218 Proceedings of t lie Royal Society of Victoria. 



tends to set itself broadside on to the direction of motion; if then 

 tlie bob be vertical and its broadest plane parallel to the plane of 

 motion, it is always trying to rotate about its vertical axis, and 

 in this way either the knife-edges will wabble on the suspension, 

 or (if the pendulum be too heavy for that) a torsional stress is 

 applied to the knife-edge, the direction of the stress being reversed 

 twice in each oscillation : of course this stress is only small, but 

 just as constant dropping wears away stones, so a constantly 

 reversed stress wears out a knife-edge. The case is very different 

 if the broadest plane of the bob be horizontal ; here the only effect 

 is a bending stress on the pendulum rod, a stress too small to 

 produce any effect on a I'od of ordinary dimensions. If we bear 

 in mind that von Sterneck's rod is about \ of an inch thick, 

 but that a decent steel knife-edge is not xoo^.o o o •^^ ^^^ viici\\. across, 

 the advantage of transferring the stress from the knife-edge to 

 the rod is at once evident. 



Another point to be attended to is the position at which the 

 starting lever gives its impulse to the rod. If this be not — as it 

 generally is not — at or near the centre of oscillation, the operation 

 of starting tends to produce a sideways shift of the head of the 

 pendulum, and so to bend over and dull the knife-edge. Now to 

 determine the dimensions of a pendulum of given form which 

 shall have its centi-e of oscillation at a given point is a matter for 

 calculation ; the calculations are rather complicated, but quite 

 manageable by known mathematical methods, so need not be 

 detailed here. 



(c) The supports of the pendulum must of course be as friction- 

 less as possible. This is secured by attending to the construction 

 of the knife-edges and planes of suspension ; and here we have 

 the advantage of the experience gained by the manufacturers of 

 chemical balance5, among whom the general consensus of opinion 

 seems to be that a steel knife-edge and agate planes affords less 

 friction than any other combination. Agate knife-edges cannot 

 be given so fine an edge as steel, and the siniilarity of the material 

 of planes and knife-edges is a further objection, and although 

 there is of course no risk of rust, there is SDine danger of splitting 

 the knife-edge ; the advantage here is on the side of the steel, for 

 a rusty steel knife-edge can be reground, but a split agate is of 

 no fui'ther use and must be replaced by a new one. 



