Ether, Light, and Matter. 943 



-should not be obliterated. The electric component there 

 ^attains for a moment an infinite speed and gets out of phase 

 with the magnetic component. When the phase difference 

 is 180° the energy-flow is reversed in sign ; when the phase 

 difference is 90° there is no energy-flow, and only a 

 stationary vibration existing inside a certain boundary. 



The problem is whether part of the magnetic circulation, 

 left stranded, could not cease to be oscillatory and become 

 continuous and permanent; and whether the synchronous 

 -electric pulses ot: myriads of successive waves could not 

 .accumulate as a separated pair of opposite electronic charges. 



Meanwhile we may roughly estimate that to generate such 

 -a pair requires energy comparable to 10~ 6 erg, which would 

 be supplied by a length of a few hundred kilometres of 

 ordinary sunshine if the area of wave contributing to the 

 result were comparable to X 2 /tt, as suggested by Lord 

 Hayleigh's tc Sound " investigation in the Phil. Mag. for 

 August 1916 (cf. 'Nature,' vol. cvii. pp. 169 and 203). 

 •But to generate a wcighable amount of matter, say a tenth 

 of a milligramme, in the laboratory, a beam of sunlight a 

 square decimetre in section, even if all of it were effective, 

 would have to shine for seventeen years. 



P.S. — In the current Proc. Roy. Soc. for February 1921, 

 vol. xcix. No. A 697, p. 118, Professor Eddington shows, in 

 a manner terribly difficult to understand, that it is " impos- 

 sible to build up matter or electrons from electromagnetic 

 fields alone, — some other form of energy must be present." 

 The other form of energy postulated above is the previously 

 existing particle on which the electromagnetic waves impinge; 

 .and an operation, such as Prof. Eddington is probably com- 

 petent to envisage, is supposed to go on in its immediate 

 neighbourhood — i. e. in a region where the electric field is 

 exceedingly intense, and where the pro 'uct of certain S 

 tensors becomes important. It may be worth just noticing 

 that the Ordinary normal surface tension (27ra 2 ! k) on an 

 electron, which the structure of the electron is somehow able 

 to resist, approaches to within a twentieth of what I have 

 elsewhere supposed to be its limiting or critical value LO 33 

 -dynes per square centimetre, or what might be called the 

 weight of a thousand earths per square inch. 



3 Q 2 



