

ANOMALOUS BEHAVIOUK OF DELICATE BALANCES, ETC. 389 



This device would assist very considerably in overcoming some of the difficulties 

 likely to arise. 



In carrying out their experiments to ascertain whether change in temperature 

 affects weight, POYNTING and PHILLIPS* observed a temporary apparent change in 

 weight equal to 0'034 mgr. per 1 C. This change they ascribed to the effects 

 produced upon the balance by convection currents and radiant energy. 



DIXON and EDGAR! placed their balance in a cellar and allowed half-an-hour to 

 elapse after adjusting the weights before the final weighings were made. 



HiCKsJ has drawn attention to BAILY'S results for the mean density of the Earth, 

 and has shown that if they are arranged in the order of the temperature of the 

 apparatus they decrease in value as the temperature increases. We agree with 

 POYNTING and PHILLIPS, and believe that the more or less uniform variations noticed 

 by HICKS are due to corresponding variations in the temperature of BAILY'S 

 apparatus, and to these alone. 



It is not improbable that the differences in the values obtained for the mean 

 density of the Earth by JOLLY, in Munich, HICHARZ and KRIGAR-MENSEL, || at 

 Spandau, and by POYNTING, in his laboratory at the Mason College, are at least 

 partly attributable to unknown and unexpected changes which took place within 

 some of the balances used by those observers. 



The supposed variations in weight observed by LANDOl/rlf during certain chemical 

 changes may, we think, be almost entirely ascribed to fluctuations occurring within 

 the beam of the instrument used for weighing. 



Preliminary Observations and Methods. 



When two weights are being compared, it is, we believe, usual to note the limits 

 of not more than five excursions of the pointer : three observations being taken on 

 the one, and two on the other side of the resting-point. From the data thus 

 obtained, the value of the resting-point (11. P.) is deduced. 



Now, it would appear to be quite reasonable to suppose that for a given weight, a 

 repetition of the observations would lead to an almost, if not quite, identical value 

 for the R.P. Therefore, with the view of testing the truth of this supposition, an 

 extended series of experiments, with first-grade balances of various types and by 

 different makers, was planned. 



When the beam of a highly finished balance is made to oscillate, the rate of the 



* 'Proc. Roy. Soc.,' 1905, p. 451. 



t 'Phil. Trans.,' A, 1906, p. 185. 



J 'Proc. Camb. Phil. Soc.,' vol. V., p. 156. 



' WIED. Ann.,' XIV., 1881, pp. 331-355. 



|| ' Abhand. der Konigl. Preuss. Akad.,' Berlin, 1898. 



It 'Preuss. Ak. Wiss.,' Berlin, Sitz. Ber. VIII., 1906. 



