ON THE KEW MAGNETOGRAPHS. 333 



balance-magnets, but in no case that I am aware of, has the variation been 

 to such an extent as is shown above. Observations have been taken of the 

 time of vibration for dififerent inclinations of the magnet, with the view of 

 ascertaining whether this diminution could be owing to variations in the 

 bearing-points of the knife-edges. These differences have in some cases been 

 considerable, but not nearly so great as to account for such excessive changes. 

 I cannot venture as yet to give any opinion as to the probable cause. With 

 the occasionally excellent performance of the magnet (as shown, for example, 

 in the magnetic disturbance of Sept. 3-4) before us, it is difficult to con- 

 ceive that it can be wholly due to imperfection of the knife-edges *. 



II. Tabulation of Numerical Results, &c. 



In preserving a numerical I'ecord of the changes as shown by the mag- 

 netographs, the objects kept in view have been, — 1st, to obtain data for 

 deducing mean results, such as the diurnal changes, daily and monthly means, 

 &c. ; 2nd, to record all the changes which can be said to come under the 

 class of disturbances ; and 3rd, generally, to possess in a numerical form, as 

 far as is practicable, the means of producing the complete curves, either as 

 originally recorded photographically, or in the true form of declination, in- 

 clination, and total magnetic force. 



The positions of all the magnets have been measured for every hour of 

 Greenwich time during which we have had records ; in almost all cases the po- 

 sition for each half-hour has also been noted. Whenever the fluctuation has 

 been at all marked, the turning-points of the fluctuation with the corre- 

 sponding epochs have been measured ; very few motions exceeding two scale 

 divisions will be found omitted. Attention has been paid, as far as possible, 

 to have simultaneous measurements of all the instruments, but especially of 

 the two components of force. Both edges of the photographic trace have 

 always been measured, and the mean of the two entered ; by this means the 

 effects of mechanical oscillation and of variable breadth of the trace from 

 whatever cause, are eliminated. The number of measurements for each in- 

 strument during twenty-four hours has of course varied very much ; it may 

 be said, however, that the lowest number is 48, whilst in some cases of great 

 disturbance as many as 150 measurements have been taken ; the average 

 is probably somewhat more than 60. Any unusual appearance in the curves, 

 such as small and rapid fluctuation with little change of mean position, has 

 been mentioned. The attempt has been made throughout to leave no phae- 

 nomenon of any consequence unrepresented. 



In taking these measurements, the edge of the scale is brought very near 

 to the surface of the plate, in order, as much as possible, to prevent error 

 from parallax ; a compound magnifying lens with a flat field being used in 

 reading off. From my own experience, supported by the opinions of several 

 gentlemen accustomed to observation, I estimate the accuracy with which 

 the better defined traces can be measured at about j-^th of an inch, or one- 

 tenth of a division of the measuring scale. In the case of the declination, 

 the trace not being so distinct, probably j-roth of an inch should be consi- 

 dered as the extent of accuracy. These estimates give for the probable error 

 of a measurement of the declination about 0'*1, and for the horizontal force 

 about 0*00003 of the whole force. The adjustments of the vertical force in- 



* Shortly after the date of this report it was discovered that a spot of rust had formed 

 upon one of the knife-edges : this had not been perceived when the magnet was examined 

 about the beginning of October ; although it may have been already in operation, but to so 

 small an extent as to be imperceptible to the eye. It seems highly probable that a consi- 

 derable share of the irregularities complained of may be ascribed to this cause. 



