6 FERGUSON, Studies in Capillarity 



an equation which can conveniently be solved for a 2 by successive 

 approximations. With a sphere of glass just under 6 inches in diameter 

 (R = 7*321 cm.), and water as the test liquid, the value of m is about 

 6 '3 5 gms., and, as far as the weighings are concerned, it is quite easy to 

 obtain T to 1 part in a thousand. 



The mean of 17 independent readings on different samples of water 

 gave, on reduction to 15 C, 



_ dynes 



73*45 ± °*° 2 3 



15 



cm. 



It may be useful to give here a table of some of the later values ob- 

 tained for the surface tension of water. The varying values obtained 

 serve to emphasise the difficulties inherent in the subject. Personally, I 

 am convinced that the most potent factor in producing this variation is 

 the difficulty of preparing a perfectly pure substance (or surface). The 

 exact conditions under which the various mathematical approximations 

 hold good are well known, and, provided these conditions are fulfilled, the 

 approximations provided will represent the experimental results with more 

 than sufficient accuracy. But the only satisfactory way of settling this 

 point is to undertake the research outlined under the heading (8) above. 



Tig. 



Observer. 



Method. 



73-26 



Volkmann 



Capillary-rise. 



73'46 



Domke 



Capillary-rise. 



7372 



Dorsey 



Ripples. 



73 '45 



Hall 



Weighing tension in film. 



7376 



Sentis 



Capillary tubes. 



74-22 



Watson 



Ripples. 



73H5 



Ferguson 



Pull on sphere. 



74-30 



Pedersen 



Waves on jet. 



7278 



Bohr 



Waves on jet. 



74-22 



Kalahne 



Rippled surface used as diffraction grating. 



73-38 



Richards and Coombs 



Capillary-rise. 



73-88 



Ferguson 



Jaeger's method. 



73'55 



Brown and Harkins 



Capillary-rise. 



Mathematical Principles. 



The evidence above detailed indicates that a considerable confusion 

 exists as to what problems have been, and what have not been exactly 

 solved ; and that where the problem is incapable of exact solution, need- 

 less doubts have arisen concerning the value of the approximations 

 given, and the conditions under which the approximations hold good. 

 I propose, therefore, in this section, to devote some little space to a 

 discussion of the more important of these problems, and I hope to show 

 that, even in those cases which are usually considered to require complex 



