32 : REPORT—1858. 
1845. Hobarton. Makerstoun. Difference. 
January. csi dvasseads., 1002565 000943 "001622 
February eteeet adsense AOOs023 °001013 002010 
March... ..ccccasesecee) °003227 000988 “002239 
April. .cscreorscecseees *003206 *000890 "002316 
May lL cveddchavbbanent PUOso aS “001340 *002253 
JUNC 5 chads eds dcecmes eeOOS920 *001645 *002275 
July capSepbibcodenspenpenOOage Ws “001572 002241 
AUguSt ...sessereeseee *003737 001407 002330 
September ........... "003625 001078 *002547 
October...s...s0.2006. *°003982 °001461 *002521 
November ........... °004312 *001851 *002461 
December .....-ee.... "004472 *001895 *002577 
An examination of the means given above will show that they follow similar laws 
at both places, the average variation of the amounts of difference from month to month 
being only about one ten-thousandth (;;4;,) of the whole force, a change that 
would be produced by a variation of half a degree Fahrenheit in the temperature of 
the magnet ; and this amount includes the differences due to different secular change 
and different effect of disturbance. The above numbers, however, are far from giving 
an idea of the great‘resemblance in the monthly mean variations at the two places. 
It will be some evidence of their agreement, that I was enabled, after projecting the 
mean annual curves for Hobarton and Makerstoun, to point out that two monthly 
means for Hobarton were probably inaccurate. An examination of the printed 
observations showed that this was the case—that one monthly mean was erroneously 
calculated, the other error being probably typographical. The former error was less 
than two scale divisions, or a quantity which would be produced by a change of 2° 
Fahr. in the temperature of the magnet. As it was desirable to compare the monthly 
mean variations more minutely than can be done from the twelve values for each year, 
I combined the observations in a manner differing somewhat from that hitherto 
adopted. The usual plan has been, in examining the variation from month to month, 
to employ only those values corresponding to the middle of the months, or the means 
of the observations from the first to the last day of each month; and in order to 
obtain from these, the law of variation, the epochs of maximum and minimum, the 
means are projected, a curve being passed through or among the points; or the 
means are considered functions of the line of the sun’s longitude and the values for 
different epochs are calculated. Where merely general results are desired, and where 
the means for a number of years are combined, either process is sufficient ; but 
neither is satisfactory when we wish to note the changes of force as they exist for 
short periods. For this purpose, I have obtained the means of the observations made 
from the Ist to the 28th inclusive, from the 2nd to the 29th, and so on; so that if 
observations had been made on every day of the week, we should have in this way 
a monthly mean corresponding for its middle point to each day of each month, each 
period containing a lunation. The result obtained was as follows :—the monthly 
mean values of the horizontal force of the earth’s magnetism at Makerstoun, Hobarton, 
Trevandrum, &c., obey the same law,—the minute variations of the monthly mean 
being shown similarly at all the places, but the relative amounts of the changes are 
sometimes greater at one place, sometimes at another. 
I was now led to consider the variations of the datly mean magnetic force. In the 
observations issued from the Colonial observatories, the daily means for each Gottingen 
astronomical day are given: at Makerstoun and in several other observatories, the civil 
day of the place has been taken. Both methods have advantages ; a disadvantage of 
the former method is, that as no observations were made on Sundays, someobservations 
made on Saturday had to be combined with some observations on Monday to form one 
of the six daily means in each week,—so that one daily mean in six was not comparable 
with the mean of any other place not in the same meridian. It was obvious, how- 
ever, that for my purpose one daily mean from twenty-four hourly observations was 
insufficient ; it was necessary to obtain a daily mean for each hour, as I had obtained 
a monthly mean for each day. Thus I combined the twenty-four observations from 
midnight on Sunday till 11 o’clock on Monday night for the first mean of the week, 
1 a.m. till 12 p.m. of Monday for the second mean, and so on. In this way 120 
