]\gt ; The Effects of the AEther-stress .
By E. CUNNINGHAM , M.A. , Fellow of , John 's College , Cambridge , Assistant Professor of Applied Mathematics in University ] , University of London .
( Communicated by Prof. Sir J. Larmor , Sec. R.S. Received October 13 , \mdash ; Read November 25 , 1909 .
) There is an outstanding gap in electromagnetic theory in respect to the attempt to reconcile the analysis of ethereal stress on the lines initiated by Maxwell with Newton 's third law and the law of the conservation of energy .
fn the present condition of theory there is assigned to the aether a certain distribution of netic energy and momentum .
The hypothetical distribution of energy is necessarily associated with the Pointing vector which measures its rate of transference .
The distribution of momentum is so defined that the rate of increase of the total amount , within any given volume supposed at rest in the aether , is equivalent to the resultant of the Maxwell stresses on the bounding surface .
There is , , no connection tablished between the transference of energy across an area and the stress across that area .
Such a connection would require that it should be possible to assign to the medium in which stress and energy reside a state of motion whereby the stresses might do the necessary amount of work , and this again would require the revision of the specification of stress , inasmuch as the ordinary expressions are computed for an element of surface which is at rest .
Numerous other questions arise as soon as such a process is attempted , bnt the present paper seeks only to analyse what type of motion must be looked for , and to specify the field of stress upon elements of area moving with the velocities obtained .
Strictly , it is incorrect to speak of the stresscs on elements of area in the aether at the same point having different velocities .
The true stress in a continuous medium can only be estimated on area moving with the medium .
All that can be done in the absence of a of the velocity of the medium is to analyse the transference of momentum across an element of area having a specified velocity .
Only when this velocity is that of the medium is itimate to interpret this transference as due to a state of stress in the medium .
Thus , unless the aether is supposed at rest , the Maxwell expressions have no significance , except as the rate at which momentum is crossing an element of area at rest .
If , however , the aether is assumed at rest , then no state of stress can rise to any transfer of energy .
Mr. E. Cunningham .
The [ Oct. 13 , , 1 .
The flux of momentum across an element of area moving with velocity differs from that across a similar element at rest by the vector per unit area , being the intensity of the electromagnetic momentum and being the component of normal to the area .
Thus with the notation of stress components , but the symbols only as fluxes of momentum , accents being added to the quantities to the moving area , For the element at rest it is known that , etc. ; the same lelations only hold for the moving element if ?
, etc. , that is , if and are in the same direction .
Let be then assumed to be in the direction of .
If may legitimately represent the velocity of the medium , the mode of transmission of electromagnetic effects being supposed unaffected by the motion , then , etc. , are the true stresses and must account for the flux of energy .
Now the rate , per unit area , at which the traction so obtained works is where is the normal to the element of area and , or where stands for the total per unit volume .
But , being in the direction of and vanish , so that Thus Now in order to satisfy principle of energy , this expression has to be made to baJance the flux of energy across the moving surface , so that theequation must be satisfied whatever the direction .
In the analysis ( AB ) , [ AB ] stand for the scalar and vector products respectively the vectors A and B. 1909 .
] ffects of the -stress .
Inasmuch as has already been taken in the direction of , this vector equation reduces to the scalar equation Given the densities of the total netic e and momentum , the medium which transmits the stress must be moving with a velocity whose magnitude is one of the roots of this equation .
2 .
The Maxwell stresses for the electrostatic field are to be reducible to the simple specification of a tension in the direction of the lines of force , together with a pressure of equal magnitude across any plane whose normal is in a direction perpendicular to the lines of force .
The reduction to this simple form , however , does hold for the total ( electric and netic ) stress on elements of area at rest in the electrodynamic field .
It may , owever , be shown that the stresses on elements moving with the velocities above obtained are such as may be specified in a similar way .
In order to obtain this result let the axis of be taken in the direction of the momentum at any point , so that Then the system of stresses is as follows:\mdash ; , The axes of and may be turned about the axis of umtil so that is also zero .
This being done , substitute in the value of in terms of and ; this gives .
As a consequence of the identity , the equation leads to Thus Mr. E. Cunningham .
The Motional [ Oct. 13 , The ambiguity in sign in this specification of the stress is associated with the ambiguity in the value of , which is given as a root of the quadratic equation .
But it is connected closely with the fact that the electric and magnetic vectors are , save for a change of sign , in reciprocal relations in the equations of the field .
For example , in a steady field where the electric and intensities are perpendicular to one another , the velocity is given by so that or If the former value be taken , would vanish in a netic field only and become infinite iri the electrostatic field , and vice if the second value be taken .
This example suggests that inasmuch as the product of the two roots is it would be a solution of the uity to take always the lesser root , which is necessarily less than .
This would make the velocity vanish in a steady field where either electric or magnetic intensity vanished .
Another example that might be given is that of a plane wave propagated in one direction in which it is found that giving equal roots , and thus making all the stresses on areas moving with this velocity vanish .
3 .
It is nportant to remark in respect of boundary conditions that the condition satisfied at a perfectly reflecting moving surface that there is no tangential stress due to the netic forces*reduces to the condition that the component normal to the surface of the velocity foumd above shall be equal to the component velocity of the surface normal to itself To show this , consider the stress on an element of area whose normal is in the direction moving with veiocity .
The component tractions are the axes being such that and , as above .
If this is normal to the area Since or must be zero .
*Cf .
Abraham , ' Theorie der Elektrizitat , ' vol. 2 , p. 334 .
1909 .
] Effects of the .-stress .
Taking zero , the further condition required to be tisfied is But , from the definition of so that ; or , since is zero , The same result follows if is taken to be zero .
The hypothetical aether velocity relative to a moving surface is therefore always tangential : in other words , a surface that does not absorb energy may be looked upon as a smooth impenetl'able boundary .
Using the above conditions , it follows at once that the pressure on such a surface is ( taking ) .
In the case of a train of waves incident normally on a movin retlector , becomes the velocity of the reflector normal to itself , so that the above result includes that given by Larmor in this restricted cas 4 .
The method adopted in the pages may be extended to the examination of rapidly fields such as those in a region traversed by radiation .
In such a field the velocity calculated as above would have a rapidly value .
The question suggests itself as to whether velocity could be assigned to the radiation at every point , by a statistical process such as that by which in the kinetic theory of ases a of independent particles is represented in some of its properties by a continuous fluid with a definite velocity at every point .
If this is possible , it will naturally be expect .
that for steady radiation within a cavity whose are fixed , such velocity will be everywhere zero ; whereas for radiation in a cavity uniformly a certain velocity , the velocity will be everywhere that of the cavity .
In what ] it will be convenient to divide the netic e into three parts as follows:\mdash ; and also to introduce the symbols 'Encyc .
Brit. Supplement , ) vol. 33 , 1900 , article 011 " " }ttion.\ldquo ; VOL. LXXXIIL\mdash ; A. Mr. E. Cunningham .
The Motional [ Oct. 13 , The mean value of any physical quantity taken over an interval of time which is too small for any appreciable cnange in any observable magnitude , but is sufficiently long to allow of the electric and netic v to go through all possible variations , will in accordance with custom be indicated by a bar over the symbol denoting the quantity in question .
Consider now the mean stresses per unit area on an element of area moving with velocity , this velocity being assumed to be a slowly changing one in the sense that its during the interval over which mean values are taken may be neglected .
Thus , with similar expressions for the other seven components oi stress .
In order that , may be equal to we must take , as before , the velocity in the direction.of the vector Let the axis of be now taken in the direction of for convenience .
So that The fluctuations of the netic and electric vectors are now to be considered statistically .
In the first place and If the mean values of the energy and momentum are given as and being other than zero , the direction is distinguished from the and directions which occur symmetrically in the given conditions .
Thus if the fluctuations are the most general in the statistical sense , and will be equal , but will differ from On inspection of the iven conditions , it is clear that ' if for given values of , a certain pair of values are , in the course of the interval of time over which the mean value is taken , attached at a certain instant to , then the same values with both signs changed must also occur , and so for It follows at once that Since , however , is not zero , the same argument will not show that .
Nevertheless , this is so , inasmuch for a given pair of values assigned to , there is nothing in the assigned conditions to prevent having at two instants in the interval of time pairs of values of which one is obtained from the other by changing the sign of both .
1909 .
] Effects of the Maxwell Ether-stress .
The result is that the stress components given above reduce to value of , so far unknown , is , of course , calculable if and are given , and if there is a definite statistical law of distribution in time of the electromagnetic vectors subject to these conditions .
What this value of must be will be seen presently .
it known for the present , it is seen that the magnitude of may be chosen so that , that is to say so that the mean stress is equivalent merely to a uniform pressure in all directions of magnitude The required value of is iven by .
The uniform pressure thus found will be denoted by , and may be otherwise expressed as It will be noted that the criterion adopted in the present instance for determining the velocity is different from that employed in the earlier case : for no consideration has yet been made of the transference of energy in the radiation .
It remains to be seen whether the above determined pressure and velocity are consistent with the given values of the energy intensity and the Pointing vector This will be so if the right-hand side being the difference between the flux and the rate at which energy is left behind , the suffix denoting that the component of the vector to which it is added is to be taken in the direction to the element of area consideled .
The equation is , however , to be true whatever this direction may be , and , inasmuch as has already been taken in the direction of , this may be so , provided only that the scalar equation is satisfied : This relation being one between magnitudes already assigned or determined , is one that has to be verified by the discussion deferred from above of the statistical value of derived from given values of and An analogy , not quite complete , is suggested with ] kinetic theory of gases .
If the symbols used above stand for the corresponding quantities in the ideal gas , where is the kinetic energy of translation .
unit volume .
Mr. E. Cunningham .
The [ Oct. 5 .
The proof here given of the consistency of the equations above stated , though it may be interpreted as purely analytical , furnishes an interesting example of the power of the transformation which forms the basis of the Lorentz-Einstein theory of relativity .
The correlation of vectors , accented magnitudes corresponding to a frame of reference in motion relative to the original , , , leads to the equations with reciprocal relations .
These equations being linear , being treated as constant , they remain true if mean values are taken .
If , now , the limitation be laid down that , the mean values of for the statistical fluctuation of these nitudes subject to this limitation only will be zero .
In terms of the original vectors , the stipulations made become , and the consequences and Conversely , if it is given only that the mean values of are zero , .
the energy in the accented system will be equally shared between the three directions .
Referring this to the o1iginal vectors , the conditions lead to the consequences These two sets of equations are then equivalent and the last equations of 1909 .
] Effects of the Illaxwell Aftloer-stress .
117 the two sets may be replaced by two equations deduced from them , which must also be equivalent to one another , , .
Putting as before , these become From these at once follow and The last equation but one identifies with the velocity of the preceding paragraph , and the last is the relation that it was required to verify , since 6 .
The result of the .
discussion is that to an ence incapable of appreciating apid changes , the mechanical properties of the radiation may be described as those of a continuous perfect fluid in which pressure , energy , momentum , and velocity are distributed , the four magnitudes being connected by the relations , ( 1 ) , ( 2 ) and the vectors and having the same direction .
The differential equations the rates of change of these quantities are .
( 4 ) .
( 5 ) , ( 6 ) *For a cial case of homogeneous radiation this equation was given by Planck , ' Ann. .
Phys vol. 26 , p. 1 .
See also recent paper by anthor , ' Proc. Lond. Math. Soc. ' Mr. E. Cunningham .
The Motional [ Oct. 13 , These equations are obtained at once by taking mean values from the exact equations of the electromagnetic field .
By means of these equations may be obtained a law which is analogous to the law constant for a gas .
From ( 1 ) , ( 2 ) , and ( 3 ) is derived the equation Multiplication of ( 4 ) , ( 5 ) , ( 6 ) by , respectively and subtraction from ( 7 ) gives an equation which after reduction , remembering that and are in the same direction , is where the operator stands for differentiation in time estimated for a point moving with the velocity , i.e. Since , etc. , we have Now let be a small volume the points of whose boundary are moving with the velocity .
It is easily shown that the rate of change of this volume is given by or .
( 10 ) ( 8 ) , ( 9 ) , and ( 10 ) thus give or whence constant .
This is the law mentioned above .
It reduces when is , or if is neglected , to the known equation connecting pressure of radiation with the volume for adiabatic changes , *Cf .
planck , ' Theorie der Warmestrahlung , ' 1909 .
] Effects of thoe -stress .
7 .
Hitherto the argument has proceeded entirely on electrodynamic foundations .
If , now , the existence of a definite temperature at every point of the region considered be admitted , the second law of thermodynamics requires that , being an increment of heat-energy , supplied from external sources , shall be a perfect differential .
The increase in the intrinsic energy of the volume is , the work used in expansion is , the work done on the element by the forces which change the momentum is .
Thus , and , using the relations ( 1 ) and ( 2 ) bove , must be such that is a perfect differential .
This is so if , i.e. if the relation , on integration this gives being a constant .
The combination of this with ( 11 ) leads to constant , in accordance with the known result in the more restricted case when the radiation is regarded as stationary , and the changes in its volume and temperature take place infinitely slowly .

