608 The Earl of Berkeley and Dr. C. V. Burton on 



equilibrium state, while the authors report a difference of 

 20 per cent, in four hours. The discrepancy is so great that we 

 cannot but conclude that there has been some important 

 oversight in the conduct of the experiments or in the presen- 

 tation of the results. This conclusion is confirmed when we 

 consider that the forces required to cause such a rapid 

 migration of the solute through the solution, even with no 

 opposing forces except those arising from internal viscosity, 

 would be enormously greater than the available driving 

 forces due to the rotation. 



Similar remarks apply to the other cases described by 

 Messrs. van Calcar and de Bruyn. 



We have also endeavoured to examine experimentally the 

 centrifugalization of solutions. Numerous trials have been 

 made, but the difficulties have proved considerable, though 

 they now appear to be largely overcome, and we hope shortly 

 to obtain more reliable results with an improved centrifuge. 



The solutions we have chiefly examined are potassium 

 iodide, iodine in potassium iodide, and a solution of about 51 

 per cent, by weight of caesium chloride. 



A brief description of the best experiment so far made may 

 not be out of place. 



The solution of the ca?sium chloride was contained in six 

 sealed glass tubes arranged radially in a massive flat circular 

 box of gun-metal ; the length of the solution column in each 

 tube was 9'3 cm. and the run was continued (at about 57*6 

 revs, per sec.) for 243*3 hours, a time long enough for a fair 

 proportion of the maximum degree of separation to be effected. 

 We had satisfied ourselves by previous tests that the internal 

 diameter of the tubes (0*34 cm.) was sufficiently small to 

 prevent any sensible re-mixing of the liquid strata during the 

 very slow stopping of the centrifuge, or in the succeeding; 

 Eew seconds which elapsed before the tubes could be stood 

 vertical. On the other hand, it was impossible to pipette the 

 contents of each tube into three samples without some slight 

 mixing taking place ; and it is partly on this account and 

 partly because the centrifuge did not revolve perfectly 

 smoothly (thus causing some slight stirring action), that we 

 regard our results as only qualitative. 



The following table gives the results : — 



%CsCl difference % CsCl difference 

 (found). (theoretical) (theoretical) 



Sample 1 51-648 . nr , 51*645 ~ RA 



9 51-709 ° 61 51-709 064 



„ 3.. 51-742 U66 51-812 1UJ 



