ON A QUARTZ THi:i:\!> ;l;.\VITV BAI.ANVF. 



corresponding to an increase of gravity by 7 X 10~* of itself. This involved reading 

 ti> one-tenth of a millimetre on a scale distant :! metres from the mirror = 3 seconds 

 of arc. The observations, of course, included a study of the daily rate of the system, 

 as well as of its temperature coefficient. We varied the form of the lever, the 

 dimensions and twist of the thread, &c. Exceedingly good mirrors were made and 

 mounted during this work, and we found that the best cementing material for thin 

 mirrors is a mixture of equal parts of white and red lead made into a paste with a 

 little boiled linseed oil. The mirrors were fastened on with a trace of this cement; 

 they were left for about a day, and then stoved at 100 C. till hard. Every kind of 

 cement deforms thin mirrors, but this paint deforms them less than anything else we 

 tried. A variation was attempted on the following lines. Two threads were 

 mounted, one many times the diameter of the other, the thicker one supports the 

 working lever, the other acting as a torsionless axle. From the end of the working 

 lever, distant, perhaps, 2 centims. from the thread, the finest possible thread also of 

 quartz was stretched to join the second lever mounted on the thin thread, and was 

 cemented to it quite close up to the thread. The second lever carried the mirror, and 

 any motion of the first lever was magnified in the ratio of the distances of attachment 

 of the connecting thread from the two main threads. We only got one-tenth the 

 sensitiveness of our best single system by this arrangement, which was besides very 

 cumbersome and difficult to set up. The finest quartz thread is too stiff to act 

 properly as a flexible connection and acts more like a rigid bar. 



During the latter part of 1891 we made an adjustable system of MICHELSON'S 

 arrangement of interference mirrors, and tried to increase the sensitiveness of our 

 angular measurements by observing the motion of the interference fringes. The 

 general result arrived at after much patient work was that the method presented no 

 advantages in practice ; and this even when we replaced our single mirror by two 

 small mirrors separated as far as the mechanical conditions permitted. 



By March, 1892, we became convinced that it was hopeless to attempt to 

 disentangle the lunar effect from the instrumental irregularities, even if we could 

 bring the sensitiveness up to the necessary point ; of which there seemed to be no 

 hope. The research was therefore abandoned. 



Investigations with the view of constructing our present form of portable 

 instrument were begun in May, 1892. For this purpose we mounted a thread on the 

 spiral head of our milling machine, and supported the fixed end of the thread on the 

 back centre. The twisting of the thread was observed by a mirror and scale, the 

 mirror being connected to the rod carrying the end of the thread to be twisted ; and 

 A lever and mirror were mounted at the centre of the thread as before. The machine 

 was provided with a special slow motion of rotation for the spiral head. The position 

 of the lever was observed by a telescope, using the method of reflected imagi-s. 

 These trials led to the present form of balance, the construction of which was 

 commenced in August, 1892. The instrument was not ready till July, 1893, and the 



