by the Red Corpuscles of the Blood. 669 



Both equations (10) now reduce to 



^=cos^(*T)-sin^|^T) + OT . . (11) 



We are now in a position to estimate c. For if we assumed 

 that the thickness, 2y, is everywhere greater than the 

 " critical " value at which T began to vary, then for such a 

 condition T would be constant and the drop would take a 

 spherical form, of radius r, say. Calling this constant value 

 of the surface-tension, T , and noticing that in such a 

 case p and yjr would have negative values, and thus 



/3= — r 



. . x 



sin ylr= 



r 



| y (.,,T) = 0, 



we should obtain 



r r 



i 2T 



and .*. c= 



r 



So equation (11) becomes 



f = C03 ir ^ (*T) - sin f|j (*T) - ^ . ,-, . (12) 



which is the equation connecting T with the form of the 

 curve. 



Before one can proceed further one must be acquainted 

 with the relation between T and y for minute values of //. 



Now all experiments indicate a decrease in T with the 

 thickness, as would be expected from the ordinary Laplace- 

 Gauss analysis of surface-tension phenomena. Riicker and 

 Remold (Phil. Trans, clxxvii. part 2, 1886) and Johannot 

 (Phil. Mag. xlvii. 1890) have shown that for soap solutions 

 the decrease is not indeed uniform, but that a decrease 

 certainly does begin somewhere about a thickness "1 yu., 

 continues for some considerable space, and then an increase 

 begins to take place, a maximum value being reached some- 

 where about a thickness '01 /jl. However, it is not necessary 

 for the purpose of our problem to assume such minute sizes 

 as the latter for the thickness of our biconcave disk cor- 

 puscles at their centre. If as a first approximation we make 



