AT KEW OBSERVATORY, 1890 TO 1900. 
239 
up to Table XXIII. was accordingly repeated, but with this difference, that the 
comparison instituted was between the Jive days of largest and the five of least 
range in the month, and it was extended to include the fourth previous day and the 
day following that to which the range belonged. 
The increase in the difference between the mean ranges in the two groups of days 
due to the reduction of the number of days in the group from 10 to 5 is shown in 
Table XXIV. The final means from the 11 years for the groups of largest and of 
Table XXIV.—Mean Absolute Ranges. 
Year. 
Groups of 10. 
Groups of 5. 
Largest. 
Least. 
Largest. 
Least. 
1890 
/ 
14-37 
/ 
7-57 
/ 
16-32 
6-88 
1891 
19-80 
9-03 
23-59 
8-24 
1892 
26-88 
11-03 
34-74 
9-97 
1893 
21-03 
11-28 
24-27 
10-29 
1894 
24-89 
10-77 
31-79 
9-84 
1895 
22-32 
10-26 
25-72 
9-29 
1896 
21-76 
9-28 
26-13 
8-40 
1897 
17-71 
7-93 
21-35 
7-00 
1898 
18-51 
7-78 
22-99 
7-02 
1899 
16-41 
7-49 
19-90 
6-77 
1900 
12-41 
6-42 
14-58 
5-94 
Means. 
19-64 
8-99 
23-76 
8-15 
least range bear the ratio of 2T8 when the monthly groups contain 10 days, but 2'92 
when they contain 5 days. Thus what we should expect in Table XXV., which 
summarises the investigation for 5-day groups, is a repetition of the phenomena 
apparent in Table XXIII., but in a considerably enhanced degree. The tables were 
prepared exactly in the same way and the headings have the same meanings; n+l 
denotes of course the day following that to which the magnetic range belongs. The 
column under the first heading, n+ 1, is spoken of as the first column. 
It will be seen that the final excesses in the means for columns 2 to 5 over those 
for columns 8 to 11 is greater in Table XXV. than in the corresponding columns of 
Table XXIII. This result is favourable to the view that the difference is not purely 
fortuitous, but betokens some physical connection. 
The marked rise in the mean of the second column, as compared to that of the first 
column in Table XXV., is also distinctly favourable to this view. On the other 
hand, the mean for the seventh column is nearly as large numerically as those for 
columns 8 to 10, and larger than the means for columns 11 and 12. This is equivalent 
