336 
PHYSICS: J. LARMOR 
But a complete isolated system travelling with velocity v, when we include 
in the system all the radiation that has issued to however great a distance from 
its nucleus, cannot create new translational momentum. Therefore the back- 
ward pressure proportional to v of the radiational momentum projected from it 
namely —v/c 2 .dE/dt on this theory, must be compensated by disappearance of 
forward momentum attached to the source itself. The momentum of the 
source is mv: its rate of change is mdv/dt-\-vdm/dt\ the first term is the usual 
kinetic reaction of the inertia m it has at the instant; the other term propor- 
tional to v is the compensating one here adumbrated. It follows that we must 
admit change of mass specified by the formulae 8m = 8E/c 2 ; so that the mass 
of a radiating body diminishes proportionally to the energy it radiates, except 
in so far as there is compensation provided by extraneous radiation which it 
absorbs. 
For example, a particle of cosmic dust describing an orbit round the Sun 
is retarded by its own radiation, but its mass is kept up by the radiation it 
gains from the Sun : thus there is no compensation to the resisting force in this 
case, and the particle will gradually be sucked into the Sun, as Poynting in an 
important memoir showed. 
The physical theory of electrodynamics, including optics, would thus, on the 
so-called classical lines — which can be held to be not yet obsolete, though 
special problems of expansion or development to adapt them to new experi- 
mental discovery are pressing — involve electrons specified in terms of charge 
and field and inertia, which are the link with matter, and it would involve also 
free radiation existing in the aether, specified as regards mechanical force by 
its convected secondary longitudinal momentum, but physically in evidence 
mainly by the type and energy of its transverse undulations. These are the 
materials out of which the science is to evolve itself : and the problem is whether 
that evolution can be gradually effected in a natural and consistent manner, and 
one which runs parallel with the course of evolution of experimental knowledge. 
The true essence of the relativity of external knowledge is that we can in- 
vestigate a system only in relation to some other system, and the most conveni- 
ent perhaps the only feasible other system has been hitherto the ideal New- 
tonian frame of reference of space and time ; for that is the canonical system, 
so to speak, with regard to which dynamical principles take on an ideal simple 
form, and it is a system which is being determined with continually expanding 
precision by the progress of astronomical science. 
The special question now in evidence is — Is it now expedient to exchange this 
frame of reference, corresponding to c infinite, for another far more com- 
plex but very slightly different continuum having a finite space-time modu- 
lus c ? The more fundamental question is — Are we to assign to either frame 
dynamical properties, typified by propagation of physical effect in space in 
terms of undulations sustained by stress and inertia, or are we to assign to it 
properties solely geometrical and regard all physical effect as merely projected 
in duration across space? The forms of special unrestricted relativity which 
