ON A QUARTZ THREAD GRAVITY BALANCE. 251 



November 9th . . . . ." 18*3 sextant minutes. ' . 



10th. . . . . 15-9 



December 15th 17'9 



16th 18-3 ,', 



20th 177 "1 Taking the mean 



22nd 18-3 J Hornsby reading. 



We may reject the observation of November 10th. Discrepancies twice as great as 

 this from the mean were discovered afterwards to be due to observing immediately 

 after too rapid temperature changes. 



When we correct for daily rate and take the mean reading at Sydney, the Sydney- 

 Hornsby difference comes out- 

 November 9th. . . . .' . . . . . 18 '5 sextant minutes. 



December 15th 18'1 



21st 18-1 



We have, therefore, determined the value of g at Hornsby relatively to that at 

 Sydney in three journeys, with a maximum difference of 0'4 sextant minute, or to 

 less than 1 part in 500,000 in the value of g. 



This sensitiveness refers, of course, only to the Sydney-Hornsby difference. For 

 the purposes of a survey the mean reading at the standard station may be determined 

 with extreme accuracy with an error at any rate negligible with respect to the 

 probable error of a single observation at any one station. Travelling does not now 

 affect the instrument, so that the accuracy of a determination of g, from a single obser- 

 vation at any station, depends on the possible deviation of a single observation 

 from the mean. 



APPENDIX B. 

 NOTES ON EXPERIMENTS MADE WITH VARIOUS FORMS OF GRAVITY BALANCES. 



In September, 1888, we made calculations as to the sensitiveness of a balance with 

 a horizontal thread twisted several times. We then constructed several trial instru- 

 ments, mounting our threads and appliances on an old watchmaker's lathe-bed. One 

 end of the thread was fastened to a rod working in a barrel against a spiral spring, 

 and the other end was attached to the centre of a bar provided with a large roughly- 

 divided circle. 



A large number of trials were made as to the right thickness of the thread, <fec., 

 and much trouble was experienced in preventing the thread from breaking after 

 having been exposed to twist and tension during several days. The thread and levers 

 were massive compared with those we now employ. One of the earliest methods of 

 cementing employed by us was by means of clean fusible metal. Attempts were also 

 made to grind the thread slightly flat so as to give a sort of key to the cement. 



2 K 2 



