OF THE EARTH’S MAGNETISM. 543 
1844. Jan. 9 to Feb. 5=27 days 
ie 2 : me rai! a Sum of 4 periods=108 days; mean= 27:00 days. 
Mars 30) 4 tApr. 26 =270;, 
Apr. 26 ,, Aug. 3 : Sum of periods= 99 days. 
mugess! ..) sept. 1=29),; 
epee, wept. 30 — so". 
Septng0i, Och .26=26-4 Sum of 4 periods =111 days; mean=27°75 days. 
Ochy26u--0 Nov. 22 —2i/ a. 
Nov. 22 ,, Feb. 5 : Sum of periods=75 days. 
1845. Feb. 5.,, Feb. 25=20 ,, 
Hebe or,... Mar, 24=27 -.. ! Sum of 3 periods=73 days; mean= 24°33 days. 
Mar 24 ,, Apr. 20=26- ;, . 
Apr. 20 ,, Aug. 3 é Sum of periods=103 days. 
Ang: 3 .,, Aug. 3027 4; 
Aug. 30 ,, Sept. 27=28 ,, 
Sept. 27 "Oct. 24=97 5, 
Oct. 24 ,, Nov. 17=24 ,, 
83. Some of the periods whose duration has been noted are by no means 
well determined. Thus the period commencing February 5, 1844, is not well 
marked, while that ending April 20, 1845, has more the appearance of ending 
April 13 or 14; any other points, however, which might be taken would not alter 
appreciably the mean period derived from these groups. If we examine them 
separately, we shall find that no period appears greater than 29 days, while some 
are only 24 days. In some cases a period appears broken into two, especially 
in the end of 1845; but this is generally on such days that we can determine, 
by the distance of preceding or succeeding minima, whether the break is the 
periodic minimum: thus December 3, 1845, is a marked minimum, but we can 
assume it to belong to the true period only by supposing some of the preceding 
periods to last 14 or 40 days, both completely without the limits of well-marked 
periods. 
84. If, then, we assume for the instant that the groups given above show the 
true number of periods, we shall have— 
Sum of 4 periods=106 days; mean=26-50 days. 
15 periods, with a sum of 398 days or 1 period= 26°53 days. 
We may suppose that the sums of days intervening betwixt each group should 
equal a whole number of periods; and if we take the value of the period approxi- 
mately as above, we must allow 4 periods to the first interval, 3 to the second, 
and 4 to the third, or 
11 periods, with a sum of 277 days and 1 period= 25°18 days. 
The mean of the whole 26 periods gives 1 period = 25°96 days. 
85. No distribution of the periods would alter this result much. The con- 
sideration of a much longer series will show, what indeed seems to follow from 
the above, that the period has a variable duration of 20 to 30 days. 
86. The first question that follows the appearance of such a period is as to its 
cause. Mr Kretiu’s results and my own seem to indicate a period of the dura- 
VOL. XXII. PART III. 7A 
