The Rev. H. Lloyd on the Results of Observations, ^x. 75 



lay before the Academy the principal conclusions already arrived at. In the 

 present paper, accordingly, I shall give the results of the observations of the 

 magnetic declination during the four years referred to ;* and in those which I 

 hope hereafter to communicate, I shall discuss in like manner the observations 

 of the other magnetic and meteorological elements made during the same 

 period. 



Diurnal Variation. 



2. A very limited series of observations is sufficient to exhibit the general 

 features of the diurnal variation ; but an extended one is necessary, if it be 

 desired to ascertain with accuracy the mean amoimt of the changes. To deter- 

 mine these with precision, observations should be taken daily, at equal intervals 

 not exceeding three hours, and be continued for one or more years. The 

 course usually adopted in the reduction of such a series is, to combine sepa- 

 rately the observations of each month, taking the arithmetical mean of all the 

 results corresponding to the same hour. In this manner the course of the varia- 

 tion (which alters considerably throughout the year) is deduced for each month 

 separately ; and when the observations extend over several years, the monthly 

 means of the separate years are to be again combined, each into a single mean. 



Even this, however, is insufficient. The mean results thus obtained are 

 deformed by the irregidar fluctuations, which are often far greater than the 

 regular changes ; and it is necessary to omit the observations taken on days of 

 disturbance, before we can deduce a correct mean from the results of any 

 practicable series. This is proved in a striking manner by the observations of 

 July, 1842. Owing to the great disturbance which took place on the 2nd and 

 4th of that month, the difference of the monthly means corresponding to 5 a.m., 

 when these observations are retained and when they are omitted, amounts to 

 5'- 76 ; so that the observations should be continued for fifty-seven years, in 

 order to reduce the error to O'-l. 



In the final reduction of the Dublin observations, accordingly, all the results 



• I have taken advantage of the delay which has occurred in the printing of this paper, to 

 introduce the monthly mean results of the three following years, in the deduction of the annual 

 and secular changes, and to make some other minor alterations of detail. The general conclusions 

 originally arrived at are, however, not affected. 



l2 



