GRAVITY AT MELBOURNE. 



Only at Melbourne has the value of " g " been accurately observed, the latest 

 values referred to Potsdam being : 



980 -003 0-0014 Hecker, 1904. 



980-003 Alessio, 1905. 



It will be observed that our final value at Melbourne differs from that obtained 

 by Hecker and that the gap between the two values is greater than the sum of the 

 probable errors in the two cases. 



The values obtained by Hecker with individual pendulums were 980-003, 980-005, 

 980-004, 980-001 and 980-003, the average difference for five pendulums, between 

 Potsdam I and Potsdam II, being 8 X 10 ~ 7 sec., compared with 9xlO~ 7 sec. in our 

 observations. 



Even if our Potsdam I observations are neglected, so that the final value is alone 

 used for comparison with the Melbourne figures, the gap between the two observations 

 for "g" at Melbourne remains 0-0045. The consistency of the values obtained by 

 Hecker with his five pendulums seems convincing evidence that his value is unlikely 

 to be wrong by more than 0-0015, but there is equally no reason to expect an error 

 of more than 0-002 in our final value. Moreover, there is a gap of 0-002 between 

 Hecker's maximum value with any pendulum and our minimum. 



So long, therefore, as one assumes that (1) the correct time of swing at Potsdam 

 lies between the extremes of Potsdam I and Potsdam II, and (2) the value of " g " at 

 Potsdam relative to that at Melbourne remains unchanged in course of time, one finds 

 it difficult to reconcile the values obtained at Melbourne in the different years. The 

 discrepancy is not a large one, but is sufficient in magnitude to raise a doubt as to 

 the accuracy of the assumptions. Is, in fact, the ratio " g " (Potsdam) to " g " 

 (Melbourne) the same at all times, or is there some other source of error, which has 

 not been evaluated and which is comparable in magnitude with the calculated probable 

 error* ? 



GRAVITY AT CAPE EVANS. 



The observations at Cape Evans are closely comparable with the earlier obser- 

 vations by Bernacchi at the "Discovery" Winter Quarters, 15 miles further south. 

 These earlier observations were not very accordant, and the difficulties connected with 

 clock rate were even greater than those experienced at our own Winter Quarters. 

 .Still, the adopted mean of his observations gave the value "g" = 982 '985, which is 



* It is important to note that the probable errors calculated from the small number of observations 

 available, can have little significance other than that of a rough measure of the consistency of the 

 observations. 



92 



