254 MESSRS. R. THRELFALL AND J. A. POLLOCK 



first thread was not mounted till September, 1893. This was a catapult -thread, and 

 was twisted with 2| turns. The aluminium lever carried a mirror, and this was 

 observed by reflexion of cross wires as before. At first the thread was intended to 

 be twisted from one end only to compensate changes of gravity exactly as in our 

 present form of instrument but as irregularities occurred we were led to put in an 

 extra set of bars so as to twist both ends of the thread at once. We had also an 

 idea that the theory would be sufficiently simplified to enable us to avoid the 

 necessity for any calibration, the total twist being capable of exact estimation in 

 consequence of the presence of a theodolite circle, which at that time occupied the 

 place since taken by the sextant arc. We also hoped that an exactly symmetrical 

 distribution of twist in the thread would reduce the viscous yielding, and so more 

 than compensate for the loss of sensitiveness due to twisting two ends of the thread 

 instead of one. 



The lever was adjusted by drops of paraffin, which are easier to regulate than 

 drops of solder. The whole structure of the instrument was mounted on a turntable 

 so as to allow us to take observations in any azimuth, for the purpose of eliminating 

 magnetic effects if such should appear. In October, 1893, this machine was taken to 

 Armidale, with the result mentioned at the commencement of the paper. The 

 temperature was taken at this time by a mercury-in-glass thermometer. 



During the time that the instrument was undergoing repair and alteration, in 

 accordance with the experience obtained, we constructed a fresh instrument in which 

 the whole of the working parts were immersed in mineral sperm oil. The machine 

 was only intended as a trial instrument, and was put together out of brass tube, &c., 

 but it was well made. We observed it in the cellar with the appliances formerly 

 used in the attempt to discover the lunar disturbance of gravity. It was found that 

 the readings went through a regular daily cycle which was ultimately traced down to 

 the action of minute convection currents. The regularity of these minute currents 

 was one of the most surprising things we have ever met with. In order to get an 

 idea as to whether it would be possible to observe at sea, we mounted the machine on 

 a swing, and satisfied ourselves that no amount of damping would enable accurate 

 observations to be made under such circumstances. A subsidiary set of experiments 

 were made in connection with this matter in order to find the resisting properties of 

 different cements when immersed in oils, i.e., to find out whether they were gradually 

 softened by the oils. These experiments lasted for two years, and showed us that the 

 resistance of shellac is surprisingly great. Mineral oils did not appear to have any 

 influence at all, and turpenes only a very slight effect, the other cements tried were 

 not nearly so resisting. 



Experiments of one sort or another with the oil balance were continued till 

 January, 1894. 



Meanwhile the portable instrument had been restored, and in November, 1893, the 

 catapult thread was replaced by a shot thread, this was under observation during the 



