314 



SCIENCE 



[N. S. Vol. LII. No. 1344 



Dr. Vernon K. Krieble, assistant professor 

 of chemistry at McGill University, succeeds 

 Dr. E. 0. Riggs as Scoville professor of clieni- 

 istry at Trinity College, Hartford, Conn. 



DISCUSSION AND CORRESPONDENCE 



ELECTRICITY AND GRAVITATION 



The action of gravitation on light is gener- 

 ally regarded as a continuous process but if we 

 consider a ray of light as the limit of a chain 

 of rectilinear rays for each of vchich the veloc- 

 ity has its upper limit value Cj we can regard 

 the gravitational action on the ray as built up 

 of a succession of impulses, each of which 

 changes the direction of the ray. To obtain a 

 definite picture of this action, let us imagine 

 the Eether to be built up of electrical doublets 

 travelling along straight lines with velocity 

 c and sometimes colliding with one another. 

 A collision in which the doublets break up 

 and their constituents secure new partners 

 leads to a temporary manifestation of free 

 electric charge. For simplicity we shall sup- 

 pose that this type of collision takes place only 

 at points where matter is present and that 

 such collisions occur continually so that the 

 manifestation of free electric charge is perma- 

 nent^ and approximately steady. At a point 

 not occupied by matter a collision may be sup- 

 posed to result simply in a change in the di- 

 rection of motion of the doublets. It is pos- 

 sible, however, that collisions are all of the first 

 type. The elementary type of electromagnetic 

 field is one in which a doublet breaks up into 

 positive and negative constituents which fly 

 away in different directions with the velocity 

 c. The field of an electric charge moving with 

 a velocity less than c can apparently be built 

 up from such elementary fields by superposi- 

 tion and so the assumption of the fundamental 



• i"We imagine one component of a doublet to be 

 momentarily separated from its feUow, when 

 another doubtlet comes along the lonely charge 

 secures a new mate and leaves another charge all 

 alone, this charge behaves in a similar manner 

 when it encounters another doublet and so on. In 

 what follows we really consider collisions between 

 doublets and free electric charges. 



character of the elementary field seems legiti- 

 mate. 



From the elementary fields it is possible to 

 build up a type of field in which the electric 

 charge associated with an electric pole fluctu- 

 ates owing to the fact that the constituents of 

 a doublet are in the neighbourhood of the pole 

 at slightly different times. We shall assume 

 that the electric action between two poles de- 

 pends on the instantaneous values of the 

 charges and shall endeavor to estimate the ef- 

 fect of the fluctuations. Let us assume that 

 the total number of doublets which break up 

 at an electric pole per unit time is proportional 

 to the mass associated with the pole. This 

 number will also be supposed to be the number 

 of doublets which are created from the con- 

 stituents of those which break up. Among the 

 doublets which arrive at the second pole B 

 there may be some that have come from A. 

 Let us suppose in the first place that there is 

 no gravitational shielding, then it seems rea- 

 sonable to assume that the percentage of B's 

 doublets which have come directly from A is 

 proportional to the number which leave A and 

 so is per unit time, proportional to the mass of 

 A. The number of doublets which pass di- 

 rectly from A to B per unit time is thus pro- 

 portional to the product of the masses of A 

 and B. The doublets themselves will be sup- 

 posed to be so small that the emission of the 

 different doublets and the arrival of others 

 may all be regarded as independent events. 

 At an instant of time t when a doublet from 

 A is arriving at B the charge on B may be 

 then regarded as equal to e' -\- f(t) when the 

 charge on A at the earlier time t — (AB/c) 

 was e — f(t). The function fit) is supposed 

 to have a mean value equal to zero so that e 

 and e' may be regarded as the mean charges 

 associated with A and B respectively. The 

 above expressions for the charges are supposed 

 to hold only for the very short periods of time 

 when the particular doublet under considera- 

 tion is in the neighborhoods of B and A, at 

 other times the values of the charges are gov- 

 erned by the presence of other doublets. 



The mean value of the electric force between 

 A and B over a small period of time, which is 



