

 of 1912 — there is a connection, and if the number of sun-spots was on 

 the increase in 1912 the number of storms is likely to be great in 1914. 

 The next three lines show coefficients of +0.41, +0.49, and +0.46. This 

 indicates a high agreement. It is noticeable here that there is more agree- 

 ment between the storms of any given year and the sun-spots of the suc- 

 ceeding year than between the storms and the sun-spots of the preceding 

 year. This is sigiiificant because it supports an idea which h^^s already 

 been suggested, namely, that the storms p.re not due directly to the spots, 

 but to certain conditions of the sun, which cause an excitement of. the 

 sun's atmosphere manifesting itself in spots, and also of the earth's at- 

 mosphere manifesting itself in storms. The disturbance seems to pro- 

 duce its maximum effect on the earth slightly sooner than on the sun. 

 The last three correlation coefficients show that when the storms of a 

 given year are compared with the sun-spots not only of that year, but of 

 that year combined with the year before or the year after, or with both, 

 the relationship is higher than when individual years are considered. It 

 reaches its maximum, +0.535, when the storms of a given year are com- 

 pared with those of that same year, together with the succeeding year. 

 From the mathematical point of view, this affords the strongest possible 

 demonstration of a relationship between the two phenomena. 



