BETWEEN SUN-SPOT FKEQUENCY AND TERRESTRIAL MAGNETISM. 175 



27. The results pointed out in the last paragraph help to explain some novel 

 features in Table XVIII. , which gives the values obtained for <i, b and b/a by 

 applying the method of groups to the sums of the 24 differences in the diurnal 

 inequalities for the several months. As in similar tables, the seasonal and yearly 

 values of a and b represent arithmetic means for the included months. 



In D and V the lowest values of a are found in winter, but in H and 1 the lowest 

 values occur in November and December, that is, at midsummer. 



In D and V, again, b is distinctly below the average in winter, though not nearly 

 to the same extent as in Europe, and the winter value for F is less than the summer 

 value ; but in H the winter value exceeds the summer. 



In D, b/a is distinctly smaller in summer than in the other seasons ; but in 1, H 

 and V the summer value is a trifle the highest. The winter and equinoctial values 

 of b/a are almost identical in all the elements. As compared to northern stations, 

 the variation in b/a throughout the year is exceedingly small. 



TABLE XVIII. Batavia. Sum of 24 Hourly Differences, L 887 1)8 (Units I' in 



D and I, ly in H and V). 



28. In applying the method of groups, it is evidently desirable that one group of 

 years should fall near the middle of the period dealt with, and that part of the second 

 group should precede, and part follow it. This arrangement helps to eliminate any 

 long-period variation, or any gradual change in the conditions. The period 1887 to 

 1898 being by no means ideal in the above respect, in dealing with the diurnal 



