AT KEW OBSERVATORY, 1890 TO 1900. 
237 
Sunspot area, it is true, shows a far from regular progression from day to day ; the 
area on one day is not infrequently a considerable multiple of that on the preceding- 
day. But, as a rule, days of large area and days of small area occur in groups. In 
one or two months the 10 days of largest area were absolutely consecutive, and it 
was rather the rule than the exception for a considerable proportion of the days of 
both Groups I. and III. to be consecutive. Thus the phenomena of Table XXII. 
might be expected to occur though there were no direct connection between 
disturbance and the simultaneous solar phenomena, provided there were an effect on 
the earth within a few days of the occurrence on the sun. The results, in short, might 
be easily reconciled with theories such as that of Arrhenius. 
§ 27. The next and much more laborious investigation was intended to throw light 
on Arrhenius’ and similar theories. If there is an influence originating in the sun, 
whose intensity increases synchronously with sunspot area, which is propagated to 
the earth in two or three days and there causes magnetic storms, then there ought 
to be a marked association between the amplitude of the absolute daily range and 
the sunspot area two or three days previously. To investigate this point the 10 days 
of largest and the 10 days of least absolute range in each month were taken to form 
two contrasted groups. The sunspot areas answering to each of these days and the 
three preceding days were entered in four successive columns, and means were formed 
for each month. One thus got for each month four mean sunspot areas S, S_j, S_ 2 , 
S_ 3 , answering : S to the 10 days of largest absolute range, S_j to the 10 days 
immediately preceding these, and so on, and four mean areas S', S'-!, S'_ 2 , S'_ 3 , 
answering to the 10 days of least absolute range, to the 10 days preceding these, 
and so on. 
What appears in Table XXIII. is the algebraic excess of the 12 monthly values of 
S, S' . . . over the mean sunspot area for the year as given by the Astronomer Royal. 
The second last line gives the algebraic mean of the above results from the 11 years. 
The last line gives the final mean obtained when the entries under each individual year 
are expressed as percentages of the mean sunspot area for that year. In the headings 
of the columns, n is intended to denote the representative day of large (or small) 
absolute range; n— 1, n — 2, and n —3 the three preceding days in order. 
If magnetic disturbance were entirely or even largely due to solar influence, whose 
activity at its source was largely dependent on sunspot area, and whose time of 
propagation to the earth varied only within narrow limits, then what one would 
expect to see would be a notable excess in the mean S_ l5 S_ 2 , or S_ 3 , and a corre¬ 
sponding deficit in the mean S'_ l5 S'_ 2 , or S'_ 3 . 
If the principal source of magnetic disturbance is of the kind postulated, but the 
time of propagation to the earth varies largely from under a day to several days, then 
what one would expect is an excess of all or most of the S’s, with corresponding 
deficits in S"s. 
It is the second hypothesis, if either, which derives support from Table XXIII. In 
