16 Proceedings of tJi: Royal Society of Victoria. 



from scale or object glass to mirror, 167"0 in. Distance from 



plane of spike to plane of support of cylinder, 43-0 in. An 



oscillation in the image of amplitude 0*1 scale division (001 in.) 



would accordingly correspond to an angular displacement in the 



cylinder expressed by 



0-01 1 0-75 648000 ^ ,^,, , „ 



-— X - X ^ ,^ X = O'iOo seconds or arc. 



167 2 43 TT 



To see whether the apparatus responded easily to small 

 disturbances the following experiments were carried out : — 



(a). To test the eftect of unsymmetrical vertical thrust. 



A brass cylinder weighing 13 oz. was placed in a vertical 

 position on the ledges of the north and south windows of the 

 cylinder alternately ; the readings obtained were 



No Load. Load on S. window. Load on N. window. 



21-8 



21-2 



21-8 



21' 



21-2 



oo-l 



[Note. — The pendulum cylinder is fully 100 times the mass of 

 the small brass cylinder employed in this experiment.] 



Hence the mean displacement on loading the S. window was 

 0-6, while the mean displacement on loading the N. window 

 was 0'4; or, in other words, 1}-, oz. on the S. window, or 3 oz. on 

 the N. window could be detected. As the line joining two of 

 the three supporting screws of the cylinder ran east and west, 

 and to the south of the third screw, this difference might be 

 reasonably expected ; for the load on tlie south window would 

 tend to relax the pressure on the third levelling screw by 

 rotating the whole cylinder about the E. and W. line through 

 the other two. 



( b) To test the effect of horizontal thrust. 



The piece of apparatus shown in Figs. (6) and (7) was con- 

 structed for this purpose'^' ; abc is a brass plate bent nearly at a 



« Fig. (6) is vertical, Fig. (7) borizontal. 



