TERRESTRIAL MAGNETISM AT KEW OBSERVATORY. 
89 
nearly constant. In this case the sunspot area was below its mean from 6 days before 
to 5 days after day n, so that here again there seems a lag of about 4 days. The 
incidence of the pulse which has its crest about day n—12 in the last line of Table V. 
is sufficiently accounted for by the high values of sunspot area in the last line of 
Table IV. for days prior to II. The difference between this crest and the 
subsequent trough represents 28 per cent, of the corresponding mean absolute daily 
range of H, or about 217y. This is about two-thirds of the corresponding amplitude 
in the line above, answering to the days of largest spot area. The ranges of the 
sunspot values in the last and second last lines of Table IV. were respectively about 
1804 and 2447 of the Greenwich units of area, so the H range and sunspot area 
ratios in the two cases are fairly similar. 
§ 13. While Table V. seems to prove to demonstration that in the average year 
there is a clear association of H ranges with sunspot area some days previously, the 
relation is either of a somewhat complex character, or else is liable to be much 
overshadowed in individual years by other influences. It is obviously desirable that 
details enabling an independent judgment to be formed should be at the disposal 
of all interested in the subject. It is hoped that Table VI. will suffice for this 
purpose. 
The figures in Table VI. are the mean absolute H ranges, in terms of ly as unit, 
from the representative 31 days n—15 to n+15 of the individual 11 years. Values 
which exceed the mean derived from the whole 31 days are in heavy type, so that one 
can see at a glance how far each year conforms to or departs from the general features 
of the 11 years combined, as exhibited in the first line of Table V. No single feature, 
it will be seen, is clearly exhibited by all the years. Three of them have the figure in 
column n + i below the average, though the deficiencies are all trifling ; and four of 
them have the figure in column n — 4 above the average. 
§ 14. One aspect of the case which suggests itself when individual H ranges are 
scrutinised calls for consideration. The number of days contributing to the mean 
value for a single year of one of the 31 columns of Table VI. is normally 60. Now 
there are a few days the H range of which is altogether outstanding. Thus in 
February, 1892, two successive days had ranges of 720y and 650y. The range of the 
average day of 1892, though larger than that of any other year, was only 84y. This 
will explain how the presence of even one outstanding range in a column sensibly 
affects the mean value, and if any “ accident ” should bring two or three such days 
into the same column for a single year the result might be to simulate a marked 
influence which had no real existence. 
R is obvious d priori that, so far as the present research is concerned, a range of 
720y must be regarded as largely an “ accident.” There are no outstanding daily 
values of sunspot area in the whole 11 years which overtop their neighbours in the 
way the larger magnetic ranges do. The recognition of this fact suggested the next 
mode of attack. It follows lines which were suggested by the procedure followed now 
VOL. C5CXII.' —A. N 
