Unit of Candle-power in White Light. 

 Table VII. 



HI 



Observer. 



Date of 

 publica- 

 tion. 



Lamp. 



Value of constants. 



Humidity. 



0057* 

 00000 



0-0071 



0-0058, 



00062, 

 0-0063 



Barometric 

 pressure. 



tfLiebentbal 



1895 

 1904 

 ] 900 

 1910 



1911 



1 c. pentane 

 10 c. „ 



»» M 



0*0005 

 0O008 

 0-0008, 

 0-0006 



0000* 

 0*0008, 



^Paterson 



oDow 



dHos'd. and Crittenden 



eButterfi'dd, JIaldane, and 

 Trotter 



Paterson and Dadding ... 



1014 



a Liebentbal, Zeitechr. fu/r InttrwmentTcumde^ 1905, p. 157. 



6 C. 0. Paterson, Proc. I. K. E. 1907, p 271. 



.]. >.. Dow, 'Electrical Jteview,' Sept.. 28, 1906. 

 cf BOM and Crittenden, Trans. 111. Eng. .Soc, N.Y., 1910, p. 75.*j. 

 e Butterileld, Haldane, and Trotter, 'Journ. Gas Lighting,' 115, p. 290 

 (1911). 



* Based on a normal humidity of 8 litres per cubic metre. 



The tentative figure of 0*006 given by Rosa and Crittenden 

 for the barometric correction mast not, as .stated by them in 

 their paper, be regarded as accurate, and the figures obtained 

 by Liebenthal were for a 1 candle lamp of entirely different 

 construction. Hence, neglecting these, it will be seen that 

 there is general agreement as to the correcting factor to be 

 applied lor barometric changes. There is not, however, the 

 same close agreement between the determinations of the 

 humidity factor. Although the actual candle-power differ- 

 ence- represented by the discrepancies between these different 

 constants are small over the range of humidity met with in 

 this country, the accuracy of most of the determinations 

 would have justified the expectation of closer agreement in 

 the value of this factor. 



The work of Etosa and Crittenden at the Bureau of Stan- 

 dard-, Washington, is of the highest accuracy, and the 

 difference between their value and those determined in this 

 country by the authors and by Messrs. Butterfield, Haldane, 

 and Trotter cannot in either case be ascribed to experimental 

 error. Jn the publication of their work f Rosa and Crit- 

 tenden make two suggestions as to how the difference might 

 be accounted for, hut they had not then before them the 

 confirmatory evidence of the more recent investigations in 

 progress in this country at the time they wrote. 



t Lac. cit. 

 Phil. Ma f/ . S. 6. Vol. 30. No. 175. July 1915. G 



