A gentle introduction to the foundations of classical electrodynamics:

arXiv:physics/0005084v2 [physics.class-ph] 8 Sep 2000

The meaning of the excitations (D, H) and the eld strengths (E, B )
Friedrich W. Hehl Institute for Theoretical Physics University of Cologne 50923 K oln, Germany and Yuri N. Obukhov Department of Theoretical Physics Moscow State University 117234 Moscow, Russia

Abstract
The axiomatic structure of the electromagnetic theory is outlined. We will base classical electrodynamics on (1) electric charge conservation, (2) the Lorentz force, (3) magnetic ux conservation, and (4) on the Maxwell-Lorentz spacetime relations. This yields the Maxwell equations. The consequences will be drawn, inter alia, for the interpretation and the dimension of the electric and magnetic elds.

Email: Email:

hehl@thp.uni-koeln.de general@elnet.msk.ru

I. INTRODUCTION

In Cologne, we teach a course on Theoretical Physics II (electrodynamics) to students of physics in their fourth semester. For several years, we have been using for that purpose the calculus of exterior dierential forms, see [1,2], because we believe that this is the appropriate formalism: It is based on objects which possess a clear operational interpretation, it elucidates the fundamental structure of Maxwells equations and their mutual interrelationship, and it invites a 4-dimensional representation appropriate for special and general relativity theory (i.e., including gravity, see [3,4]). Our experimental colleagues are somewhat skeptical; and not only them. Therefore we were invited to give, within 90 minutes, a sort of popular survey of electrodynamics in exterior calculus to the members of one of our experimental institutes (group of H. Micklitz). The present article is a worked-out version of this talk. We believe that it could also be useful for other universities. Subsequent to the talk we had given, we found the highly interesting and historically oriented article of Roche [5] on B and H , the intensity vectors of magnetism. . . . Therein, the corresponding work of Bamberg and Sternberg [6], Bopp [7], Ingarden and Jamiolkowski [8], Kovetz [9], Post [10], Sommerfeld [11], and Truesdell and Toupin [12], to drop just a few names, was neglected yielding a picture of H and B which looks to us as being not up of date; one should also compare in this context the letter of Chambers [13] and the book of Roche [14], in particular its Chapter 9. Below we will suggest answers to some of Roches questions. Moreover, ...any system that gives E and B dierent units, when they are related through a relativistic transformation, is on the far side of sanity is an apodictic statement of Fitch [15]. In the sequel, we will prove that we are on the far side of sanity: The absolute dimension of E turns out to be magnetic ux/time and that of B magnetic ux, see Sec. IV. According to the audience we want to address, we will skip all mathematical details and take recourse to plausibility considerations. In order to make the paper self-contained, we 2

present though a brief summary of exterior calculus in the Appendix. A good reference to the mathematics underlying our presentation is the book of Frankel [16], see also [6] and [17]. For the experimental side of our subject we refer to Bergmann-Schaefer [18]. As a preview, let us collect essential information about the electromagnetic eld in Table I. The explanations will follow below. Table I. The electromagnetic eld Field name math. independent related reecobject components D H E electric excitation magnetic excitation electric to tion absolute dimension

odd D23 , D31 , D12 area 2-form odd 1-form even E1 , E2 , E3 line H1 , H2 , H3 line

D q = electric charge H E q/t

0 /t

eld strength 1-form B magnetic even B23 , B31 , B12 area B 0 = magnetic ux

eld strength 2-form

It was Maxwell himself who advised us to be very careful in assigning a certain physical quantity to a mathematical object. As it turns out, the mathematical images of D , H, E, B are all dierent from each other. This is well encoded in Schoutens images of the electromagnetic eld in Fig.1.

FIGURES

FIG. 1. Schoutens images of the electromagnetic eld, see [17] p. 133.

II. ELECTRIC CHARGE CONSERVATION

The conservation of electric charge was already recognized as fundamental law during the time of Franklin (around 1750) well before Coulomb discovered his force law in 1785. Nowadays, when one can catch single electrons and single protons and their antiparticles in traps and can count them individually (see, e.g., Dehmelt [19], Paul [20], Devoret et al. [21], and Lafarge et al. [22]), we are more sure than ever that electric charge conservation is a valid fundamental law of nature. Therefore matter carries as a primary quality something called electric charge which only occurs in positive or negative units of an elementary charge e (or, in the case of quarks, of 1/3th of it) and which can be counted. Thus it is justied to introduce the physical dimension of charge q as a new and independent concept. Ideally one should measure a charge in units of e/3. However, for practical reasons, the SI-unit C (coulomb) is used in laboratory physics. 4

Let us start with the 3-dimensional Euclidean space in which we mark a 3-dimensional domain V . Hereafter, the local coordinates in this space will be denoted by xa and the time by t, with the basis vectors ea := a and a, b, . . . = 1, 2, 3, see Fig.2. The total charge in the domain V is given by the integral Q=
V

,
1 3!

(2.1) abc dxa dxb dxc = 123 dx1 dx2

where the electric charge density is the 3-form =

dx3 . Here summation is understood over the indices a, b, c and abc = bac = bca = . . ., i.e., the components abc of the charge density 3-form are antisymmetric under the exchange of two indices, leaving only one independent component 123 . The wedge denotes the (anticommutative) exterior product of two forms, and dx1 dx2 dx3 represents the volume element. For our present purpose it is enough to know, for more details see the Appendix, that a 3-form (a p-form) is an object that, if integrated over a 3-dimensional (p-dimensional) domain, yields a scalar quantity, here the charge Q.

x3 V e3 e1 x
1

part of

e2

x2

FIG. 2. A volume V with its boundary V .

The dimension of Q is [Q] = q . Since an integral (a summation after all) cannot change the dimension, the dimension of the charge density 3-form and its components are, respectively, [] = q , and [abc ] = q/3 , with = length. In general, the elementary charges are not at rest. The electric current J owing across a 2-dimensional surface S is given by the integral J=
S

j,
1 j 2! ab

(2.2) dxa

see Fig.3. Accordingly, the electric current density j turns out to be a 2-form: j =

dxb = j12 dx1 dx2 + j13 dx1 dx3 + j23 dx2 dx3 , with jab = jba . If t = time, then the dimensions of the current integral and the current 2-form and its components are [J ] = [j ] = q/t and [jab ] = q/(t 2), respectively.

S S
FIG. 3. A surface S with its boundary S .

If we use the abbreviation t := /t, the global electric charge conservation can be expressed as 6

t
V

+
V

j=0

(Axiom 1) ,

(2.3)

where the surface integral is evaluated over the (closed and 2-dimensional) boundary of V , which we denote by V , see Fig.2. The change per time interval of the charge contained in V is balanced by the ow of the electric current through the boundary of the domain. The closed surface integral
V

j can be transformed into a volume integral

d j by

Stokess theorem (9.4). Here d denotes the exterior derivative that increases the rank of a form by one, i.e. d j is a 3-form. Thus (2.3) translates into (t + d j ) = 0 .
V

(2.4)

Since this is valid for an arbitrary domain, we arrive at the local form of electric charge conservation, d j + t = 0 . (2.5)

III. EXCITATIONS

Since the charge density is a 3-form, its exterior derivative vanishes: d = 0. Then, by a theorem of de Rham, it follows that can be derived from a potential: d = 0 = = dD. (3.1)

In this way one nds the electric excitation 2-form D . Its absolute dimension is [D ] = [] = q , furthermore, for the components, [Dab ] = [D ]/2 = q/2 . Substituting (3.1)2 into charge conservation (2.5) and once again using the de Rham theorem, we nd another potential for the current density: d (j + t D) = 0 = j + t D = d H . (3.2)

The magnetic excitation H turns out to be a 1-form, see Fig.4. Its dimension is [H] = [j ] = q/t , [Ha ] = q/(t ). 7

.a
FIG. 4. A line with its boundary , i.e., its end points a and b.

Consequently, the excitations (D , H) are potentials of the sources (, j ). All these (additive) quantities (How much?) are described by odd dierential forms. In this way, we nd the inhomogeneous Maxwell equations (the Gauss law and the Oersted-Amp` ere law): dD = , d H t D = j . (3.3) (3.4)

Electric charge conservation is valid in microphysics. Therefore the corresponding Maxwell equations (3.3) and (3.4) are valid on the same microphysical level as the notions of charge density and current density j . And with them the excitations D and H are microphysical quantities of the same type likewise in contrast to what is stated in most textbooks. Before we ever talked about forces on charges, charge conservation alone gave us the inhomogeneous Maxwell equations including the appropriate dimensions for the excitations D and H. Under the assumption that D vanishes inside an ideal electric conductor, one can get rid of the indeterminacy of D which is inherent in the denition of the excitation as a potential of the charge density, and we can measure D by means of two identical conducting plates 8

(Maxwellian double plates) which touch each other and which are separated in the D eld to be measured. The charge on one plate is then measured. Analogously, H can be measured by the Gauss compensation method or by a superconductor and the Meissner eect (B = 0 H = 0). Accordingly, the excitations do have a direct operational signicance.
IV. FIELD STRENGTHS

So far, conserved charge was the notion at center stage. Now energy enters the scene, which opens the way for introducing the electromagnetic eld strengths. Whereas the excitations (D , H) are linked to (and measured by) the charge and the current (, j ), the electric and magnetic eld strengths are usually introduced as forces acting on unit charges at rest or in motion, respectively. Let us consider a point particle with electric charge e and velocity vector v . The force F acting on it is a 1-form since its (1-dimensional) line integral yields a scalar, namely the energy. Thus F carries the absolute dimension of an energy or of action over time: [F ] = energy = h/t, where h denotes the dimension of an action. Accordingly, the local components Fa of the force F = Fa dxa possess the dimension [Fa ] = h/(t ) = force. In an electromagnetic eld, the motion of a test particle is governed by the Lorentz force: F = e (E v B ) (Axiom 2) . (4.1)

The symbol denotes the interior product of a vector (here the velocity vector) with a p-form. It decreases the rank of a form by 1 (see the Appendix), and since v B is to be a 1-form, then B is a 2-form. Newly introduced by (4.1) are the electric eld strength 1-form E and the magnetic eld strength 2-form B . They are both intensities (How strong?). The dimension of the velocity is [v ] = 1/t. With the decomposition v = v a a , we nd for its components [v a ] = /t. Then it is straightforward to read o from (4.1) the absolute dimension of the electric eld strength [E ] = h/(t q ) = 0 /t, with 0 := h/q . For its components we have [Ea ] = 0 /(t ). Analogously, for the dimension of the magnetic eld 9

strength we nd [B ] = (h/t)/(q/t) = h/q = 0 and [Bab ] = 0 /2 , respectively. The eld B carries the dimension of a magnetic ux 0 . In superconductors, magnetic ux can come in quantized ux tubes, so-called uxoids, underlining the importance of the notion of magnetic ux. The denition (4.1) of the eld strengths makes sense only if the charge of the test particle is conserved. In other words, axiom 2 presupposes axiom 1 and should not be seen as a stand alone pillar of electrodynamics.

V. MAGNETIC FLUX CONSERVATION

Taking into account the rank (as exterior forms) of the eld strengths, the only integral we can build up from E and B , respectively, are line integrals and surface integrals and
surface line E

B . Apart from a factor t, the dimensions are equal. Hence, from a dimensional

point of view, it seems sensible to postulate the conservation theorem (see Fig.5), t
S

B+
S

E=0

(Axiom 3) .

(5.1)

Magnetic ux conservation (5.1) has to be seen as an analog of electric charge conservation (2.3). Magnetic ux, though, is related to a 2-dimensional surface whereas electric charge is related to a 3-dimensional volume. Thus the integration domains in the conservation theorems (5.1) and (2.3) dier by one spatial dimension always. Axiom 3 gains immediate evidence from the dynamics of an Abrikosov ux line lattice in a superconductor. There the quantized ux lines can be counted, they dont vanish nor are they created from nothing, rather they move in and out crossing the boundary S of the surface S under consideration.

10

FIG. 5. Faradays induction law.

Again, by means of Stokess theorem (9.4), we can transform the boundary integral:
S

E=

d E . Taking into account the arbitrariness of the surface S , we recover Faradays

induction law d E + t B = 0 , (5.2)

which is experimentally very well established. We dierentiate Faradays law by means of d and nd t (d B ) = 0. Since an integration constant other than zero is senseless (recall the relativity principle), we have dB = 0. (5.3)

The homogeneous Maxwell equations (5.2) and (5.3) (Faradays induction law and the closure of the magnetic eld strength) nearly complete the construction of the theory. We nd the 3 + 3 time evolution equations (3.4) and (5.2) for the electromagnetic eld (D , H; E, B ), i.e., for 6 + 6 components. Before we can nd solutions of these equations, 11

we have to reduce the number of variables to 6, i.e., we have to cut them in half. Such a reduction is achieved by Axiom 4.

VI. THE MAXWELL-LORENTZ SPACETIME RELATIONS

In Axiom 4 we assume linear, isotropic, and centrosymmetric relations between the (additive) quantities and the intensities [23]: D = 0 E and H= 1 B 0 (Axiom 4). (6.1)

The proportionality coecients 0 , 0 encode all the essential information about the electric and magnetic properties of spacetime. The Hodge star operator is needed, since we have to map a 1-form into a 2-form and vice versa or, more generally, a k -form into a (3 k )-form. Then the operator in (6.1) has the dimension of a length or its reciprocal. Note that the Hodge star depends on the metric of our Euclidean space, see the Appendix. Recalling the dimensions of the excitations and the eld strengths, we nd the dimensions of the electric constant 0 and the magnetic constant 0 as [0 ] = qt 1 = 0 c 0 and [0 ] = 0 t 0 = , q c (6.2)

respectively. They are also called vacuum permittivity and vacuum permeability, see the new CODATA report [24]. Here we dene 0 := 0 /q = h/q 2 and velocity c := /t. Dimensionwise, it is clearly visible that 1 0 0 =c and 0 0 = 0 . (6.3)

Obviously, the velocity c and the resistance 0 are constants of nature, the velocity of light c being a universal one, whereas 0 , the characteristic impedance (or wave resistance) of the vacuum [25], seemingly refers only to electromagnetic properties of spacetime. Note that 1/0 plays the role of the coupling constant of the electromagnetic eld which enters as a factor into the free eld Maxwell Lagrangian. 12

The Maxwell equations (3.3)-(3.4) and (5.2)-(5.3), together with the Maxwell-Lorentz spacetime (or aether) relations (6.1), constitute the foundations of classical electrodynamics. These laws, in the classical domain, are assumed to be of universal validity. Only if vacuum polarization eects of quantum electrodynamics are taken care of or hypothetical nonlocal terms should emerge from huge accelerations, Axiom 4 can pick up corrections yielding a nonlinear law (Heisenberg-Euler electrodynamics, see [4]) or a nonlocal law (VolterraMashhoon electrodynamics, see [26]), respectively. In this sense, the Maxwell equations are more universal than the Maxwell-Lorentz spacetime relations. The latter ones are not completely untouchable. We may consider (6.1) as constitutive relations for spacetime itself.

VII. SI-UNITS

The fundamental dimensions in the SI-system for mechanics and electrodynamics are (, t, M, q/t), with M as mass. And for each of those a unit was dened. However, since action we denote its dimension by h is a relativistic invariant quantity and since the electric charge is more fundamental than the electric current, we rather choose as the basic units (, t, h, q ) , (7.1)

see Schouten [17] and Post [10]. Thus, instead of the kilogram and the ampere, we choose joulesecond (or webercoulomb) and the coulomb: (m, s, Wb C, C ). Numerically, in the SI-system, one puts (for historical reasons) 0 = 4 107 Wb s Cm (magnetic constant). (7.3) (7.2)

Then measurements ` a la Weber-Kohlrausch yield 0 = 8.85 1012 Cs Wb m (electric constant). (7.4)

The SI-units of the electromagnetic eld are collected in Table II. 13

Table II. SI-units of the electromagnetic eld Field SI-unit of eld SI-unit of components of eld D H E B C A = C/s Wb/s = V Wb C/m2 A/m = C/(sm) ( oersted) V /m = Wb/(sm) Wb/m2 = tesla = T ( gauss)

VIII. ELECTRODYNAMICS IN MATTER

It should be needless to remark that while from the mathematical standpoint a constitutive equation is a postulate or a denition, the rst guide is physical experience, perhaps fortied by experimental data. C. Truesdell and R.A. Toupin (1960)

In a great number of the texts on electrodynamics the electric and magnetic properties of media are described following the macroscopic averaging scheme of Lorentz (1916). However, this formalism has a number of serious limitations, see the relevant criticism of Hirst [27], e.g.. An appropriate modern presentation of this subject has been given in the textbook of Kovetz [9]. Here we follow a consistent microscopic approach to the electrodynamics in media, cf. [27]. The total charge or current density is the sum of the two contributions originating from the inside and from the outside of the medium: = mat + ext , j = j mat + j ext . (8.1)

Hereafter, the bound charge [28] in matter is denoted by mat and the external charge [29] by ext. The same notational scheme will also be applied to the excitations D and H. Bound charge and bound current are inherent characteristics of matter determined by the medium itself. They vanish outside the medium. In contrast, external charge and external current in general do not vanish outside matter. They can be prepared for a specic purpose (such as 14

the scattering of a current of particles by a medium or a study of the reaction of a medium in response to a prescribed conguration of charges and currents). We assume that the charge bound by matter fullls the usual conservation law: d j mat + t mat = 0 . (8.2)

Taking into account (2.5), this means that there is no physical exchange (or conversion) between the bound and the external charges. As a consequence of this assumption, we can repeat the arguments of Sec.III that will give rise to the corresponding excitations D mat and Hmat as potentials for the bound charge and the bound current. The conventional names for these newly introduced excitations are polarization P and magnetization M , i.e., D mat P , Hmat M . (8.3)

The minus sign is conventional, see Kovetz [9]. Thus, in analogy to the inhomogeneous Maxwell equations, we nd d P = mat , d M + t P = j mat . (8.4)

The identications (8.3) are only true up to an exact form. However, if we require D mat = 0 for E = 0 and Hmat = 0 for B = 0, as we will do in (8.8), uniqueness is guaranteed. The Maxwell equations are linear partial dierential equations. Therefore we can dene D ext := D D mat = D + P , Hext := H Hmat = H M . (8.5)

The external excitations (D ext , Hext ) can be understood as auxiliary quantities. In terms of these quantities, the inhomogeneous Maxwell equations for matter nally read: d D ext = ext , d Hext t D ext = j ext , D ext = 0 E + P [E, B ] , Hext = (8.6) (8.7)

1 B M [B, E ] . 0

Here the polarization P [E, B ] is a functional of the electromagnetic eld strengths E and B . In general, it can depend also on the temperature T , and possibly of other thermodynamic 15

variables specifying the material continuum under consideration; similar remarks apply to the magnetization M [B, E ]. The system (8.6)1 and (8.7)1 looks similar to (3.3) and (3.4). However, the former equations refer only to the external elds and sources. The homogeneous Maxwell equations (5.2) and (5.3) remain valid in their original form. In the simplest cases, we have the linear constitutive laws P = 0 E E , M= 1 B B , 0 (8.8)

with the electric and magnetic [30] susceptibilities (E , B ). With material constants := 0 (1 + E ) , one can rewrite the material laws (8.8) as D ext = E , H ext = 1 B. (8.10) := 0 , 1 B (8.9)

For the discussion of the concrete applications of the developed microscopic theory in modern condensed matter physics, we refer to the review of Hirst [27].

IX. CONCLUSION

The Maxwell equations dD = , dB = 0, d H t D = j , d E + t B = 0 , (9.1) (9.2)

are the cornerstones of any classical theory of electromagnetism. As an expression of charge and ux conservation, they carry a high degree of plausibility as well as solid experimental support. The Maxwell equations in this form remain valid in an accelerated reference frame and in a gravitational eld likewise, without any change. The Maxwell-Lorentz spacetime relations D= 1 E c 0 and 16 H= c B 0 (9.3)

are necessary for developing the Maxwellian system into a predictive physical theory. They depend, via the star operator, on the metric of space and are, accordingly, inuenced by the gravitational eld. They are valid in very good approximation, but there are a few exceptions known (if the Casimir eect is to be described, e.g.). For the description of matter, the sources (, j ) and the excitations (D , H) have to be split suitably in order to derive, from the microscopic equations (9.1) and (9.2), appropriate macroscopic expressions. Summing up, we can give an answer to one of the central questions posed by Roche [5]: The need for the dierent notations and dierent dimensions and units for the excitation H and the eld strength B (and, similarly, for D and E ) is well substantiated by the very dierent geometrical properties and physical origins of these elds, see Table I and Fig. 1. Even in vacuum, these dierences do not disappear.

ACKNOWLEDGMENTS

We are grateful to H. Micklitz (Cologne) for arranging this lecture. One of the authors (FWH) would like to thank W. Raith (Berlin-Bielefeld) for an extended exchange of letters on the fundamental structure of Maxwells theory. Moreover, he is grateful to R.G. Chambers (Bristol), A. Kovetz (Tel Aviv), J. Roche (Oxford), and S. Scheidl (Cologne) for most useful remarks.

APPENDIX: THE ABC OF EXTERIOR CALCULUS

The formalism of exterior dierential forms is widely used in dierent domains of mathematics and theoretical physics. In particular, in electromagnetic theory, exterior calculus oers an elegant and transparent framework for the introduction of the basic notions and for the formulation of the corresponding laws. Here, we will give a very elementary description of the objects and operations used above. 17

We will conne ourselves only to the case of a 3-dimensional space. Let be given the set of local coordinates xa = {x1 , x2 , x3 }; hereafter Latin indices a, b, . . . will run over 1, 2, 3. Then the vectors ea = {1 , 2 , 3 } will serve as a basis of the tangent vector space at every point. The symbol dxa denotes the set of linear 1-forms dual to the coordinate vector
a basis, dxa (eb ) = b . An arbitrary k -form can be described, in local coordinates, by its

components: = a dxa = 1 dx1 + 2 dx2 + 3 dx3 for 1-forms and =

1 2

ab dxa dxb =

12 dx1 dx2 + 23 dx2 dx3 + 31 dx3 dx1 for 2-forms. Any 3-form has a single nontrivial component, = 123 dx1 dx2 dx3 . When is smoothly dened on the whole space, it is called a volume form. Zero-forms are just the ordinary dierentiable functions. It is often stated that the exterior product generalizes the vector product. However, one should be careful with such statements, because the vector product in the standard 3-dimensional analysis is, strictly speaking, a superposition of the wedge product and of the Hodge duality operator. Thus, the vector product necessarily involves the metric on the manifold. In contrast, the exterior product is a pre-metric operation, although it resembles the vector product. For example, the exterior product of the two 1-forms and with the components a and a yields a 2-form with the local components {(2 3 3 2 ), ( 3 1 1 3 ), ( 1 2 2 1 )} . The exterior dierential d increases the rank of a form by 1. It is most easily described in local coordinates, see Table III. Thus, d naturally generalizes the grad operator which leads from a scalar to a vector and, at the same time, it represents a pre-metric extension of the curl operator. The exterior dierential is a nilpotent operator, i.e., dd = 0. Table III. Operators acting on an exterior form k -form =
1 k!

a1 ...ak dxa1 . . . dxak , with k = 0, 1, 2, 3

d (k + 1)-form d =

[a1 a1 ...ak+1 ] dxa1 . . . dxak+1 v a aa1 ...ak1 dxa1 . . . dxak1

(k 1)-form v = (3 k )-form =

1 (k 1)! 1 k!

a1 ...ak g a1 b1 . . . g ak bk ebk . . .eb1

Complementary to d, one can dene an operation which decreases the rank of a form 18

by 1. This is the interior product of a vector with a k -form. Given the vector v with the components v a , the interior product with the coframe 1-form yields v dxa = v a , which is a sort of a projection along v . By linearity, the interior product of v with a k -form is dened as described in Table III. The Hodge dual operator maps k -forms into (3 k )-forms. Its introduction necessarily requires the metric which assigns a real number g (u, v ) = g (v, u) to every two vectors u and v . In local coordinates, the components of the metric tensor are determined as the values of the scalar product of the basis vectors, gab := g (ea , eb ). This matrix is positive denite. The metric introduces a natural volume 3-form := det gab dx1 dx2 dx3 which underlies the denition of the Hodge operator . The general expression is displayed in Table III. Explicitly the Hodge dual of the coframe 1-form reads, for example: dxa = det gab (g a1 dx2 dx3 + g a2 dx3 dx1 + g a3 dx1 dx2 ), where g ab is inverse to gab . The notions of the odd and even exterior forms are closely related to the orientation of the manifold. In simple terms, these two types of forms are distinguished by their dierent behavior with respect to a reection (i.e., a change of orientation): an even (odd) form does not change (changes) sign under a reection transformation. These properties of odd and even forms are crucial in the integration theory, see, e.g., [16]. For a k -form an integral over a k -dimensional subspace is dened. For example, a 1-form can be integrated over a curve, a 2-form over a 2-surface, and a volume 3-form over the whole 3-dimensional space. We will not enter into the details here, limiting ourselves to the formulation of Stokess theorem which occupies a central place in integration theory: =
C C

d .

(9.4)

Here is an arbitrary k -form, and C is an arbitrary (k + 1)-dimensional (hyper)surface with the boundary C . For a deeper and a more rigorous introduction into exterior calculus, see, e.g., [6,16].

19

REFERENCES
[1] F.W. Hehl and Yu.N. Obukhov, Electric Charge and Magnetic Flux: On the Structure of Classical Electrodynamics. Tex-script of 230 pages. March 1999 (unpublished). [2] M.R. Zirnbauer, Elektrodynamik. Tex-script of a course in Theoretical Physics II (in German), July 1998 (Springer, Berlin, to be published). [3] Yu.N. Obukhov and F.W. Hehl, Space-time metric from linear electrodynamics, Phys. Lett. B458 (1999) 466-470. [4] F.W. Hehl and Yu.N. Obukhov, How does the electromagnetic eld couple to gravity, in particular to metric, nonmetricity, torsion, and curvature? Preprint IASSNSHEP-99-116, Institute for Advanced Study, Princeton, see also http://arXiv.org/ abs/gr-qc/0001010. [5] J.J. Roche, B and H, the intensity vectors of magnetism: A new approach to resolving a century-old controversy, Am. J. Phys. 68 (2000) 438-449. [6] P. Bamberg and S. Sternberg, A Course in Mathematics for Students of Physics. Vol.2 (Cambridge University Press, Cambridge 1990). [7] F. Bopp, Prinzipien der Elektrodynamik, Z. Physik 169 (1962) 45-52. [8] R. Ingarden and A. Jamiolkowski, Classical Electrodynamics (Elsevier, Amsterdam 1985). [9] A. Kovetz, Electromagnetic Theory. Oxford University Press, Oxford, UK (2000). [10] E.J. Post, Formal Structure of Electromagnetics General Covariance and Electromagnetics. North Holland, Amsterdam (1962) and Dover, Mineola, New York (1997). [11] A. Sommerfeld, Elektrodynamik. Vorlesungen u ber Theoretische Physik, Band 3. Dietrischsche Verlagsbuchhandlung, Wiesbaden (1948).

20

[12] C. Truesdell and R.A. Toupin, The Classical Field Theories. In Handbuch der Physik, Vol. III/1, S. Fl ugge ed.. Springer, Berlin (1960) pp. 226-793. [13] R.G. Chambers, Units B, H, D , and E , Am. J. Phys. 67 (1999) 468-469. [14] J.J. Roche, The Mathematics of Measurement A Critical History. The Athlone Press, London (1998). [15] V.L. Fitch, The far side of sanity, Am. J. Phys. 67 (1999) 467. [16] T. Frankel, The Geometry of Physics An Introduction. Cambridge University Press, Cambridge (1997). Now also available as paperback. [17] J. A. Schouten, Tensor Analysis for Physicists, 2nd ed.. Clarendon Press, Oxford (1954) and Dover, Mineola, New York (1989). [18] W. Raith, Bergmann-Schaefer, Lehrbuch der Experimentalphysik, Vol.2, Elektromagnetismus, 8th ed.. de Gruyter, Berlin (1999). [19] H. Dehmelt, Experiments with an isolated subatomic particle at rest, Rev. Mod. Phys. 62 (1990) 525-530. [20] W. Paul, Electromagnetic traps for charged and neutral particles, Rev. Mod. Phys. 62 (1990) 531-540. [21] M.H. Devoret, D. Esteve, and C. Urbina, Single-electron transfer in metallic nanostructures, Nature 360 (1992) 547-552. [22] P. Lafarge, P. Joyez, D. Esteve, C. Urbina, and M.H. Devoret, Two-electron quantization of the charge on a superconductor, Nature 365 (1993) 422-424. [23] Alternatively, between the extensive and the intensive quantities. In electrodynamics, the distinction between these two types of quantities goes back to Mie. He also suggested the name excitation (in German: Erregung) for D and H, see G. Mie: Lehrbuch der Elektrizit at und des Magnetismus. 3rd ed.. Enke, Stuttgart (1948), see also [11] and [6]. 21

[24] P.J. Mohr and B.N. Taylor, CODATA recommended values of the fundamental physical constants: 1998, Rev. Mod. Phys. 72 (2000) 351-495. [25] 1/0 is also sometimes called the admittance of free space, see, for instance, [10] p. 184. [26] U. Muench, F.W. Hehl, and B. Mashhoon, Acceleration-induced nonlocal electrodynamics in Minkowski spacetime, Phys. Lett. A271 (2000) 8-15. [27] L.L. Hirst, The microscopic magnetization: concept and application, Rev. Mod. Phys. 69 (1997) 607-627. [28] Also called polarization charge. [29] Also called free, true, or real charge. [30] In older texts, the magnetization M was usually expressed in terms of H, namely M = H H . For reasons of relativistic invariance, this is inappropriate, provided we start with P = 0 E E , as we do in (8.8)2 . Note that = 0 (1 + H ).

22