474 
MR. 0. HEAVISIDE ON THE FORCES, STRESSES, AND 
way. But when both stresses are allowed for, we see that when g. 2 >g 1 the pull 
is towards the first medium in all parts of the interface, and that this becomes a push 
in all parts when g 1 > g 2 . 
A definite Stress only obtainable by Kinetic Consideration of the Circuital Equations 
and Storage and Flux of Energy. 
§ 38 . We see that the stress considered in the last paragraph gives a rationally 
intelligible interpretation of the attraction or repulsion. The same may be said of 
other stresses than that chosen. But the use of Maxwell’s stress, or any stress 
leading to a force on inductively magnetised matter as this stress does, leads us into 
great difficulties. By (198) we see that there is first a bodily force on the whole of the 
g 2 medium, because it is magnetised, unless g 2 =1. When summed up, the resultant 
does not give the required attraction. For, secondly, the g l medium is also magnetised, 
unless g 1 = 1 , and there is a bodily force throughout the whole of it. When this is 
summed up (not counting the force on the magnet), its resultant added on to the 
former resultant still does not make up the attraction (i.e., equivalently, the force on 
the magnet). For, thirdly, the stress is discontinuous at the interface (though not in 
the same manner as in the last paragraph). The resultant of this stress-discontinuity, 
added on to the former resultants, makes up the required attraction. Tt is unneces¬ 
sary to give the details relating to so improbable a system of force. 
Our preference must naturally be for a more simple system, such as the previously 
considered stress. But there is really no decisive settlement possible from the theo¬ 
retical statical standpoint, and nothing short of actual experimental determination of 
the strains produced and their exhaustive analysis would be sufficient to determine 
the proper stress-function. But when the subject is attacked from the dynamical 
standpoint, the indeterminateness disappears. From the two circuital laws- of variable 
states of electric and magnetic force in a moving medium, combined with certain dis¬ 
tributions of stored energy, we are led to just one stress-vector, viz. (136). It is, in 
the magnetic case, the same as (188); that is, it reduces to the latter when the 
medium is kept at rest, so that J„ and Gr 0 become J and G-. 
It is of the simple type in case of isotropy (constant tensor), but is a rotational 
stress in general, as indeed are all the statically probable stresses that suggest 
themselves. The translational force due to it being divisible conveniently into (a) 
the electromagnetic force on electric current, ( b ) the ditto on the fictitious electric 
current taking the place of intrinsic magnetisation, (c) force depending upon space- 
variation of g ; we see that the really striking part is (6). Of all the various ways of 
representing the forcive on an intrinsic magnet it is the most extreme. The magnetic 
“matter” does not enter into it, nor does the distribution of magnetisation; it is 
where the intrinsic force h 0 has curl that the translational force operates, usually on 
