450 Messrs. S. E. Sheppard and C. E. K. Mees. [Dec. 20, 



six readings were taken, from 3*000 to O'lOO, the figures are trust- 

 worthy to three units on the third decimal place. 



The next consideration is as to how far D represents the mass of 

 silver per unit area. In treating the image as obeying the absorption 

 law, no account is taken of the light diffusely scattered.* It is true 

 that so long as all densities are measured at the same distance, the 

 scattered light lost is probably proportional to the scattering surface, 

 i.e., to the density. In this case the result does not affect the form 

 of the curve, for D = < (T) = <' (mass of silver), but only the slope, 

 i.e., the photometer-constant. But the error due to this cause with 

 the Hiifner instrument is, for negatives, inappreciable. If proper 

 precautions with regard to obstruction of light by the absorption 

 stand are taken, and adjustment made so that the zero does not alter, 

 it is found that the measured density does not alter on changing the 

 distance of the light source. Further, if a diffusing medium, such as 

 milk-glass, be interposed, the density readings are not altered beyond 

 the probable error. On the other hand, if this be done with an un- 

 developed plate, quite different readings are obtained. When measuring 

 highly diffusing media, such as haloid of silver emulsions, it is necessary 

 to use a highly diffusing light source, and also to utilise only a narrow strip 

 of the transmitted bundle of rays. This may be accomplished by means 

 of the carriers described in our former paper. 



The Photometric Constant. As was stated above, Hurter and Driffield 

 showed that the amount of silver per unit area was directly proportional 

 to the density D = - log T where T is the photometric transparency. 

 This was found for densities from 0-525 to 2'0, the mean value of P 

 being 0*0121, P is the mass of silver in grammes per 100 cms. 2 , 

 corresponding to a density of 1 '0, and is henceforth referred to as the 

 photometric constant. A redetermination of P in 1898 gave P = 0"0131 

 for ferrous oxalate. The photometric constant varies slightly with 

 the developer, especially if organic stain be deposited. Dr. Eder also 

 confirmed this ratio for densities from 0'5 to 2'0, and obtained the 

 value 0-0103 for P (ferrous oxalate). As this relation is of great 

 importance in the theory of development and sensitometry, we made 

 a redetermination, extending the range of densities to 3 '5, as it seemed 

 possible divergencies might occur at higher densities. The photometric 

 constant allows quantities of silver to be estimated much below analy- 

 tical measurement. 



Hurter and Driffield, after the necessary density-measurements, 

 removed the film from the plate and dissolved the silver in nitric acid : 

 it was then precipitated as AgCl, and weighed. Dr. Eder converted 

 the silver directly by means of bichromate and HC1 into AgCl. 



These gravimetric methods are somewhat disadvantageous in dealing 

 with the small quantities of silver present. If plates of small area are 



* Abney, ' Phil. Mag.,' 1875. C. Jones, ' Phot. Journal,' 1898 to 1899, p. 99. 



