392 



Dr. C. Cliree. Collimator Magnets and the 



determined specially in each observation of X. The accuracy of the 

 determination thus depends more on the observer than on the in- 

 strument. 



P has never been regarded at Kew as a constant to be determined 

 once for all, but observers are recommended to determine it for them- 

 selves, employing, when possible, a series of observations made about 

 the same time and in one neighbourhood. In other words, P is treated 

 as a quantity which probably alters slowty with the time, and which 

 may vary with X. In the Kew unifilar itself the mean value of P 

 calculated from a year's observations varies slightly but somewhat 

 irregularly from year to year. 



The degree of variation is best seen by consulting the following 

 table : — ■ 



Table XL 



Year. 



1860 



1875 



1879 



18821 



1883 J 



1884 



1885 



_lQ5p/ r 2 



at 30 em. 



198 

 185 

 112 



133 



172 



95 



1886 

 1887 

 1888 

 1889 

 1890 

 1891 



-10 5 x P/> 

 at 30 era. 



153 

 192 

 174 

 211 

 177 

 121 



18921 



1893 J 



1894 



1896 



1897 



1898 



Each one of these results is based on a large number of observations, 

 but I should hesitate to say that observational error plays no part in 

 the apparent fluctuations. There are of course frequently minor 

 magnetic disturbances during horizontal force observations, and two 

 or three outstanding values of P sensibly affect a mean though derived 

 from forty observations. Though the annual mean has invariably 

 made P negative, positive values from individual observations are by 

 no means uncommon. 



If we consider that the data in the above table are based each on 

 numerous experiments, taken at a fixed station under the best conditions, 

 we must, I think allow that uncertainty in the P correction is a very 

 probable source of trouble in survey work. 



Criticism of Formula' from Mathematical Standpoint. 



§ 25. If 9 be the angle made by the axis of the collimator magnet 

 with the magnetic meridian at time r, the equation of motion is 



K' c !~l+®6+{mX(l-qt-q't 2 ) + iiX.X}smO = ... (6), 

 dr' 1 



