104 
MARVIN. 
face of the globe cannot be accurately measured. The spring- 
balance naturally suggests itself in this connection, but what 
spring can we make that will be sufficiently elastic and con¬ 
stant for this purpose ? The very best steel springs have, in 
seems, proved quite unsatisfactory in every case. The dis¬ 
covery by Mr. Boys * of the remarkable elastic properties of 
fused quartz fibers and the bold and ingenious method he 
invented for making these fibers by shooting arrows from a 
cross-bow with bits of fused quartz attached places physicists 
in possession of what is probably the most perfectly elastic 
material available for the construction of delicate springs. 
Messrs. Threlfall and Pollock have recently published f a 
description of a quartz-thread gravity balance perfected by 
them after experiments extending over several years. The 
writers state, as a result of their investigations thus far, 
that: 
‘‘We have determined the value of ‘gravity’at Hornsby, a station 
21 miles from Sidney and 472 feet above the laboratory, relatively to that 
at Sidney, in three journeys, with a maximum difference less than one 
part in 500,000. For the purposes of a survey the evidence shows that a 
single observation with the balance will enable g to be determined rela¬ 
tively to a standard value to within one part in 100,000. There is a great 
probability, however, that the error would be less than one part in 
200 , 000 .” 
This quotation shows the authors’ faith in their apparatus. 
At the same time, from their discussion of the influence of 
temperature on the quartz thread, it appears that an error 
of 0.1 degree in determining the temperature means an 
error of one part in about 67,000 in the value of g. To attain 
an accuracy of one part in 500,000, or even one part in 
100 ,000, therefore, means a very high order of accuracy in 
temperature measurements. 
The variations of gravity per degree of latitude at the 
equator and poles is about one part in 123,000; at 45° of 
latitude the variation is a little more than one part in 11,000. 
* Phil. Mag., June, 1887, p. 489. 
f Phil. Trans. Roy. Soc., A, 1900, vol. 193, p. 215. 
