330 BELL SYSTEM TECHNICAL JOURNAL 



If ju 5^ 1 everywhere and if we write 



u; = u + curl M = \E + curl M (V) 



equation (8) of my paper becomes ^ 



1 , too r W I iwr\ , ^ , 1 , njr 



_^ + _J_exp^--jJ. = G+^curlM 



(2') 



X 



and correspondingly equation (9) becomes 

 f{w'-G") - {w"-G')]dv 



+ I ^ {u;'-curl M") - (w;"-curl M')]dv = 0. (3') 



If now in (3') we replace u; by w + curl M and note that u/X = E, 

 (3') reduces to 



f{iu'-G") - {u"-G')]dv 



- f{{G'- curl M") - (G" • curl M') } dv (4') 



+ f{E' -cml M") - (E"-cur\ M')]dv = 0. 



Finally since E - G = -—A, (4') reduces to 



f{{u'-G") - {u"-G')}dv 



^l^fUA'- curl M") - {A" ■ curl M') ]dv ^ {). (5') 

 c 



But 



/(A' -curl M")dv = /(M"-curl A')dv 



= T- f- ^(5"-curl^0^i; 



47r J II 



= -T- f ^^-^ (curl A" ■ curl A')dv, 

 47r J n 



SO that the second integral of (5') vanishes and 



f{{u'-G") - {u"-G')}dv = 0, (60 



which is equation (9) of the original paper. The rest of the proof of 

 the theorem is now simply that of the original paper. 



It will be observed the theorem is stated for the current u = \E; 

 that is the conduction (plus polarization) current. Ballantine ^ in 



^ The paper itself must be consulted for the significance of the symbols and the 

 method of attack and proof. 



6 June 1929 issue of Proc. I. R. E. 



