PROFESSOR KEELY’S MiiGNETlCAL OBSERVATIONS, 
207 
From the above Table it is obvious that both needles lost considerable force between 
February and June, and that after that time L(2) continued to lose, while L(l) 
retained its force. I have therefore reduced all observations made with L(2) to the 
30th of June, by considering the loss as proportional to the time, the coefficient of 
reduction being (1 +’000005.c?), d being the number of days after the 30th of June. 
Column (3.) in Table IV. contains the logarithms corrected in this manner for L(2), 
and those of L(l) as originally found. The numbers ‘8707137 and ‘9070629 are 
the values of <p at Waterville on the 30th of June by L(l) and L(2) respectively; 
if the factor 1 ‘04 175 which connects them is compared with the corresponding 
factors in column (4.) (omitting that for Grand River for a reason already stated), 
these factors will all be found to exceed it by a small quantity : waiving discussion, 
at present, on the cause of this difference, I shall assign half the difference between 
the mean of these factors and 1 ‘04 175, or 0‘00036 as a correction common to both, 
positive for L(l) and negative for L(2), which, while it leaves the mean determina- 
tions where both needles were used unaltered, will secure a more just comparison 
for stating where but one needle was used. I believe the difference, however, has 
hitherto been considered wholly within the errors of observation. Column (5.) con- 
tains the corrected logarithms, that is, the values of for L(l) increased in the ratio 
of 1 to 1 ‘00036, and those of <p for L(2) reduced in the same ratio. Column (7.) 
contains the total forces for both needles, considering ‘8707137, or the number 
expressing the relative force by L(l) on the 30th o-f June at Waterville, as the unit 
of force ; and column (8.) contains the means. 
