Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Springer is collaborating with JSTOR to digitize, preserve and extend access to Archive for History of Exact
Sciences.
http://www.jstor.org
Contents
1. Introduction 235
2. ExperimentalResults 237
3. The PhenomenologicalApproach 240
4. The BritishApproach.Faraday 243
5. W.Thomson 244
6. The Developmentof Maxwell's VortexTheory 248
7. The ApplicationoftheVortexTheoryto the Faraday Effect 255
8. The ContinentalApproach.De la Rive 261
9. C. Neumann'sTheory 262
10. Conclusion 271
Appendix 273
1. Introduction
On September13, 1845, Michael Faraday made one of his fundamental
discoveries.Having placed a piece of heavy glass in a strongmagneticfieldhe
passed a ray of linearlypolarized lightthroughthe glass in a directionparallel
to thefieldand foundthattheplane ofpolarizationofthelightraywas somewhat
turned1.This phenomenonhas since been called by various names,such as the
magneto-opticrotationor magneticcircularpolarization,but is now known
simplyas theFaraday effect.
The discoverydid not come to Faraday out oftheblue. It had been preceded
11 yearsearlierbya seriesofattemptsto discoversomerelationbetweenelectricity
and light2.These attemptswere all fruitless, but Faraday's fundamentalbelief
in theunityofthevariousforcesofnatureremainedunshakenovertheyearsand
in August 1845 a letterfromWilliam Thomson led him to repeat his old ex-
periments 3. Stillunable to detectany action of electricity
on lighthe hit on the
idea oflookingfora magneticaction,withsuccessfollowingalmostimmediately4.
1 Michael Faraday : " On the
magnetizationof lightand the illuminationof magneticlines of
force",reprintedin M.Faraday: ExperimentalResearches in Electricity,vol. Ill, London 1855,
Dover ed. 1965,ser.XIX, §§2146-2242.
1 Faraday:
Experimental Researches,vol. I, §§ 951-955.
L. Pearce Williams : Michael Faraday, London 1965, pp. 381 et seqq. See also J.Brookes
Spencer: "On theVarietiesofNineteenth-Century Magneto-OpticalDiscovery",Isis 61(1970)34-51.
ThomasMartin, ed.: Faraday'sdiary,vol. IV, London 1933,pp. 256 etseqq.
2. ExperimentalResults
8
Faraday, op. cit.note 1.
y
Athenœum,Nov. 8, 1845, 1080. The "gossiper thoughtthat Faraday had discoveredalso
"thatelectro-magnetic rotationsmaybe producedby theagencyof light"and wentintoecstasyover
thiswhiledevotingonlythreelinesto therealdiscovery.Faraday's paperwas readto theRoyal Society
in the meetingson November20 and 27, and a competentabstractof it was givenin the Athenœum,
Dec. 6, 1845,1176.In a letterto A. de la Rive Faraday refersto "a verygood abstract"in the Times
of Nov. 29. (L.P.Williams, ed.: The selectedcorrespondence of Michael Faraday, vol. 1, Cambridge
1971, p. 469).
1U R. Boettger: Ueber
Faradays neueste Entdeckung,die Polarisationsebeneeines Licht-
strahlsdurcheinenkräftigen Elektromagneten abzulenken".Ann.d. Phys.u. Chem.67 (1846) 290-293.
The paper is dated Dec. 21, 1845.Boettger statesthathe learntabout thediscoverythroughnotices
in the Athenaeum,Nov. 8, and in theRheinischer Beobachter,Dec. 7.
Fig. 1
3. The Phenomenological
Approach
The firstmathematicaltreatment oftheFaraday effectwas givenby George
Biddell Airy,BritishAstronomerRoyal from1835,who seemsto have worked
it out as soon as he heard of the discovery24. AlthoughAiry, not surprisingly,
proceededfromthe generallyaccepted theoryof lightas transversewaves in an
elasticether,his treatmentwas purelyphenomenologicalin so faras he restricted
himself to introducinga newtermin thewaveequationoflight,withoutattempting
tojustifyeitherthe existenceor theparticularmathematicalformofthistermfrom
more fundamentalpostulates.ParallellingMacCullagh's previous work on
naturalrotation,Airy provedthata rotationoftheplane ofpolarization,withthe
propertiesfoundbyFaraday, wouldfollowif,in thesimplewave equationforthe
propagationalong thez-axisoflightin emptyspace,
- ' (3>1)
c2 dt2 dz2
Taking(3.3)ratherthan(3.1)as hisstarting
point,Verdet nowproposedthefollow-
inggeneralwave equation :
.-K-H-^-^ÍSg.
2 2k(p(k2)
(3.7)
wherek is a solutionofthe"non-magnetic"dispersionrelation(3.4) and q>'is the
derivativeofthefunctionq>.Introducingtheindexofrefraction n and thevacuum
values k and c of wave lengthand speed correspondingto thefrequencyco,from
(3.4) one finds
^4 (■>-*.£)
and,further, thisinto(3.7),
byinserting
S=m-4^{n-Xll)^k2)- (3-9)
This expressionallows one to compare any hypothesisfor the function'j/
with experimentalresults,provided n is known as a functionof k. Using his
data forcarbon disulphideand creosote,Verdet investigatedthe threesimplest
cases
,6í3
whenusedin(3.9),giveas results
which,
ml dn'
2n2mn2I dn'
(311)
£=piHn-Aďi)
2n2mc/ dn'
~jr~'n~Aü)'
n(X)VerdettookCauchy'sapproximation
Forthefunction
n= (3.12)
A+^+^
determining thevaluesofthethreeconstants A,B and C fromhis ownmeas-
urements ofthedispersive powerofthetwosubstances.
Of thethreeformulae in (3.11)thefirstcouldclearlynotbe madeto fitthe
approximate validityofthe¿~2-law,so thecorresponding expression for'¡/could
be immediately discarded.Closerinspection showedthatthe secondformula
in(3.11)wasdefinitely thanthethird;however,
better inthecaseofcreosoteitstill
ledtodiscrepancies whichweretoolargetobeexplained awaybyanyconceivable
sourceoferror.Verdetwas therefore forcedto concludethat'j/couldnotbe a
universalfunction butthatitsformmustdependon thenatureofthesubstance.
Henceanytheory whichled to one definite expression for'¡/mustbe,ifnotre-
jected,at leastregardedas incomplete. Since thiswas true bothofNeumann's
theory, whichimplied thefirstof the expressions (3.10),and of Maxwell's,
in
whichimpliedthesecond,Verdetcouldnot,as he perhapshad hoped,declare
oneofthementirely although
satisfactory, Maxwell's resultseemedto be prefer-
able to Neumann's.As I shallshowlater,Maxwell's explanation had rather
seriousflawsofanotherkind,whichVerdet,however, seemsnotto havenoted,
probably becauseherestricted hisdiscussiontothedependence onwavelength.
the
Despite phenomenological characterof the work of Airy and Verdet,
neitherofthemwasinanysenseofthewordpositivistic inattitude.Airyhadinthe
early183O'sbeena partisan forthewavetheory oflightagainsttheemissionists,
andVerdet,editorofFresnel's Œuvres,belongedwhollyto thegroupofpeople
whoseworkin opticswas basedon Fresnel'stheory oflightand hismolecular
conception ofetherandmatter. Thus,although Verdetdeclared thathewishedhis
analysisto be as independent as possibleofanyparticular theory ofdispersion,
hisworkwasinfactentirely based on Cauchy's theory which in turnrestedon a
pictureofmatter ofbilliard
as consisting ballmolecules scatteredamongthepoint
oftheether.LikeCauchy,Verdetneverhitupontheideathatmaterial
particles
molecules might haveinternaldegreesoffreedom whichwouldmakeitnecessary
to coupletheequationsofmotionoftheetherwithequationsdescribing relative
motionsof molecularparts.A harmonicoscillatormodelof the kindwhich
Sellmeierand othersintroduced fordispersion,and whichlaterbecamean im-
portantfeature oftheelectrontheory,would,I believe,lie welloutsidetherealm
ofVerdet'sscientific concepts.Henceit is fairto say thatcontemporary ideas
abouttheconstitution ofetherand matterhad a morerestrictive influence on
Verdet'streatment its
thanone wouldgatherfrom general character and from
Verdet'sownassessment ofit.I mayadd thatina longfootnote tohisdiscussion
ofNeumann's theoryVerdet describedhisown basic ideasin somedetail.I shall
returnto thispointin§ 9 below.
4. The BritishApproach.Faraday
5. W. Thomson
modelswas notsharedbyhisprom-
molecular
Faraday'sdislikeofdefinite
inentBritish William
disciples, Thomson and JamesClerk Maxwell. To them
theuse of mechanicalmodelsbecamean important researchtool whichthey
28 Ibid.
§§2253-2274.
29 In Researches,III, at least 500 pages are devoted to these subjects,while the
Experimental
magneticrotationis dealtwithin thefirst26 pages. See also L. P.Williams: MichaelFaraday,chapters
9-10.
30
Verdet, comparing Faraday to other experimenters,said that his experimentswere
"... beaucoup plus nombreuseset plus variées,mais destinéessimplementà montrerla généralitédu
phénomène,sans prendreaucune mesure..." (Ann.de chim.et phys.,3e sér.,52 (1858),p. 129).
31 L. P. Williams: Michael
Faraday,pp. 409-411.
A (j) B
im
Fig. 2
42 J.C.Maxwell: "On
PhysicalLines ofForce", Phil.Mag. 21 (1861) 161-175,281-291,338-348;
23 (1862) 12-14, 85-95; repr.in: The ScientificPapers of James Clerk Maxwell, Cambridge 1890,
Dover Pubi. 1965,vol. I, pp. 451-513.
43 J.C.Maxwell: "On
Faraday's Lines of Force", Trans. Camb. Phil. Soc. 10 (1864) 27-83;
Papers,I, pp. 155-229.The paperwas read on December10,1855,and February11,1856.
44 Letterto W.Thomson,
May 15, 1855,printedin J.Larmor: The Originsof Clerk Maxwells
ElectricIdeas, as Described in Familiar Lettersto W.Thomson", Proc. Camb. Phil. Soc. 32 (1936)
695-750,p. 705.
45 Letterto W.Thomson,
September13, 1855,ibid.p. 711.
*° "Now I
thoughtthatas everycurrentgeneratedmagneticlines & was acted on in a manner
determinedby the lines throwh: it passed thatsomethingmightbe done by considering'magnetic
polarization'as a propertyofa 'magneticfield'or space and developingthegeometricalideas accord-
ingto thisview".Letterto W.Thomson,November13,1854,ibid.p. 702.
satisfied
thesamemathematical equationsas streamlinesinan idealincompress-
iblefluid.He tookgreatpainsto spelloutto hisreadersthathe did notbelieve
ofsucha fluidandthathewasnotattempting
intherealexistence togivea physical
theory ofelectric
or explanation and magnetic phenomena butmerely tryingto
assisthisownimagination,and thatofothers, bypointingto an analogy.At the
sametimeheexpressed hisbeliefthat47
a maturetheory,in whichphysicalfactswillbe physically willbe
explained,
formed by those who Nature
by interrogating can
herself obtainthe only
whichthemathematical
truesolutionofthequestions theorysuggests.
Itis veryclearfromMaxwell's statements thata physical analogyis a temporary
tool whichcan be moreor lessuseful, whilea physicaltheoryis eithertrueor
false;andalso thatthereis oneandonlyonetruetheory ofelectricity
andmagne-
tism.
In thesecondpartofhis paperMaxwell turnedto electrodynamics. Here
he did notevenhavea physicalanalogyto offer, buthe had beenimpressed by
Faraday'suseoftheconceptofan"electrotonic state"as thebasisfortheformula-
tionofthelawsofelectrodynamics. He showedthatiftheelectrotonic statewas
defined as a vectorfield,Faraday's verbaldescriptioncouldbe condensedinto
a tightmathematical formalism.He madeitquiteclearthat,sincehecouldgiveno
physicalinterpretationof the state,histheorywas a purelyformal
electrotonic
one:48
I do notthinkthatitcontainseventheshadowofa truephysicaltheory; in
fact,itschiefmeritas a temporary
instrument
ofresearchis thatit does not,
evenin appearance, accountforanything.
In theintroductionMaxwell hadsaidthatthepurposeofhispaperwas
... toshewhow... thelawsoftheattractions
andinductiveactionsofmagnets
andcurrents be
may clearly conceived,withoutmakinganyassumptionsas to
thephysicalnatureofelectricity,
or addinganything
to thatwhichhas been
alreadyprovedbyexperiment.
and further
... to avoidthedangersarisingfroma premature
theory to explain
professing
thecauseofthephenomena.49
Towardstheendofthepaperhe contrasted hisformaltheoryoftheelectrotonic
statewithWeber's"professedly physicaltheoryof electrodynamics"and tried
to defend himself
in advanceagainsttheobviousdangerthathisworkwouldbe
seenas an unnecessaryandincompleteduplicationofWeber's:50
Whatis theuse thenofimagining an electro-tonic
stateofwhichwe haveno
distinctly
physical instead
conception, of a formula
ofattraction
whichwecan
47 Maxwell:
Papers,I, p. 159.
48 Ibid. 207.
p.
49 Ibid. 159.
p.
50 Ibid. 208.
p.
wherep is the densityof the mediumand the xi} are the elementsof the stress
tensorin themedium.Withoutthe vorticesthetwo last equations would lead to
the usual wave equation69,
|př-«^". (7.4)
wherea denotesthewave velocity,but in our case thereis a furthercontribution
to the stresstensor,arisingfromthe change in L caused by the displacementu.
The torqueon a surfaceelementda, withunitnormalvectorn,is x x tda, wheret
is thestresson da, givenby
=
ti Tijnj-
= (7-5)
-jr
at sijk ix/TjkiMo"
sv
^-f^S (7-8)
and
dx'ij 'ir 93wk
I^j=~J^SijkVl^d^dt'
Whenthistermis added to therighthand side ofequation (7.4),theequationsof
motionbecome
( ' }
%np 2ka2 4p X2c
wherec and X are vacuum values of velocityand wave lengthof the lightwave,
and n is the indexof refraction.
Maxwell modifiedthisexpressionby adopting
Fresnel's relationbetweenp, thedensityoftheetherin thetransparentsubstance,
and 5,thedensityin vacuum,
p = s • n2
and furtherby changingto absoluteunitsofmagneticforce.His finalexpression,
at theveryend ofthepaper,correpondsto
2 s3'2cã2- (LU)
Ì ,
[ paramagnetics
= 11 in I vacuum,
I><J (diamagnetics.
71 /bid., 507.
p.
"--?H£)
andconfines himselftoa comparison ofthedependence on nandXwithVerdet's
resultsof 1863,rightlyconcluding that (7.15) agrees betterwiththeseresults
thanC.Neumann'sformula.
Maxwell's silenceabouttheconstantC, and themoresignificant lack of
anyattempt in the Treatiseat the
relating vortexexplanation of theFaraday
effect
to theWeberiantheoryofdia- and paramagnetism, probablyreflect his
failureto overcomethedifficulties inherent in his earliertreatment where,as
we haveseen,thetwosubjectsweretiedtogether the
through quantity ¡i. The
followingpassage from a letterto Thomson, written in 1868,shows thathe was
wellawareofthesedifficulties:72
I wanttoknowwhatyouthinkofT [i.e.P.G.Tait] inhisbookon Heat§ 125
wherehe saysthatVerdet'sdiscovery thatparamagnetics act oppositely on
lightfrom diamag8 constitutes
a proofthatthepolarityofbothclassesofbodies
is thesame.Thiscertainly requires explanationwhichI wouldbe gladof,for
myself & fortheHeatbook.He also givesyourproofoftheimpossibility ofa
diamagnetic acquiringitsreverse polaritygraduallylikea paramagnetic. But
Weber'sdiamagn.c hypothesis of inducedmolecularcurrents does not lead
to yourabsurdconclusionanymorethanour King'sCollegecoil could be
madea perpetual motion.
T' also speaksofthe'Fact' established byFaraday'thata diamag.takes
the same polarityas a paramag.in the same position.I cannotfindthat
Faradaythought he had establishedthisas a fact.He certainly
showedthat
thelinesofmagnetic as relatedto induction
force, ofcurrentsrunin thesame
general directioninbismuth Iron& steel(thatis,notintheoppositedirection)
buthe also showedthatin a steelmagnetplacedin theoppositedirection to
72 Letter fromJ.C.Maxwell to
W.Thomson, dated "Glenlair, Dalbeattie, 1868 July 18",
Library,MS. Add. 7342,Box 2. The quotationis takenfrompp. 7-8 oftheletter.
CambridgeUniversity
1 3A
expressionsare,apartfromconstantsand signs,
Maxwell's corresponding
8A
E = wx B - gradcp,
F = <?E,
8. The Continental
Approach.De la Rive
On theContinent theFaraday effect did nothavethefundamental impor-
tanceitacquiredin Britain. To Frenchand Germanphysicists itdidnotbecome
a startingpointforthedevelopment ofa newkindofphysicaltheory. Theysaw
it ratheras an interestingnewphenomenon whichwouldrequirethatexisting
notionsoftheluminiferous etherand itsrelationwiththemolecules ofmatter be
extended by an additional
hypothesis in theformofa newtypeofactionon the
ether.In speculating aboutthenatureofthisnewactionthey,muchmorethan
Thomsonand Maxwell, emphasizedFaraday's observationthatthe effect
seemedtobe due,notto a directactionofmagnetism on light, butto an actionin
whichthemolecules ofmatteractedas intermediaries. Itis therefore notsurprising
thatwhenCarl Neumannsucceededin tightening such speculations into a
precisemathematical he
theory, gave this aspectof the a
phenomenon veryfull
treatment.
One ofthefirstto publishhisideason thenatureofthenewmagnetic action
was Faraday's friendand frequent correspondent, Auguste de la Rive,pro-
fessorof physicsin Geneva.Like Boettger in Germany, de la Rive learned
about Faraday's discovery fromtheAthenaeum notice.His firstreactionwas
to connecttheeffect withhis own discovery thatan oscillating magnetic field
couldsetnon-magnetic bodiesin motion, a resultwhichhe interpreted as a sign
ofa magnetic modification
ofmolecular structure73.However, inthefirst volume
ofhis Traitéď électricité,
whichappearedin 1854,he triedinsteadto relatethe
magneticrotatory powerof a substanceto its indexof refraction 74. He now
arguedthattheeffect couldnotbe due to a modification ofmolecular structure,
sinceonecouldagitatea liquid,orpasselectric currents through itinanydirection,
without disturbing theeffect.On theotherhand,theeffect did notconsistin a
directactionofmagnetism on light,sinceit appearedonlyin liquidsand solids,
butnotin gasesor in a vacuum.AdoptingFresnel'sexplanation ofrefraction
as due to a greateretherdensityin densebodies,de la Riveconcludedthatthe
effectwas causedby a magnetic actionon thesurplusofetherin a body,and
thatitmustbe thegreater themorethedensity ofetherin a substance exceeded
thedensity inair.Hencetherotatory powerought to follow theindex ofrefraction,
a hypothesis whichdela Rive foundto be confirmed by the"stillverylimited
73
L.P.Williams, ed.: The selectedcorrespondence of Michael Faraday, vol. 1, Cambridge1971,
letters316,318,326,328,and 335.
'* A. de la Rive: Traitéd électricité
théoriqueet appliquée,tomeI, Pans 1854,pp. 529-579.
9. C. Neumann'sTheory
Carl Neumannfirst publishedhistheoryoftheFaraday effect in 1858,in
theformof a shortLatininauguraldissertation to theuniversityof Halle77.
Five yearslaterhe publisheda smallbook in whichhe gave a moredetailed
accountofhistheory, thoughwithout changing itsbasicassumptions78. These
assumptions came fromtwo sources.One was Weber'stheoriesof electro-
dynamicsand diamagnetism, the other,as Neumannexplicitly statedin the
preface to hisbook,washisfather's, Franz Ernst Neumann's, theory ofoptical
dispersion. F. Neumannhad in theperiodbefore1841developeda theoryof
dispersion whichdiffered from thatofCauchyinonesignificant aspect.Cauchy's
theorywas based on an assumption thatthe distance betweenetherparticles,
whichhe regarded as mass pointsinteracting distance, of the same
at a was
orderofmagnitude as thewavelength oflight.Underthisassumption thetheory
of elasticityled to a replacement of the simple,non-dispersive waveequation
(3.1),validin the case of negligible distancesbetweenetherparticles, by the
dispersive equation(3.3). This theory, whichexplaineddispersion by a modifica-
tionoftheinternal structure oftheether,wasopento theobviousobjectionthat
it did notaccountfortheabsenceofdispersion in air and emptyspace.It was
possibly forthis reason thatF. Neumann, while accepting Cauchy'sassumption,
it by
supplemented postulating that themolecules of matterexertedan elastic
forceon theparticles oftheether.Assuming thatthemassofa moleculewas so
greatthatitsdisplacement he obtainedthefollowing
couldbe neglected, equation
75 A. de la
Rive, op. cit. tome III, Paris 1858,pp. 715-719.
76 A. de la Rive,op. cit.tomeI, p. 578.
77 Carolus Neumann:
Explicare tentaturquomodofiat ut lucis planumpolarisationisper vires
eléctricasvel magnéticasdeclinetur. Halis Saxonum 1858.(In thefollowingreferred to as Dissertation.)
78 Carl Neumann: Die des Lichtes. Versucheiner
magnetische Drehungder Polarisationsebene
mathematischen Theorie.Halle 1863.(In thefollowingreferred to as MagnetischeDrehung.)
(9.2)
K=-gm^-=-^P,(pV)u.
79 F.E.Neumann: Die Gesetze der des Lichts in comprimirtenoder ungleich-
Doppelbrechung
förmigerwärmten Körpern,Berlin1843,pp. 28-32,footnote.See also E. T. Whittaker :
unkrystallinischen
A HistoryoftheTheoriesofAetherand Electricity, vol. I, London 1951,pp. 165-167.
80 C.Neumann:
MagnetischeDrehung,p. III.
81
Ibid.,p. IV.
m -^
im1)
Fig.3
' doy
Fig. 4
dP=
4mG;r[(vr)(d(T.grad(^)+^vdff].
This can be transformed
to
dP = 4m Gj % v x (da x r)+ dP' (9.8)
on mbythewholemolecular as
current
p = |dP=_vxA (9.9)
wherethevectorA is givenby
A=4Gm[{-17--){-^-)7-17v]- (910)
while
to thecentreofthemolecule,
Herer nowdenotesthepositionofmrelative
a' is themolecular moment,
magnetic givenby
a' =j kn
wherek is theplaneareaenclosedbythemolecular currentandn is a unitvector
normalto theplaneofthe current.
We mustnowfindthemeantotalforceon mbyaddingthecontributions (9.9)
fromall moleculesinsidea sphereroundm,withradiusequal to therangeof
actionoftheforceF, and byaveragingoverall etherparticles
insidea sphereof
similardimensions. the
Assuming transparent substanceto be non-crystalline,
isotropie,and uniformlymagnetizedinsidethe sphere,Neumannfoundthe
followingexpressionforthemeantotalforceon m:
mE= mLvxH (9.11)
where(9.4)has beenused to replacethemeanmagnetic momenta by H. The
coefficient
L, whichwill an
play important rolein thelater is givenby
discussion,
L=^U (9.12)
n
Herek is themagnetic the
constantdefinedby(9.4); N and n are,respectively,
numberof moleculesand etherparticlesin unitvolume;and S is a quantity
depending ofetherparticlesrounda moleculeand defined
on thedistribution
the
by following :
expression
^,^^,...„4 «,14,
WeneednotfollowNeumann'scalculation since(9.14)obviously
anyfurther,
corresponds ofthethreecasesinvestigated
to thefirst byVerdet;cf.equations
(3.5)and (3.10).Setting
m=-LHz,
= o),
i¡/((o,k2)
(p(k2)=C + Clk2 + C2{k2)2+ ---,
wegetfrom(3.7)thatVerdet'sconstant
in thiscase is givenby
V= (9'15)
~2k(C1+2C2k2 + -)
from(3.11)
or,equivalently,
"-è(-'£)•
It is obviousfromeitherof thesetwoexpressions thattherotationis pro-
portional L whichis relatedto themagnetic
to thecoefficient constantk through
(9.12).Neumanngavea detaileddiscussionofthispointand showedhow(9.12)
and (9.13)couldbe usedto explainVerdet'sresultsconcerning therotationin
composite substances.
His discussionwas basedon thefactthatthequantities k
and S reflected characteristic
propertiesof an individualelementarymolecule,
namely itsmagnetizability
andthedistributionofetherparticles
initsneighbour-
hood.Thus,fora substance containingin unitvolumeNa9Nb9.,., Ncmolecules
oftheelementary substancesA9B,...,C, thequantityL mustbe givenby
+ +
L_GNakgSa NbkbSb :.+NckcSc^
n
whileitsmagnetic
constant
musthavethevalue
K = Naka+ Nbkb
+ ...+Nckc. (9.18)
Sincethefunction <P(r)is a universal equation(9.13)showsthatS must
function,
alwayshavethesamesign,whileitsnumerical valuemayvaryiftheetherparticles
are distributeddifferently aroundthedifferent kindsofmolecules. As all N are
itfollows
clearlypositive, thatifall k areeitherpositiveornegative,
L andK will
havethesamesign.However, ifsomekarepositive, othersnegative,
thenumerical
valuesofSa9Sb,..., Sc maybe suchthatthesignofL becomestheoppositeto
thatofK. ThusNeumanncouldclaimthathistheory hadthefollowingimportant
consequences95:
Alle Substanzen,derenBestandtheile sämmtlich paramagnetisch sind,
drehendie Polarisationsebene in gleichem Sinne;undalle Substanzen, deren
Bestandtheile sämmtlich diamagnetisch sind,imentgegengesetzten Sinne.
Bei einerSubstanz,derenBestandtheile theilspara-theilsdiamagnetisch
sind,kannmanaus dempara-oderdiamagnetischem CharakterderSubstanz
keinenSchlussaufdenSinnmachen,in welchem die Polarisationsebenevon
derSubstanzabgelenkt wird.
95 C.Neumann:
MagnetischeDrehung,p. 77.
10. Conclusion
Oneofthepurposesofthisarticlehasbeento bringout,as clearlyas possible,
theroleplayedbytheFaraday effect in thedevelopment ofMaxwell's electro-
magnetic theory.It is obvious thatit was W. Thomson's analysisoftheFaraday
effectwhichledtotheinvention ofthevortexmodelbymeansofwhichMaxwell
first
derivedhiselectromagnetic equations.I havetriedto showthatMaxwell,
whileregarding the"idle wheel"particles as onlya crudemodelforelectricity,
wasconvinced thattheconception ofa magnetic fieldas a fieldofvortices
wasa
truedescriptionofthefactualconstitution oftheether,and thathe retained this
convictionunalteredthroughall his revisionsof the electromagnetic theory.
I havearguedfurther thatthisviewof thenatureofthemagnetic fieldforced
Maxwell to thinkof electricity as a "connecting mechanism" couplingthe
motionofthevortices, and thatthisledto hiswellknowndifficulties concerning
thenatureofelectricity. Thesedifficultieswerea majorobstacleto theunder-
standing ofelectricity as a constituentofmaterialmoleculesand thereby to the
invention ofa workablemodelfortheinteraction betweentheelectromagnetic
fieldandmatter.Thisis seennowhere moreclearly thaninMaxwell's application
ofhis theoryto theFaraday effect itself,wherethedescription was eitherin-
consistentorin seriousdisagreement withexperience.
It waspartlybecauseofthisaspectof Maxwell's theorythatW.Thomson,
whileconvinced oftheessentialcorrectness ofsomeofitsfeatures, nevergaveit
hisfullacceptance.In one ofhisBaltimore lectures
Thomsonsaid102
99
Ibid.,p. 462.
100
Ibid.,p. 475.
101
Ibid.,p. 463.
102 Kelvin: BaltimoreLectureson Molecular
Dynamicsand the Wave Theoryof Light,London
1904,p. 376.
Acknowledgements
While workingon this paper I have receivedfinancialsupportfromthe Danish Council for
ScientificResearch,and fromthe Departmentof Historyof Science and Medicine,Yale University,
whereI spenta fruitful period of two months.I have had valuable discussionswith E. Frankel,
K.M. Pedersen, and M. J.Klein. The latteralso read the manuscriptand suggesteda numberof
improvements. K.H: Wiederkehrprovidedme witha copyofhisdissertation on Weber,and H. Kan-
gro helpedme obtain a copy of the "RheinischerBeobachter"noticeon the Faraday effect. I have
also benefitedfromthe workofmystudentsO. W. Nielsen, G. Riber and J.Rechendorff.
103 Ibid.
Appendix
Thomson'sand Maxwell's Proofof theExistenceof Vortices
In §§ 5 and 6 1 have severaltimesreferred to theproof,firstoutlinedby Thom-
son in 1856 and laterworkedout in detail in Maxwell's Treatise,thattheexis-
tenceand characteristic propertiesof the Faraday effectentail the existenceof
microscopicvorticeswherevermagneticforcesare present.I have repeatedly
stressedthe importanceof Maxwell's belief,inheritedfromThomson,thatthis
was indeeda conclusiveproof,and forthisreasonI shallgivea somewhatdetailed
expositionof it,followedby a presentationof Maxwell's derivationof equation
(7.14). Because of its lengthand technicalnaturethis discussionis given as an
appendix.
Thomson'sformulation of the proofpresupposescompletefamiliarity on the
part of the readerwiththe analysisof the optical rotationin termsof circularly
polarizedrays.This analysisis givenin Maxwell's Treatise,more or less along
thefollowinglines104.
We considerfirsta linearlypolarized plane wave travellingalong the z-axis,
whichis parallelto thedirectionofa magneticfield.It entersa lumpofmagnetized
materialat z = 0, and leaves it again at z = l withits plane of polarizationturned
througha positiveangle 6. If we choose thex-axisto be in the originaldirection
ofpolarization,thewave willbe describedby
u = (2acos(-cot),0,0) at z = 0, (A.I)
u = (2acos(- cot+ cp)cos 0, 2a cos{-œt + (p)sin6,0) at z = l. (A.2)
This maybe explainedon theassumptionthatinsidethematerialleft-and right-
velo-
hand circularlypolarized waves of equal frequenciestravelwithdifferent
cities.Hence,for0 rgz ^ /we describethewave by
u = U!+u2, (A3)
ux= (a cos (^z-co t),a sin(^z-co t),0), (A.4)
u2= (a cos (k2z + cot),a sin(k2z + cot),0) (A.5)
where
kx>09 k2<0, IkJ+ l^l. (A.6)
It is easilyseen thattheseequations reduce to equation (A.I) forz = 0. For z = l
we get
ux= acos(k1l-œt) + acos(-k2l-œt) = 2acos(k0l-œt)cos(ôk'l)
( A 7^
uy= asm(kll-cot)-asin(-k2l-œt) = 2acos(k0l-œt)sm(ôk'ï)
where
*o = 2(^1-^2) and àk=±(kx+k2)
or
kx=k0+ õk and k2= -ko + ôk. (A.8)
104 Maxwell:
Treatise,§§811-817.
We seethatequations(A.7)and(A.2)agreefor
q>= kol (A.9)
and
9 = ôk-l. (A.10)
Thus,thesuperposition ofthetwocircularly polarizedwavesuxandu2, travelling
withthevelocities and
vx=co/k1 v2=co/'k2'9 is equivalentto a linearly
polarized
wavetravellingwiththevelocityvo= co/ko, but whoseplaneofpolarization is
turnedthroughtheangle6.
Now theessentialdifference betweenthemagnetic rotationand thenatural
rotationproducedin opticallyactivesubstances consistsinthefactthatifthe
atz = /,itwillemerge
waveisreflected atz = 0 witha rotation of29 inthemagnetic
case,butwithno netrotationin thenaturalcase. Fromthisit followsthatthe
forthevaluesoffrequency
variouspossibilities and wavenumberofa circularly
polarizedwave,givenbythegeneralexpression
u= (a cos(fez- œř),a sin(fez- œt' 0), (A.I1)
mustbe relatedas showninTable 1.
Table 1
thatthenaturalrotationis causedby
Thomsonthenarguestheprobability
an intrinsic ofthemoleculesoftheopticallyactivesubstances:
spiralstructure
... and it is certainthatanyspiralheterogeneousness ofa vibrating
medium
must, ifeither right-handedor spiralspredominate, a finite
left-handed cause
rotationof theplaneof polarization ofall wavesof whichlengthsare not
infinitelygreatmultiples ofthestepsofthestructural spirals.
It is otherwise,however, withthemagnetic rotation:
Butthemagnetic influenceon lightdiscovered byFaradaydependson the
direction ofmotionofmovingparticles. For instance, ina mediumpossessing
it,particlesin a straightlineparallelto thelinesofmagnetic force,
displaced
to a helixroundthislineas axis,and thenprojectedtangentially withsuch
velocitiesas to describecircles,willhavedifferent velocitiesaccording as their
motionsareroundin onedirection (thesameas thenominaldirection ofthe
galvaniccurrent inthemagnetizing coil),or in thecontrary direction.Butthe
elasticreactionofthemediummustbe thesameforthesamedisplacements,
whatever be thevelocities and directions oftheparticles;thatis to say,the
forceswhichare balancedby centrifugal forceof thecircularmotionsare
equal, whilethe luminiferous motionsare unequal.The absolutecircular
motionsbeingtherefore eitherequal or suchas to transmit equal centrifugal
forcesto the particlesinitially considered, it followsthatthe luminiferous
motionsare onlycomponents ofthewholemotion;and thata lesslumini-
ferouscomponent in one direction,compoundedwitha motionexisting in
themediumwhentransmitting no light,givesan equal resultant to thatofa
greater luminiferous motionin thecontrary direction
compounded withthe
samenon-luminous motion.I thinkitis notonlyimpossible to conceiveany
otherthanthisdynamical explanationof thefactthatcircularly polarized
lighttransmitted through magnetized glassparallelto thelinesofmagnetizing
force,withthesamequality,right-handed always,or left-handed always,is
propagated at rates
different according as its courseis in the directionor is
contrary to the direction in which a north magneticpole is drawn; but I
believeit can be demonstrated thatno otherexplanation ofthatfactis pos-
sible.
Thomson'sargument is,ofcourse,basedon theelasticsolidtheory oflight.
Hencethevectoru is an elasticdisplacement. For thecircularly polarizedwave
(A.ll) the end-pointsof u willlie on a helixwhich rotatesuniformly aboutthe
z-axiswithangularvelocity œ. The theoryofelasticity the
gives components of
straininthemedium as combinations ofthespatialderivativesofthecomponents
of u. The straindepends,therefore, on thevalueof k (cf.equations
essentially
(A.15) and (A.17) below).The relationbetweenœ and k,whichdetermines the
wavevelocity, mustthenbe thought ofas a dynamicalexpression ofthebalance
betweencentrifugal forces,whichdependon a>9and theelasticreactionsofthe
medium, depending on k. The gistofThomson'sargument is, thatsincein the
case ofnaturalrotationthewavevelocitydependson theconfiguration ofthe
helix(i.e.on thesignoffc),heterogeneousness ofstrainis themostobviousas-
sumption; whilein themagnetic case,wherethewavevelocity dependson the
W=iA(en+e22+e33)2 + fi(efl+e222
+ el3 + 2ef2+ 2e223+2e2i3' (A.14)
whereX and 'i are elasticconstantsand theelementsof strain,eV},are definedby
T=ya2œ2K (A.16)
and theonlynon-zeroelementsof strainare
1 ¡duv duz' 1
(A.17)
1 /6ux du,' i, . „
**««»(**-«")■
«i3-T(-ã7+83f)-
Hence we have,in thiscase,
W=±fjLk2a2 (A.18)
and
U=±iik2a2V. (A.19)
For a wave ofconstantintensitytheamplitudea musthave a constantvalue.The
conditionforthisis, by Lagrange's equationsof motion,
-M-0
da da
(A.20)
106 Maxwell: I have added a fewresultsfromthetheory
Treatise,§§818-821. For clarification,
These maybe foundin Sokolnikoff,op. cit.note 69, pp. 21-22 and 85.
ofelasticity.
'k'=yï.'œ'. (A.22)
alwayshave thewell-knownvalue fortransverse
The wave velocitywilltherefore
elasticwaves
°Jë-']/1
|fe| ' P
<A-23>
independently of thesignsof k and co.
Now when the material,throughwhichthe wave travels,is magnetized,we
know thattheremustbe two different values of |fc|forany givenvalue of 'co'. In
thiscase equation (A.21) musttherefore containa termwhichis linearin eitherk
or co.It was clearlythefirstof thesealternativeswhichThomsonhad in mindas
theexplanationofnaturalrotation,and a similaropinionis foundin thefollowing
passage of Maxwell's :
The potentialenergy,[[/], ofthe systemdependson itsconfiguration, thatis,
on therelativepositionofits parts.In so faras it dependson thedisturbance
due to circularly-polarized light,it mustbe a functionof [a], the amplitude,
and [fc],the coefficientof torsion,only.It may be different forpositiveand
negative values of of
[fe] equal numerical value, and it probablyis so in the
case ofmedia whichof themselvesrotatetheplane ofpolarization.
For the case of magneticrotation,Maxwell dismissesthis possibilityand as-
sumes,withoutfurther comment,thatequation (A.21) mustcontaina termlinear
in co.That thisis indeeda necessarychoice is easilyseen,forfromthetable above
it followsthatfora definitevalue ofco,say co= + 'co',our equation musthave the
two rootsk=±'kl'9 i.e. it musthave theform
= /c2-|/c1|2=0.
+ |/c1|)
(fc-|fc1|)(/c
Hence therecan be no termlinearin k.
Since co representsan angular velocityaround the z-axis, the additional
co-termmustcome fromtheexpression(A.16) forthe kineticenergyT. We must
thereforeadd a lineartermto the right-handside of thisequation. But T must,
like any kineticenergy,be a homogeneousquadratic functionof velocities,so
thistermcan onlyhave theform
Ti = const,-co-eoi (A.24)
wherecoxmust be a quantityof the same physicalnatureas co,i.e. dinangular
velocityaround the z-axis. Furthermore,cox must vanish when the external
magneticfieldis zero, and its directionmustbe relatedto thatof the magnetic
field.Maxwell's proofis now complete,and he ends thissectionwiththe state-
ment:
¿« = (io-F)n. (A.28)
Anothereffectof the disturbanceu is that each volume elementacquires an
angularvelocity,givenby
ß=iFxii. (A.29)
The nextstepis to assume thatthekineticenergyof themediumcontainsa term
oftheform
T^lClS-QdV. (A.30)
v
Maxwell commentsbriefly
This is equivalent to supposing that the angular velocityacquired by the
elementof the mediumduringthe propagationof lightis a quantitywhich
mayenterintocombinationwiththatmotionby whichmagneticphenomena
are explained.
In orderto obtain the equations of motion forthe mediumwe insert(A.29) in
(A.30) and performa partialintegration.
The resultis
T^CjÇ-iVxuidV
v
(A.31)
= C J(ùx{).nda + CjMFx{)d7
sv v
wheretheclosed surfaceSv is takento be at infinityso thatthefirsttermmay be
neglected in thefollowing.
At this point Maxwell makes an interestingdigression.By analogy with
(A.26) he takes£ to be thetotalmagneticfieldand uses the"Maxwell equation"
Fx{=FxH = 4tcJ (A.32)
whereJ is the densityof total current,includingdisplacementcurrent.He then
writes(A.31) as
T1=4nC'hJdV (A.33)
and says v
VxÇ=Vx(AÇ)
andby(A.28)wehave
T1= C$h.{Vx(l;0-V)u)dV=C$h-l(Ç0-V)(Vxu)-]dV. (A.34)
v v
For theplanewave
u(r,í)= (wx(z,í)?Mz'ř)'°)
thisgives
=CH°l
Tl K)dV
(S"Ùy~l$ (A35)
wherewe havereplaced£0by Ho. The totalkineticenergydependent
on u is
therefore
(cf.(A.12))
T='±pù2dV+T1. (A.36)
v
polarizedwave(A.ll) wefinally
For thecircularly get
T=&pa2 œ2+ CH0 a2 k2œ) V. (A.37)
Maxwell replacesthesimpleexpression
In orderto takeaccountofdispersion,
(A.19)byonederived fromCauchy'stheory:
U=$a2Q(k2)V (A.38)
Thecondition
whereg is a polynomial. (A.20)forconstant nowgives
amplitude
pco2+ 2 CH0 k2œ= Q(k2) (A.39)
weget
ofcoand Ho. Bydifferentiation
inwhichk is a function
'dH0/Ho=0 pœ dœ
to replacek0and œ bythewavelengthin a vacuumA,
relations
Usingfamiliar
n, and thevelocityof lightc, we get Maxwell's final
theindexof refraction
fortherotation
formula angle
to equation(7.14).
whichis equivalent
Historyof ScienceDepartment
UniversityofAarhus
Denmark
(ReceivedDecember13, 1975)