Ixxii 
General E-esults of the Makerstoun Observations. 
spring. In the winter months, the principal or westeniloop is formed hy the motion from 8^ a.m. till 5*^ p.m, ; 
in the months from April to August, three-fourths of the whole diurnal motion occur between 6^ a.m. and 
6^ P.M., the remaining fourth forming a slightly inflected side to each of the figures : it is this side which is 
gradually twisted up to foi-m the eastern loop of the winter months. The figures for means of groups of 
months, as in Tables 13 and 58, have been projected in Plate VIII. on a larger scale, the diurnal mo- 
tions fi'om the days selected as nearly free from irregular disturbance have been projected with dotted outlines 
along with the others. In these figures the actual values in Tables 13, 14, 58, 59 have been employed. In 
the winter months the undisturbed diurnal variation presents a series of convolutions instead of the eastern 
loop, and in the other months the general form of the figures is not much altered. 
155. It is evident that no pi'oper comparison can be made of the areas of these figures, on account of the 
involved forms in the winter months ; the areas, however, of the figures from April to August, differ very 
little. * 
156. Perimeters of the Figures. — The twisting of the perimeters, which renders a comparison of the areas of 
little value, does not appear to affect the length of the motion, and this therefore seems a fair subject for ex- 
amination. The following are the values of the angular motion, or length of the perimeter, for each month, as 
obtained approximately from Plate VII. 
Jan. Feb. March. April. May. .June. July. Aug. Sept. Oct. Nov. Dec. 
5'-60 6'-16 9'-22 12'-18 12'-04 12'-00 ll'-56 ll'-64 10'-48 9'-78 7'-22 5'-84 
December and .Tanuary shew the least perimeters, April, May, and June, the greatest, though the perimeters 
for the months from April to August are nearly constant. 
157. The following are the approximate perimeters of the five independent figures of Plate VIII. : — 
Dec. -Feb. March, April. May, June. July, Aug. Sept.-Nov. 
Mean of all, 6'-19 11'.58 ll'-88 11''92 9'-04 
Mean of undisturbed days, 4'-34 9'-86 10'-68 ll'-28 7'-76 
158. Hourly Angular Motions. — Having obtained the approximate motion from hour to hour for each of 
the monthly figures of Plate VII., we find that, on the whole, they follow nearly the same law, that indicated 
in the following numbers, which are the means of the motions from the 12 separate months, and from other 
groups of months. 
Table 67. — Mean Angular Motions, from Hour to Hour, of the north end of a Needle supposed freely 
suspended in the direction of the Magnetic Force, as obtained (1.), from the Monthly Figures of 
Plate VII. (2.), from the 5 Independent Continuous Figures of Plate VIII. ; and (3.), from the 1 
5 Independent Dotted Figures of Plate VIII. 
Time. 
12 
Figures. 
Means from 
5 
Continuous 
Figures. 
5 
Dotted 
Figures. 
Time. 1 
12 
Figures. 
Jleans from 
5 
Continuous 
Figures. 
5 
Dotted 
Figures. 
h. h. 
li. h. 
12—1.3 
0-19 
6^10 
6^20 
0— 1 
6^64 
6^59 
6^62 
13—14 
•23 
•10 
•20 
1— 2 
•49 
•48 
■50 
14—1.5 
•25 
•14 
•16 
2— 3 
■59 
•53 
■60 
15—16 
•22 
•18 
•25 
3- 4 i 
•51 
•51 
•53 
16—17 
•24 
•24 
•23 
4— 5 1 
■48 
•34 
•48 
17—18 
•25 
•23 
•24 
3 — 6 \ 
•40 
•19 
•35 
18—19 
•37 
•28 
■38 
6— 7 
•20 
•14 
■28 
1 9—20 
•58 
•53 
•55 
7— 8 
•36 
•19 
■30 
20—21 
•61 
■58 
■63 
8— 9 
■21 
■17 
•22 
21—22 
•69 
•76 
•69 
9 — U) 
■15 
• 13 
■1.3 
22—23 
•91 
•85 
•88 
10—1 I 
•14 
•17 
•14 
23— 0 
•70 
•67 
•73 
11 — 12 
•24 
•22 
•16 
It may not be unimportant to remark here, tliat the processes usually adopted in order to determine the epochs of maxim* 
and minima for tlie separate elements of declination and dip, are not strictly accurate ; and that is the case whether the process be 
one of inter|)olation from grapliic projection, where the time is the abscissa, or one of computation, where the variable is a func- 
tion of the hour angle. Tliis is evident, when we examine the figures in Tlates VII. and \"III., where the dip and declination are 
the co-ordinates. The error, however, will not aiVect any of the comparative conclusions for these element* in the previous p.iges- 
A similar exception may be taken to the accuracy of comparisons of areas of declination curves, where time is the abscissa. 
