Annual Variations for the Horizontal Component. xxxiii 



considerable distinctness in each of the six years ; the variations from it are not greater than might be expected 

 when we take into account the large effect of disturbances, the irregular value of the secular change in some 

 years, and in others the fewness of the daily observations, and consequent imperfect nature of the corrections. 

 These corrections, as deduced from the observations for 1844 and 1845, vary so much as to account fully for 

 many of the minor differences from the mean law in the years 1842 to 1847, and for even the larger differences 

 in the years 1848 and 1849. Upon the whole the summer maximum appears rather greater than the winter 

 maximum, and the autumnal minimum than the spring minimum ; although as this is not the case in the mean 

 for the first four years (column 10, Table 24), and as very large disturbances occurred in the end of the years 

 1846 and 1847, which pi'obably increased the autumn minimum of these years, this difference in the minima 

 is perhaps accidental, and might be removed or considerably diminished in a larger series of observations. 



57. When w-e compare the monthly means, as deduced from the 10 days selected in each month as least 

 affected by ii-regular disturbances, with those deduced from all the hourly observations in the same months, we 

 find the latter less ( — ) or greater ( -f ) than the former, by the following quantities : — 



Vear. Prefix. 



1844, 0-000 



1845, 0-000 



Mean, 0-000 



Jan. Feb. March. April. May. June. July. Aug. Sept. Oct. Nov. Dec. 



-109 -346 -399 -315 -029 +018 -021 -108 -062 -507 -164 -234 



-301 -160 -260 -143 -032 +021 +003 -119 -109 -251 -111 -377 



-205 -253 -329 -229 -030 +019 -009 -113 -085 -379 -137 -305 



In each month, with the exception of June, the mean deduced from the undisturbed days is greater than 

 that deduced from all the observation days ; and the excess is greatest in March and October, the months of 

 greatest disturbance : the effect of disturbance on the means for the months of May, June, and July, is very 

 small, nearly zero. 



58. As the above effects of disturbance seem to obey a law similar to that of the annual period of the 

 mean, it will be interesting to consider the law for the latter, as deduced from the nearly undisturbed 10 days 

 selected in each month of the years 1844 and 1845. The following are the mean variations, deduced from all 

 the daily observations in 1844 and 1845, as in Table 24. 



I Jan. Feb. March. Ajjril. May. June. .luly. Aug. Sept. Oct. Nov. Dec. 



Meanofall, 0-000 I +082 +044 -224 -280 +132 +494 +373 +022 -314 -294 -065 +028 



Correcting these by the mean quantities in No. 57, we obtain the variations of the monthly means from the 

 nearly undisturbed days of 1844 and 1845. 



Means of 

 20 days 



^ ] 0-000 j +116 +126 -066 -222 -009 +304 +211 -036 -400 -086 -099 +162 



The monthly means, deduced fi-om 10 nearly undisturbed days in each month of the two years, give the 

 same law as has already been deduced from the means of all the observations (as in Table 24) : in the undis- 

 turbed means, the maximum at the winter solstice is rendei-ed more marked, and it appears probable that the 

 difference between the values of the two maxima may be wholly a result of disturbance, which appears to 

 diminish the winter means considerably, while it rather tends to increase those at midsummer. When a more 

 careful selection of undisturbed days is made, as in that oi five days in each month of 1844, (see p. 365 of the 

 volume for that year) it is found, that the effect of disturbance in diminishing the winter means, and in increas- 

 ing the summer means, is even more considerable than that found above No. 57. 



Differences of the Dailtj Means of the Horizontal Component from, the Means for the corresponding Montlis. — 

 The discussion for 1844, will be found in the volume for that year, page 357, the results for 1845 and 1846, 

 are obtained from Tables XVIII. and L VII. of this volume. 



Lament's observations at Munich (1843-5), although by the combinations which he had employed he had failed in detecting the fact. 

 The following are the variations of the Munich numbers as corrected for secular change in the note cited, the horizontal component 

 at the place being taken as unity. 



Sept. Oct. Nov. Dec. 



-421 -328 -045 +172 



The value of this confirmation of the Makerstoun law is increased by the fact, that the Munich instrument has a unifilar suspen- 

 sion, and that the processes of observation, reduction, and correction, are completely different from those employed at Makerstoun. 



Prefix. 



Dec. 



Jan. 



Feb. 



Mar. 



Apr. 



May. 



June. 



July. 



Aug. 



0000 



+ 172 



-029 



-086 



-029 



-158 



+ 265 



+ 238 



+ 280 



+ 136 



