Report of the Gravity Survey Counnittee. 217 



1st. The difficulty of setting the knife-edges accurately in line 

 is entirely got rid of ; for they can be attached, once for all, to 

 their seats in the suspending apparatus, and ground up together 

 in situ. 



2nd. The difficulty, never yet overcome, of setting the knife- 

 edges accurately perpendicular to the axis of the pendulum is 

 (lone away with : for the setting of the agate planes can be tested 

 by optical means to any required order of accuracy, and the head 

 to which the planes are attached can be altered till they attain 

 their proper positions. This fact removes the only possible advan- 

 tage of a flexible pendulum bar. Kater's object in making the bar 

 flexible was to ensure its verticality even if the knife-edges were 

 not quite accurately set ; this advantage has always seemed to me 

 rather a doubtful one,* even with Kater's original pattern, but 

 whether it be so or no, the attachment of the agate planes to the 

 pendulum makes any flexibility in the bar quite superfluous. 



(b) Not merely should the pendulum be invainable, its shape 

 is a matter of importance ; for it must be so designed as to 

 render the resistance of the air to its motion a minimun. This is 

 a serious objection to a form which would otherwise be highly 

 advantageous; Captain Basevif suggested that a rigid cylindrical 

 rod with a spherical bob would be a good form, because for it, 

 alone of all known figures, the pressure connection could be 

 determined directly by calculation; unfortunately this form offers 

 a great deal of air resistance, and such a pendulum would have 

 its oscillations damped too rapidly for convenient observing. A 

 lenticular figure is probably the best, but it is difficult to make 

 symmetrical ; and on the whole, the most convenient seems to be 

 the double-cone pattern of von Sterneck, which can be figured in 

 the lathe with all desired accuracy. 



Another point in connection with the shape is the position to 

 be assigned to the pendulum bob. The usual practice is to set 

 its broadest plane vertical and in the plane of swing ; but von 

 Sterneck puts it horizontal. I believe this was done by him for con- 

 structional reasons, but a further advantage is incidentally secured 

 in this way. It is well known that a flat body moving in a fluid 



Sir G. G. Stokes however thinks diffierejitlj'. Vide appendi 

 t G. T. Survey of India, vol. v., p. 92. 



