MESSRS. R. THRELFALL AND J. A. POLLOCK 



From the unsatisfactory comparison in June, 1897, remembering that the sensi- 

 tiveness is doubled since that date, 68 sextant minutes. 



From the comparison in October, 1898, 63 sextant minutes. 



Until the difference has been more accurately determined, we may take say 

 60 sextant minutes as the difference of readings at Sydney and at Melbourne. 



Mr. LOVE considers that the value of g at Melbourne is 979 '9 16, and that the 

 Sydney-Melbourne difference is 0'277. If the change in the reading of our instrument 

 between Sydney and Melbourne is 60 sextant minutes, then a change in the value 

 of g of 1 part in 100,000 would be represented by a change of reading of 2'12 sextant 

 minutes. 



There is now ample evidence before the reader in the plots of the observations 

 taken in November, 1897, and from the 2nd to the 18th October, 1898, to show 



Plot 6. 



Day of month 21 2 23 4 ZS 26 27 ZB 23 



October 



Readings reduced to 21" C. 



that, apart from discontinuities due to travelling, it is extremely unlikely that 

 any single observation would differ from the mean reading at any one station by 

 1 sextant minute. 



There is only one observation, that on the afternoon of October 3rd, 1898, which 

 differs from the mean Sydney reading by 2 sextant minutes. If, therefore, the 

 instrument can be carried from place to place without altering its behaviour, we may 

 say that the value of g at any station may be determined relatively to that at some 

 standard station by a single observation, with extreme probability to 1 part in 

 200,000, and with certainty to 1 part in 100,000. 



We had, therefore, at this time to determine the cause of the effect produced by 

 travelling the instrument. We determined to take the balance to Hornsby Junction, 

 a station 21 miles north of Sydney, 592 feet above sea level. 



The observations at Hornsby were made at night in the lamp room of the station, 



