You are on page 1of 7
‘mended that undergraduate physics programs be made ‘more flexible so that itis easier for students to enter the ‘major after the freshman year and still complete degree re- quirements in four college years. ‘Susanne D. Ellis and Patrick J. Mulvey, Enrollment and Degrees (American Tnsttute of Physics, New York, 1989), AIP Pub. No. Re 151.26. This report is issued annually. The number of students going into high-school teaching declined sharp- ly when fewer jobs were availabe, There is some evidence of renewed Interest ina high-school teaching career. "A detailed description of the curriculum and differences in approach is found in APT's Guidelines forthe Review of Baccalaureate Physics Programs, prepared by the Committees on Profesional Concerns and on Undergraduate Education of the American Association of Physics ‘Teachers, 112 Berwyn Road, College Park, MD 20740 (1987), “For example, heat may be approached from the viewpoint of classical thermodynamics or statistical physis: the examples emphasized in quantum physies may dif. Various courses that do not require concurrent calculus usually are available for students in still other majors. Ocasionally physies majors, begin ther study in such courses "African-American, Hispanic, Native American, and Alaskan Native students are underrepresented in science and engineering, A numberof efforts to revise parts, or even the whole, ofthe introduc tory college physics course are underway. The American Journal of Physics published 7 articles, editorials or guest comments, and 12 let- ters devoted tothe calculus-level introductory course in 1988 and 1989, ‘One well-publicized effort thet is promoting controversial and vigorous, discussion of the introductory course isthe Introductory University Physics Project. "Research participation, which sof great benefit tothe student, also may ‘beasgniican help for faculty membersin pit of the expensein faculty “Sec, for example, LC. MeDermott, M. L, Rosenquist, and E. H. van “Zee, “Strategies to improve the performance of minority students inthe ‘sciences," in New Directions in Teaching and Learning No. 16, eited by JH. Cones IM, J. F, Noonan, D. Janha (Jossey-Bass, San Francisco, 1983); LC. MeDermort, LK. Piterick, and M. L, Roseaquit, “Help. ‘ng minority students succeed in science: I. Development of a currcu umn physics and biology; I. Implementation ofa curriculum in phys ies and biology; IT. Requirements forthe operation of an academic ‘program in physics and biology," J. Coll. Sc. Teach. 9(1), 138-140; 9(3), 201-208; 9(5), 261-265 (1980). ‘Report of the Denver conference on curricula for undergraduate ma- ors in physics," Am. J. Phys. 30, 153-162 (1962); "Recommendations ‘of the second Ann Arbor conference on undergraduste curricula for physics majors,” Am. J. Phys. 31, 328-335 (1963); E. Leonard Jossem, "Undergraduate curricula in physics: A report onthe Princeton confer. ence on curriculum," Am, J. Phys, 32, 491-497 (1964), "Physics forthe 19903: AAPT Conference of Department Chairsin Physics, ‘edited by M. N. MeDermott and, M. Wilson (American Assocation of Physies Teachers, College Park, MD, 1989). "The Undergraduate Curriculum for Non-graduate Bound Majors report “ofthe department heads meeting, group G, Ref. 11 Time-dependent generalizations of the Biot-Savart and Coulomb laws David J. Griffiths Department of Physics, Reed College, Portland, Oregon 97202 Mark A. Heald Department of Physics and Astronomy, Swarthmore College, Swarthmore, Pennsylvania 19081 (Received 12 June 1989; accepted for publication 2 September 1990) Strictly, Coulomb's law and the Biot-Savart law determine the electric and magnetic fields for static sources only. Time-dependent generalizations of these two laws, introduced by Jefimenko, are used here to explore the applicability of Coulomb and Biot-Savart outside the static domain. The fact that Biot-Savart holds for some classes of time-dependent currents has led to much confusion in the literature, which this paper endeavors to dispel. To illustrate Jefimenko's ‘equations, the standard Liénard-Wiechert fields of a point charge are derived. I. INTRODUCTION Over the years many papers in this Journal have been concerned with aspects of the Biot-Savart law, Baie [2c ar, a ar and the Ampere-Maxwell law, VXB = fiod + Hoe 2E, @ a ‘and the relationship between them. Most of these contribu- tions address one or more of three recurring questions (1) Can one derive the Maxwell displacement-current term, €, 9 E/dt, in Eq. (2) from Biot-Savart?™ 111 Am.J. Phys. $92), February 1991 (2) Should the d E/at term be included in the integrand of Biot-Savart (and/or the integrand of the vector poten- tial)?" (3) What relationship exists between the two “sources” of VB in Eq, (2); for instance, can one be derived from the other?!" This vast literature contains its share of confusions, re- dundancy, and lack of cross referencing. It bears testimony to the fact that these questions of basic physics are not “universally understood by referees and editors, as well as by authors. The problematic issues arise when time depen- denceis introduced into the strictly static limit. We hope in this paper to add clarity to these questions (a) by calling attention to two little-known but fundamental formulas re- cently presented by Jefimenko!? and (b) by pointing out © 1991 American Association of Physics Teachers 111 that there isa hierarchy of distinct special cases that fall in what is simplistically called the “quasistatic” or “slowly varying” regime? ‘The Ampere-Maxwell law can of course be recast in its integral form, § Bdl=n f (+62) @) ‘The magnetic field is constrained by the magnetic version of Gauss’ law (the nonexistence of magnetic monopoles) VB=0, 4) or, in integral form: f Bea -0. We shall also consider the electric analog: the relation between Coulomb's law, 2) ©) and the differential and integral forms of Gauss" law VE = (1/e.)p, om fbaa~ @) together with Faraday’s law: vxe= - 28, co) ar aB -di— — [2B 10 § * 9) Somewhat surprisingly, the literature has given much less attention to the status of Eq. (6) under time-dependent conditions than it has to Eq, (1) although the electric case is similarly problematic. The set of Eqs. (2), (4), (7), and (9) [or equivalently the set (3), (5), (8),and (10) ] constitute Maxwell's equa- tions, which are complete, general, and relativistically cor- rect, in classical physics.** By contrast, Biot-Savart (1) and Coulomb (6) are strictly valid only in the static limit (J and p independent of time)—although, as we shall see, they continue to hold under certain special dynamic condi- tions. Maxwell’s differential equations are “local” in the sense that they apply at one space-time point; hence the issue of retardation does not arise. (In the integral form, the sources p, J and the fields F, B are evaluated at all spatial poinis af the same time in one inertial frame.) On the other hand, Biot-Savart and Coulomb link the field at one spatial point to the distribution of sources over all space. Therefore, the issue of retardation becomes perti- nent if they are to be applied somehow to nonstatic sources, ‘Standard pedagogy bases the mathematical foundation of electromagnetic theory on Helmholtz’s theorem, which establishes that it s necessary and sufficient (given appro- priate boundary conditions” ) to specify both the diver- gence and the curl ofa vector field in order to determine the field.”*** In spite of the familiarity of this argument, there is opportunity for terminological confusion even in the static limit of electromagnetism. For instance, consider the relation between the magnetostatic field B and the current density J. A mathematician, following Helmholtz, could say that Ampere’s law [ (2), with E/ét—~0] and the Biot Savart law (1) are not equivalent—equivalence requires 112, Am.J. Phys, Vol. $9, No.2, February 1991 that Ampere be supplemented by the constraint V-B = 0, and that Biot-Savart be supplemented by the constraint ¥-J =0.% A physicist, however, could say that Ampere and Biot-Savart are equivalent since the mathematician’ constraints are implicit by the Maxwellian assumption, Eq, (4), that there are no magnetic monopoles, the static as- sumption that dp/@r is zero, and the equation of continuity expressing conservation of charge: v= -# ap Similarly, Gauss’ law (7) and Coulomb's law (6) are “equivalent” only when the former is supplemented by the constraint ¥XE = 0—which is to say, only in the static limit of Faraday’s law (9) Helmholtz’s theorem is by no means restricted to time- independent fields. However, it is of limited utility in the nonstatic context, because the Ampere/Maxwell law (2) and Faraday'slaw (9) donot specify thecurls of Eand Bas such in terms of the sources p and J—rather, they specify the hybrid quantities. [VXE+AB/ar] and [VXB —s1y¢, 9E/@t]. The formal mathematical exten- sion of Helmholtz’s theorem to include the induction terms is considerably more complicated.” Il. GENERALIZED BIOT-SAVART AND COULOMB LAWS In the nonstatic case, neither the Biot-Savart law (1) nor Coulomb's law (6) holds, in general. They do not con- tain the induction phenomena represented by Maxwell's 4 E/at termin (2) and Faraday’s B/@r termin (9). Heur- istically, one might suspect that the sources, J and p, in the integrands of Biot-Savart and Coulomb should now be evaluated at the retarded time, 1 ata rle, 2) (where ~ = |r —r'[)* as suggested by the standard re- tarded potentials of electrodynamics."*”” Unfortunately, sucha simple modification is not enough. What, then, isthe proper generalization? We seck a universally valid formula expressing the magnetic field [or the electric field] as an integral over the source currents [or charges], which re- duces directly to the Biot-Savart [or Coulomb] law in the static limit, In a formal sense, invoking Helmholtz’s theorem, one can simply combine the “displacement current” 3 E/dt with the true current J, to obtain a Biot-Savart-like for- mula!” pate [4a dB/Irs 4, a3) 4n This is correct and completely general {the integrand is evaluated at simultaneous (nonretarded) time]. Tt obvi- ously reduces directly to the Biot-Savart la the static case. However, it does not express B in terms of its true source, which is the current distribution J(r’,t). The Max- well term E/dt is necessary in the local equation (2) as a surrogate for the source currents at other places. Since Biot-Savart integrates over al spac, all the source currents can come in directly, without mediation by the electric field. In principle, J E/dtis no more a source of VxB, than ¥xXB is a source of @E/dt.”"” For practical computation, Eq. (13) isnot useful because it is self-referential: To calcu- late B at a point, one must know E everywhere—but you D.s.Grifithsand M.A, Heald 112 can't determine E unless you already know B (every- where), because of Faraday induction. Also, the absence of retardation invites confusion; like the scalar potential in the Coulomb gauge, Eq. (13) is ostensibly acausal—the physical retardation is buried deep, in the form of cancella- tions between the true current and the displacement eur- rent.” Jefimenko has recently called attention to the proper ‘generalized Biot-Savart law"! (DHE ean, «4 where the square brackets mean that the contents are to be evaluated at the retarded time, Eq. (12). This formula is readily obtained from the familiar retarded vector poten- tial?” Ho tf 181 (3 gp The ite field is the curl of A: He f [Luxist -tixe(4)] (16) Notice that [J] depends implicitly on the field point r, through the retarded time. Thus, 1 4123) x2, ar =(2)xv0, and Eq, (14) follows. Incidentally, the partial time deriva- tivein the second term of Eq, (14) can be taken outside the square brackets, or even outside the integral. The electric analog of Eq. (13) would be “fe = 69 BOX: a «sy B=¥x, ¥x[J] = ay «sy Ge, which again is formally correct but self-referential.™»”” Je- fimenko's generalized Coulomb law is"” 2, [dosar\z _ (as/ar This is obtained from the retarded scalar potential,”*2° 20 = -£S¢ Viel + lev He f [as/ae] gy, an But viel =(2)vun = -Y2)s, (22) from which Eq, (19) follows. The provenance of Eqs. (14) and (19) is curious. They are obvious enough if one has the wit to calculate them. But, until Jefimenko’s recent paper," the present authors were not aware of them, and circumstantial evidence indi- cates that the authors (and the referees) of our Refs. 5-11, 113 Am.J. Phys, Val. 59, No.2, February 1991, 14, and 17 were equally ignorant—otherwise most of these papers would surely have quoted them. As far as we can determine, these remarkable equations were discovered by Jefimenko in the late 1950s, and first published in his 1966 textbook.” They appear in the third edition (1988) of Lor- rain, Corson, and Lorrain. It seems extraordinary that they would not have been found much earlier. Probably they appear somewhere in the works of Jeans, Heaviside, Schott, Sommerfeld, et al., but they have certainly van- ished from the current literature and from the minds of the present generation of physics teachers. We think they greatly clarify many of the sometimes confused discussions in our Reference list. For completeness, we mention another pair of formulas for the fields, which do seem to be moderately well known: te [2 ar, (23) [Vp + (03/at)/c*) dr, 24) x nceeeeemede ee where as usual the brackets signify retardation. These may be obtained by solving the inhomogeneous wave equations for E and B in terms of the sources p and J.""* However, wwe regard Jefimenko’s equations as superior generaliza- tions of Biot-Savart and Coulomb because they contain no spatial derivatives (which are encumbered by retarda- tion), and because they reduce more transparently to the familiar static forms. III, UNDER WHAT NONSTATIC CONDITIONS, DOES THE BIOT-SAVART LAW STILL HOLD? By “static” we mean a configuration in which and J are constant in time [and hence J is solenoidal, VJ = O by Eq (11) ]. In the static case Ampere’s law [Eq. (2), omitting AE/dr] and the Biot-Savart law (1) both apply. But a glance at Eq. (14) reveals that the Biot-Savart law is more general than this: It holds whenever J is independent of t, regardless of whether p is constant or not. Unfortunately, this doesn’t buy us much, for the continuity equation (11) requires that if J is constant, p is at most linear in t: UE) = ple.0) + 1914.0), (25) where p = dp/at. Interestingly, in this case Coulomb's law also holds, by virtue of a fortuitous cancellation of the first two terms in Eq. (19):° lol + +) =plr0) + (« Spo += p(n = prt), (26) so that E-Z [Har 4re,) 7 with no retardation! The electric field is a function of time, so Ampere’s law fails (i.e., the Maxwell term is needed) — and yet both Coulomb's law and the Biot-Savart law apply. This is consistent with the principle that radiative effects enter only when charges are acceleroted. ‘Meanwhile, exactly the same cancellation occurs in Eq. (14) whenever J is linear in time (in which case is at most quadratic), so the Biot-Savart la holds here too, though in this case B is a function of time, and hence Coulomb's Taw (which would require VXE = 0) fils If pi in fact, Qn) D.J.Gritiths and M.A.Henld 113, constant, then so is E, and hence Ampere’s law holds. ‘There is a kind of hierarchy here: (a) Static (p and J constant in t): Biot-Savart, Cou- lomb, Ampere all hold. (b) Semistatic (p linear, J constant):"" Biot-Savart, Coulomb hold; Ampere fails. (©) Demisemistatic (p constant, J linear): Biot-Savart, Ampere hold; Coulomb fails. (A) Hemidemisemistatic (p linear or quadratic, J lin- car): Biot-Savart holds; Coulomb, Ampere fail. (©) General (arbitrary p and 3): Biot-Savart, Coulomb, Ampere all fail The fact that the Biot-Savart law holds in circumstances slightly more general than Ampere's has led many people over the years to think that they have “derived” the dis- placement current term in the Ampere-Maxwell law from Biot-Savart. A priori, one has no business using Biot-Sa- vart outside of magnetostatics; the fact that it happens to work in a marginally broader (“hemidemisemistatic”) context should be regarded as a curiosity, of no profound significance. In any event, there is clearly no way that the restricted Biot-Savart law can prove the completely gen- eral Ampere-Maxwell law, ‘Somewhere between the static regime and the general case lies the elusive quasistatic region, in which p and J change slowly enough that Coulomb's law and the Biot Savart law yield suitable approximations to the exact fields. In rea life, of course, there is no such thing asa truly stat (oreven hemidemisemistatic) configuration, To the extent that Coulomb’s law and the Biot-Savart law can be used at all (apart from purely pedagogical applications), it is be- ‘cause quasistatic conditions prevail. Fortunately, the gen- eral formulas for B and E, Eqs. (14) and (19), are struc- tured just right so as to make Coulomb's law and the Biot Savart law much better approximations than we had any right to anticipate. For if we expand the retarded current density (81 =a(n- 2) =e, 2 dee +4(4)aeo + (28) ake and plug this into Eq, (14), the by-now-familiar conspir- acy between the two terms effects a perfect cancellation of the first-order correction to the Biot-Savart law: aff b-aea (sien ~ 20-254 frcKar cv 1/+y. F Ho 3-4) : ) dr fO-sepe)e If Tis the “characteristic time” for changes in J, so that 5J~J/T°, then the dominant surviving correction to the Biot-Savart law is of order (~/eT)*, which is small as long asthe current changes little in the time takes alight signal to travel from the source to the field point ‘A similar analysis, applied to Eq, (19), yields |. (30) defo 114 Am.J. Phys, Vol $9, No.2, ebrusry 1991 (29) ne Once again, the first-order correction has canceled out; the largest surviving correction to Coulomb's law is of order G-/eT), where T is now the characteristic time for changes in p. The third term in Eq. (30) is ordinarily smaller still, for if we use the continuity equation (11), p= —V-J~I/R, where R is the characteristic distance for changes in J, then the third term isof order (R /») with respect to the second, and for a localized source R is typi- cally less than »."* Thus Coulomb's law and the Biot-Sa- vart law hold up to second order in the small parameter (-/eT), and in this sense the quasistatic approximation is “unreasonably” good. ‘Theapplication of Ampere’s law in quasistatic situations can be an extremely delicate matter. The problem arises when we convert Ampere’s law from differential to integral form, G Bel =ne f See, G1 for the surface integral “sniffs out” re quasistatic, even when the line integral is safely in the qua- sistatic zone. In differential form, then, Ampere’s law is a suitable quasistatic approximation, but in integral form cone must be extremely cautious. ‘We have seen how the appropriately generalized Biot Savart law (14) reduces to the Biot-Savart law for “hemi- demisemistatic” configurations. But how about Eq. (13)? This too should reduce to the Biot-Savart law—evidently the displacement current term must integrate to zero when is linear in ¢. This is easily confirmed, as follows: a Sans J exe -2fft (xe) —vx(E) dr --2 (By ar) [We invoked Faraday’s law, Eq. (9), in the first term; the second term becomes a surface integral, and vanishes for localized sources. But in the hemidemisemistatic case, B like J) is linear in t, and hence the second derivative is In the quasistatic regime the displacement-current term, (32) is not precisely zero; nevertheless, as a correction to the Biot-Savart law it is at most of second order in the small parameter (~/cT) [Barta ard ery. (an tisdoes not akin acount the decreassin Bas get farther from the source). Thus, if we work with Eq. Eib)itfe appropriate to onthe dopant current the quite apprimaton Cast seven worag with (14)—one omits the second term and ignores retarda- tion). Direct observation of a magnetic field associated with shpat cuentas reel epee oy Bae nd Corle.” Hovey has given a simplified description of the ‘experimental validation of the Biot-Savart law.*! Aninteresting example of the quasistatic approximation iste "Bax Seta ew for apm chargeqtovelng wth i. (32) Hof D.S.Grifths and M.A.Heald 114 velocity ¥ B fe BE G4) Since a moving charge is not hemidemisemistatic even when the velocity is constant, this often-quoted formula cannot be exact. Nevertheless, it provides a good approxi- ‘mation to the field, the error being of order (v/c)*. More- over, if Eq. (34) is integrated over the continuum of infini- tesimal charges making up a hemidemisemistatic configuration, the result is exact, even though the starting equation is not.” Evidently the “extra” terms integrate to zero, and hence in this application it does not matter whether one uses the true field of a point charge or its Biot- Savart approximation. IV, DERIVATION OF THE FIELDS OF A POINT CHARGE FROM JEFIMENKO’S EQUATIONS ‘When one goes beyond some form of quasistatic limit, the problem of greatest interest is to find the fields of an arbitrarily moving point charge. The usual approach is to introduce the Liénard-Wiechert potentials, and from them, derive the electric and magnetic fields." We now show how the Liénard-Wiechert fields can be obtained from Jefi- menko's equations (14) and (19). Let the position of the point charge q be w(1), and its velocity v = dw/dt. Then the retarded charge and current, densities are given in terms of the delta function by (el =e.) =98 (1 = Set, = [pv] w(t), (3s) QV 8(F —WUi)), G36) where w({,) is the position of the particle at the retarded time ole. an) ‘Thus, substituting in Eq. (19), we have Fr —w(t,))F E(r, 4 Te a ll 2 1a (OME? y c a - 1a Pune 4) ay oe ot ” Because of the delta function, the various factors of », 2, and v come outside the integral, where they are now evalt- ated at the specific retarded position and time defined, for given field coordinates r and z, by Eq. (37) andr’ = w(t). Tomark the fact that the source-point-to-field-point vector is now a function of time, we introduce the notation roR=RAi wr), (9) Hence Eq, (37) becomes t, = 1— R /e. ‘There remains the integral of the delta function, which is done formally by changing the variable of integration from ¥ to z=1'—w(t,) using the Jacobian K = d(2)/A(r') 1 wife = 1 — Bai fee —wandr= 5 Ba where fi = (v/c) and fiare (again) evaluated at the retard- 115 Am.J. Phys, Vol.$9, No.2, February 1991 closion an tine. SB (8) bomen =k (eel + cole -2a ney (lee cc alkR| at Similarly, from Eq, (14) we obtain aon a(Relt cater) Equations (41) and (42) are essentially in the form made famous by Feynman.***° To obtain the conventional Liénard-Wiechert formulas, ‘we must carry out the time derivatives, which are not tri- vial because of the subtle relation between present and re~ tarded time: aR_@ =f2r—wer, (r—w0,)) al) (41) (44) (45) (46) a’ But it follows from the definition of retarded time, R=c(t—1,), that OR fi %) : «ap a a ‘which combines with Eq. (44) to give % od (48) a 1-pa K Putting all this together, we obtain (after some algebraic simplification) po 2 [G@=B) , axa B) xa) 49. dre, (FRR? ERR G2 and B= (1/e) [A] XE. (50) ‘These are the well-known formulas for the fields of a point charge in arbitrary motion, with square brackets denoting retarded values, and using the common shorthands B= vey = (1 —B?)~'?, and K = (1 — Ba). 8" Wechallenge the reader to reverse the calculation of this, section, that is, to derive Jefimenko's equations (19) and (14) directly from the point-charge fields (49)-(50), us- ing the superposition principle. V. CONCLUSION We return now to the three “recurring questions” cited in our introduction: (1) Can one derive the displacement current term in the Ampere-Maxwell law from Biot-Savart? Answer: No, not in general. But it can be done in certain very special cases (nicknamed “‘semistatic and hemidemisemistatic" in our hierarchy), and this possibility accounts for some of the confusion in the literature. (2) Should the @ E/at term be included in the integrand of Biot-Savart? Answer: It can be, as in Eq. (13), but while technically correct this equation is seldom useful and po- D.J.Grifthsand M.A.Heald 115, tentially misleading. In the quasistatic limit the extra term integrates to zero, or is swamped by the (ordinary) current 4, whereas under rapidly varying conditions one rarely knows the electric field in advance, and one may be tempt- ed (as many authors have been) to use an E obtained from an inapplicable Coulomb law. (3) What relation exists between the two “sources” of YXB in Eq, (2)? Answer: None. The two terms are inde- pendent and both are needed, in general: J represents the ordinary current ar the point where the curl is evaluated, while the “displacement current” 4 E/@t is a surrogate for ordinary currents at other locations. When the Ampere ‘Maxwell aw is recast in its integral form, Eq. (3), the do- ‘main of the surface integral on the right is arbitrary (pro- Vided only that its perimeter matches the integration con- tour on the left). Therefore, in a specific case one may find some integration surfaces that are dominated by ordinary currents and others that are dominated by displacement currents, but ultimately, the source of every magnetic field isJ. In the course of this study we have found Jefimenko's equations (14) and (19) extraordinarily powerful and illu- inating. We hope they will become more widely known. Not only do they clearly establish the domains of applica- bility of Coulomb's law and the Biot-Savart law, but they offer a clean and efficient way to derive the fields of a point charge. Jefimenko himself has used them to obtain the fields of oscillating electric and magnetic dipoles."* We find it hard to believe that these equations were not discov- ered until the late 1950s, and we would be interested to learn of any earlier references. 'P. A. Mello, “A remark on Maxwell's displacement curtent.” Am. J Phys. 40, 1010-1013 (1972). Compare Ref. 4 °T, Biswas, "Displacement current—A direct derivation 56, 373-374 (1988). Compare Ref. & °N- Gauthier, “A diret derivation ofthe displacement current, Phys $6, 871-872 (1988), famias,"Comment on ‘Displacement current—A dice derivation,” bby T. Biswas,” Am. J. Phys. 87, 857-558 (1989) ‘8.3. Raf, “Ampere’s law and the vector potenti,” Am. J, Phys 26, 454-460 (1958), A. P. French and Jack R. Tessman, “Displacement curtents and mag- netic feds," Am. J. Phys. 3, 201-204 (1963), °W. G. V. Rosser, “Interpretation ofthe displacement current,” Am. J Phys. 31, 807-808 (1983), “Robert M. Whitmer, "Calculation of magnetic and electric fds from displacement cureens," Am. J. Phys 33, 481-484 (1968), “AW. G. V. Rosser, "Displacement current and Maxwells equation,” Am, 4. Phys. 43, 802-805 (1975), W.G.V. Rosser, "Does the displacement current in fee space produce magnetic eld?” Am. J. Phys. 44, 1221-1223 (1976), VT. A Weber and D. J, Macomb, “On the equivalence ofthe laws of Biot-Savart and Ampere,” Am. J. Phys. 87, 37-89 (1989). Compare Ref. 12 "Oleg D. Jeimenko, “Comment on “On the equivalence of the las of ‘Biot-Savart and Ampere’ by Weber and Macomb,” Am. J. Phys. $8, 505 (1990) "John P. Vini and D. J. X. Montgomery, "Note on the presentation of “Maxwell's equations,” Am.J. Phys. 17, 298-300 (1949), "Wiliam K. Terry, “The connection between the charged-particle cur ‘ent and the displacement current,” Am. J Phys. 80, 742-745 (1982) * A.J. Dahm, “Calculation of the displacement current using the integral form of Ampere’ law,” Am.J. Phys. 46, 1227 (1978). N, Gauthier, "Displacement current, transport current, and charge conservation," Am. J. Phys $1, 168-170 (1983) "W.G. V. Rosser, “The displacement current,” Am. 3. Phys. 51, 1149 (1983); William K. Tey, “Reply to “The displacement current” "Am. J. Phys Am. 116 Ama J. Phys, Vol $9, No. 2, Febraary 1991 ‘Am.J. Phys. 81, 1150 (1983). "B.R. Russell, "A different approach to Ampere's law,” Am.J-Phys.83, 918 (1985), "G. B. Walker, “The axioms underlying Maxwel's electromagnetic equations.” Am. J. Phys, 53, 1169-1172 (1985); John P. Vintiand D3 X. Montgomery, “Note on “The axioms underiying Maxwell's clectro- ‘magnetic equations’ [Walker], Am. J. Phys. 84,872 (1986); G. B. Walker, "Reply tothe note by Vinti and Montgomery.” Am. J. Phys. 54,872 (1986). 2 Harcld 8. Zapolsky, "Does charge conservation imply the displacement ‘current?” Am. J. Phys. 55, 1140 (1987) 2D, F Bartlett, "Conduction current and the magnetic eld in circular capaciter.” Am. J. Phys. 88, 168-1172 (1990), Our identification of Eqs. (1) and (2) by the names Biot-Savart and Ampere[ Maxwell) is by no means unversally accepted. We do not ‘wish to argue the merits of this debate; however, believing that it is ppedagogically useful o dignity fundamental physical laws with proper ‘names, we follow here what appears to be the most common practice. ‘We use the name Ampere's law [without Maxwell} to designate the reduced forms of Eas. (2) and (3) omitting the 9E/d¢ term, In Eq (1) and throughout this paper we use a script» to denote the ‘ecto from thesource point (7), at which in—Eq,(1)—J evaluate, (0 the field point (¢), at which B is evaluated: » =-2-=r—'. The element of volume at the source is dr = d';u and dare elements. of area and length influx and circulation integrals. For simplicity, we assume that poarizable/magnetizable media are ‘ot present. We use the term “dgplacement current” for ¢, JE/@t. IF ‘one wishes to include material media, one simply interprets J to include <4 /ar and ¥ XM in addition othe fee current density Jn and p to include — VP in addition 1 the free charge density pa. ‘Wedonotaddressheretheasymmetryoftheforcebetweenclementsof two current circuits, using Biot-Savart and the Lorentz force law, See Cynthia Koll Whitney, “On the Ampere/Biot-Savart discussion.” Am. J. Phys, 6,871 (1988) and C, Christodoulides, “On the equivalence of the Ampere nd Biot-Savart magnetosaticforcelaws,” Am.J.Phys. 87, 1680 (1989), The usual argument assumes that charge and current densities g0 to zero at large distances, so that all volume integrals converge, and all, surface integrals “a ininity” dropout. Iimposes “natural” boundary ‘conditions on E and B: They go t0 zero at lage distances from the source. The argument applies as well to such artificial nonlcalized ‘configurations asthe infinite wire and the infinite plane, But there exist pathological cases (e.g, all space filled with uniform and constant, charge or current density) for which the Coulomb and Biot-Savart, integral are hopelessly indefinite, and ao natural boundary conditions ‘or symmetry arguments ae avilable o resolve the ambiguities in ap plying Gauss law or Ampere's la “Wolfgang K. H. Panofaky and Melba Phillips, Classical Electricity ond Magnetism (Adsison- Wesley, Reading, MA, 1962), 2nd ed, Se. 1-1; George Arfken, Mathematical Methods for Physician (Academic, New York, 1985), 3rd ed, See. 115. © David 1. Grif, Introduction to Electrodynamics (Prentice-Hall, En- slewood Cli, NJ, 1989), 2nd ed, Se. 1.6 "Reference 28, Se. 5.3.2 Donald H. Kobe, "Helmholt’s theorem revisited," Am. J. Phys. 84, 552-554 (1986); Jose A. Heras, "A short proof of the generalized Helmholtz theorem,” Am. J. Phys. 58, 154-155 (1990), ery B. Marion and Mark A. Heald, Clasial Electromagnetic Ro ‘ation (Academic, New York, 1980), 2nd ed, Sec. 72 ® Paul Lorrain, Dale P, Corson, and Frangois Lorrain, Electromagnetic Fields and Waves (Freeman, New York, 1988), 3rd ed, Sec. 374 Equation (13) ean be treacherousin certain quasistatic problems where ‘the integration ofthe 4/3 term over all space vanishes identically. ‘the integration, typically of Fringe fl, isnot done correctly, one is actually worse off to have included the “Maxwell correction See French and Tessman, Ref. 6 “Oleg D. Hefmenko, Eletricity and Magnetism (Appleton-Century ‘Crafts, New York, 1966), See 15-7;orsamesectionin2nded, (Electret Scientific, Stat City, WY, 1989), © Francis W. Sears, “Faraday’s law and Ampere'sIaw,” Am. Phys 31, 439-443 (1963), D.J.Griffiths snd M.A.Heald 116 © Inthe study ofinductane, tis common to invoke a quasistatic approxi> ration which amount to ignoring the “Maxwell” term 3/21 in Eq, (13), while carefully preservingthe "Faraday" @B/@r termin Eq, (18) Typically, at low frequencies, Maxwell induction is swamped by the current term J in Eq, (2), bu there is no analogous term in Eq (9) to ‘conceal Faraday induction. See Edward M. Purcell, Electricity and Magnetism (MeGraw-Hill, New York, 1985), 2nd ed, Sc. 9.2. Reference 28, See. 376 ° Reference 29, Sees. 27.8 and 37.5 Fora somewhat diferent approach to thiseasesee Prob. 7.51 in Ref. 25, ps "Some authors would regard the “semistaic™ case as magnetostatic (be cause 35/91 ~0) but not electrastatic (because Jp/3t #0). For them ‘the Biot-Savart law holds in magnetostatic, but Ampere’ aw applies ‘only in lectostatis, We think tis wiser to reserve the word "magneto- stati” fo full static configration, in whieh E and B are both inde pendent of time, and Ampere's Iw and the Biot-Savart law both hold When feldsare tobe calculated inside a charge configuration, quasisat- ie conditions require that (-/eT)? and -R /(eT)?~ (0/e)(o/eT) are both sal For a more physical discusion of this point see Ref. 33, Sec. 9.2; and also Refs. 6 and 20, The situation is reminiscent of the caleulation of Faraday emf asthe line integral of E around a closed loop: E can be taken tobe either the total eld or the nonconservative portion alone, since the conservative part will integrate 10 zero. Another analogous situation s the well-known ambiguity in defining electromagnetic ener ay density (se, fr instance, Ref, 28, Prob. 412, p. 128) “©D. F Bartlett and T. R Corle, "Measuring Maxwell's displacement ‘current inside a capacitor,” Phys. Rev. Let $8, 9-62 (1985). See also D. F. Bartlett and Glenn Gengel, “Measurement of quasistatic Max- wells displacement current,” Phys. Rev. A 39, 98-945 (1989); Ref, 21; and Thomas R, Carver and Jan Rajhel, “Direct ‘itera? demonstra- tion ofthe eet of a displacement current,” Am. J. Phys. 42, 246-249 (1974), “Arthur Hovey, “On the magnetic fed generated by a short segment of current,” Am. J. Phys. $7, 613-616 (1989), ‘John David Jackson, Clasical Electrodynamics (Wiley, New York, 1975), 2nd ed, footnote p. 170, See Ref. 28, Prob, 7-3, p. 222 “© For instance, Ref 25, See. 9.2; Ref. 24, Chaps. 19-20 “Reference 42, See. 141, and Jacksons firs edition (1962), See. 141, Which presents the discussion in a more familiar notation; Alan M. Portis, Electromagnetic Fields (Wiley, New York, 1978), Appendix H. “Richard P, Feynman, Robert B. Leighton, and Matthew Sands, The Feynman Lectures on Physics, Vol 1 (Addison-Wesley, Reading, MA, 1964), Chap. 21; AR Janab, T. Padmanabhan, and T.P. Singh, "On Feynman's formula for the eleetomagnetic field of an arbitrarily mov ingcherg," Am. J. Phys, 6, 1036-1038 (1988) “Oliver Heaviside, Electromagnetic Theory, Vol. Il (Chelsea, New York, 1971 [1912]), Sec. 534 (esp. p. 438); J. 3. Monaghan, "The Heaviside-Feynman expression fr the elds of an accelerated dipole.” J-Phys. A, 112-117 (1968) “Tris ofcourse well known that the point-charge magnetic felis related to its electric eld by the rematkably simple relation Eg. ($0). One right expect thatthe integrands ofthe Jefmenko field equations (19), and (14) would be related by the analogous relation, ¢ dB = 7d E, ‘ut they are not. Te reason fortis curiosity les in the subtle distinc tion between the soure-point-to-eld-point vector» inthe volume inte- srals (whichis not dependent on time or retardation) and the core- Sponding vector R in the point-charge formulas (which is time dependent and retarded) “Reference 31, Sec. 168 nT ELECTROSTATICS EXPERIMENTS IN DAMP WEATHER ‘There are dozens of experiments in the electrostatic category that are spectacular when they work, but rather dismal if they do not. I remember hearing of an incident in which a lecturer was due to talk toa large group of firemen about the part played by electrostatic sparks in starting fires and explosions. He decided to start with a dramatic demonstration; he was to walk into the room without a word and to peel off a woolen jumper, worn over a silk shirt, and to apply his finger toa as et, thereby producing a spark and igniting the jet. It worked successfully every time he tried at rehearsal. But at the actual lecture nothing happened and so his lecture started with a rather limp explanation of what should have happened. What went wrong? He had clearly omitted to realize that the absence of several hundred people emitting water vapor at the rehearsal enabled him to perform the experiments in a dry atmosphere, but at the lecture proper the enormous amount of ‘water vapor produced by the audience was enough to ruin any electrostatic demonstration. Charles Taylor, The Art and Science of Lecture Demonstration (Hilger, Bristol, 1988), pp. 85-86. ‘Am. J. Phys. Vol. $9, No.2, February 1991 DJ. Grifiths and M.A. Heald "7

You might also like