On a Quartz-thread Gravity Balance. 



125 



nation of gravity of the various observational errors which are 

 possible. The instrument of course only refers differences of gravity 

 to a known difference. The results are expressed in round numbers, 

 gravity being taken at its value in the latitude of Sydney. 



Our temperature observations may be inconsistent by at most one- 

 hundredth of a centigrade degree ; this would correspond to an uncer- 

 tainty of one part in 700,000 in the value of " g." 



The accuracy with which the microscope can be set on the lever is 

 much greater than the accuracy of reading the sextant arc. If our 

 estimate of the latter is wrong by 5" the resulting value of "g " is 

 affected to the extent of one part in 1,300,000. 



The errors of levelling may amount to one part in 700,000 in the 

 value of " g." This gives a possible maximum uncertainty of one part 

 in 300,000 in the value of " g." The daily rate of the instrument does 

 not introduce an uncertainty of anything like the amounts mentioned 

 above, and can in any case be eliminated by observing alternately at 

 two stations, the difference in the value of gravity between them being 

 the subject of observation. 



Observations. — Two observers are required, one for the balance and 

 one at the thermometer resistance box. 



It is only possible to observe with sufficient accuracy when the tem- 

 perature is nearly steady ; we always observe therefore at a time when 

 the temperature passes through a maximum or a minimum value. 

 With the instrument as constructed of various metals it is also neces- 

 sary to avoid observing too soon after any great and sudden variation 

 of temperature. 



Journeys. — We have travelled with the balance from Sydney to Mel- 

 bourne by train, from Melbourne to Hobart by steamer, from Hobart 

 to Launceston (in Tasmania) by train, back to Melbourne by steamer, 

 and to Sydney by train. We have also made many less extended 

 journeys in Xew South Wales, having travelled over more than 6000 

 miles with the instrument. 



Most of these journeys led to our making improvements in the 

 instrument, and therefore are not to be regarded as forming surveys. 



If, however, a consistency of one part in 50,000 in the value of " g " 

 be considered satisfactory, then the Tasmanian stations may be consi- 

 dered as surveyed, and the values assigned to gravity at these stations 

 to be referred to the Melbourne-Sydney difference. Since this journey 

 was undertaken the instrument has been so much improved in detail 

 that we do not discuss its results from a gravitational point of view. 



We have, however, made three test journeys between . Sydney and 

 Hornsby in New South Wales under proper conditions, and the result 

 of these observations shows that at Hornsby the thread has to be 

 untwisted at one end by the following amounts as referred to the 

 reading at Sydney : — 



