33 



exist, one finds that on the average the R periods exceeded the D in pendulums 37 and 38 by 17 x 10~ T 

 and 15 x lO" 7 second respectively, whereas in No. 39 the D period exceeded the R by 9 x 10" 7 . 

 Individual differences, however, fluctuate largely, and corresponding results, based on a greater nurnlMT 

 of observations at Kew, make the difference much less for No. 37, and both numerically loss and of 

 opposite sign for No. 38. It has thus been decided to neglect any difference that may jjossibly havn 

 existed between the D and E positions in dealing with the observations at Melbourne, and those on 

 February 6 at Winter Quarters. 



Tables II., III., and IV. summarise the results of Table I. 



TABLE II. Results at Winter Quarters. 



TABLE III. Results at Winter Quarters. 



