TERRESTRIAL MAGNETISM AT KEW OBSERVATORY". 
103 
The rise to the crest was clearly more rapid than the subsequent fall, the secondary 
pulse resembling closely in this respect the primary, for which successive character 
totals ran as follows :— 
Column . 
71-3. 
77 - 2 . 
77-1. 
77 . 
77 - 1 - 1 . 
77 4- 2 . 
77 4- 3. 
77 - 1 - 4. 
77 4- 5. 
Sum . . . 
423 
510 
690 
996 
731 
570 
511 
463 
438 
Assuming a uniform progression between days ri—3 and n—4, and again between 
days n+i and ?i+5, we find 
Value 430 occurring at —2‘92 as well as at n + 24, 
,, 440 ,, + 4 92 ,, ,, 7i + 33. 
These two values come near the beginning and end of both pulses, and we may thus 
regard 7’84 days in the primary pulse as represented by the somewhat increased width 
of 9 days in the secondary pulse. If now we assume the widening exhibited by the 
secondary pulse to be contril)uted to in like proportion from the parts which precede 
and follow the crest, and take 27 +a; as the time of the crest, we have 
(3+£c)-^{5 + (l-a;)} = 2-92/4-92, 
whence x = 0’35, and so period = 27'35 days. 
From these and other similarly rough calculations I should assign to the period as 
indicated by the 11-year data the duration 27’3±0‘1 days. Of any period shorter 
than this there seems not even a suggestion in the 11-year'figures. 
§ 24. Let us now return to a consideration of the results from the three shorter 
groups of years in Table XI. The figures are naturally less smooth than those in the 
first line, but all the groups show the 27-28-day period. Analogous figures were 
really got out for each half-year separately, and of these 22 sets of data there was not 
one that did not show enlarged values in the immediate neighl30urhood of days n + 27 
and n + 28. In every instance the mean from columns n + 26 to ti-l-30 exceeded the 
mean from columns ?^-l-20 to n + 25, and in 14 of the half-years the largest value in 
any column subsequent to n + 5 — t.e., subsequent to columns clearly aftected by the 
preliminary pulse—occurred in one or other of the columns n + 27 and Ji-f-28. There 
was only one whole year, 1894, in whicli the figures in columns n + 27 and ';i-l-28 were 
both exceeded by the figure in any other column subsequent to n + 5. 
If we take the ratio borne by the larger of the two figures in columns n -t- 27 and n -I- 28 
to the figure in column as a measure of the prominence of the 27—28-day period, we 
find 0'62 for the whole 11 years, 0’59 for the sunspot maximum period, and 0’G7 for the 
group composed of 1891, 1895, and 1896. The 27—28-day period was thus considerably 
most in evidence in the group of years which gave the faintest indication of the 
magnetic pulse following 4 days after the sunspot area pulse, and it was least in 
