16 Lord Rayleigh on Reflexion from Liquid Surfaces 

 of the contamination. The readings then became 



+ 30° 

 -30° 



37, 44, 43 

 17, 14, 12 



It would appear that the first of these were premature, 

 insufficient time having been allowed for the contamination to 

 spread. The difference, reckoned as before, may now be 

 taken to be about — 30', and is in the opposite direction to the 

 small effect of the clean surface. For the contaminated sur- 

 face k= -'0026, and for the clean k= +'00018*. 



Although the above results, and others of a similar nature, 

 obtained both with tap-water and with distilled water, render 

 it practically certain that k is positive for pure water, I do not 

 regard with the same confidence the numerical value above 

 recorded. It is difficult to feel sure that the cleansing was 

 sufficient. A theoretical objection to the hoop method has 

 already been alluded to; and the more perfect methods de- 

 pending upon the use of convection currents! are scarcely 

 applicable here. Attempts were indeed made to work with 

 a surface cleaned by an ascending column of fluid, the column 

 being expanded by heat communicated to it from an immersed 

 platinum spiral, itself warmed by an electric current. But 

 the readings were not accordant; and it appeared that the 

 observations were prejudiced by the deformations of the 

 surface which are the necessary accompaniment of such a 

 flow. Doubts as to the perfection of the cleanliness actually 

 attained lead me to think it possible that the true value of k 

 for an ideal water-surface may be even twice as great as that 

 actually found. 



Opportunities for useful work upon clean surfaces have been 

 very few, for it is hopeless to attempt observations without 

 a prospect of at least an hour's almost uninterrupted sun- 

 shine. But shorter and more uncertain periods may be 

 utilized for observations upon contaminated surfaces, as these 

 do not demand the same care or amount of repetition. As an 

 example of such I will record the readings of June 6, from a 

 water-surface slightly greased with oil of cassia. 



* The observations so far did not of themselves determine which of the 

 two surfaces has the positive k according to Jamin's convention. It was 

 evident, however, that it must be the contaminated and not the clean 

 surface which corresponds to Jamin's determination of a negative k. 

 Subsequent observations upon reflexion from glass verified this as- 

 sumption. 



t See Roy. Soc. Proc, " On the Superficial Viscosity of Water," 

 vol. xlviii. p. 133. 



