TEERESTEIAL MAGNETISM AT KEW OBSEEVATOEY. 
79 
each of these and the three previous days. In this instance, there was a deficiency in 
the mean derived from each of the columns headed n, n—1, n—2, n — 2> as compared to 
the average day of the 11 years, and the deficiency was greatest in column 7^ —3. 
The calculations were repeated, limiting the days of largest and of least D range 
each to 5 a month. The two groups of days thus represented more extreme conditions 
than in the previous investigation, and consistently with this the phenomena proved 
to be of the same general character as before, only more pronounced. In the 
investigation which employed 5-day groups there were in reality six columns headed 
n-\-\, 71, 71—1, n — 2, n — 3, and 4, thus representing six successive days, extending 
from four days before to one day after the representative day of large (or small) D 
range. 
§ 6. As another line of attack, the initial selection of days was based on the sun¬ 
spot area,* the one group of days for any month consisting of the 10 days of largest 
area, the other of the 10 days of least area. The D ranges were put down for each 
day of either group, and for the three subsequent days in columns headed respective!}^ 
n, n+l, n-\-'2, and 7i + 3. Taking again a mean from all months of the 11 years, the 
mean D ranges from the four columns for the group of days of large spot area were 
all above the mean derived from all days of the year, and the excess was greatest in 
column -t- 3. 
In short, when the 11 years as a whole were combined, all the lines of investigation 
pointed to a relationship between the size of the absolute D range on individual days 
and sunspot areas on one or more preceding days. It appeared, however, that if 
Arrhenius’s views were correct, the time requisite for the electrified particle to travel 
from the sun to the earth must vary from less than 1 to more than 3 days, and it 
seemed that 4 days must be a more common interval than 3, and 3 days a more common 
interval than 2. The phenomena were, however, at least as favourable to the view 
that magnetic conditions on any one day represent an integral to which a number of 
previous days contribute. 
The similarity in the nature of the results derived by these different investigations 
from the 11 years as a whole, and the fact that the apparent sunspot influence 
indicated was far from infiniteshnal, seemed to preclude the possibility of the 
phenomena being wholly accidental. There was, however, this remarkable fact, that 
when the years were treated individually, some of them, notably 1895, gave results 
which appeared to be of an opposite character to those derived from the whole 11 years, 
there being an apparent association of large D ranges with small, not large, spot areas. 
Further investigation thus appeared necessary, and it was clear that it ought to be of 
a much more comprehensive character, calling for a large expenditure of time. It is 
only within the last twelve months that a suitable opportunity has presented itself. 
§ 7. The first question now to be considered is whether D is the best magnetic 
element on which to base the enquiry. The investigations which involved applications 
* National Physical Laboratory, ‘ Collected Eesearches,’ vol. 5, 1908, p. 55. 
