98 REPORT — l88i. 



Altliotlgli we had thus learnt that changes of temperature formed 

 the great obstacle in the way of success, thei'c were a good many things 

 to be learnt from the instrument as it existed at that time. 



After the box and pipes had been filled for some days the plate-glass 

 front cracked quite across, and a slow leakage began to take place ; we 

 were thus compelled to dismount tlic whole appai-atus and to make a 

 fresh start. 



It is obvious that to detect and measure displacements of the pendulum 

 in the N. and S. direction, the azimuth of the silks by which the mirror 

 is suspended must be E. and W., and that although any E. and W. dis- 

 placement of the pendulum will be invisible, still such displacement will 

 alter the sensitiveness of the instrument for the N. and S. disiDlacements. 

 In order to obviate this we determined to constrain the pendulum to 

 move only iu the N. and S. azimuth. 



Accordingly wc had a T-piece about 4 inches long fixed to the end of 

 the iron rod from which the pendulum hung. The two ends of a fine 

 copper wire were soldered into the ends of the T-picce ; a long loop of 

 wire was thus formed. The square-headed plug at the top of the 

 pendulum-bob was replaced by another containing a small copper wheel, 

 which could revolve about a horizontal axis. The bearings of the wheel 

 were open on one side. 



When the wheel was placed to ride on the bottom of the wire loop, 

 and the pendulum-bob hooked on to the axle of the wheel by the open 

 bearings, we had our pendulum hanging by a bifilar suspension. The 

 motion of the pendulum was thus constrained to take place onlj'^ perpen- 

 dicular to the plane of the wire loop. 



The iron tripod was replaced by a slate slab large enough to entirely 

 cover the hole in the lintel of the gallows. Through the centre of the 

 slab was a round hole, of about one inch in diameter, through which 

 passed the ii'on rod with the T-jjiece at the lower end. The iron rod was 

 supported on the slate by means of the flanged nut above referred to. 

 There was also a straight slot, cut quite through the slab, running from 

 the central hole to the margin. The purpose of this slot will be explained 

 presently. 



In the preceding experiment we had no means of determining the 

 absolute amount of displacement of the pendulum, although, of course, 

 we knew that it must be very small. There are two methods by which 

 the absolute displacements are determinable ; one is to cause known small 

 displacements to the pendulum and to watch the effect on the mirror ; 

 and the second is to cause known small horizontal forces to act on the 

 pendulum. We have hitherto only employed the latter method, but we 

 are rather inclined to think that the former may give better results. 



The following- plan for producing small known horizontal forces was 

 suggested by my Ijrother. 



Suppose there be a very large and a very small pendulum hanging by 

 wires of equal length from neighbouring points in the same horizon ; and 

 suppose the large and the small pendulum to be joined by a fibre which 

 is a very little shorter than the distance between the points of suspension. 

 Then each pendulum is obviously deflected a little from the vertical, but 

 the deflection of the small pendulum varies as the mass of the larger, and 

 that of the larger as the mass of the smaller. If m be the mass of the 

 small pendulum, and If of the large one, and if a be the distance between 

 the points of suspension, then it may be easily shown that if a be in- 



