The Classical Theory of Fields Course of Theoretical Physics Volume 2 L.D. Landau and E.M. LifshitzTHE CLASSICAL THEORY OF FIELDS Fourth Revised English Edition L. D. LANDAU AND E. M. LIFSHITZ Institute for Physical Problems, Academy of Sciences of the U.S.S.R. Translated from the Russian by MORTON HAMERMESH University of Minnesota UTTERWORTH EINEMANN AMSTERDAM BOSTON HEIDELBERG LONDON NEW YORK OXFORD PARIS SAN DIEGO SAN FRANCISCO SINGAPORE SYDNEY TOKYOCONTENTS EXCERPTS FROM THE PREFACES TO THE First AND SECOND EpItions PREFACE TO THE FourtH ENGLisH EDItion Eprror’s PREFACE TO THE SEVENTH RUSSIAN EpITion Notation Charter 1. THE Principle oF RELATIVITY Velocity of propagation of interaction Intervals Proper time The Lorentz transformation ‘Transformation of velocities Four-vectors Four-dimensional velocity yauaAwne Cuapter 2. RELAtivistic MECHANICS 8 The principle of least action 9 Energy and momentum 10 Transformation of distribution functions 11 Decay of particles 12 Invariant cross-section 13 Elastic collisions of particles 14 Angular momentum (Cuaprer 3, CHARGES IN ELECTROMAGNETIC FIELDS. 15 Elementary particles in the theory of relativity 16 Four-potential of a field 17 Equations of motion of a charge in a field 18 Gauge invariance 19 Constant electromagnetic field 20 Motion in a constant uniform electric field 21 Motion in a constant uniform magnetic field 22. Motion of a charge in constant uniform electric and magnetic fields 23 The electromagnetic field tensor 24 Lorentz transformation of the field 25 Invariants of the field pe) pa) 26 30 32 36, 38 42 46 46 47 49 a 53 5S 56 59 64 66 67vi CONTENTS Cuapter 4. THE ELECTROMAGNETIC FIELD EQUATIONS 26 The first pair of Maxwell’s equations 27 The action function of the electromagnetic field 28 The four-dimensional current vector 29 The equation of continuity 30 The second pair of Maxwell equations 31 Energy density and energy flux 32 The energy-momentum tensor 33. Energy-momentum tensor of the electromagnetic field 34 The virial theorem 35. The energy-momentum tensor for macroscopic bodies ‘Cuapter 5. CONSTANT ELECTROMAGNETIC FIELDS 36 Coulomb's law 37 Electrostatic energy of charges 38 The field of a uniformly moving charge 39 Motion in the Coulomb field 40 The dipole moment 41 Multipole moments 42. System of charges in an external field 43 Constant magnetic field 44 Magnetic moments 45 Larmor’s theorem Cxaper 6. ELECTROMAGNETIC Waves 46 The wave equation 47 Plane waves 48 Monochromatic plane waves 49 Spectral resolution 50 Partially polarized light 51 The Fourier resolution of the electrostatic field 52 Characteristic vibrations of the field Chapter 7. THE PRopAGATION OF LiGHT 53 Geometrical optics 54 Intensity 55 The angular eikonal 56 Narrow bundles of rays 57 Image formation with broad bundles of rays 58 The limits of geometrical optics 59 Diffraction 60 Fresnel diffraction 61 Fraunhofer diffraction 70 70 oe ae 16 7B 80 82 86 92 95 ae 96 98 100 103 105 108 110 11 113 116 116 118 123 128 129 134 135 140 140 143 145 147 153 154 156 162 165‘CONTENTS (Cuapter 8. THE FIELD OF MOVING CHARGES 62. The retarded potentials 63 The Lienard—Wiechert potentials 64 Spectral resolution of the retarded potentials 65 The Lagrangian to terms of second order (Cuaprer 9. RADIATION OF ELECTROMAGNETIC WAVES 66 The field of a system of charges at large distances 67 Dipole radiation 68 Dipole radiation during collisions 69 Radiation of low frequency in collisions 70. Radiation in the case of Coulomb interaction 71 Quadrupole and magnetic dipole radiation 72 The field of the radiation at near nces 73 Radiation from a rapidly moving charge 74 Synchrotron radiation (magnetic bremsstrahlung) 75 Radiation damping 76 Radiation damping in the relativistic case 77 Spectral resolution of the radiation in the ultrarelativistic case 78 Scattering by free charges 79 Scattering of low-frequency waves 80 Scattering of high-frequency waves Ciaprer 10. PARTICLE IN A GRAVITATIONAL FIELD 81 Gravitational fields in nonrelati ic mechanics 82 The gravitational field in relativistic mechanics 83 Curvilinear coordinates 84 Distances and time intervals 85 Covariant differentiation 86 The relation of the Christoffel symbols to the metric tensor 87 Motion of a particle in a gravitational field 88 The constant gravitational field 89 Rotation 90 The equations of electrodynamics in the presence of a gravitational field Cuaprer 11. THe GRAVITATIONAL FELD EQuations: 91 The curvature tensor 92 Properties of the curvature tensor 93 The action function for the gravitational field 94 The energy-momentum tensor 95. The Einstein equations 96 The energy-momentum pseudotensor of the gravitational field 97 The synchronous reference system 98 The tetrad representation of the Einstein equations 17 171 173 176 179 184 184 187 191 193 195 203 206 210 215 222 226 230 233 238 240 243 243 244 247 251 255 260 263 266 273 aaee 278 278 281 287 290 295 301 307 313Vili ‘CONTENTS Cnaprer 12. THE Fre.o or Gravitatinc Bopies: 99 Newton's law 100 The centrally symmetric gravitational field 101 Motion in a centrally symmetric gravitational field 102 Gravitational collapse of a spherical body 103 Gravitational collapse of a dustlike sphere 104 Gravitational collapse of nonspherical and rotating bodies 105. Gravitational fields at large distances from bos 106 The equations of motion of a system of bodies in the second approximation Carter 13. GravitatiONAL WAVES 107 Weak gravitational waves 108 Gravitational waves in curved space-time 109 Strong gravitational waves 110 Radiation of gravitational waves Cuaprer 14. Revarivistic CosMoLocy, 111 Isotropic space 112 The closed isotropic model 113 The open isotropic model 114 The red shift 115 Gravitational stability of an isotropic universe 116 Homogeneous spac: 117 The flat anisotropic model 118 Oscillating regime of approach to a singular point 119 The time singularity in the general cosmologit Einstein equations INDEX 316 316 320 328 331 338 344 353 360 368 368 370 373 376 382 382 386 390 394 400 406 412 416 420 425EXCERPTS FROM THE PREFACES TO THE FIRST AND SECOND EDITIONS Tus book is devoted to the presentation of the theory of the electromagnetic and gravitational fields, i.e. electrodynamics and general relativity. A complete, logically connected theory of the electromagnetic field includes the special theory of relativity, so the latter has been taken as the basis of the presentation. As the starting point of the derivation of the fundamental relations we take the variational principles, which make possible the attainment of maximum generality, unity and simplicity of presentation. In accordance with the overall plan of our Course of Theoretical Physics (of which this book is a part), we have not considered questions concerning the electrodynamics of continuous media, but restricted the discussion to “microscopic electrodynamics”—the electrodynamics of point charges in vacuo. The reader is assumed to be familiar with electromagnetic phenomena as discussed in general physics courses. A knowledge of vector analysis is also necessary. The reader is not assumed to have any previous knowledge of tensor analysis, which is presented in parallel with the development of the theory of gravitational fields. Moscow, December 1939 Moscow, June 1947 L. Lanpau, E. LirsHrrzPREFACE TO THE FOURTH ENGLISH EDITION ‘The first edition of this book appeared more than thirty years ago. In the course of reissues over these decades the book has been revised and expanded; its volume has almost doubled since the first edition. But at no time has there been any need to change the method proposed by Landau for developing the theory, or his style of presentation, whose main feature was a striving for clarity and simplicity. I have made every effort to preserve this style in the revisions that I have had to make on my own, ‘As compared with the preceding edition, the first nine chapters, devoted to electrodynamics, have remained almost without changes. The chapters concerning the theory of the gravitational field have been revised and expanded. The material in these chapters has increased from edition to edition, and it was finally necessary to redistribute and rearrange it. I should like to express here my deep gratitude to all of my helpers in this work—too many to be enumerated—who, by their comments and advice, helped me to eliminate errors and introduce improvements. Without their advice, without the willingness to help which has met all my requests, the work to continue the editions of this course would have been much more difficult. A special debt of gratitude is due to L. P. Pitaevskii, with whom I have constantly discussed all the vexing questions. ‘The English translation of the book was done from the last Russian edition, which appeared in 1973. No further changes in the book have been made. The 1994 corrected reprint includes the changes made by E. M. Lifshitz in the Seventh Russian Edition published in 1987. I should also like to use this occasion to sincerely thank Prof. Hamermesh, who has translated this book in all its editions, starting with the first English edition in 1951. The success of this book among English-speaking readers is to a large extent the result of his labour and careful attention. E, M. Lirsurrz, PUBLISHER'S NOTE, As with the other volumes in the Course of Theoretical Physics, the authors do not, as a rule, give references to original papers, but simply name their authors (with dates). Full bibliographic references are only given to works which contain matters not fully expounded in the text.EDITOR’S PREFACE TO THE SEVENTH RUSSIAN EDITION E. M. Lifshitz began to prepare a new edition of Teoria Polia in 1985 and continued his work on it even in hospital during the period of his last illness. The changes that he proposed are made in the present edition. Of these we should mention some revision of the proof of the law of conservation of angular momentum in relativistic mechanics, and also a more detailed discussion of the question of symmetry of the Christoffel symbols in the theory of gravitation. The sign has been changed in the definition of the electromagnetic field stress tensor. (In the present edition this tensor was defined differently than in the other volumes of the Course.) June 1987 L. P. Prmazvskt xiNOTATION Three-dimensional quantities Three-dimensional tensor indices are denoted by Greek letters Element of volume, area and length: dV, df, dl Momentum and energy of a particle: p and # Hamiltonian function: 7 Scalar and vector potentials of the electromagnetic field: @ and A Electric and magnetic field intensities: E and H Charge and current density: p and j Electric dipole moment: d Magnetic dipole moment: m Four-dimensional quantities Four-dimensional tensor indices are denoted by Latin letters i, k, J, ... and take on the values 0,1,2,3 We use the metric with signature (+ - — -) Rule for raising and lowering indices—see p. 14 Components of four-vectors are enumerated in the form A’ = (A°, A) Antisymmetric unit tensor of rank four is e”", where e'?3 = 1 (for the definition, see p. 17) Element of four-volume dQ = dx°dx!dx"dx? Element of hypersurface dS’ (defined on pp. 20-21) Radius four-vector: x! = (ct, r) Velocity four-vector: ué = dx‘/ds Momentum four-vector: p = (#/c, p) Current four-vector: j= (cp, pv) Four-potential of the electromagnetic field: A’ = (¢, wm a (for the relation of the components of Electromagnetic field four-tensor Fy, = © in the formulas of relativistic mechanics. Inclassical mechanics distance is already relative, i.e. the spatial relations between different events depend on the system of reference in which they are described. The statement that two nonsimultaneous events occur at one and the same point in space or, in general, at a definite distance from each other, acquires a meaning only when we indicate the system of reference which is used. On the other hand, time is absolute in classical mechanics; in other words, the properties of time are assumed to be independent of the system of reference; there is one time for all reference frames. This means that if any two phenomena occur simultaneously for any one observer, then they occur simultaneously also for all others. In general, the interval of time between two given events must be identical for all systems of reference. It is easy to show, however, that the idea of an absolute time is in complete contradiction tothe ein principle of relativity. For this it is suffcient to recall that in classical mechanics, based on the concept of an absolute time, a general law of combination of velocities is valid, according to which the velocity of a composite motion is simply equal to the (vector) sum of the velocities which constitute this motion. This law, being universal, should also be applicable to the propagation of interactions. From this it would follow that the velocity of$2 INTERVALS. 3 propagation must be different in different inertial systems of reference, in contradiction to the principle of relativity. In this matter experiment completely confirms the principle of relativity. Measurements first performed by Michelson (1881) showed complete lack of dependence of the velocity of light on its direction of propagation; whereas according to classical mechanics the velocity of light should be smaller in the direction of the earth’s motion than in the opposite direction. Thus the principle of relativity leads to the result that time is not absolute. Time elapses differently in different systems of reference. Consequently the statement that a definite time interval has elapsed between two given events acquires meaning only when the reference frame to which this statement applies is indicated. In particular, events which are simultaneous in one reference frame will not be simultaneous in other frames. To clarify this, it is instructive to consider the following simple example. Let us look at two inertial reference systems K and K’ with coordinate axes XYZ and X’Y’Z’ respectively, where the system K” moves relative to K along the X(X’) axis (Fig. 1) z Zz BeAsc Fic. 1. Suppose signals start out from some point A on the X’ axis in two opposite directions. Since the velocity of propagation of a signal in the K’ system, as in all inertial systems, is equal (for both directions) to c, the signals will reach points B and C, equidistant from A, at one and the same time (in the K” system) But it is easy to see that the same two events (arrival of the signal at B and C) can by no means be simultaneous for an observer in the K system. In fact, the velocity of a signal relative to the K system has, according to the principle of relativity, the same value c, and since the point B moves (relative to the K system) toward the source of its signal, while the point C moves in the direction away from the signal (sent from A to C), in the K system the signal will reach point B earlier than point C. Thus the principle of relativity of Einstein introduces very drastic and fundamental changes in basic physical concepts. The notions of space and time derived by us from our daily experiences are only approximations linked to the fact that in daily life we happen to deal only with velocities which are very small compared with the velocity of light. § 2. Intervals In what follows we shall frequently use the concept of an event. An event is described by the place where it occurred and the time when it occurred. Thus an event occurring in a certain material particle is defined by the three coordinates of that particle and the time when the event occurs. It is frequently useful for reasons of presentation to use a fictitious four-dimensional4 ‘THE PRINCIPLE OF RELATIVITY g2 space, on the axes of which are marked three space coordinates and the time. In this space events are represented by points, called world points. In this fictitious four-dimensional space there corresponds to each particle a cetain line, called a world line. The points of this line determine the coordinates of the particle at all moments of time. It is easy to show that to a particle in uniform rectilinear motion there corresponds a straight world line. We now express the principle of the invariance of the velocity of light in mathematical form. For this purpose we consider two reference systems K and K’ moving relative to each other with constant velocity. We choose the coordinate axes so that the axes X and X’ coincide, while the Y and Z axes are parallel to Y’ and Z’; we designate the time in the systems K and K’ by f and 1’. Let the first event consist of sending out a signal, propagating with light velocity, from a point having coordinates x1y;z; in the K system, at time t; in this system. We observe the propagation of this signal in the K system. Let the second event consist of the arrival of the signal at point x2y2z) at the moment of time t. The signal propagates with velocity c; the distance covered by it is therefore c(t, ~f9). On the other hand, this same distance equals, [0q- x)? + (2—y)? + (zz — 1). Thus we can write the following relation between the coordinates of the two events in the K system: (2-41) + 02-1? + @~ AP - Ah - HY = 0. 1) the propagation of the signal, can be observed from the K’ The same two events, i. system: Let the coordinates of the first event in the K’ system be x{y{zjt{, and of the second: x4y3z3t}. Since the velocity of light is the same in the K and K’ systems, we have, similarly to 2. (ag — af)? + 8 — yi)? + - at)? — 703 - HH)? =0. (2.2) If x12 fy and x»y2Z9t are the coordinates of any two events, then the quantity 1 An — 02-41? - 02-1 ~ (@ - a) P (2.3) is called the interval between these two events. Thus it follows from the principle of invariance of the velocity of light that if the interval between two events is zero in one coordinate system, then it is equal to zero in all other systems. If two events are infinitely close to each other, then the interval ds between them is ds? = Ad? ~ dx? — dy? d2. (24) The form of expressions (2.3) and (2.4) permits us to regard the interval, from the formal point of view, as the distance between two points in a fictitious four-dimensional space (whose axes are labelled by x, y, z, and the product ct), But there is a basic difference between the rule for forming this quantity and the rule in ordinary geometry: in forming the square of the interval, the squares of the coordinate differences along the different axes are summed, not with the same sign, but rather with varying signs. As already shown, if ds = 0 in one inertial system, then ds’ = 0 in any other system. On si =I - + The four-dimensional geometry described by the quadratic form (2.4) was introduced by H. Minkowski, in connection with the theory of relativity. This geometry is called pseudo-euclidean, in contrast to ordinary euclidean geometry.§2 INTERVALS 5 the other hand, ds and ds’ are infinitesimals of the same order. From these two conditions it follows that ds? and ds’? must be proportional to each other: ds? = ads” where the coefficient a can depend only on the absolute value of the relative velocity of the two inertial systems. It cannot depend on the coordinates or the time, since then different points in space and different moments in time would not be equivalent, which would be in contradiction to the homogeneity of space and time. Similarly, it cannot depend on the ditection of the relative velocity, since that would contradict the isotropy of space. Let us consider three reference systems K, K,, Kz, and let V, and V2 be the velocities of systems K, and K; relative to K. We then have: ds? =a(Vi)ds?, ds? = a(V,)ds} Similarly we can write ds? = a(Vj2)ds3, where Viz is the absolute value of the velocity of K> relative to Kj. Comparing these relations with one another, we find that we must have a(V2) ay; ) But Vip depends not only on the absolute values of the vectors V, and V>, but also on the angle between them. However, this angle does not appear on the left side of formula (2.5). Itis therefore clear that this formula can be correct only if the function a(V) reduces to a constant, which is equal to unity according to this same formula, Thus, =a(Vi2). (2.5) “ds? = ds”, (2.6) and from the equality of the infinitesimal intervals there follows the equality of finite intervals: 5 = s’. Thus we arrive at a very important result: the interval between two events is the same in all inertial systems of reference, i.e. it is invariant under transformation from one inertial system to any other. This invariance is the mathematical expression of the constancy of the velocity of light. Again let x,y, Zit; and x2y2Z9f2 be the coordinates of two events in a certain reference system K. Does there exist a coordinate system K’, in which these two events occur at one and the same point in space? We introduce the notation + (2) = Mh. b-hahy — (2-x) + Or~y, ‘Then the interval between events in the K system is: sh =Cth — lp and in the K’ system si = en} — 12 whereupon, because of the invariance of intervals,6 ‘THE PRINCIPLE OF RELATIVITY §2 72, a Ope eed ot - = ne - We want the two events to occur at the same point in the K’ system, that is, we require {2 = 0. Then = 02, - 1, =c%/} >0. Consequently a system of reference with the required property exists if s2, > 0, that is, if the interval between the two events is a real number. Real intervals are said to be timelike. Thus, if the interval between two events is timelike, then there exists a system of reference in which the two events occur at one and the same place. The time which elapses between the two events in this system is z = T = “2. 27) te c If two events occur in one and the same body, then the interval between them is always timelike, for the distance which the body moves between the two events cannot be greater than cfjp, since the velocity of the body cannot exceed c. So we have always hz < cha. Let us now ask whether or not we can find a system of reference in which the two events occur at one and the same time. As before, we have for the K and k’ systems ¢71?, — 1, = c?#(? 1{2. We want to have f/f, = 0, so that sh =- 12 <0. Consequently the required system can be found only for the case when the interval sy) between the two events is an imaginary number. Imaginary intervals are said to be spacelike. Thus if the interval between two events is spacelike, there exists a reference system in which the two events occur simultaneously. The distance between the points where the events occur in this system is Ip = hh - 71 = isn. (2.8) The division of intervals into space- and timelike intervals is, because of their invariance, an absolute concept. This means that the timelike or spacelike character of an interval is independent of the reference system. Let us take some event O as our origin of time and space coordinates. In other words, in the four-dimensional system of coordinates, the axes of which are marked x, y, z, f, the world point of the event O is the origin of coordinates. Let us now consider what relation other events bear to the given event O. For visualization, we shall consider only one space dimension and the time, masking them on two axes (Fig. 2). Uniform rectilinear motion of a particle, passing through x = 0 at 1 = 0, is represented by a straight line going through O and inclined to the ¢ axis at an angle whose tangent is the velocity of the particle. Since the maximum possible velocity is c, there is a maximum angle which this line can subtend with the r axis. In Fig. 2 are shown the two lines representing the propagation of two signals (with the velocity of light) in opposite directions passing through the event O (i.e. going through x = (at f= 0). All lines representing the motion of particles can lie only in the regions aOc and dOb. On the lines ab and ed, x = + ct. First consider events whose world points lie Within the region aOc. It is easy to show that for all the points of this region c”” — 1? > 0.§3 PROPER TIME 7 In other words, the interval between any event in this region and the event O is timelike. In this region r > 0, i.e. all the events in this region occur “after” the event O. But two events which are separated by a timelike interval cannot occur simultaneously in any reference system. Consequently it is impossible to find a reference system in which any of the events in region aOc occurred “before” the event O, i.e. at time ¢ <0. Thus all the events in region 4a0c are future events relative to O in all reference systems. Therefore this region can be called the absolute future relative to O. t Absolute d past b Fic. 2 In exactly the same way, all events in the region bOd are in the absolute past relative to O; i.e. events in this region occur before the event O in all systems of reference. Next consider regions dOa and cOb. The interval between any event in this region and the event O is spacelike. These events occur at different points in space in every reference system, Therefore these regions can be said to be absolutely remote relative to O. However, the concepts “simultaneous”, “earlier”, and “later” are relative for these regions. For any event in these regions there exist systems of reference in which it occurs after the event O, systems in which it occurs earlier than O, and finally one reference system in which it occurs simultaneously with O. Note that if we consider all three space coordinates instead of just one, then instead of the two intersecting lines of Fig. 2 we would have a “cone” x? + y? + 2 ~ 2° = 0 in the four dimensional coordinate system x, y, zt, the axis of the cone coinciding with the ¢ axis. (This, cone is called the light cone.) The regions of absolute future and absolute past are then represented by the two interior portions of this cone ‘Two events can be related causally to each other only if the interval between them is timelike; this follows immediately from the fact that no interaction can propagate with a velocity greater than the velocity of light. As we have just seen, it is precisely for these events that the concepts “earlier” and “later” have an absolute significance, which is a necessary condition for the concepts of cause and effect to have meaning. § 3. Proper time Suppose that in a certain inertial reference system we observe clocks which are moving relative to us in an arbitrary manner. At each different moment of time this motion can be considered as uniform. Thus at each moment of time we can introduce a coordinate system8 ‘THE PRINCIPLE OF RELATIVITY $3 rigidly linked to the moving clocks, which with the clocks constitutes an inertial reference system. In the course of an infinitesimal time interval df (as read by a clock in our rest frame) the moving clocks go a distance dx? + dy? + dz”. Let us ask what time interval df’ is indicated for this period by the moving clocks. In a system of coordinates linked to the moving clocks, the latter are at rest, ie., dx’ = dy’ = dz’ = 0. Because of the invariance of intervals ds? = dt? — de — dy -d2 = cdt?, from which _ dx? + dy? + de? edt? But dx? + ay" + dz? =¥, dt where vis the velocity of the moving clocks; therefore a= Bea h-¥. GB.) a Voe Integrating this expression, we can obtain the time interval indicated by the moving clocks when the elapsed time according to a clock at rest is ty — ty: 4 aaai= far The time read by a clock moving with a given object is called the proper time for this object. Formulas (3.1) and (3.2) express the proper time in terms of the time for a system of reference from which the motion is observed. As we see from (3.1) or (3.2), the proper time of a moving object is always less than the corresponding interval in the rest system. In other words, moving clocks go more slowly than those at rest. Suppose some clocks are moving in uniform rectilinear motion relative to an inertial system K. A reference frame K” Jinked to the latter is also inertial. Then from the point of view of an observer in the K’ system the clocks in the K” system fall behind. And conversely, from the point of view of the K’ system, the clocks in K lag. To convince ourselves that there is no contradiction, let us note the following. In order to establish that the clocks in the K’ system lag behind those in the K system, we must proceed in the following fashion. Suppose that at a certain moment the clock in K” passes by the clock in K, and at that moment the teadings of the two clocks coincide. To compare the rates of the two clocks in K and K’ we must once more compare the readings of the same moving clock in K’ with the clocks in K. But now we compare this clock with different clocks in K—with those past which the clock ‘in K’ goes at ths new time. Then we find that the clock in K” lags behind the clocks in K with which it is being compared. We see that to compare the rates of clocks in two reference ye ez (3.2)84 ‘THE LORENT2 TRANSFORMATION 9 frames we require several clocks in one frame and one in the other, and that therefore this process is not symmetric with respect to the two systems. The clock that appears to lag is always the one which is being compared with different clocks in the other system. If we have two clocks, one of which describes a closed path returning to the starting point (the position of the clock which remained at rest), then clearly the moving clock appears to lag relative to the one at rest. The converse reasoning, in which the moving clock would be considered to be at rest (and vice versa) is now impossible, since the clock describing a closed trajectory does not carry out a uniform rectilinear motion, so that a coordinate system linked to it will not be inertial, Since the laws of nature are the same only for inertial reference frames, the frames linked to the clock at rest (inertial frame) and to the moving clock (non-inertial) have different properties, and the argument which leads to the result that the clock at rest must lag is not valid. The time interval read by a clock is equal to the integral ' Al ds, c taken along the world line of the clock. If the clock is at rest then its world line is clearly a line parallel to the t axis; if the clock carries out a nonuniform motion in a closed path and returns to its starting point, then its world ine will be a curve passing through the two points, on the straight world line of a clock at rest, corresponding to the beginning and end of the motion. On the other hand, we saw that the clock at rest always indicates a greater time interval than the moving one. Thus we arrive at the result that the integral 2 fa taken between a given pair of world points, has its maximum value if it is taken along the straight world line joining these two points.} § 4. The Lorentz transformation Our purpose is now to obtain the formula of transformation from one inertial reference system to another, that is, a formula by means of which, knowing the coordinates x, y, zt, of a certain event in the K system, we can find the coordinates x’, y’, z’, / of the same event in another inertial system K’. In classical mechanics this question is resolved very simply. Because of the absolute nature of time we there have ? = 1’; if, furthermore, the coordinate axes are chosen as usual (axes X, X’ coincident, Y, Z axes parallel to Y’, Z’, motion along X, X’) then the coordinates y,z clearly are equal to y’, z’, while the coordinates x and x’ differ by the distance traversed by one system relative to the other. If the time origin is chosen as the moment when the two coordinate systems coincide, and if the velocity of the K’ system relative to K is V, then this, distance is V1. Thus + It is assumed, of course, that the points and b and the curves joining them are such that all elements ds along the curves are timelike. ‘This property of the integral is connected with the pseudo-euclidean character of the four-dimensional ‘geometry, In euclidean space the integral would, of course, be a minimum along the straight line.