C>2 Dr. J. W. Nicholson on the Bending of 



that the exact value of their mean is not very important, but 

 the later experiments strongly support the general result 

 before obtained, that the he it evolution is at least 6xl0~ 5 

 calorie per hour per gram, instead of 4*4 x 10~ 5 , which is 

 the figure obtained on the assumption that each gram of radium 

 generates 110 calories per hour. 



Physical Laboratory, 



Trinity College, Dublin. 



October 1910. 



X- On the Bpndinq of Electric Waves round a Large Sphere. 

 III.* B/j'. W. Nicholson, M.A., B.Sc'.1[ 



Determination of the constant (3. 



rj^HE present section of the paper takes up the problem of 

 J- the determination of #, the numerical coefficient which 

 is necessary to a final tabulation of the intensity at any point 

 of the surface of the obstacle. This coefficient is defined by 

 the fact that the first root of 



3/a*.-«»K^*)=0, •• . . . (102) 



where the Bessel function involved is regarded as a function 

 of m, has an imaginary portion —iz%(3, and a real portion 

 whose most significant term is z. It was shown before that 

 no root of this equation occurs whose real part has not an 

 order z at least (if the real part is to be preponderant as 

 supposed), and therefore that the first possible root should 

 be sought within such a region of variation of m that 

 | m—z ] is of order zh. 



Now for values of m and z corresponding in this way, a 

 development of K TO (Y), suitable for the present purpose, can 

 be deduced at once from the results of a previous paper % 

 to which reference has already been made. For we may 

 write, if p= (m — *?)(6/V)5, and if the terms of the series 

 converge, 



J "«-^)'«»«< + ^©»4 + ^(i>4'- ■•} 



J - mW= ^© s { r (l) sin r os ( m,r -f) + i[ r (l) sin T cos ( m,r "T) + 



the error involved being of order z~ x at most. 



* Part I., Phil. Mag. April 1910. Part II., Phil. Mag. July 1910. 

 + Communicated bv the Author. 

 X Phil. Mag. August 1908. 



