ON A QUARTZ THREAD GRAVITY BALANCE. 247 



its reading on the 1 9th. It was considered that the breaks in the readings when the 

 instrument was " travelled " were due to a looseness in the joint fixing the lever to the 

 thread, which might have been caused by the accident after coining from Armidale, 

 .UK! observations were discontinued until the joint could be re-made. 



In 1'Vhriiary, 1898, the bars carrying the thread were taken out of the outer case 

 and the lever was re-soldered to the thread. This necessitated the little lump of 

 fusible metal at the end of the lever being altered in order to keep the twist in the 

 thread the same as before. In putting the instrument together again the further end 

 of the irftenial framework was wedged to the copper tube. Previously permanency 

 of relative position between the microscope and the thread depended on the attach- 

 ment of the inner framework to the copper tube by the screws at the circle end only. 



Other work did not permit observations to be commenced until the end of Sep- 

 tember. The temperature coefficient was now 267 sextant minutes per degree 

 centigrade. Plot 5 shows that the daily rate due to elastic after-working had 

 decreased from 2'5 minutes in September, 1897, to 0*3 minute in September, 1898. 

 (The present thread was mounted in September, 1896.) Satisfactory observations 

 were made in Melbourne on October 8th, 9th, and 10th. The observations on our 

 return to Sydney show a difference of 5 '5 minutes compared to the ones taken before 

 going to Melbourne, but the change, instead of being permanent as it was in 1897, 

 completely came out by the 17th. 



The journey to Melbourne was undertaken with a view to finding the sensitiveness 

 of the instrument. The rate, as given by the observations at Melbourne, which were 

 completely satisfactory, is the same as that at Sydney. We may suppose it likely 

 then that the change in the Sydney reading is due to something which occurred on 

 the return journey. The difference between the readings at Sydney and at Melbourne, 

 taken from the plot, is 63 sextant minutes. 



If three whole turns in the thread, or 2160 (sextant degrees), are required to keep 

 the lever horizontal against a g of 980, assuming that the relation between the 

 necessary twist and g is a linear one, in spite of the accumulated after-working, 

 72 sextant minutes would be required to compensate for a change of g of 0'277, 

 which is the difference betweeu g at Melbourne and Sydney, as given by Mr. LOVE.* 



We have then for the difference of reading between Sydney and Melbourne for the 

 instrument, as at present constructed, calculated from the number of twists in the 

 thread, known only approximately, 72 sextant minutes. 



* Mr. E. F. J. LOVK, who has given attention to the determination of gravity at Sydney and Melbourne, 

 both experimentally and by discussing the results by other observers, was good enough to give us the 

 following values of <j as the most probable : 



Values of gravity at Melbourne Observatory 979-916 



Values of gravity at Sydney Observatory 979-639 



1 >itrerenco 0'277 



