534 PROFESSOR W. E. AYRTON, MR. T. MATHER AND MR. F. E. SMITH : 



to be 26xlO~ 7 second, the period at Teddington being the greater. Excluding 

 observations made over 30 years ago, only two comparisons have been made 

 interconnecting Kew with a station where g is believed to be known in absolute 

 measure. The first of these comparisons was made by VON STERNECK in 1893, and 

 the second by Mr. G. R PUTNAM (U.S. Coast and Geodetic Survey) in 1900. The 

 former of these observers assigned the value 981 '160 to Kew and 981 '200 to Greenwich ; 

 Mr. PUTNAM'S values are 98T199 and 98T187 respectively. It will be observed that 

 VON STERNECK makes the value at Kew less than that at Greenwich, but all other 

 observers make it greater.* Also the differences found between Kew and Greenwich 

 by the latest and most complete observations (those by PUTNAM and by BURRAND, 

 CONSTABLE and LENOX-CONYNGHAM) are very close to that given by theory. 

 VON STERNECK observed on only two days at Kew as against six at Greenwich ; thus 

 the probabilities of serious error are much greater for Kew than for Greenwich. 

 VON STERNECK'S value for Greenwich exceeds PUTNAM'S by 0*013, but this, if we may 

 judge from the difference 0'019 between their values for Potsdam, represents largely 

 a difference in what answers to their base values. HELMERT has accepted for Kew 

 the value 981 '200, t and it appears that no serious error is introduced by our 

 acceptance of this value. From Mr. CONSTABLE'S observations the value of g at the 

 National Physical Laboratory would therefore appear to be 981"19 centims./sec 2 . 



The theoretical difference between Kew and Teddiugtou may be obtained from 

 VON HELMERT'S formula. The places are very similarly situated with respect to 

 surface strata and surroundings, and the only corrections it is necessary to apply are 

 those for difference of latitude and difference of level. The latitude of Kew is 

 51 28' 6", and of the National Physical Laboratory 51 25' 20" approximately; 

 the level of PUTNAM'S observations at Kew was 17 feet above mean sea-level, and at 

 Teddington the mean level of -Mr. CONSTABLE'S observations was about 34 feet. The 

 correction for difference of latitude is 0'0044, and for the difference of level it is 

 0-0010 ; the theoretical value is therefore 981'19 6 if Kew is 981'20 . The probable 

 error of any accepted value depends, of course, on the errors of the intercomparisons 

 and on the error of the absolute determination at the base station. It appears that 

 these are not very large, and that we may accept the value 981'19 ceutims./sec 2 as 

 correct to 3 in 100,000. 



The determination of current by means of the ampere balance is therefore subject 

 to errors of the following magnitude : 



(1) Due to uncertainty of dimensions of coils : possible error about 0'001 per cent. 



(2) Due to uncertainty in the value of g : possible error about 0'001 5 per cent. 

 All the other sources of error introduce uncertainties less than 0'001 per cent., 



and may be disregarded. The total error of an estimation is therefore of the order 

 0-002 per cent., or 2 in 100,000. 



* See G. P. LENOX-CONYNGHAM, ' Roy. Soc. Proc.,' A, vol. 78, p. 246, 1906. 

 t 'Report, Geodetic Conference of 1900,' p. 321. 



