34 



TABLE IV. Results at Melbourne and Christchurch. 



6. The periods observed at Kew in 1901 and 1904 have been already given. 

 uncorrected for flexure, are 



Their mean values, 



Pendulum 

 Period . 



37 



0-5087776 



38 

 0-5087732 



39 

 0-5088064 



It has been judged best to employ these mean Kew values for comparison with the results obtained at 

 Winter Quarters and Christchurch, but to employ only the 1901 results for comparison with Melbourne. 



The observations made at Christchureh in 1904 gave lower values than those obtained in 1901, the 

 differences in the seventh place of decimals being - 20 for No. 37, - 61 for No. 38, and - 138 for No. 39. 

 The differences for Nos. 37 and 38 give a mean which is closely similar to the corresponding mean 

 difference observed at Kew, thus suggesting that any change that took place in these two pendulums 

 occurred at Winter Quarters, and so influenced the Kew and Christchurch observations alike, leaving the 

 Melbourne observations unafl'ected. The comparative brevity of the interval between the observations 

 made at Kew and Melbourne in 1901 is an argument pointing in the same direction. 



7. If /i and (> denote the periods of a pendulum at two places where 171 and g<> are the values of gravity, 

 then, assuming the pendulums unchanged, and the conditions as to temperature, pressure, &c., the' same at 

 the two places, we have 



<7i<i 2 = g-J-2-, 



or 



Accepting 981-200 (centimetre/second 2 ) as the value at Kew,* the values deduced by the above formula 

 for Melbourne, Christchurch, and Winter Quarters are those given in Table V. under the heading 

 "Observed values." In the probable mean the results from pendulum No. 39 have been allowed only half 

 weight as compared to those from either 37 or 38. 



TABLE V. 



* ' Roy. Soe. Proc.,' A, vol. 78, 19('fi, ],. 245, 



