144 ON THE VARIATION OF THE EARTH'S MAGNETISM, &c. 



values for Sundays were used. These were obtained by mterpolation ; the mean of 

 the three days preceding, and of the three days succeeding, were taken as the 

 means for the Sundays. The want of approximate means for these days would 

 tend to destroy the regularity and distinctness of the results, owing to the varia- 

 tions, due to different causes, which it is desired to eliminate, and that the blank 

 days enter irregularly into the days of the mean monthly periods. 



22. The results of these summations (Curves No. 4, Plate VI.) are, that each year 

 indicates maxima of the vertical component near the quadratures, and minima near 

 the syzygies. In oi-der to render the fact more distinct, and the curves somewhat 

 more regular, eleven days* of greatest disturbance were eliminated in each of the 

 years 1844 and 1845, namely, those days on which the mean value of the verti- 

 cal component was greater than the mean of the previous and succeeding days 

 by more than 26 micrometer divisions (about three times the resulting range). 

 The principal minimimi in 1844 occurs at the period of full moon ; in 1845 it 

 occurs about three days before the new moon. In the mean of both years the 

 principal minimum occurs at the time of fuU moon, the secondary minimum about 

 two days before new moon ; the maxima occur between these periods, and they 

 are nearly equal. The lowest curve of No. 4 is the projection of the mean for 

 the two years without eliminating disturbances. It differs httle in regularity from 

 the other in which the large disturbances are eliminated. The irregularities in 

 these curves may be partially due to the cause of error already stated (17), 

 namely, an insufficiency in the temperature correction. 



23. Whether due to the cause just mentioned or not is uncertain, but the 

 results for the relation of the variations of the vertical component to the moon's 

 declination are neither distinct nor consistent for the two years, and the elimina- 

 tion of the days of large disturbance does not improve them. The mean for the 

 two years seems to shew something like the law found for the horizontal com- 

 ponent, namely, maxima about the periods of gi-eatest north and south declina- 

 tion ; but I do not place any trust at present in this result. 



Makeestodn, April 13. lSi6. 



% 



* Each of these days was actually observed (observations were made) at the time as a day of dis- 

 turbance. 



t 



