312 C. Bar us — Slower-like Distortion of Coronas, etc. 



No. 2 is more oval, the members having risen above No. 1. 

 Both curves are closed. No. 2' is the interesting transitional 

 case between closed and open curves concerning which pres- 

 ently. No. 3 is already quite open and bell-shaped ; No. 4 

 more so ; No. 5 is basin-shaped, and succeeding curves would 

 more and more nearly approach the horizontal Hue through the 

 source. Naturally all curves pass through the same two points 

 in this line. 



Moreover equation (6) shows that s becomes imaginary when 

 l<^(s* /A) sin <f> since A is negative. The final values of s and 

 <f) are thus given by sin <f> m = — A/s 2 , so that on reduction 

 s m = 2s = 2*88. These data are also given in the table. It is 

 further apparent that the corona will just begin to open on 

 top when 1 = — s\ / A, or sin = 1. Since A = — -00072/a 

 and s = 1'44: the gradient a = -0035. This is the curve No. 2' 

 between the conditions of Nos. 2 and 3. In all these cases the 

 equation given strikingly interprets the opening of a harebell 

 or what would be called campanulate efflorescence in botany. 



I may add in conclusion that all the types of curves given 

 are continually and repeatedly met with in working with vola- 

 tile liquids, among which I have now examined gasoline, 

 benzine, benzol, and carbon disulphide, at length. The law of 

 distribution reproduces the cases as nearly as they can be tested 

 in the fleeting coronas, though the real law is not liable to be 

 linear and will have to be specially worked out. 



Brown University, Providence, E. I. 





