]\gt ; [ 423 ] XI .
the Forces , Stress .of Energy in the agnetic Field .
OLIVER HEAVISIDE , F.R. Received June 9 , \mdash ; Read June 18 , on the of ) S1 .
THE remarkable experimental work of late years has inaugurated a new era in the development of the Faraday-Maxwellian theory of the ether , considered as the primary concerned in eIectrical phenomena\mdash ; eIectric , magnetic , and electro- magnetic .
MAXWELL'S theory is no longer entirely a paper theory , istl i with unproved possibilities .
The reality of eIectromagnetic waves has been thoroughly demohstrated by the experiments of HERTZ and LODGE , FITZGERALD and TROUTON , J. J. , and others ; it appears to follow that , although MAXWELL'S theory may not be fully correct , even regards the ether ( as it is not fully comprehensive as regards material bodies ) , yet the true eory must be one of the same type , and may probably be merely an extended form of MAXWELL'S .
No excuse is therefore now needed for investigations tending to exhibit and elucidate this theory , or to extend it , even though they be of a very abstract nature .
Every part of so important a theory deserves to be thorougbly examined , if only to see what is in it , and to take note of its unintelligible parts , with a view to their future explanation or elimination .
S2 .
Perhaps the simplest view to take of the medium which such a necessary part , as the recipient of energy , in this theory , is to regard it as continuously filling all space , and sessing the mobility of a fluid rather than the rigidity of a solid .
If whatever possess the property of inertia be , then ] medium is a of matter .
But away from matter it is , .
obvious reasons , to call it as usnal by a separate name , the ether .
Now , a difficult and hig.hly speculative question , at present , is the connection between matter ( in the ordinary sense ) and ether .
When the lnedium transmitting the electrical disturbances consists of ether and matter , do they move together , or does the matter only forward the ether which immediately surrou1lds it ?
reasons may lead us to conclude , though only tentatively , that the latter may be the case ; present , for the purpose of fixing the data , and in the pursuit of investigations not having specialIy * Typographical troubIes have delayed the publication of this paper .
The { ootnotes of date Nay 11 , 1892 .
11 .
892 optical it is convenient to assume that the matter and the ether in contact with it move .
This is the working hypothesis made by H. in his recent of the e]ectrodynamics of moving bodies ; it is , in fact , what we tacitly assume in a f ; traightforward and consistent working out of MAXWELL'S principles without any plainly-expressed statement on the question of the relative motion of tter and ether ; for the part played in MAXWELL'S theory by matter is merely ( and , of course , roughly ) nularised by supposing that ic causes the ethereal constants to take different values , whilst introducing new properties , that of dissipating energy being the most prominent and important .
We may , therefore , think of merely one medium , the most of which is uniform ( the ether ) , whilst certain portions ( matter as well ) have different powers of supporting electric displacement and netic induction from the rest , as well as a host of additional properties ; and of these we can include the power of supporting conduction current with dissipation of energy according to law , the change from isotropy to eolotropy in respect to the distribution of the several fluxes , the presence of intrinsic sources of energy , \amp ; .
We do not in any way form the equations of motion of such a medium , even regards the nniform simple ether , away from grcss matter ; we have only to discuss it as the electric and magnetic fluxes it supports , and the stresses and huxes of energy thereby necessitated .
First , we suppose the medium to be stationary , and vine t flux of electromagnetic energy .
This is the POINTING flux of energy .
Next we set the medium into motion of an unrestricted kind .
We have now necessarily a convection of the electric and magnetic energy , as well as the pointing flux .
Thirdly , there must be a similar convection of the kinetic energy , \amp ; c. , of the translational motion ; and fourthly , since the motion of the medium involves the working of ordinary ( Newtonian ) force , there is associated with the previous a flux of energy due to the activity of the corresponding stress .
The question is therefore a complex one , for we have to properly fit together these various fluxes of energy in harmony with electromagnetic equations .
A side issue is the determination of the proper measure of the activity of intrinsic forces , when the medium moves ; in another form , it is the determination of the proper meaning of " " true current\ldquo ; in MAXWELL'S sense .
S4 .
The only general principle that we can bring to our assistance in interpreting romagnetic results relating to activity and flux of , is that of the permps it is best to say as little as possible at present about connection between matter cthcl ; but to take the electromaguetic equations in an abstract manner .
This will Jeave us greater froedom for future modifications without contradiction .
here are , also , cases in which it is obviously impossibIe to suppose thnt matter in bulk carries on with it the ethel in bulk which permeates it .
ithcr , then , the mathematical machinery must work between the molecules ; or else , we must make such alterations in the equations ] eferring to bulk as will be practically in eflect .
example , the motional magnebic force of equations ( 88 ) , ( 92 ) , ( 98 ) may be modified eilher in or in , by nse of a smallclctive velocity , or by the substitution in or of a modified reckoning of for the effcctive permittivity .
FLUXES ENERGY IN THE FILLED .
eistence of energy .
But it )yould be quite inadequate in its older sense to integral nounts ; the definite isation by MAXWE , of electric nd magnetic energy , and of its waste , necessitates the similar localisation of sources of energy ; and in the consideration of the supply of energy at certain places , combined with the continuous transmission of electrical bances , therefore of the associated energy , the idea of a flux of energy through space , and therefore of the continuity of energy in space and in time , becomes forced upon us as simple , useful , and necessary principle , which cannot be avoided .
When energy from place to place , it trayerses the intermediate space .
Only by the use of principle can we ] derive the etic stress from the equations of the field the two laws of circuitation of the electric and magnetic forces : and this again becomes permissible only by the postulation of the definite localisation of the and magnetic energies .
But we need not go so far as to assume the objectiyity of energy .
This is an exceedingly difficult notion , and seems to be rendered inadmissible by mere fact of the relativity of motion , on which kinetic energy depends .
We cannot , therefore , definitel .
energy in the same way as is done with matter .
If be the density of a quantity wIlose total amount is invariable , and which can change its distribution actual motion from place to place , its tion of continuity is COllv , . . . . .
( l ) where is its velocity , and the flux of .
That is , the convergence of the flux of equals the rate of increase of its density .
Here may be the density of matter .
But it does not appear that we can apply the same method of representation to the flux of energy .
We may , indeed , write conv , . . . . . . . . . . . .
( 2 ) if X be the flux energy from all causes , and the density of localisabIe energy .
But the assumption would involve the assumption that moved about like matter , with a definite velocity .
A part of may , indeed , do this , when it confined to , and is carried by matter ( or ether ) ; thus we may write conv , . . . . . . . . . .
( 3 ) where is energy which is simt ) carried , whilst X is the total flux of energy other sources , and which we cannot symbolise in the form Tq ; the energy comes to us from the Sun , for example , or radiated eIlergy .
It is , , often impossible to carry out the principle in this form , from a want of knowledge of how energy gets to a certain place .
This is , for example , particularly evident in the case ' gravitational energy , the distribution of which , it is commun icated to matter , MR. O. ON THE FORCES , STRESSES , AND its kinetic energy , is highly speculative .
If it come from the ether ( and where else can it come from ?
) , it should be possible to symbolise this in , if not in ; but in default of a knowledge of its distribution in the ether , we cannot do so , and must therefore turn the equation of continuity into conv ( , . . . . .
( 4 ) where indicates the , rate of supply of energy per unit volume from the gravitational source .
whatever that may be .
A similar form is convenient in the case of intrinsic stores of energy , which we have reason to believe are positioned within the element of volume con cerned , as when heat gives rise to thermoelectric force .
Then is the activity of the intrinsic sources .
Then , in special applications , is conveniently divisible into different kinds of energy , potential and kinetic .
Energy which is dissipsted or wasted comes under the same category , because it may either be regarded as stored , though irrecoverably , or passed out of existence , so far as any immediate useful purpose is performed .
Thus we have as a standard practical form of the equation of continuity of energy referred to the unit volume , conv { , . . . . .
( 5 ) where is the energy supply from intrinsic sources , potential energy and kinetic en ergy of localisable kinds , its convective flux , the rate of waste of energy , and X the flux of energy other than convective , e.g. , that due to stresses in the medium and representing their activity .
In the netic application we shall see that and must split into two kinds , and so must X , because there is a flux of energy even when the medium is at rest .
S5 .
Sometimes we meet with cases in which the flux of energy is either wholly or partly of a circuital character .
There is nothing essentially peculiar to electromagnetic problems in this strange and apparently useless result .
The electromagnetic instances are paralleled by similar instances in ordinary mechanical science , when a body is in motion and is also strained , especially if it be in rotation .
This result is a necessary consequence of our ways of reckoning the activity of forces and of stresses , and serves to still further cast doubt upon the " " thinginess\ldquo ; of energy .
At the same time , the flux of energy is going all around us , just as certainly as the flux of matter , and it is impossible to avoid the idea ; we should , therefore , make use of it and formularise it whenever and as lonoo as it is found to be useful , in spite of the occasional failure to obtain readily understandable results .
The idea of the flux of energy , apart from the conservation of energy , is by no means a new one .
Had gravitational energy been less obscure than it is , it might have found ) statement long .
Professor brought the principle , ' Phil. Trans 1884 .
FLUXES OF ENERGY THE ELECTROMAGNETIC FIELD .
into prominence in 1884 , by making use of it 10 determine the electromagnetic flux of energy .
Professor LODGE* gay very distinct and emphatic expression of the principle generally , from its electromagnetic aspect , in 1885 , and pointed out how much more simple and Qatisfactory it makes the principle of the conservation of energy become .
So it would , indeed , could we only understand grayitational energy ; but in that , and similar respects , it is a matter of faith only .
But Professor LODGE attached , I I think , too much importance to the identity of energy , as well as to another principle he enunciated , that energy cannot be transferred without being transformed , and conversely ; the transformation being from potential to kinetic energy or conyersely .
This obviously cannot apply to the convection of , which is a true flux of energy ; nor does it , seem to apply to cases of wave motion in which the energy , 1 potential and kinetic , of the disturbance , is transferred through a medium unchanged in relative distribution , simply because the disturbance itself travels without change - of type ; though it may be that in the unexpressed internal actionS associated with the wave propagation there might be found a better applicatiolL It is impossible that the ether can be fully represented , even merely in its transmissive functions , by the electromagnetic equations .
Gravity is left out in the cold ; and although it is convenient to ignore this fact , it may be sometimes usefully remembered , even in special electromagnetic work ; for , if a medium have to contain and transmit gravitational energy as well as electromagnetic , the proper systern of equations should show this , and , therefore , include the electromagnetic .
It seems , therefore , not unlikely that in discussing purely netic speculations , one may be within a stone 's throw of the explanation of gravitation all the time .
The consummation would be a really substantial advance in scientific knowledge .
On the Algebra Analysis of Vectors )ithont Q. Outline of S6 .
The proper language of vectors is the algebra of vectors .
It is , therefore , quite certain that an extensiye use of vector-analysis in mathematical physics generally , and in electromagnetism , which is swarming with vectors , in particular , is coming and may be near at hand .
It has , in my opinion , been retarded by the want of 1 special treatises on vector analysis adapted for use in mathematical physics , Professor -v TAIT'S well-known profound treatise being , as its name indicates , a treatise on Quaternions .
I have not found the HAMILTON-TAIT notation of vector operations convenient , and have employed , for some years past , a simpler systeln .
It is not , however , entirely a question of notation that is concerned .
I reject the quaternionic xi basis of vector.analysis .
The anti-quaternionic argument has been recently ably stated by Professor WILLARD GIBBS .
He distinctly separates this from the question * LODGE , ' Phil. Mag June , 1881 " " On the Identity of ergy.\ldquo ; Professor GIBBS'S letters will be found in ' Nature , ' vol. 43 , vol. 44 , ; and ProfesRo .
O. VISlD ON THh1 FORCES , , AND of notation , ) this may considered fortunate , for whilst I can fully appreciate and ( Jfroln practical experience ) the -quaternionic , I am unable to appreciate is notation , and think that of HAMILTON and TAIT is , in some respects , preferable , ient in others .
In 's system the quaternion is the fundamental idea , and everything cvolves it .
This is exceedingly unfortunate , it renders the establishment of of veclors withont physics a very difficult matter , an in its application to analysis is a tendencv for the algebra to get more more complex as the ideas concerned get simpler , and the quaternionic basis forms real difficulty ) kind in attempting to work in harmony with ordinary methods .
, I can recommend , as a really practical working system , the tion I lave m. It has , and not the least amongst them is fact that the quaternion does not appear in it at all ( though it may , without mnch be ought in sometimes ) , and also that the notation is arranged to rmonise Cartesian mathematics .
It rests entirely a few definitions , may be ( one point of view ) as a abbreviated Cartesian method of investigation , and be understood and practically used by any one omed to Cartesians , without any study of the difficult science of Quaternions .
It is sinlply the elements of Quaternions without the quaternions , with the notation ified to the uttermost , and with the very inconvenient before scalar products done with .
's in vol. 4 , 608 .
This one-sided discu out of ' stigmatising 11 " " of ress.\ldquo ; This 1 be very true ; but Professor GIBBS is retardcr of ress in vector and its pplication to physics .
*SS 7 , 8 , itlt introducbiou to vecfor-analysis ( without the quaternion ) , which is sufficient for the purposes of the papel , and , I may add , for general use in mathematical physics .
It is an expansion of that given in " " the omagnetic Wave Surface ' Phil. Mag June , 1886 .
The and notation substantially those employed in all my papers , especially in " " Electromag etic lnduction and its Prop , gation , \ldquo ; ' Electrician , ' Professor GlBI , work is reely known , and deserves bo be well .
In June , 1888 , I recve.d from him a littlc book of singulat imprint NOT Newhaven , 1881-4 .
It is indeed odd that the author should not have published what he had been at the trouble of having printed .
His treatment of linear vccbor operator is specially deserving of notice .
Altbough ' ' for ths use of students in physics I am bound to say that 1 think the work much too condensed for a first to subjcct .
In The Nov. 13 , 1891 , p. 27 , I commenced few articles on elemental.y vcctoralgebra and analyhiLq , specially meant to explain to readers of my papel .
S how to work vectors .
I flm givcn to nndert , tand that ones , on the ) much the later ones , difficult .
But is identically same in botb , and is of quite a kind .
The difference is , that the later ones are concerned with analysis , with rarying vectol'S ; it is the same as the difference between common algebra and differential calcnlus .
The difficulty , whether or not , does not indic.ate any difficulty in the vect .
I mention this on of the great lico which xists agaiust vector-algebra .
FLUXES OF ENERGY IN CTRO MAGNETIC FIELD .
S7 .
Quantities divided into scalars and vectors , I denote the scalars , as usual , by ordinary letters , and put the vectors in the plain black type , known , I believe , as Clarendon type , rejecting Germall letters on account of their being hard to read .
A special type is inly not essential , but it ] reading of printed complex vector investigations to be to see at a glance which quantities are scalars and which are vectors , and the on the But in MS .
work there is no occasion formed Thus A stan ds for a vector .
tensor of a vector may be denoted by the same letter plain ; thus A is the tensol of A. ( In .
the tensor is ) Its rectangular scalar components are .
A unit vector parallel to A be denoted by , so that .
But little things of sort very llllch matters of taste .
What is important is to avoid as far as possible of letter prefixes , which , when they come two ( or even three ) together , in Quaternions , very The scalar product of a pair of vectors A is denoted by AB , and is defined to be . . . . . . . . .
( 6 ) The addition of vectors being as in the polygon of displacements , or velncities , or forces ; i.e. , such that the vector length of any closed circuit is zero ; either of the vectors A and may be split into the sum of any number of others , and the multiplication of the two sums to form AB is done as in commoll If be a unit vector , NN or ; similarly .
for any vector .
The reciprocal of a vector A has the same direction ; its tensor is the eciprocal of the tensor of A. Thus ; and . . . . .
( ) The vector product of a pair of vectors is denoted by VAB , and is defined to be the vector whose tensor is AB , and whose direction is perpendicular to the plane and B. Or , . . .
( 0 ) where , ore any three mutually rectangular unit vectors .
tensor of is ; or . . . . . . . .
( 10 ) Its components are iVAB , jVAB , kVAB .
In accordance with the itions of the scalar and vector products , we have , ( 11 ) Vij and from these we prove at once that and so on , for any number of component vectors .
The order of the letters in each product has to be preserved , since Vab Vba .
Two very useful formulae of transformation are AVBC BVCA CVAB ; . . . .
12 and VAVBC ( ) \mdash ; ( ) .
( 13 ) Here the dots , or the brackets in the alternative notation , merely act as separators , separating the scalar products CA and AB from the vectors they multiply .
A space would be equivalent , but would be obviously unpractical .
As is a scalar product , so in harmony therewith , there is the vector product V Since VAB , it is now necessary to make a convention as to whethex the denominator comes first or last in V .
Say therefore , .
Its tensor is .
( 14 ) S 8 .
Differentiation of vectors , and of scalar and vector functions of vectors with respect to scalar variables is done as usual .
Thus , AB .
( 15 ) AVBC same pplieS with complex scalar differentiators , e.g. , with the differentiator FLUXES OF ENERGY IN THE ELECTROMAGNETIC FIELD .
used when a moving particle is followed , being its velocity .
Thus , AB . . . . . . . . .
( 16 ) Here is a scalar differentiator given by ( 17 ) so that is the scalar product of A and the vector ; the dots here again act essentially as separators .
Otherwise .
we may write it B. The fictitious vector given by ( 1S ) is very important .
Physical mathematics is very largely the mathematics of .
The name Nabla seems , therefore , ludicrously inefficient .
In virtue of , the operator behaves as a vector .
It also , of course , differentiates what follows it .
Acting on a scalar , the result is the vector , . . . . . .
( 19 ) the vector rate of increase of with length .
If it act on a vector , there is first the scalar product , . . . . .
( 20 ) or the divergence of A. Regarding a vector as a flux , the divergence of a vector is the leaving the unit volume .
The vector product is , curl A. . . . . . . . . . . . . .
( 21 ) The line-integred of A round a unit area equals the component of the curl of perpendicular to the area .
We may also have the scalar and vector products and , where the vector is not differentiated .
These of course , require a function to follow them on which to operate ; the previous of ( 16 ) illustrates .
The Laplacean operator is the scalar product or ; or ; . . . . . . . . . . .
MR. O. AyISI D ON THE FORCES , ] SSES , AND example of ( 13 ) is or , . . . . .
( 23 ) which is an important formula .
Other important formulae are the next hree .
( 24 ) scalar .
Here note and ( the latter being the scalar product of A and ) identical .
This is not true when we substitute a vector .
Also \mdash ; A curl ; . . . . . . . . .
( 26 ) which is of ( 12 ) , noting that both A and have to be differentiated .
And , . . . . .
( 26 ) This is an example of ( 13 ) .
S9 .
When one vector is a linear function of another vector that is , connected by equations of the form , ( 27 ) in terms of the rectangular components , we denote this simply by , . . . . . . . . . .
( 28 ) where is the linear operator .
The conjugate function is given ] by , . . . . . . . . .
( 29 ) where is got from by exchanging , \amp ; c. Should the nine coefficients reduce to six by , \amp ; c. , and are identical , or is a self-conjugate or symmetrical linear function of E. But , in , it is the sum of and which is a symmetrical function of and the difference is a simple vector-product .
Thus ( 30 ) yhere c is a self-conjugate , and is the vector given by FLUXES OF ENERGY IN THE ELECTROMAGNETIC FIELD .
The important characteristic of a self-conjugate operator is or ( 32 ) where and are any two , and the correspondinlgD 's .
But when there is not symmetry , the corresponding property is or ( 33 ) Of these operators we have three or four in electromagnetism connecting forces and { fluxes , and three more connected with the stresses and strains concerned .
As it .
seems impossible to avoid the consideration of rotational stresses in electromagnetism , : and these are not usually considered in works on elasticity , it will be desirable to .
briefly note their peculiarities here , rather than later on .
On Stresses , irrotational rotational their Activities .
S .
Let be the vector stress on the plane , or the plane whose unit is N. It is a linear function of N. This will fully the stress on ) Thus , if are the stresses on the planes , we hall have Let , be the conjugate stress ; then , similar]y , ( 35 ) Half the sum of the stresses and is an ordinary irrotational stress ; so that ( 36 ) / Wbere is conjugate , and . . . . . . . .
( 37 ) Here is the torque per unit volume arising from the stress P. The force , , per unit volume is ( by inspection of a unit cube ) . . . . . . .
( S8 ) ; . . . .
( 39 ) or , in terms of the self-conjugate stress and the torque , \mdash ; curl , . . . . .
( 40 ) where \mdash ; curl is the translational force due to the rotational stress alone , as in Sir W. THOMSON'S latest theory of the mechanics of an " " ether.\ldquo ; * Next , let be the unit normal drawn outward from any closed surface .
Then , . . . . . . . . . .
( 41 ) where the left summation extends over the surface and the right summation throughout the enclosed region .
For ; ( 42 ) so the -known theorem of divergence gives immediately , by ( 39 ) , . . . . . . . .
( 43 ) Next , as regards the equivalence of rotational effect of the surface-stress to that of the internal forces and torques .
Let be the vector distance from any fixed origin .
Then is the vector moment of a force , , at the end of the arm .
Another ( not so immediate ) application of the divergence theorem gives , . . . . . . . .
( 44 ) Thus , any distribution of stress , whether rotational or irrotational , may be regarded as in equilibrium .
Given any stress in a body , terminating at its boundary , the body will be in equilibrium both as regards translation and rotation .
Of course , the dary discontinuity in the stress has to be reckoned as the equivalent of internal divcrgence in the appropriate manner .
Or , more simply , let the stress fall off continuously from the finite internal stress to zero through a thin surface-layer .
We * ' Mathematical and Physical Papers , ' vol. 3 , Art .
99 , p. 436 .
FLUXES OF LNERGY IN THE ELECTROMAGNETIC FIELD .
then have a distribution of forces and torques in the surface.layer which equilibrate the internal forces and torques .
To illustrate ; we know that MAXWELL arrived at a peculiar stress , compounded of a tension parallel to a certain direction , and an equal lateral pressure , which would account for the mechanical actions apparent between electrIfied bodies , and endeavoured similarly to determine the stress in the interior of a gnetised body to harmonise with the similar external magnetic stress of ) simple type mentioned .
This stress in a magnetised body I believe to be thoroughly erroneous ; nevertheless , so far as accounting for the forcive on a magnetised body is concerned , it will , when properly carried out with due attention to surface-discontinuity , answer perfectly well , not because it the stress , but because any stress would do the same , the only essential feature concerned being the external stress in the air .
Here we may also note the very powerful nature of the stress-function , considered merely as a mathematical engine , apart from physical reality .
For example , we may account for the forcive on a magnet in many ways , of which ) two most promiDent are by means of forces on imaginary magnetic matter , and by forces on imaginary electric currents , in the magnet and on its surface .
To prove the equivalence of these two methods ( and the many others ) involves very complex surface- and volumeintegrations and transformations in the general case , which may be all avoided by the use of the stress-function instead of the forces .
S11 .
Next as regards the activity of the stress and the equivalent translational , distortional , and rotational activities .
The activity of is per unit area , if be the velocity .
Here , . . . . . . . . .
( 45 ) by ( 42 ) ; or , re-arranging , , . . . . . . . . . .
( 46 ) where is the conjugate stress on the ] .
That is , or is the negative of the vector flux of energy expressing the .
For we choose so as to mean a pull when it is positive , and when the stress works in the same sense with energy is transferred against the motion , to the matter which is pulled .
The convergence of the energy-flux , or the diyergence of , is therefore the activity per unit volume .
Thus ( 47 ) . . . .
( 48 ) where the first form ( 47 ) is generally most useful .
Or ; . . . . .
( 49 ) where the first term on the right is the translational activity , and the rest is the sum of the distortional and rotational activities .
To separate the latter introduce the strain velocity vectors analogous , , ; . . . .
( 60 ) and generally .
( 51 ) Using these we obtain curl curl curl curl . . . . . . .
62 ) Thus is the distortional activity and curl the rotational activity .
But since the distortion and the rotation are quite independent , we may put Pp for the distortional activity ; or else use the self-conjugate stress , and write it S .
In an ordmary " " elastic solid when isotropic , there is elastic resistance to compression and to distortion .
We may also imaginably have elastic resistance to translation and to rotation ; nor is there , so far as the mathematics is concerned , any reason for excluding dissipative resistance to translation , distortion , and rotation ; and kiuetic energy may be associated with all three as ell , instead of with the translation alone , as in the ordinary elastic solid .
Considering only three elastic moduli , we have the old and of THOMSON and TAIT ( lLesistance to compression and rigidity ) , and a new coefficient , say , such that curl , . . . . .
( 53 ) if be the displacement and the torque , as before .
The stress on the plane ( any plane ) is curl D. ( 54 ) and its conjugate is ( \mdash ; S ) ; ( b5 ) from which is the component of the translational force ; the complete force is- therefore FLUXES OF ENERGY IN THE ELECTROMAGNETIC ; ( 57 ) or , in another form , if being the isotropic pressure , , . . . . .
( 58 ) remembering ( 23 ) and ( 53 ) .
We see that in ( 57 ) the term involving may vanish in a compressible solid by the relation ; this makes , . . . . . . . . .
which are the moduli , longitudinal and lateral , of a simple longitudinal strain ; that is , multiplied by the extension , they give the longitudinal ction , ana ' the lateral traction required to prevent lateral contraction .
The activity per unit volume , other than translational , is curl ; . . . . .
or , which is the same , curl \mdash ; , ( 61 ) where the quantity in square brackets is the potential energy of an infinitcsimal distortion and rotation .
The italicised reservation appears to be necessary , as we shall see from the equation of activity later , that the convection of the potential energy destroys the completeness of the statement if be the potential energy , In an elastic solid of the ordinary kind , with , we have ( 2 curl curl D ) , ( 62 ) curl2 D. In the case of a medium in which is zero but finite ( Sir W. THOMSON'S rotational .
' ( 63 ) Thirdly , if we have both and , then ( url D ) , ( 64 ) , the sums of the previous two stresses and forces .
S13 .
As already observed , the vector flux of energy , due to the stress , is . . . . . . . .
( 65 ) Besides this , there is the flux of energy by convection , where is potential and kinetic energy .
Therefore , . . . . .
( 66 ) represents the complete energy flux , so far as the stress and motion are concerned .
Its convergence increases the potential energy , the kinetic energy , or is dissipated .
But if there be an impressed translational force its activity is fq .
This supply of energy is independent of convergence of W. Hence . . . . .
( 67 ) is the equation of activity .
But this splits into two parts at least .
For ( 67 ) is the same as , . . . . .
( 68 ) and the translational portion may be removed altogether .
That is , .
( 69 ) if the quantities with the zero suffix are only translationally involved .
For example , if ' . . . .
( 70 ) as in fluid motion , without frictional or elastic forces associated with the translation , then FLUXES OF ENERGY IN THE ELECTROMAGNETIC FIELD .
if , the kinetic energy per unit volume .
The complete form ( 69 ) comes in by the addition of elastic and frictional resisbing forces .
So deducting ( 69 ) from ( 68 ) there is left , . . . . .
( 72 ) where the quantities with suffix unity are connected with the distortion and the rotation , and there may plainly be two of dissipative terms , and of energy ( stored ) terms .
Thus the relation curJ D will bring in dissipation and kinetic energy , as well as the former potential energy of rotation associated with That there can be dissipative terms associated with the distortion is clear enough , remembering STOKES'S theory of a viscous fluid .
Thus , for simplicity , do away with the rotating stress , by putting , making and identical .
Then take the stress on the plane to be given ) , ( 74 ) and similarly for any other plane ; where When , we have the elastic solid with rigidity and compressibility .
When , we have the viscous fluid of STOKES .
When only , we have a viscous elastic solid , the viscous resistance being purely distortional , and proportional to the speed of distortion .
But with , all finite , we still associate kinetic energy with the potential energy and dissipation introduced by and We have for infinitesimal strains , omitting the effect of convection of energy ; where . . . . . . .
( 76 ) .
( 77 ) Observe that and only differ in the exchange of ; bnt , the potential energy , is not the same function of and that is of and .
But if we take , we produce similarity .
An elastic solid having no resistance to compression is also one of Sir W. THOMSON'S ethers .
When / , we come down to the frictionless fluid , in which , ( 78 ) and , ( 79 ) with the equation of activity fq , . . . . . . . . . .
( 80 ) the only parts of which are not always easy to interpret are the Pq term , and the proper measure of U. By analogy , and oonformably with more general cases , we should take , and reckoning the expansion or compression some mean condition .
The Electromagnetic Equations in Moving S14 .
The study of the forms of the equation of aotivity in purely mechanical cases , and the interpretation of the same is useful , because in the electromagnetic problem of a moving medium we have still greater generality , and difficulty of safe and sure interpretation .
To bring it as near to abstract d.ynamics as possible , all we need say regarding the two fluxes , electric displacement and magnetic induction , is that they are linear functions of the electric force and magnetic force , say ( 81 ) where and are linear operators of the symmetricA kind , and that associated with them are the stored energies and , electric and magnetic respectively ( per unit volume ) , given by , . . . . .
( 82 ) In isotropic media is the permittivity , the inductivity .
It is unnecessary to say more regarding the well-known variability of and hysteresis than that a magnet is here an ideal magnet of constant inductivity .
As there may be impressed forces , is divisib]e into the force of the field and an impressed part ; for distinctness , then , the complete may be called the " " force of the flux\ldquo ; D. Similarly as regards and B. OF ENERGY IN There is also waste of energy ( in conductors , namely ) at the rates , . . . . . . . . . .
( 83 ) where the fluxes and are also linear functions of and respectively ; thus , . . . .
( 84 where , when the force is parallel to the flux , and is scalar , it is the electric conductivity .
Its magnetic analogue is , the magnetic conductivity .
That is , a magnetic conductor is ( fictitious ) body which cannot support magnetic force without continuously dissipating energy .
Electrification is the divergence of the displacement , and its analogue , magnetification , is the divergence of the induction ; thus . . . . .
( 85 ) are their volume densities .
The quantity is probably quite fictitious , like K. According to MAXWELL'S doctrine , the true electric current is always circuital , and is the sum of the conduction current and the current of displacement , which is the time rate of increase of the displacement .
But , to preserve circuitality , we must add the convection current when electrification is moving , so that the true current becomes . . . . .
( 86 ) where is the velocity of the electrification .
Similarly . . . .
, . . .
( 87 ) should be the corresponding magnetic current .
S15 .
equation of electric current in terms of magnetic force in medium at rest , say , curl where is the force of the field , shouId be made , using instead , curl and here will be the force of intrinsic magnetis tion , such that is the intensity of intrinsic magnetisation .
But I have shown that when there is motion , another impressed term is .
required , , the motional magnetic force , ( 88 ) making the first circuital law become curl . . . . . .
( 89 ) MAXWELL'S other connection to form the equations of propagation is made through vector-potential A scalar potential .
Finding this method not practically workable , and also not sufficiently general , I have introduced instead a companion equation to ( 89 ) in the , . . .
( 90 ) where , expresses intrinsic force , and is the motional electric force given by , . . . . .
( 91 ) which is one of the terms in MA XWELL'S equation of electromotive force .
As for it includes not merely the force of intrinsic electrification , the analogue of intrinsic netisation , but also the sources of energy , voltaic force , thermoelectric force , \amp ; c. ( 89 ) and ( 90 ) are thus the working equations , with ( 88 ) and ( 91 ) in case the medium moves ; along with the linear relations before mentioned , and the definitions of energy and waste of energy per unit volume .
The fictitious and are useful in symmetrizing the equations , if for no other purpose .
Another way of writing the two equations of curl is by removing the and terms to the right side .
Let . . . . .
( 92 ) -curl Then ( 89 ) and ( 90 ) may be written rlcm l ( 93 ) So far as circuitality of the current goes , the is needless , and still further complicates the make-up of the true current , supposed now to be .
On the other hand , it is a simplification on the left side , deriving the current from the force of the flux or of the field more simply .
A question to be settled is whether or should be the true current .
There seems ] one crucial test , viz. , to find whether or is the rate of supply of energy to the electromagnetic system by an intrinsic force .
This requires , however , a full and rigorous examination of all the fluxes of energy concerned .
OF ENERGY IN THE ELECTROMAGNETIC The Etectroma.rjnetic of Energy in a S16 .
First let the medium be at rest , giving us the equations curl , . . . . . . . . .
( 94 ) -curl . . . . . . . .
( 95 ) Multiply ( 94 ) by , and ( 5 ) by , and add the results .
cIlrl which , by the formula ( 25 ) , becomes ; or , by the use of ( 82 ) , ( 83 ) , , . . . . .
( 96 ) where the new vector is given by . . . . .
The form of ( 96 ) is quite explicit , and the interpretation sufficiently clear .
The left side indicates the rate of supply of energy from intrinsic sources .
These shows the rate of waste and of storage of energy in this unit volume .
The remainder , therefore , indicates the rate at which energy is passed out from the unit volume , : and the flux represents the flux of energy necessitated by the localisation of energy and its waste , when and are connected in the manner shown by ( 94 ) and .
There might also be an dependent circuital flux of energy , but , being useless , it would be superfluous to bring it in .
The very important formula ( 97 ) was first discovered and interpreted by Professol POINTING , and independently discovered and interpreted little later by myself in an extended form .
It will be observed that in my mode of proof there is no limitation as to homogeneity or isotropy as regards the perlnittivity , inductivity , and conductivity .
But and should be symmetrical .
On the other hand , ) do not require this limitation in deducing ( 97 ) * The method of treating MAXWELL'S scheme employed in the texb ( hrst introdDced in " " Electromagnetic Induction and its Propagation The Electrician , ' January 3 .
and Iater ) is important to recognize tlJat this flux of energy is not dependent upon the motion of the medium , for it is assumed explicitly to be at rest .
The vector therefore , be a flux of the kind before discussed , unless possibly it be merely a that is concerned .
The only dynamical analogy with which I am acquainted which seems at all f , isfactory is that furnished by Sir W. fHOMSON'S theory of a rotational ether .
'Take bhe case of , and and constants , that is , pure ether uncontaminated by ordinary matter .
Then , . . . . .
( 98 ) . . . . . . . . . .
( 99 ) Now , let be velocity , density ; then , by ( 99 ) , \mdash ; curl is the translational force due to the stress , which is , therefore , a otating stress ; thus , ; . . . .
100 and is the torque .
The coefficient represents the compliancy or reciprocal of the quasi-rigidity .
The kinetic energy represents the magnetic energy , and the potential energy of the rotation represents the electric energy ; ] the flux of energy is VEH .
For the activity of the torque is curl and the translational activity is curl E. Their sum is may , , be tely tcrmed the Duplex method , since its characteristics are the exhibition of the electric , magnctic , and gnetic relations in a duplex form , symmetrical with espect to the electric and netic sides .
But it is not merely of exhibiting the relations in a manner suitable to the snbject , inging to light useful reIations which were formerly hidden from view by the intet'velltion of the vector-potential and its pal'asites , but constitutes a method of WOl.king as well : The ] are considerable difficulties in the way of the practical employment of MAXWELL'S equations of propagalion , as they stand in his treatise .
These difficulties are greatly nified when we pro ceed to general cases , involving heterogeneity and eolotropy and motion of the medium supporting the fluxes .
The duplex method supplies what is wanted .
Potentials do not appear , at least initially .
They are as auxiliary functions which do not represent any physical stat ' , of the medium .
In special problems they may be of great service for calculating purposes ; but in general iuvestigations their avoidance simplifies mattel'S greatly .
The slate of the field is settled by and and these the primary objects of attention in the duplex system .
As the papers to ich I have referred are not eadily accessible , I may take this opportunit .
of mentioning bhat a Reprint of my ' Electrical Papers ' is the press ( MACMII , and Co and that tho first volume is ready .
FLUXES OF ENERGY IN THE ELECTROMAGNETIC FIELD .
-di making VEH the flux of energy .
* All attempts to construct an elastic solid analogy with a distortional stress fail to give satisfactory results , because the energy is wrongly localised , and the flux of energy incorrect .
Bearing this in mind , the above analogy is at first very enticing .
But when we come to remember that the in ( 98 ) should be , and find extraordinary difliculty in extending the analogy to include the conduction current , and also remember that the electromagrnetic stress to be accounted for ( in other words , the known mechanical forces ) , he perfection of the analogy , as far as it goes , becomes disheartening .
It would further seem , from the explicit assumption that in obtaining above , that no analogy of this kind can be sufficiently comprehensive to form the basis of a physical theory .
We must go altogether beyond the elastic solid with the additional property of rotational elasticity .
I should mention , to avoid misconception , that Sir W. THOMSON does not push the analogy even so far as is done above , or give to and the same interpretation .
The particular meaning here given to is that assumed by Professor LODGE in his " " Modern Views of Electricity on the ordinary elastic solid theory , however .
I have found it very convenient from its making the curl of the electric force be a Newtonian force ( per unit volume ) .
When essed electric force produces disturbances , their ] source is , as I have shown , not the seat of , but of curl .
So we may with facility translate problems in electromagnetic waves into elastic solid problems by taking the electromagretic source to represent the mechanical source of motion , impressed Newtonian force .
Examination of the Flux of in Medium , stabti of the Measof " " True\ldquo ; Cu S17 .
Now pass to the more general case of a moving medium with the equations curl . . . . .
( 101 ) \mdash ; cjurl , . . . . . .
( 102 ) where is , for brevity , what the torce of the flux becomes after deducting the intrinsic and motional forces ; and similarly for From these , in the same way as before , we deduce ; . . .
( 103 ) and it would seem at first sight to be the same case again , but with impressed forces * This form of application of the rotating ether I gave in ' The ician , ' January 23 , 1891 , p. 360 .
and instead of and , whilst the POINTING flux requires us to reckon only and as the effective electric and magnetic forces concerned in i But we must plainly first .
We have , by ( 86 ) , ( 87 ) , used in ( 103 ) , . . .
( 104 ) Now here we have ( 105 ) will be observed that the constant , which usually appears in the electrical equations , is absent from the above investigations .
This demands a few wordB of explanation .
The units employed in the text are rational units , founded upon the principle of continuity in of vector functions , and the corresponding appropriate measure of discontinuity , , by the amount of divergence .
In popular language , the unit pole sends out one line of force , in the rational system , instead of lines , as in the irrational systcm .
The effect of the rationalisat , ion is to introduce into the formulae of central forces and potentials , and to abolish the swarm of that appear in the practical formulae of the practice of theory on lines , which receives its fullest and most appropriate expression in the rational method .
The system was explained by me in ' The Electrician , ' in 1882 , and applied to bhe general theol.y of potentials and connected functions in 1883 .
( Reprin , vol. 1 , p. 199 , and later , especially p. 262 .
) I then eturned to irrational muJae because I did not think , then , that a reform of the units was practicable , partly on account of the labours of the B. A. Committee on Electrical Units , and partly on accoullt of the ignorance of , and indifference to , tlJeoretical matters which prevailed at that time .
But the circumstances have greatly changed , and I do think a change is now practicable .
There has been great advance in the knowledge of the meaning of MAXWELL'S theory , and a diffusion of this owledge , not scientific men , but amongst a large body of practicians called into existence by the extension of the practical applications of electricity .
Electricity is becomiug , not only a scicnce , but also a very practical one .
It is fibhing , therefore , that learned traditions should not be allowed to control matters too eatly , and that the units should be ationalised .
To make a beginning , I am employing rational units throughout in my work on ' Electromagnetic Theory commenced in ' Electrician , ' in January , 1891 , and continued as fast as circumstances will permit ; to be republished in book form .
Section XVII .
October , will be found stated more fully the natore of the challge proposed , and the reasons for it .
I oint out , in conclusion , that as regards theoretical eatises and is no difficulty in the way , since the connection of the rational and irrational units may be explained separately ; and I express the belief that when the merits of the rational system arc fully recognised , there will arise a demand for the rationalisation of the practical units .
We are , in the opinion of men qualified to judge , within a measurable distance of adopting the system in England .
Surely the smaller reform I advocate should precede this .
To put the matter })Jainly , the present system of units contains an absurdity running all through it of the nature would exist in the metric system of common units were we to define the unit area to be the area of a circle of unit diameter .
The absurdity is only different in being less obvious in the electrical case .
It would not matter much if it were nob hab electricity is a practical science .
OF IN THE FIELD .
Comparison of the third with the second form of ( 105 ) defines the generalised meaning of when is not a mere scalar .
Or thus , ' . . .
( 106 ) representing the time-variation of due to riation in the 's only .
Similarly . . . . . . .
( 107 ) with the equivalent meaning for generalised .
Using these in ( 104 ) we have the result .
( 108 ) Here we have , besides , terms indicating the activity of a translational force .
Thus is the force on electrification , and its activity .
Again , ; so that we have and , similarly , .
( 109 ) the generalised meaning of which is indicated by \mdash ; ; ( 110 ) where , in terms of scalar products involving and . . . . . . . . . . .
( 111 ) This is also the activity of a translational force .
Similarly , is the activity of a translational force .
Then again GVDq . . . . .
( 113 ) expresses a activity .
them all in ( 108 ) it becomes .
( 1]4 ) It is clear that we should make the factor of be the complete translational force .
But that has to be found ; and it is equally clear that , although we appear to have usted all the terms at disposal , the factor of in ( 114 ) is not the complete force , because there is no term by which the force on intrinsically magnetised or electrized matter could be exhibited .
These involve and .
But as we have , . . .
115 a possible way of bringing them in is to add the left member and subtract the right member of ( 115 ) from the right member of ( 114 ) ; bringing the translational force to , say , where . . . .
( 116 ) But there is still the right number of to be accounted for .
have , . . . . .
( 117 ) and , by using this in ( 114 ) , through ( 115 ) , ( 116 ) , ( 117 ) , we bring it to ( 118 ) or , transfel.ring the terms from the right to the left side , ( n9 ) Here we see that we have correct form of activity equation , though it may not be the correct form .
Another form , equally probable , is to be obtained by bringing in Veh ; thus curl \mdash ; cnrl , . . . .
( 120 ) which converts ( 119 ) to ( 121 ) where is the translational force curl H. curl E. , . . .
FLUXES OF ENERGY IN THE ELECTROMAGNETIC which is perfectly symmetrical as regards and , and in the vector products utilises the fluxes and their complete forces , whereas former forms did this only partially .
Observe , too , that we have only been able to bring the activity equation to a correct form ( either ( 119 ) or ( 122 ) ) by making be the activity of intrinsic force , which requires that should be the true electric current , according to the energy criterion , not J. S 18 .
Now , to test ( 119 ) and ( 121 ) , we ) interpret the flux in ( 121 ) , or say , . . .
( 12.3 ) which replaced the POINTING flux when , along with the other changes .
Since reduces to when , there must still be a POINTING flux when is finite , though we do not know its precise form of expression .
There is also the stress flux of energy and the flux of energy by convection , making a total flux , .
, . .
( 124 ) where is the POINTING flux , that of the stress , whilst means convection of energy connected with the translational force .
We should therefore have . . . . .
to express the continuity of energy .
More explicitly But here we may simplify by using the result ( 69 ) with , however , , making ( 126 ) becolne , . . .
( 127 ) where is the torque , and a the spin .
Comparing this with ( 121 ) , we see that we require , . . .
with a similar equation when ( 119 ) is used instead ; and we have now to separate the right member into two parts , one for the POINTING flux , the other for the stress flux , in such a way that the force due to the stress is the force in ( 121 ) , ( 122 ) , or the force in ( 119 ) , or similarly in other cases .
It is unnecessary to give the failures ; the only one that stands the test is ( 121 ) , which satisfies it completely .
I argued that ( \mdash ; ( 129 ) the probable of the POINTING flux in the case of a moving not , because when a medium is endowed with a translational motion , the transllission of nces through it takes place just as if it were at rest .
With this expression ( 129 ) for , we have , identically , . . . .
( 130 ) Therefore , by ( 128 ) and ( 130 ) , we get , . . .
( 131 ) to represent the negative of the Stress flux of energy , so that , finally , the fully significant equation of activity is . . . . .
( 132 ) This is , of course , all identity , subject to the eleotromagnetic equations we started from , is only one of the multitude of forms which may be given to it , many being far simpler .
But the particular importance of this form arises from its being the only form apparently possible which shall exhibit the principle of continuity of energy without outstanding terms , and without loss of generalityj and this is only possible by taking as the proper flux for to work upon *In the original an el.roneous estimate of the of was used in some of the above tions .
This is COl.rected .
The following contains full details of the calculation .
We require the value of , or of -1E2 is the linear operator whose components are the iations ( for the same matter ) , of those of .
The calculation is very lengthy in terms of these components .
But it is not difficult .
Tn ( 27 ) , ( 28 ) , we have if the vecfors al.egivell by We , thercfo ] , have .
The part played by the dots is to clearly separate the scalar products .
Now suppose that the eolotropic property symbolised by is intrinsically unchanged by the shift of the matter .
The mere translation does not , therefore , affect it , nor does the distortion ; but the rotation FLUXES OF NERGY IN THE ELEOTRO ) IELf ) .
Derivation of the Electric and resses and Forces from the Flux S19 .
It will be observed that the convection of energy disappears by occurring twice oppositely signed ; but as it comes necessarily into the expression for the stress flux of , I have preserved the cancellino , terms in ( 132 ) .
A comparison of the stress flux with the PoYNTrrG flux is interesting .
Both are of form , vector products of the electric and magnetic forces with convection terms ; but whereas in the latter the forces in the vector product are those of the field ( i.e. , only intrinsic forces deducted from and H ) , in the former we have the motional forces and combined with the complete and of the fluxes .
Thus the stress depends does .
For if we turll round an eolotropic portion of matber , keeping unchanged , the value of is altered by the rotation of the principal axes of .
with the matter , so that a torque is required .
In equation , then , to )roduce ( , we keep constant , and let the six vectors , , C3 l.otate as a crid body with the spin curl .
But when a vector magnitude is turned round in this way , its rate of time-change is .
Thus , for , we may put Va throughout .
Therefore , by , . . .
In this use the parallelepidedal transformation ( 12 ) , and it becomes , . . . . . . . . .
( 13 ) by , if is conjugate to ; that is , .
So , when , as in the case , we have and similarly DVEa aVDE , Now the tol'quoal.ising from tho strcss is ( see ( 139 ) ) so we have Sa torquc spin .
The variation allowed to may seem to conflict with lStalCy renco vectors general .
But they merely vary for a temporary purpose , bein fixed in the mattel instead of in spacc .
But we may , perhaps better , discard altogethel , and use any independent , 1 , instead , making wherein the 's are propc.l .
lie choson to suit the new axes .
The calculation then procoods beforc , entirely on the fluxes , however they be ) , in this respect resembling the electric crIletic energies .
To exhibit the stress , we , or . . . . .
( 1.3.3 ) In this use the for and , giving VEVDq ; . . . . . . . . . . . . .
( 1.34 ) where observe the singularity that has changed its sign .
The first set belongs to the netic , the second to the electric stress , since we see that the complete stress is thus divisible .
The divergence of being the activity of the stress-variation per unit volume , its component is the activity of the stress per unit surface , that is , ( .Hq \mdash ; Nq .
T ) .
( 135 ) The stress itself is therefore the valne of arising from the variation of 1 , , and the other half from the c 's , provided is otation Or we may choose the thrce principal axes of in the body , when 1 , ] ] coincide with , and therefore move with them .
Lastly , we may proceed thus:\mdash ; EVaD -DVaE aVDE .
is , replace by Va when he operands are and D. This is the correct resuIt , but it is not easy to the process ectly and plainly ; although the clue is given by observing that what we do is to take a difference , from which the time-variation of disappears .
If it is that is kept constant , the resulb is aVED , the negative of the above .
It is also worth noticing that if we split np into we shall have These Ec ; so that , in the expansion of the torque , the cross-torqucs not , which are unequal , but aro each equal to half the sum of products .
FLUXES Oli ENERGY IN THE CTROMAGNETIC F ( E.DN \mdash ; NU ) : . . . . . . .
divided into electric and magnetic portions .
This is with restriction to symmetrical and , and with persistence of their forms as a particle moves , but is otherwise unrestricted .
Neither stress is of the symmetrical or irrotational type in case of eolotropy , and there appears to be no getting an irrotational stress save by arbitrary assumptions which destroy the validity of the stress as a correct deduction from electromagnetic equations .
But , in case of isotropy , with consequent directional identity of and , and of and , we see , by taking in turns parallel to , or perpendicular to in the electric case , and to in the magnetic case , that the electric stress consists of a tension parallel to combined with an equal lateral pressure , whilst the magnetic stress consists of a tension parallel to combined with an equal lateral pressure .
There are , in fact , MAXWELL'S stresses in an isotropic medium homogeneous as regards and .
The difference from MAXWELL arises when and are variable ( including abrupt changes from one value to another of and ) , and when there is intrinsic magnetisation , MAXWELL'S stresses and forces being then different .
The stress on the plane whose normal is VEH , is , ( 1.37 ) reducing simply to a pressure in lines parallel to in case of isotropy .
S .
To find the force , we have \amp ; .
\amp ; .
curl E.D . . . . . . . . .
( 188 ) where the unwritten terms are the similar magnetic terms .
This being the component of , the force itself is given by ( 122 ) , as is necessary .
It is V curl that expresses the translational force on intrinsically magnetised matter , and this harlnonises with the fact that the flux due to any impressed force depends solely upon curl Also , it is that explains the forcive on elastically magnetised , e.g. , FARADAY'S motion of matter to or from the places of greatest intensity of the field , independent of its direction .
If be the torque , it is given by VSN ; therefore ( 139 ) But the matter is put more plainly by considering the convergence of the stress flux of energy and dividing it into translational and other parts .
Thus , . . . . .
( 140 ) where the ierms following Fq express the sum of the distortional and rotational activities .
Shorter of going from the to the flux of Energy , Stresses , Forces .
S 21 .
I have given the investigation in SS17 to 19 in the form in which it occurred to me before I knew the precise nature of the results , being uncertain as regards the true measure of current , the proper form of the POINTING flux , and how it worked in harmony with the stress flux of energy .
Bub knowing the results , a short demonstration may be easily drawn up , tbough the former course is the most instructive .
Thus , start now from curl ( curl ( . . . . .
( 141 ) on the understanding that and are the currents which make and the activities of and the intrinsic forces .
Then , . . .
( 142 ) where ; . . .
( 143 ) and we now take this to be the proper form of the pointing flux , Now develop and thus : \mdash ; curl e , by ( 93 ) ; curl curl , by ( 88 ) and ( 91 ) ; , ( 26 ) , magnetic terms , magnetic terms . . . . . . . . . . . . . . . . .
( 144 ) FLUXES OF ENERGY IN THE ELECTROMAGNETIC FIELD .
Now here so that the terl1ls in the third pair of brackets in ( 144 ) represent with the generalised meaning before explained .
So finally ( ) , ( 14.5 ) which brings ( 142 ) to , ( 146 ) which has to be interpreted in accordance with the principle of continuity of Use the form ( 127 ) , first , howeyer , eliminating Fq by means of which brings ( 127 ) to ; , . . .
and now , by comparison of ( 147 ) with ( 146 ) we see that ; ( 148 ) from which , when and do not change intrinsically , we conclude that ( 149 ) as before .
In this method we lose sight altogether of the translational force which formed so prominent an object in the former method as a guide .
Some Remarks on HERTZ'S relating to th Stresses .
S22 .
Variations of and in the same portion of matter may occur in different ways , and altogether independently of the strain variations .
Equation ( 146 ) how their influence affects the energy transformations ; but if we consider only such changes as depend on the strain , i.e. , the small changes of value which and undergo as the strain changes , we may express them by thirty-six new coefficients each there being distortion elements , elements , and so reduce the expressions for and in ( 148 ) to the form suitable for exhibiting the change in and in the stress function .
As is usual in such cases of secondary corrections , the magnitude of the resulting formula is out of all proportion to the importance of the correction terms in relation to the imary formula to which they are added .
Professor H. HERTZ has considered question , and also refers to VON HELMOLTZ'S previous investigation relating to a fluid .
The and can then only depend on the density , or on the compression , so that , single coefficient takes the place of the thirty-six .
But I cannot quite follow HERTZ'S stress investigation .
First , I would remark that in developing the expression for the distortional ( plus rotational ) activity , he assumes that all the coefficients of the spin vanish identically ; this is done in order to make the stress be of the irrotational type .
But it may easily be seen that the assumption is inadmissible by examining its consequence , for which we need only take the case of and intrinsically constant .
By ( 139 ) we see that it makes and therefore ( since the electric and magnetic stress are separable ) , VHB , and VED ; that is , it produces directional identity of the force and the flux and of the force and the flux B. This means isotropy , and , therefore , breaks down the investigation so far as the eolotropic application , with six and six coefficients , goes .
Abolish the assumption made , and the stress will become that used by me above .
Another point deserving of close attention in HERTZ'S investigation , relates to the principle to be followed in deducing the stress from the electromagnetic equations .
Translating into my notation it would appear to amount to this , the a priori assumption that the quantity , . . . . . . . . . . . .
( 150 ) where indicates the volume of a moving unit element undergoing distortion , may be taken to represent the distortional ( 'plus rotational ) activity of the magnetic stress .
Similarly as regards the electric stress .
Expanding ( 150 ) we obtain .
( 151 ) Now the second circuital law ( 90 ) may be writtep -curl .
( 152 ) * ' WIgDEMANN'S Annalen , ' FLUXES OF IN THE LECTROMAGNETIC FILLED .
Here ignore ignore the curl of the force , ) we obtain , by using ( 152 ) in ( 151 ) , , . . .
( 163 ) which represents the distortional activity ( my form , not equating to zero the coefficients of curl in its development ) .
We , therefore , derive the magnetic stress in the manner indicated , that is , from ( 150 ) , with the special .
of later stated , and the ignorations or nullifications .
In a similar manner , from the first circuital law ( 89 ) , which may be written curl , ( 154 ) we can , by ignoring the conduction current and the curl of the force , obtain \mdash ; , ( 155 ) which represents the distortional activity of the electric stress .
The difficulty here seems to me to make it evident priori that ( 150 ) , with the special reckoning of should represent the distortional activity ( plus rotational understood ) ; this interesting property should , perhaps , rather be derived from the magnetic stress when obtained by a safe method .
The same applies to the electric stress .
Also , in ( 150 ) to ( 155 ) we overlook the POYNrJNG flux .
I am not sure how far this is intentional on Professor HEBTZ'S part , but its neglect does not seem to give a sufficiently comprehensive view of the subject .
The complete expansion of the magnetic distortional activity is , in fact , ; ( 156 ) and similarly , that of the electric stress is .
( 157 ) It is the last term of ( 156 ) and the last term of ( 157 ) , together , which bring in the POINTING flux .
Thus , adding these equations , ' where ; so we come round to the equation of activity , in the form ( 146 ) , by using ( 159 ) in .
of Stress-vector , and Application to the Surface two Regions . .
The electromagnetic stress , PN of ( 149 ) and ( 136 ) may be put into another interesting form .
We may write it . . . . . .
( 160 ) Now , ND is the surface equivalent of and NB of ; whilst VNE and VNH the surface equivalents of curl and curl H. We may , therefore , write , . . . . . . .
( 161 ) and this is the force , reckoned as a ) , on unit of the surface whose normal is N. Here the accenbed letters are the surface equivalents of the same quantities unaccented , which have reference to unit volume .
Comparing with ( 122 ) we see that the type is preserved , except as regards the terms in due to variation of and in space .
That is , the stress is represented in ( 101 ) as the translational force , due to and , on the fictitious electrification , netification , electric current , and magnetic current produced by imagining and to terminate at the surface across which is the stress .
The coefficient which occurs in ( 161 ) is understandable by supposing the fictitious quantities matter\ldquo ; and " " current to be distributed uniformly within a very thin layer , so that the forces and which act upon them do not then terminate quite abruptly , but fall off gradually through the layer from their full values on one side to zero on the other .
The mean values of and through the layer , that is , and are thus the effective electric and magneti forces on the layer as a whole , per unit volume density of matter or current ; or and per unit surface density when the layer is indefinitely reduced in thickness .
Considering the electric field only , the quantities concerned are electrlfication and etic current .
In the magnetic field only they are magnetification and electr current .
Imagine the medium divided into two regions A and , of which A is internal , external , and let be the unit normal from the surface into the external region .
The mechanical action between the two regions is fully represented by the stress over their interface , and the forcive of upon A is fully represented by the and in acting upon the fictitious matter and current produced on the boundary of , on the assumption that and terminate there .
If the normal and be drawn the other way , thus negativing them both , as well as the fictitious matter and current on the then it is the forcive of A on that is repreFLUXES OF ENERGY IN THI LECT ROMAGNETIC 1 sented by the action of and in A on the new interfacial matter current .
That is , the in the region A may be done away with altoo-ethelb , because their abolition will immediately introduce the fictitious matter and current so far as is concerned , to the influence of the region A. Similarly and in may be abolished altering them in A. And , generally , ) of the medium may be taken by itself and regarded as being subjected to an system of forces , when treated as a rigid body .
S24 .
When and do not vary in space , we do away with the forces and \mdash ; , and make the form of the surface and volume forces agree .
We may then regard every element of or of as a source out from itself displacement and induction isotropically , and every element of or as causing induction or displacement according to AMPERE'S rule for electric current and its analogue for magnetic curren , .
Thus , ( 1613 ) where is a unit vector drawn from the infinitesimal unit volume in the summation to the point at distance where or is reckoned .
Or , introducing potentials , ' curl These apply to the whole medium , or to any portion of the same , with , the latter case , the surface matter and current included , there being no or outside the region , whilst within it and are the same as due to the matter and current in , he whole region matter and ' current and G ) .
But is no known general method of finding the potentials when and We may also divide and into two parts each , say and due to und current in the region , and due to matter and current in ) the gion B surrounding it , determinable in the isotropic homogeneous case by the above formulae .
Then we may ignore and in estimating the forcive on the matter and current in the region ; thus , , . . . . .
( 166 ) where , and curl Hcurl H in region , is the resultant force on the region , and , . . . . .
( 167 ) is the resultant force on the region ; the resultant force on A due to its own and being zero , and similarly for B. These resultant forces are equal and opposite , and so are the equivalent surface-integrals and . . . . .
( 169 ) taken over the interface .
The quantity summed is that part of the stress-vector , which depends upon products of the of one region and the of the other , \amp ; c. Thus , for the magnetic stress only , , . . .
( 170 ) and it is the terms in the second and fourth brackets ( which , be it observed , are not equal ) which together make up the magnetic part of ( 168 ) and ( 169 ) or their negatives , according to the direction taken for the normal ; that is , since , . . .
( 171 ) where the first six expressions are interfacial summations , and the four last summations throughout one or the other region , the last summation applying to either region .
No special reckoning of the sign to be prefixed has been made .
The notation is such that , \amp ; c. , \amp ; c. The comparison of the two aspects of electromagnetic theory is exceedingly curious ; namely , the precise mathematical equivalence of ' ' explanation\ldquo ; by roeans of instantaneous action at a distance between the different elements of matter and current , each according to its kind , and by propagation through a medium in time at a finite velocity .
But the day has gone by for any serious consideration of the former view other than a mathematical curiosity .
OF ENERGY IN IAGNLTIC 1 I1 .
Quaternionic ) of Stress-Vector . .
We may also notice the Quaternion form for the stress function , which is vital a part of the mathematics of forces varying as the inverse square of the distance , and of potential theory .
Isotropy being understood , the electric stress may be written , . . . . . . . . . .
( 172 ) where the quantity in the square brackets is to be understood quaternionically .
It is , howeyer , a pure vector .
Or , , ( 173 ) that is , not counting the factor , the quaternion is the same as the quaternion ; or the same operation which turns to also turns to .
Thus , and are in the same plane , and the angle between and equals that between and ; and and are on the same side of when makes an acute angle with N. Also , the tensor of is [ I , so that its normal and tangential components are and , if Otherwise , . . . . . . . . .
( 174 ) since the quaternionic reciprocal of a vector has the reverse direction .
The sponding volume translational force is , . . . . .
which is also to be understood quaternionically , and expanded , and separated into parts to become physically significant .
I only use the square brackets in this paragraph to emphasise the difference in notation .
It rarely occurs that any advantage is gained by the use of the quaternion , in saying which , I merely repeat what Professor WILLARD GIBBS has been lately ] us ; and I further believe the disadvantages usually far outweigh the advantages .
Nevertheless , apart from practical application , and at it from the purely quaternionic point of view , I ought ro also add that the invention of quaternions must be regarded as a most remarkable feat of human ingenuity .
Vector analysis , without quaternions , could have been found by any mathematician by carefully examining the mechanics of the Cartesian mathematics ; but to find out quaternions required a genius . .
O. HEAVISl D ON THE FORCES , STRESSES , AND Remarks the Force in Free Ether : S26 .
The little vector Veh , which has an important influence in the activity equation .
where and are the motional forces has an form , , by expansion , qVDB qVEH , . . . . . . . .
( 176 ) if be the speed of propagation of disturbances .
We also have , in connection therewith , the equivalence , . . . . . . . . . . .
( J77 ) always .
The translational force in a non-conducting dielectric , free from electrification and intrin S , is or , approximately , VDB .
( 178 ) The vector VDB , or the flux of energy divided by the square of the speed of propagation , is , therefore , the momentum ( trans]ational , not magnetic , which is quite a differe , nt thing ) , provided the force in the complete force from all causes acting , and we neglect the small terms and But have we any right to safely write , ( 179 ) where is the density of the ether ?
To do so is to assume that is the only force acting , and , therefore , equivalent to the time-variation of the momentum of a moving particle .
* Now , if we say that there is a certain forcive upon a conductor supporting electric current ; or , equivalently , that there is a certain distribution of stress , the magnetic stress , acting upon the same , we do not at all mean that the accelerations of momentum of the different parts are represented by the translational force , the electromagnetic force It is , on the other hand , a dynamical problem in which the electromagnetio force plays the part of an impressed force , and similarly as regards the magnetic * Professor J. J. THOMSON has endeavoured to makc practical use of the idea , ' Phil. March , 1891 .
See also my alticle , ' The Electrician , ' January 15 , 1886 .
ENERGY IN THE ELECTROM .
AGNLTIC FIELD .
stress ; the actual forces stresses being only determinable from a knowledge of the mechanical conditions of the conductor , as its density , elastic constants , and the way it is constrained .
Now , if there is ally dynamical meaning at all in the electromagnetic equations , we must treat the ether in precisely the same .
But we do not know , and have not formularised , the equations of motion of the ether , but only the way it propagates disturbance through itself , with due allowance made for the effect thereon of given motions , and with formularisation of eaction between electromagnetic effects and the motion .
Thus the theory of the stresses and forces in the ether and its motions is an lltisolved problem , only a portion of it being known so far , i.e. , assuming that the Maxwellian equations do express the known part .
When we assume the ether to be motionless , there is a partial similarity to the theory of the propagation of vibrations of infinitely small range in elastic bodies , when the effect thereon of the actual translation of the matter is neglected .
But in ordinary electromagnetic phenomena , it does not seem that the tion of can make any sensible difference , because the speed of propagation of disturbances through the ether is so enormous , that if the were stirred about round a example , there would be an almost instantaneous adjustment of the netic induction to what it would be were the ether at rest .
Consideration of the Str.esses.\mdash ; Inde minateness .
S27 .
In the following the stresses are considered from the static point of view , principally to examine the results produced by changing the form of the stress function .
Either the electric or the magnetic stress alone may be taken in hand .
Start then , from a knowledge that the force on a magnetic pole of strength is , where is the polar force of any distribution of intrinsic magnetisation in a medium , the whole of which has unit inductivity , so that conv h . . . . .
( 180 ) measures the density of the fictitious ' magnetic\ldquo ; matter ; being the intrinsic force , or , since here , the of magnetisation .
The induction is .
This rudimentary theory locates the on a magnet its poles , or internal , by . . . . . . . . . . .
( 181 ) The component of is because curl .
Therefore is the appropriate stress , of irrotational type .
Now , tain we may be .
O. ON THE FORCES , STRESSES , AND about the stress in the interior of a magnet , there can be no question as to the possible validity of this stress in the air outside our magnet , for we know that the force is then a polar force , and that is all that is wanted , and being merely auxiliaries , derived from R. Now consider a rion A , containing magnets of this kind , enclosed in , the rest of space , also containing magnets .
The mutual force between the two regions is expressed by over the interface , which we may exchange for through either region A or , still on the assumption that remains polar .
But if we l.emove this restriction upon the nature of , and allow it to be arbitrary , say in region or in any portion thereof , we find NF curl R. ; or if curl R. This gives us , from a knowledge of the external magnetic field of polar magnets only , the mechanical force exerted by a magnet on a containing whatever that Irlay be , provided it be measurable as above ; and without any experimental knowledge of electric currents , we could now predict their mechanical effects in eve1y respect by the principle of the equality of action and reaction , not merely as regards the mutual influence of a magnet and a closed current , but as regards the mutual influence of the closed currents themselves ; the magnetic force of a closed current , for instance , being the force on unit of , is equivalently the force exerted by on the closed current , which , by the above , we know .
Also , we see that according to this magnetic notion of electric current , it is necessarily circuital .
At the same time , it is to be that our real knowledge must cease at the boundary of the region containing electric current , a metallic conductor for instance ; the surface over which is reckoned , on one side of which is the magnet , on the other side electric current , can only be pushed up as far as the conductor .
The stress may tlJerefore cease altogether on reaching the conductor , where it forms a distribution of surface force fully representing the action of the magnet on the conductor .
Similarly , we need not continue the stress into the interior of the magnet .
Then , so far as the resultant force on the magnet as a ] , in translating or rotating it , and , similarly , so far as the action on the conductor , is concerned , the simple stress of constant tensor , varying from a tension parallel to to an equal pressure laterally , acting in the medium between the magnet and conductor , accounts , by its terminal pulls or pushes , for the mechanical forces on them .
The lateral pressure is especially prominent in the case of conductors , whilst the tension goes more or less out of sight , as the immediate cause of motion .
Thus , when parallel currents appear to attract one another , the conductors are really pushed together by the lateral pressure on each conductor being greater on the side remote from the other than on the near side : whilst if the currents are oppositely directed , the pressure on the near sides is greater than on the remote sides , and they appear to repel one another , FLUXES or ENERGY IN THE ELECTROMAGNETIC FIELD .
The effect of continuing the stress into the interior .
a conductor of unit inductivity , according to the same law , instead of stopping it on its boundary , is to distribute the translational force bodily , to the formula , instead of superficially , according to .
In either case , of course , the conductor must be strained by the magnetic stress , with the consequent production of a mechanical stress .
But the strain ( and associated stress ) will be different in the two cases , the applied forces being differently localised .
The effect of the stress on a straight portion of a wire supporting current , due to its own field only , is to compress it laterally , and to lengthen it .
Besides this , there will be resultant force on it arising from the different pressures on its opposite sides due to the proximity of the return conductor or rest of the circuit , tending to move it so as to increase the induction through the circuit per unit current , that is , the inductance of the circuit . .
If now , we bring an stically magnetisable body into a magnetic field , it modifies the field by its presence , causing more or less induction to go through it than passed previously in the air it replaces , according as its inductivit .
exceeds or is less than that of the air .
The forcive on it , considered as a rigid body , is completely accounted for by the simple stress in the air outside it , reckoned according to the changed field , and supposed to terminate on the surface of the disturbing body .
This is true whether the body be isotropic or heterotropic in its inductivity ; nor need the induction be a linear function of the magnetic force .
It is also true when the body is intrinsically magnetised ; or is the seat of electric current .
In short , since the external stress depends upon the netic force outside the body , when we take the external field as we may find it , that is , as modified by any known or unknown causes within the body , the corresponding stress , terminated upon its boundary , fully represents the forcive on the body , as a whole , due to magnetic causes .
This follows from the equality of action and reaction ; the force on the body due to a unit pole is the opposite of that of the body on the pole .
If we wish to continue the stress into the interior of the body , surrounded on all sides by the unmagnetised medium of unit inductivity , as we must do if we wish to arrive ultimately at the mutual actions of its parts , and how they are modified by variations of inductivity , by intrinsic magnetisation , and by electric current in the , we may , so far as the resultant force and torque on it are concerned , do it in any way we please , provided we do not interfere with the stress outside .
For the internal stress , of any type , will have no resultant force or torque on the body , and there is merely left the real external stress .
Practically , however , we should be guided by the known relations of magnetic force , induction , magnetisation , and current , and not go to work in a fanciful manner ; furthermore , we should always choose the stress in such a way that if , in its expression , we take the inductivity to be unity , and the intrinsic magnetisation zero , it must reduce to the simple Maxwellian stress in air ( assumed to represent etber here ) . .
O. THLt FORCES , STRESSES , AN I ) But as we do nob know definite]y the forcive arising from the magnetic stress in the interior of a magnet , there are several mulae that suggest themselves as possible .
Kinds of Stress statically suggested . .
Thus , first we have the stress ( 183 ) ; let this be quite general , then ( 1 ) Here is the netic force of the lield , not of the flux B. If is the density of magnetic matter , the convergence of the intrinsic magnetisation , but not otherwise .
In general , it is the density of the matter of the magnetic potential , calculated on the assumption .
The force on a magnet is located in this system at its , whether the magnetisation be intrinsic or induced .
The second term in ( 185 ) represents the force on matter bearing electric current curl R , but has to be supplemented by the first term , unless at the place .
S3O .
Next , let the stress be times as great for the same magnetic force , but be still of the same simple type , being the inductivity , which is unity outside the body , but having any positive value , which may be variable , within it .
Then we shall have ( 2 ) where conv is the density of magnetic matter , being the intensity of intrinsic magnetisation .
The electromagnetic force is made times as great for the same magnetic force ; the force on an intrinsic magnet is at its poles ; and there is , in addition , a force wherever varies , including the intrinsic magnet , and not forgetting that a sudden change in , as at the boundary of a magnet , has to count .
This force , the third term in ( 187 ) , explains the force on inductively magnetised matter .
It is in the direction of most rapid decrease of S31 .
irdly , let the stress be of the same simple type , but taking instead of being the force of the flux , where is as before .
We now haVe ( 3 ) , . . . . . .
where curl h is the distribution of fictitious electric current which produces the same induction as the intrinsic magnetisation , and is , as before , the real current .
It is now -electromagnetic force that acts on an intrinsic magnet , with , ever , force due to , since a magnet has usually large compared with air .
FLUXES OF ENERGY IN THE FIELD .
The above three stresses are all of the simple type ( equal tension and perpendicular pressure ) , and are irrotational , unless be the eolotropic operator .
No change is , the latter case , needed in ( 186 ) , ( 188 ) , whilst in the force formulae ( 187 ) , ( 189 ) , only change needed is to give the generalised meaning to .
Thus , in ( 189 ) , instead of , use or , or , or HB , or , showing the components .
Similarly in the other cases occurring later .
The following stresses are not of the simple type , though all consist of a tension parallel to or combined wit .
an isotropic pressure .
S 32 .
Alter the stress so as to locate the force on an intrinsic magnet bodily upon its magnetised elements .
Add to the stress ( 186 ) , and therefore to its conjngate ; then the divergence of the latter must be added to the -component of the force .
Thus we get , if ( 4 ) But here the sum of the first two terms in may be put in a different form .
Thus , Also iVJI .
These bring ( 191 ) t , o , . . .
( 192 ) where the first component ( the bracketted part ) is MAXWELL'S force on intrinsic magnetisation , and the second his electromagnetic force .
The third , as before , required where varies .
S 33 .
To the stress ( 190 ) add , without the conjugate stress making ( 5 ) we need not discuss , as it is merely a transition to the next form .
34 .
To the stress ( 193 ) add ; we then get , . . . . . . . . . . . .
( 195 ) ( 6 ) , ( 196 ) where are the components of Now if to this last stress ( 195 ) we add , we shall come back to the third stress , ( 188 ) , of the simple type .
Perhaps the most instructive order in which to take the six stresses is ( 1 ) , ( 2 ) , ( 4 ) , ( 5 ) , ( 6 ) , and ( 3 ) ; merely adding on to the force , in passing from one stress to the next , the new part which the tion in the stress necessitates .
To the above we should add MAXWELL'S general stress , which is ( 7 ) where intensity of induced magnetisation .
There is a good , deal to be said against this stress ; some of which later .
Remarks on MAXWELL'S Generul Stress .
S35 .
All the bove force formulae refer to the unit volume ; whenever , therefore , a discontinuity in stress occurs at a surface , the corresponding expression per unit surface is needed ; i. e. , in making a special application , for it is sted labour else .
It might be thought that MAXWELL gives the force ( 198 ) , and in his treatise usually gives surface expressions separately , so none is required in the case of this his force system ( 198 ) .
But this formula will give entirely erroneous results if carried out literally .
It forms no exception to the rule that the expressions require surface MAXWELL'S general stress has the apparent advantage of simplicity .
It merely equires an alteration in the ension parallel to , from to , whilst the lateral pressure remains , when we pass from unmagnetised to magnetised matter .
The FLUXES OF ENERGY IN THE ELECTROMAGNETIC force to which it gives rise is also apparently simple , being merely the sum of two forces , one the electromagnetic , , the other a force on magnetised matter whose component is , both per unit volume , the latter being accompanied ( in case of eolotropy ) by a torque .
Now I the intrinsic and the induced sation , so the force is made irrespective of the proportion in which the magnetisation exists as intrinsic or induced .
In fact , MAXWELL'S " " magnetisation\ldquo ; is the sum of the two without reservation or distinction .
But to unite them is against the whole behaviour of induced and intrinsic magnetisation in the electromagnetic scheme of MAXWELL , as I interpret it .
Intrinsic magnetisation ( using Sir W. THOMSON'S term ) should be regarded as impressed ( I , where is the equivalent impressed magnetic force ) ; on the other hand , " " induced\ldquo ; magnetisation depends on the force of the field .
Intrinsic magnetisation keeps up a field of force .
Induced magnetisation is kept up by the field .
In the circuital law I and therefore behave differently .
There may be absolutely no difference whatever between the magnetisation of a molecule of iron in the two cases of being in a permanent or a temporary magnet .
That , however , is not in question .
We have no concern with molecules in a theory which ignores molecules , and whose element of volume must be large enough to contain so many molecules as to swamp the characteristics of individuals .
It is the resultant magnetisation of the whole assembly that is in question , and there is a great difference between its nature according as it disappears on removal of an external cause , or is intrinsic .
The complete amalgamation of the two in MAXWELL'S formula must certainly , I think , be regarded as a false step .
We may also argue thus against the probability of the formula .
If we have a system of electric current in an unmagnetisable medium , and then change everywhere in the same ratio , we do not change the magnetic force at all , the induction is made times as great , and the magnetic energy times as great , and is similarly distributed .
The mechanical forces are , therefore , times as great , and are similarly distributed .
That is , the translational force in the medium , or VJR , becomes in the second case in which the inductivity is , without other change .
But there is no force brought in on magnetised matter se .
Similarly , if in the medium we have intrin sic magnetisation I , and then alter in any ratio everywhere alike , keeping I unchanged , it is now the induction that remains unaltered , the magnetic force becoming times , and the energy times the former values , without alteration in distribution ( referring to permanent states , of course ) .
Again , therefore , we see that there is no translational torce brought in on magnetised matter merely because it is magnetised .
Whatever formula , therefore , we should select for the ress function , it would certainly not be NAXWELL'S , for cumulative reasons .
When , some six years ago , I had occasion to examine the subject of the stresses , I was unable to arrive at any very nite results , except outside of magnets or conductors .
It was a perfectly indeterminate problem to find the magnetic stress inside a body from the existence of a known , or highly probable , stress outside it .
All one could do was to examine the consequences of assuming certain stresses , and to reject those which did not work well .
After going into considerable detail , the only two which seemed possible were the second and third above ( those of equations ( 186 ) and ( 188 ) above ) .
As regards the seventh ( MAXWELL'S stress equation ( 198 ) above ) , the apparent simplicity produced by the union of intrinsic and induced magnetisation , turned out , when examined into its consequences , to lead to great complication and unnatnralness .
This will be illustrated in the following example , a simple case in which , we can readily and fully calculate all details by different methods , so as to be quite sure of the results we ought to obtain .
eoorked-out Example to Exhibit the Forcives contained in Different Stresses .
S36 .
Given a fluid medium of inductivity , in which is an intrinsic magnet of the same inductivity .
Calculate the attraction between the magnet and a large solid mass of different inductivity .
Here it is only needful to calculate the force on a single pole , so let the magnet be infinitely thin and long , with one pole of strength at distance from the medium , which may have an infinitely extended plane boundary .
By placing a fictitious pole of suitable strength at the optical image in the second medium of the real pole in the first , we may readily obtain the solution .
Let PQ be the interface and the real pole be at A and its image at B. We have first to calculate the distribution of , magnetic force , in both media due to the pole , as disturbed by the change of inductivity .
We have in the first medium , and in the second , therefore has divergence only on the interface .
Let be the surface density of the fictitious interfacial matter to correspond ; its force goes both ways ; the continuity of the normal induction therefore gives , ab distance ?
from , the condition OF NhRGY IN THL FlhLD .
because is the tensor of the magnetic force due to in the medium when of infinite extent .
Therefore .
( 200 ) The magnetic potential , such that is the polar force in either region , is therefore the potential of at A and of over the interface .
But if we put matter at the image , of amount ' ( 201 ) the normal component of on the side due to and the pole will be , ( 202 ) the same ] as before ; the force on the side is , therefore , the same as that due to matter at A and matter at ; whilst on the side the force is that due to matter at A and matter also at , that is , to matter \mdash ; at A. Thus in the medium the force is radial from A as if there were no change of inductivity , though altered in intensity .
The repulsion between the pole and the solid mass is not the repulsion between the matters and of the potential , but is magnetic force at A due to matter at magnetic force at due to matter at ' ( 203 ) becoming an attraction when , making negative .
When , the repulsion is ; when , it is turned into an attraction of equal amount .
Similarly , if we consider the attraction to be the resultant force between and the interfacial matter , we shall get the same result by , . . . . . . . . . .
the quantity summed ( over the interface ) being normal component of magnetic force due to matter in a medium of unit inductivity , or the normal component of duction due to in its own medium .
For this is Another way is to calculate the variation of energy made by displacing either the pol or the mass .
The potential energy is expressed by , . . . . .
( 205 ) where and ; the potentials of matter and , where is the distance from or from to the } ) oint where and are reckoned .
The value of the second part in ( 205 ) , depending upon , comes to , ( 206 ) and its rate of decrease with respect to expresses the repulsion between the pole and the region .
This gives ( 203 ) again .
A fourth way is by means of the quasi-electromagnetic force on fictitious interfacial electric current , instead of matter , the current being circular about the axis of symmetry AB .
The formula for the attraction is curl , . . . . .
207 ) if be the radial .
magnetic force from in its own medium , tensor Here the curl of is represented by the interfacial discontinuity in the tangential induction , or Also the tangential component of is .
Therefore the repulsion is ( 208 ) as before , equation ( 203 ) .
This method ( 207 ) is analogous to ( 204 ) .
S37 .
There are several other ways ofrepresenting the attraction , employing fictitious matter and current ; but now let us change the method , and observe how the attrac tion between the magnetic pole and the iron mass is accounted for by a stress tribution , and its space-variation .
The best stress is the third , equation S31 .
Applying this , we have simply a tension in the first medium UXES OF IN THE ELECTROMAGNETIC FILLED .
and in the second , parallel to and respectively , each combined with an equal lateral pressure , so that tensor of the vector is constant .
But , so far a , s the attraction is concerned , may ignore the stress in the second medium altogether , and consider it as the of the stress-vector in the medium over the surface of the second medium .
The tangential component sulnmed has zero resultant ; the attraction is therefore the sum of the normal components , or where is the angle between and , he normal .
This is the same as ) , if and are the normal and tangential components of ; or ; ( 209 ) which on evaluation gives the required result ( 203 ) .
Bub this method does not give he true distribution of force due to the stresses .
In the first medium there is no translational force , except on the magnet .
Nor is there any translational force in the second medium .
But at the interface , where changes , there is the force \mdash ; per unit volume , and this is represented by the stress-difference at the interface .
It is easily seen that the tangential stressdifference is zero , because , . . . . .
( 210 ) and both the normal induction and the tangential netic force are continuous .
The real force is , therefore , the difference of the normal components of the stressvectors , and is , therefore , normal to the interface .
This we could conclude from the expression \mdash ; .
But since the resultant of the interfacial stress in the second medium is zero , we need not reckon it , so far as the attraction of the pole is concerned .
The normal traction on the interface , due to both stresses , is of amount ( 211 ) per unit area .
Summed up , it gives ( 203 ) again .
That ( 211 ) properly represents the force - when is discontinuous , we may also verify by supposing to vary continuously in a very thin layer , and then proceed to the limit .
The change from an attraction to a repulsion as changes from being greater to being less than , depends upon the relative importance of the tensions parallel the magnetic force and the lateral pressures operative at different parts of the interface .
In the extreme case of , we have tangential , with , therefore , pressure everywhere .
For the other extreme , is normal , and there is pull on the second medium everywhere .
When is finite there is a certain circular are on the interface within which the translational force due to the stress in the medium containing the pole is that medium , whilst outside it the is the ) .
O. HEAyISIDE ON THE FORCLS , , AND way .
But when both stresses are allowed for , we see that when the pull towards the first medium in all parts the interface , and that , becomes a push in parts when A by Consideration of the Circuital Storage and Flnx of Energy . .
We see that the stress in 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 stress , or any stress leading to ( 1 ) force on inductively magnetised matter as this stress does , leads us into grreat difliculties .
By ( 198 ) we see that there is first a bodily force on the whole of the , because it is magnetised , unless .
When summed up , the resultant does not the required ction .
For , secondly , the medium is also magnetised , lnless there is a bodily throughout the whole of it .
When this is summed up ( not counting the force on the magIlet ) , its resultant added on to the former ) still does not make up the attraction , 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 lired attraction .
It is unneces- sary to give the details relating to so improbable a system of force .
Our preference must natm.ally be for a more simple system , such as the previously considered stress .
But there is really no decisive settlement possible from the theoretical statical standpoint , and nothing short of actual experimental determination of the strains produced and their exhaustive analysis would be sufficient to determine the prcper 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 moying medium , combined with certain disbributions of stored energy , we are led to.just one stress-vector , viz. ( 136 ) .
It is , in the etic case , the same as ( 188 ) ; that is , it reduces to the latter when the medium is kept at rest , so that and become 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 the electromagnetic force on electric current , ( b ) the ditto on the fictitious elec current taking the place.of intrinsic magnetisation , ( c ) force depending upon spacevariation of ; we see that the really striking part is ( b ) .
Of all the various ways of representing the forcive on an int1insic it is the most extreme .
The magnetic " " matter\ldquo ; does not enter into it , nor does the distribution of magnetisation ; it is where the intrinsic force has curl that the translational force operates , usually on OF IN ELECT R the sides of a magnet .
From actual experiments with bar magnets , lleedles , \amp ; c. , one would naturally prefer to the polar regions as the seat of translational force .
But the equivalent forcive has one striking recommendation ( apart from t'ne dynamical method of deducing it ) , viz. , that the induction of an intrinsic magnet is determined by curl , not by itself ; and this , I have shown , is true when is imagined to vary , the whole varying states of the fluxes due to impressed force being determined by the curls of and which the sources of the disturbances ( though not of the energy ) .
The rotational peculiarity eolotropic nces d not seem to be a very formidable objection .
Are they not solid ?
As regards the assumed constancy of , a more complete theory must , to be correct , reduce to one assuming constancy of , because , as Lorcl has shown , the assumed law has a limited range of validity , and is therefore justifiable as a preparation for more complete views .
Theoretical requirements are not identical with those of the practical engineer .
But , for quite other reasons , the dynamically determined stress be entirely wrong .
Electric and magnetic " " force\ldquo ; and ] energies are facts .
But it is the total of the energies in concrete cases that should be regarded as the , rather then their distribution ; for example , that , as Sir W. THOMSON proved , the ' mecl ) anical value\ldquo ; of a simple closed current , where is the inductance of the circuit ( coefficient of netico capacity ) , rather than that its distribution in space is given by HB per unit volume .
Other distributions may give the same total amount of energy .
For example , the energ ) of distortion of an elastic solid may be expressed in terms of the square of the rotation and the square of the expansion , if its boundary be held at rest ; but this does not correctly localise the energy .
If , then , we choose some other distribution of the energy for the displacement and induction , we should find quite a different flux of energy .
But I have not succeeded in making any other arrangement than MAXWELL'S work practically , or without an immediate duction of great obscurities .
Perhaps the least certain part of MAXWELL'S scheme , as modified by myself is the estimation of magnetic energy as HB in intrinsic magnets , as well as outside them , that is , by , however may be caused .
Yet , only in this way are thoroughly consistent results apparently enable when the electromagnetic field is considered * il .
' , 188 .3 [ R. HE .
ON , ANI ) APPENDIX .
Received Juue 27 , 1891 .
of the Kinetic Mcthod of rriving at the Stresses to cases of Non-linear the Electric Magnetic Forces the Fluxes .
Preserva- from of 1r)of the of .
It may be worth while to give the results to which we are led regarding the stress and flux of energy when the restriction of simple proportionahty between " " forces\ldquo ; and " " fluxes electric and magnetic respectively , is removed .
The course to be followed , to an interpretable form of the equation of activity , is sufficiently clear in the light of the experience gained in the case of proportionality .
First ssume that the two circuital laws ( 89 ) and ( 90 ) , or the two in ( 93 ) , hold good generally , without any initially stated relation between the electric and its associated fluxes and , or between the magnetic force and its associated { luxes and B. When written in the form most convenient for the present applition , these laws are curl , ( 212 ) \mdash ; .
( 213 ) Now derive the equation of activity in the manner previously followed , and arrange it in the particular form conv V , ( 214 ) which wiJl best facilitate interpretation .
Although independent of the relation between and \amp ; c. , of course the dimensions must , be suitably chosen so that this equation may really represent activity per unit volume in every term .
Now , guided by the previous investigation , we can assume that represents ) rate of supply of energy from intrinsic sources , and also that V , which is a flux of energy independent of , is the correct form in general .
Also , if there be no other intrinsic sources of energy than , and no other fluxes of energy besides thatjust mentioned except the convective flux and that clue to the stress , f , he equation of activity should be representable by Fr , OF IN .
conv [ conv Q , . . . . . . . . . . . . . . . . .
( 215 ) where is the conjugate of the stress vector , the translational force , and the rate of waste and the stored .
energies , whatever they be .
Comparing with the preceding equation ( 214 ) , we see that we require . . . . .
( 216 ) Now assume that there is no waste of energy except by conduction ; then . . . . .
Also assume that These imply that the relation between and is , for ) same particle of matter , an invariable one , and that the stored electric energy is , ' . .
( 21S ) where is a function of D. Similarly , , . . .
( 219 ) expresses the stored magnetic energy , and must be a definite function of B. On these assumptions , ( 216 ) reduces to , . . .
( 220 ) from which the stress-vector follows , namely , Or , ( VDVEN ) VBVHN . . .
( 222 ) Thus , in case of isotropy , the stress is a tension parallel to Ecolnbined with a lateral pressure ; and a tension parallel to combined with a lateral pressure The corresponding translational force is ( HB \mdash ; T ) , . . .
which it is unnecessary to put in terms of the currents .
Exchange , and and , in ( 221 ) or ( 222 ) to obtain the conjugate vector ; from which we obtain the flux of energy due to the stress , ( \mdash ; ) VEVDq VHVBq , . . .
( 224 ) or , ( 225 ) where are the motional electric and )gnetic forces , of the same form as before ( 88 ) and ( 91 ) ; so that the complete form of the equation of activity , showing the fluxes of energy and their convergence , is conv [ \mdash ; conv , ( 226 ) where the above meaning .
There is thus a remarkable preservation of form as compared with the corresponding formulae when there is proportionality between force and flux .
For we produce harmony by means of TTING flux of identical expression and a flux due to the stress , which is also of identical expression , although and now have a more general meanin , of course .
of the above , and on Intrinsic tiomvhen there steresis .
S40 .
In the stress-vector itself ( for either the electric or the magnetic stress ) the relative magnitude of the tension and the lateral pressure vaties unless the curve * As hc invcstigation in Appendix has some pretensions to generality , we should try to settle the amonnt it is irly entihled to .
No objection is likely to be raised to the use of the circuital equations ( 212 ) , , with tlJe restriction of strict proportionality botween and and the fluxes and , or and entirely rellloved ; nor to the estimation of and as the " " true\ldquo ; currents ; nor to the use of the same of flux of electromagnetic hell the medium is stabionary .
For these things are obviously suggested by the preceding investigations , and their justification is in their being found to continue to work , which is the case .
But the use in the text of language appropriate to linear functions , which arose the notation , \amp ; c being the same as before , is unjustifiable .
We may , however , remove this misuse of language , and make the equation ( 226 ) , showing the flux of energy , rest entirely upon the two circuital equations .
In fact , if we substitute in ( 226 ) the relations , , it becomes a way of .
It is , therefore , to , that should look for limitations .
As regards , there does llot seem to be ally limitatiou necessax.y .
That is , there is no kind of imposed between and and and K. This seems to merely from meaning energy wasted for good , and having no further entry into the system .
But as regards , the case is .
For ib seems necessary , in old ] to exclude erlns corresponding to in linear theol.y , when there FLUXES OF ENERGY IN connecting the force and the induction be a line .
Thus , if the curve be of the type shown in the firsb figure , the shaded area will represel1t the stored energy and the tension , and the remainder of the rectangle will represent the lateral pressure .
They are equal when is small ; later on the pressure prepoIJderates , and more and more so the bigger becomes .
But if the curve be of the type shown in the second figure , then , after initial equality the tension prepon derates ; though , later on , when is very big the pressure preponderates .
To obtain idea of the effect , take the concrete example of an infinitely long rod , uniformly axially inductized by a steady current in an overlapping solenoid , and consider the forcive on the rod .
Here both and are axial or longitudinal ; and so , by equation ( 223 ) , the translational force would be a normal on the surface of the rod , acting outwards , of amount per unit area ; being the excess of the lateral pressure in the rod over , the lateral pressurejust outside it .
In case of proportionality of force bo flux , the first pressure is HB , and if there is no intrinsic magnetisation and are equal .
The outward force is therefore positive for paramagnetic , and negative for substances , and the result would be lateral expansion or contraction , since the infinite length would prevent elongation .
rotation , that and should be paralleI , and likewise and B. At any rate , if such terms owed , some modification may be required in the subsequent jkoning of the mechanical force .
other respects , ib is merely implied by that and are definitely connected , likewise and that there is no waste of energy other than that expressed by Q. But if the curve in rod be of the type of the first figure , and the straight line be the air curve to correspond , it is the area that now represents the force per unit area ] the magnetic force has the value .
If the straight can cross the curve , we see that by sufficiently increasing we can make the external air pressure preponderate , so that the rod , after initially expanding , would end by contracting .
If the rod be a ring of large diameter compared with its thickness , the forciye would be approximately the same , , an outward surface force equal to the diflerence of the lateral } ) ressures in the rod and air .
The result would then be elongation , with final retraction when the external pressure came to exceed the internal .
BIDWELL found a pheIjomenon of this kind in iron , but it does not seem possible that the above supposititious case is capable of explainin it , though of course the true explanation may be some respects of a similar nature .
But the circumstances are not the same as those supposed .
The assumption of a definite connexion between , and elastic storage of the energy , is very inadequate to represent the facts of magnetisation of iron , save within a small range .
Magneticians usually plot the curye connecting and , not between and or would be the same , between and , where is the intrinsic magnetisation .
Now when an iron ring is subjected to a given gaussage ( or magneto motive force ) , going through a sequence of values , there is no definite curve connecting nd , on account of the intrinsic magnetisation .
But , with proper allowance for , it might be that the resulting curve connecting and in a given specimen , would be approximately definite , at any rate , far more so than that connecting and B. Granting perfect definiteness , however , there is still insufficient information to make a tbeory .
The energy put into iron is not wholly stored ; is , in increasmg the coil current we increase as well as , and in doing so dissipate energy ; and although we know , by EWING'S experiments , the amount of waste in changes , it is not so clear what the rate of waste is at a given moment .
There is also the further peculiarity that the energy of the intrinsic magnetisation at given moment , though apparently locked up , and really locked up temporarily , however loosely it may be secured , is not wholly irrecoverable , but comes into play again when is reversed .
Now it may be that the energy of the intrinsic magnetisa tion , in ] ation to the stress , an entirely different part from that of the elastic etisation .
It is easy to make up formulae to express special phenomena , but very difficult to make comprehensive theory .
Rut in any , apart from the obscurities connected with iron , it is desirable to be apologetic in making any application of MAXWELL'S stresses or similar ones to practice the actual strains produced are in question , bearing in mind the difficulty of interpreting and harmonising with theory the of KERR , QUINCKE , and others .

