252 MASK'S. ];. THRELFALL AX!> .1. A. |f)LI.OCK 



We then began to attempt to compensate the natural increase in stiffness which 

 occurs as the temperature of a quart/ thread rises. It was clear that compensation to 

 the requisite extent could only be attained by the use of some form of compound 

 lever. Professor GURNEY suggested to us to try a lever consisting of three bars. 

 Two of these were to be horizontal and situated at a short distance one above the 

 other in the same vertical plane, they were to be of metals of as dissimilar expansi- 

 bilities as possible, the less expansible one being uppermost. One end of each lever 

 was to be attached rigidly to a cross piece to which the thread could be fastened. At 

 the free ends of the levers the following disposition was to be made : a bar was to be 

 pivoted from the end of the lever of the less expansible metal, and was to bear against 

 the wedge-shaped end of the lever made from the more expansible metal. When the 

 system of two bars was horizontal, the pivoted bar was vertical. As the temperature 

 rose the more expansible bar would deflect the pivoted bar, and so increase the 

 moment round the thread. Many levers were made on this principle, for which 

 purpose we instructed ourselves in the art of the watchmaker, and finally managed 

 to make very small levers, using aluminium and platinum-iridium as the two dissimilar 

 metals. We finally made a compound lever weighing less than a decigram, and 

 having all its dimensions correct to produce the compensation required. After some 

 experimenting with this lever, it was abandoned on account of the inevitable want of 

 permanency of form to which its hinged joint gave rise. 



During the experiments with compound levers an iron box was made which would 

 contain the whole appliance, and which was air-tight and could be exhausted. Most 

 of our earlier trials were made in this box ; it lent itself especially to the investiga- 

 tion of the flotation of the levers by air of different densities. We had hoped to be 

 able to deduce the sensitiveness from the flotation effects, but the cement of uncertain 

 density occupied so large a relative volume in comparison with the lever that these 

 attempts failed. 



By May, 1891, we had sufficient experience to hope to detect the lunar disturbance 

 of gravity. At this time we used a lever of aluminium shaped like a cross, the thread 

 being cemented across the shorter arm by shellac. A mirror was mounted on the 

 lever so as to be vertical when the lever was horizontal ; the mirror was close to the 

 thread, being carried by the head of the cross. The balance, in the air-tight box, 

 was placed in a cellar and supported on a very heavy stand, made by filling a large 

 iron cylinder (a mine case for 500 Ibs. of guncotton) with sand and stones. The 

 temperature was taken by means of a very fine mercury in glass thermometer, the 

 bulb of which was placed inside the iron box. The temperature of the cellar was 

 very uniform, but was disturbed considerably by the presence of the observer. The 

 sensitiveness of the instrument was ascertained by weighting the lever with shreds of 

 fine wire ; the calculation involved a knowledge of the distance of the shred of wire 

 from the thread, a quantity which could not be got with any great accuracy. 



The highest sensitiveness attained was that we got an observable deflection 



