Measuring the Surface- Tension of Liquids. 413 



In these cases there can be no doubt that the surface was 

 as pure as the interior of the water. Small bits of cork and 

 dust were sprinkled on the surface after the last measurement, 

 as a test, and the surface was swept clean in a few seconds, 

 all visible particles (except the largest) being carried of! in the 

 overflow. In finding the maximum weight, in no case was 

 an observation accepted as final when there was the least 

 visible oscillation of the balance or variation of the pointer 

 from zero just before the final overturn. 



With the same arrangements for overflow, except that the 

 glass reservoir was replaced by a large galvanized iron tank 

 and that the connecting-tube was in two pieces united by a 

 new perforated cork, a large number of measurements were 

 made by thin plates at various times on tap-water. The 

 highest value in each set of observations is given below. In 

 each case very great care was taken to secure a zero contact- 

 angle. 



Plate. 



Weight (grm.). 



Temperature. 



T (dynes). 



T at 18° O. 



Platinum 



Silver 



1-9735 

 1-1383 

 1-8421 

 1-1358 

 1-9904 

 0744 



18-95 

 19-01 

 19-85 

 20-08 

 20-16 

 20-50 



7219 

 73-22 



73-29 

 73-06 

 72-81 

 71-91 



72-33 

 73-36 

 73-55 

 73-35 

 7312 

 72-26 



Zinc 





Platinum 



Glass 





These results in connexion with the foregoing point to 

 very distinct differences in the surface-tension of different 

 specimens of water. The very great differences in the values 

 of the temperature-coefficient found by different investigators 

 are probably also due to real differences in the specimens ex- 

 amined. Water free from gases has a temperature-coefficient 

 not far from 0*14 dyne for each degree C. But if care be 

 not taken to free the water from dissolved gases, the coefficient 

 may be as high as O20 dyne, or even higher. The great 

 differences in determinations of the critical temperature of 

 water and of alcohol by different observers may be due to 

 the same cause. 



My thanks are due to the authorities of Clark University 

 for the facilities afforded me, and to Professor Michelson 

 and Assistant-Professor Webster for valuable suggestions 

 during the progress of this work. 



Clark University, 

 Worcester, Mass., U.S.A. 



