Surface Separation from Solutions o; Saponin, fyc. 561 



solutions, such an obviously fundamental research as the 

 above has never yet been carried out. The chief difficulty 

 has been the lack of any suitable means of quantitative 

 investigation of the excessively thin and fragile surface-film. 

 Methods similar to that of Plateau*, involving a sweeping 

 up of the surface-film, interfere too much with the process 

 under investigation. The same objection applies to the 

 torsion-balance method of measuring the shearing stress 

 necessary to break the filmf. The method of measuring the 

 surface rigidity statically by means of the torsion-balance £,. 

 though not open to this objection, is not suited to the solu- 

 tions under investigation. In the case of fairly fresh surfaces, 

 the elastic forces called into play by the shearing of the 

 film exhibit the phenomenon of elastic relaxation, so that, 

 it is impossible to get any satisfactory balance between 

 these forces and the elastic forces of the torsion-balance 

 wire. In the case of older surfaces the method is unsuitable 

 owing to the smallness of the elastic limit of the surface 

 pellicle. 



The method adopted in the present work is one which I 

 devised in a previous work on the surface-elasticity of 

 saponin solutions § . I found that the surface-film, apparently 

 inelastic if tested in the ordinary way by means of the 

 torsion-balance (i. e. by twisting the torsion-head so as to 

 impress an initial torsion on the wire), exhibits an elastic 

 effect if the motion is started by rotating the vessel con- 

 taining the solution (i. e. by impressing an initial shear on the 

 surface- film). It is possible in this way to make the disk of 

 the torsion-balance execute torsional oscillations under the 

 action of the surface-film. These oscillations can be obtained 

 even when the surface is so fresh that the surface-film yields 

 rapidly under the action of the wire. In this case the film 

 is viscous, i. e. it shows elastic relaxation, but the relaxation 

 time is sufficiently great for the film to exhibit an elastic 

 effect when the vessel is rotated, though when the torsion- 

 head is twisted (and an approximately constant stress applied) 

 the film behaves like a viscous liquid, such as castor oil, 

 which, according to Natanson||, has a relaxation time of 

 about # 001 sec, and whose behaviour, therefore, approximates 

 to that of an ideal viscous medium for all ordinary motions. 



* Statique cles Liquidcs, Oberbeck, Wied. Ann. xi. p. 634 (I860), 

 t Schiitt, Ann. der Phys. xiii. p. 714 (1904) ; Rohde, Ann. der Phys. 

 xix. p. 935 (1906). 

 X Schiitt, he. cit. 

 § Phil. Mag. Feb. 1906. 

 j| Phil. Mag. Nov. 1901, p. 469. 



Phil. Mag. S. 6. Vol. 17. No. 100. April 1909. 2 Q 



