Observations ivith Kater's Pendulinns. 13 



insular situation raises it still more : consequently M - S should 

 be smaller, G - S greater, and G- M still greater than the 

 calculated value, as is the case. 



If, however, we take the U.S. Coast Survey result, all three 

 diflerences fall below the calculated value ; M-S in particular 

 becomes absurdly small, unless we are prepared to assert that 

 g has an abnormally high value at Sydney, owing to local 

 peculiarites which cannot well be allowed for. Such an assertion 

 appears to me to be quite unwarrantable in view of the agree- 

 ment between the recent determinations of Elblein and myself ; 

 and I am in consequence reluctantly compelled to assume that 

 the U.S. Coast Survey determination must not be employed 

 differentially in connection with recent observations with the 

 Kater pendulums. 



"What the source of the discrepancy may be it is not so easy to 

 determine. The following suggestions may be made : — 



(a) That some change has taken place in all the pendulums 

 since 1883. 



(b) That some error has crept into the reduction of the 

 American observations. 



(c) Errors of observing. 



Of these (c) is practically out of the question, as all three pen- 

 dulums give nearly the same result ; besides the known skill 

 and ability of the American observers would negative such a 

 hypothesis. 



(a) is inadmissible ; for Herschel's measurements with these 

 pendulums at Kew and Greenwich in 1882 agree very well with 

 those made by Hollis and Constable in 1889, consequently no 

 such change can have occurred during that interval ; while the 

 values of G - M, M-S, and G - S, as recently determined, 

 negative the supposition of any serious change between 1889 and 

 1894. 



(b) seems the only possible solution ; and it is noteworthy that 

 the error, whatever it be, affects all the pendulums alike. The 

 possible soui'ces of error which could give such a result are limited 

 in number ; the only one which suggests itself to me is a syste- 

 matic change in the sign of the clock rate. Whether anything of 

 this kind has occurred I cannot tell, as I have of course been 

 unable to refer to the original notes of the American observers, 

 which alone could be relied on to settle the matter. 



