ON THE ELLIPTICITY OF THE EARTH. 459 



When the object of experiment is simply the de- 

 termination of the ellipticity, the invariable pendulum 

 unquestionably presents the most direct and the 

 most safe method of ascertaining the acceleration in 

 different latitudes, from which the ellipticity is an im- 

 mediate deduction. It is far preferable to an inde- 

 pendent determination, at each station of experiment, 

 of the absolute length of the seconds pendulum, either 

 by Borda's method, or by Kater's convertible pen- 

 dulum; and still more to an endeavour to combine 

 results at different stations which have been arrived at 

 by the employment of dissimilar apparatus and pro- 

 cesses. The determination of the absolute length de- 

 mands a far more elaborate and delicate process of 

 experiment than is required simply for the acceleration ; 

 and, since Bessel's remarkable discovery in 1828 

 (Kosmos, Bd. iv. Anm. 20), it has been known that one 

 part of the process, whereby the absolute lengths were 

 supposed to be determined before that date, was founded 

 on an erroneous theoretical supposition ; and hence has 

 followed the necessity, wholly unforeseen and unsus- 

 pected by the eminent persons who devised the methods 

 or who employed them previous to 1828, of experi- 

 mentally investigating and applying corrections by no 

 means insignificant in amount, and affecting different 

 forms of apparatus in very different proportions. 



Although it is obvious on a very slight consideration, 

 that the corrections required by Bessel's discovery to be 

 applied to the acceleration, derived from experiments 

 with an invariable pendulum in which the same form of 

 pendulum had been employed throughout, must be very 

 small in comparison with those required for the absolute 



