THE QUANTUM THEORY
 
 THE 
 QUANTUM THEORY 
 
 
 
 BY 
 
 FRITZ REICHE 
 
 PROFESSOR OF PHYSICS IN THE UNIVERSITY OF BRESLAU 
 
 TRANSLATED BY H. S. HATFIELD, B.Sc., Ph.D.. AND 
 HENRY L. BROSE. M.A. 
 
 WITH FIFTEEN DIAGRAMS 
 
 NEW YORK 
 
 E. P. BUTTON AND COMPANY 
 PUBLISHERS 
 
 66038
 
 PRINTED IN GREAT BRITAIN BY 
 THE ABERDEEN UNIVERSITY PRESS
 
 c 
 
 o 
 
 CONTENTS 
 
 ^ CHAP. PAGE 
 
 ^ INTRODUCTION 1 
 
 j, I. THE ORIGIN OF THE QUANTUM HYPOTHESIS .... 2 
 
 k\ II. THE FAILURE OF CLASSICAL STATISTICS .... 13 
 
 III. THE DEVELOPMENT AND THE RAMIFICATIONS OF THE QUANTUM 
 
 THEORY. * 16 
 
 IV. THE EXTENSION OF THE DOCTRINE OF QUANTA TO THE Mo- 
 
 ^ LECULAR THEORY OF SOLID BODIES .... 29 
 
 V. THE INTRUSION OF QUANTA INTO THE THEORY OF GASES . 68 
 
 VI. THE QUANTUM THEORY OF THE OPTICAL SERIES. THE DE- 
 VELOPMENT OF THE QUANTUM THEORY FOR SEVERAL 
 
 DEGREES OF FREEDOM 84 
 
 VII. THE QUANTUM THEORY OF RONTGEN SPECTRA . . . 109 
 
 VIII. PHENOMENA OF MOLECULAR MODELS i ..... 117 
 
 ) IX. THE FUTURE . . . ' 125 
 
 W MATHEMATICAL NOTES AND REFERENCES .... 127 
 
 A INDEX . . 181
 
 THE QUANTUM THEORY 
 
 INTRODUCTION 
 
 r ~r" v HE old saying that small causes give rise to great effects 
 J[ has been confirmed more than once in the history of 
 physics. For, very frequently, inconspicuous differences be- 
 tween theory and experiment (which did not, however, escape 
 the vigilant eye of the investigator) have become starting- 
 points of new and important researches. 
 
 Out of the well-known Michelson-Morley experiment, 
 which, in spite of the application of the most powerful 
 methods of exact optical measurement, failed to show an 
 influence of the earth's movement on the propagation of 
 light as was predicted by classical theory, there arose the 
 great structure of Einstein's Theory of Kelativity. In the 
 same way the trifling difference between the measured and 
 calculated values of black-body radiation gave rise to the 
 Quantum Theory which, formulated by Max Planck, was 
 destined to revolutionise in the course of time almost all 
 departments of physics. 
 
 The quantum theory is yet comparatively young. It is 
 therefore not surprising that we are confronted with an 
 unfinished theory still in process of development which, 
 changing constantly in many directions, must often destroy 
 what it has built up a short time before. But under such 
 circumstances as these, in which the theory is continually 
 deriving new nourishment from a fresh stream of ideas and 
 suggestions, there is a peculiar fascination in attempting to 
 review the life-history of the quantum theory to the present 
 time and in disclosing the kernel which will certainly out- 
 last changes of form. 
 1
 
 CHAPTEE I 
 The Origin of the Quantum Hypothesis 
 
 i. Black-Body Radiation and its Realisation in Practice 
 
 THE Quantum Theory first saw light in 1900. When, in 
 the years immediately preceding (1897-1899), Lummer 
 and Pringsheim made their fundamental measurements 1 of 
 black-body radiation at the ReichsanstaU, they could have 
 had no premonition that their careful experiments would 
 become the starting-point of a revolution such as has seldom 
 occurred in physics. 
 
 In the field of heat radiation chief interest at that time was 
 centred in the radiation of a black body (briefly called " black- 
 body " radiation), that is, of a body which absorbs completely 
 all radiation which falls on it and which thus reflects, trans- 
 mits, and scatters 2 none. We may shortly call to mind 
 the following facts. It is known that any body at a given 
 temperature sends out energy in the form of radiation into 
 the surrounding space. This radiation is not energy in a 
 single simple form but is made up of a number of single 
 radiations of different colours, i.e. of different wave-lengths A 
 or of different frequencies 3 v. In other words, it forms in 
 general a spectrum in which radiations of all frequencies 
 between v = and v = oo are represented. Further, these 
 radiations are present in varying " intensities." We define 
 this term thus. Consider the radiation emitted from unit 
 surface of the body per_second in a certain direction ; break 
 it up spectrally and cut out of the spectrum a small frequency 
 interval dv such that it contains all frequencies between v and 
 v + dv. The energy of radiation E v thus sliced out (namely, 
 the emissivity of the body for the frequency v) may be defined 
 in the following terms : * 
 
 E v = ^K v d v , . . (1) 
 2
 
 BLACK-BODY RADIATION 3 
 
 provided that as we shall assume for the sake of simplicity 
 the surface. of the body emits uniform and unpolarised 
 radiation in" all directions. 
 
 The magnitude K,, thus defined is called the intensity of 
 radiation of the body for the frequency v. It is in general a 
 more or less complicated function of the frequency v, of the 
 absolute temperature of the body T, and of the inherent 
 properties of the body. The black body alone is unique in 
 this respect. For its radiation and therefore its K v is, as 
 G. Kirchhoff* was the first to point out, dependent only 
 on the frequency .v and the absolute temperature T, that is, 
 mathematically, 
 
 K v =f(v,T) .... (2) 
 
 This formula which gives the relation between the intensity 
 of radiation from a black body, the temperature, and the 
 " colour " is called the radiation formula or the law of radia- 
 tion of a black body. 
 
 To calculate this- relationship on the one hand and to 
 measure it on the other were unsolved problems at that 
 time. Unimpeachable measurements were of course possible 
 only if one could succeed in constructing a black body which 
 approached sufficiently near the theoretical ideal. This im- 
 portant step, the realisation of the black body, was taken by 
 0. Lummer and W. Wien* on the basis of KirchhofTs 1 
 Law of Cavity Eadiation, which states : In an enclosure 
 or a cavity which is enclosed on all sides by reflecting walls, 
 externally protected from, exchanging heat with its surroundings, 
 and evacuated, the condition of " black radiation " is auto- 
 matically set up if all the emitting and absorbing bodies at the 
 walls or in the enclosure are at the same temperature. In a 
 space, therefore, which is hermetically surrounded by bodies 
 at the same temperature T and which is prevented from ex- 
 changing heat with its surroundings, every beam of radiation 
 is identical in quality and intensity with that which would be 
 emitted by a black body at the temperature T. 
 
 Lummer and Wien, therefore, had only to construct a 
 uniformly heated enclosure wi.th blackened walls having a 
 small opening. The radiation emitted from this opening was 
 then " black " to an approximation which was the closer
 
 4 THE QUANTUM THEORY 
 
 the smaller the opening, that is, the less the completeness of 
 the enclosure was disturbed. The manner in which the 
 intensity Kv of the black radiation thus realised depended 
 on the frequency v and the temperature T had next to be 
 determined. The above-mentioned investigation of Lummer 
 and Pringsheim was devoted to this purpose. 
 
 2. The Stefan-Boltzmann Law of Radiation and Wien's 
 Displacement Law 
 
 While experimental research was proceeding on its way, 
 theory was not idle, for valuable pioneer work was being 
 done inasmuch as two fundamental laws were set up. In 
 the first place, L. Boltzmann* proved, with the help of 
 thermodynamics, the law previously enunciated by Stefan, 9 
 that the sum-total of the radiation from a black body, 
 taking all the frequencies together, namely, the quantity 
 
 K = \ K. v dv, is proportional to the fourth power of its absolute 
 
 Jo 
 temperature : 10 
 
 K = y . T 4 (y = const.) . . . (3) 
 
 The laws proposed by Wien n entered more deeply into the 
 question. Wien imagines the black radiation enclosed in a 
 closed space with a perfectly reflecting piston as one wall, 
 and then supposes the radiation to be compressed adiabatically, 
 as in the case of gases (that is, no passage of heat to or from 
 the cavity is allowed during the process), by infinitely slow 
 movements of the piston. Now, if we express the change 
 which this process causes in the energy of a definite colour 
 interval dv in two ways, and if we take into consideration 
 that the waves reflected at the piston undergo a change of 
 colour according to Doppler's principle, we succeed in limiting 
 very considerably the unknown functional dependence of the 
 quantity K v on v and T. There is thus obtained a re- 
 lation of the form 12 
 
 in which c is the velocity of light in vaciio, the function F 
 being left undetermined. From this, Wien's Displacement 
 Law, the conclusion 13 may be drawn that the frequency
 
 WIEN'S LAW OF RADIATION 5 
 
 I'max for which 1C (plotted as a function of v) is a maximum 
 is displaced towards higher values proportional to T as the 
 temperature increases : 
 
 Vmax = COnst. . T . . . (4ft) 
 
 If, as is usual in physical measurement, we use the wave- 
 length A = - instead of the frequency as the variable, Wieris 
 
 Law assumes a somewhat different form. For if we consider 
 the radiant energy of a narrow range of wave-length d\ cor- 
 responding to the frequency range dv, and write it in the 
 
 form E^dX, then EidK = K^dv, that is : E^ - K*> . ^. In 
 
 A 
 
 place of (4) and (4a) we then get the relations : 
 
 >-max . T . const. = S . . (5a) 
 
 3. Wien's Law of Radiation 
 
 To formulate the law of radiation it was therefore neces- 
 sary only to evaluate the unknown function F in (4) or (5). 
 But this was just the central point of the whole question, 
 and the most difficult part of the problem. 
 
 Here, too, Wien made the first successful attack. On the 
 basis of not entirely unobjectionable calculations, which were 
 founded on Maxwell's law of distribution of velocities among 
 gas molecules, he arrived at the following specialised form 14 
 of the function F : 
 
 p a e -pf ( a an d ft are two constants). 
 
 Thus the law of radiation (4) assumes the form 
 
 K,a.-4 . . . . (6) 
 
 which is called Wien's Law of Kadiation. 
 
 How far did experiment confirm these theoretical results ? 
 While the Stefan-Boltzmann Law and Wien's Displacement 
 Law were confirmed to a large extent by the observations of 
 Lunimer and Pringsheim, both experimenters found Wien's
 
 6 THE QUANTUM THEORY 
 
 Law of Eadiation confirmed only for high frequencies, that is, 
 for short wave-lengths (more precisely, for large values of 
 
 m), and detected, on the other hand, systematic dis 
 
 for small frequencies, that is, for long wave-lengths. They 
 maintained with unswerving persistence that these discrepan- 
 cies were real in spite of objections from authoritative quarters. 
 For while F. Paschen 17 imagined that he had proved by his 
 work that Wien's Law of Radiation was universally valid, 
 Max Planck, in his detailed theory of irreversible processes 
 of radiation, 18 had arrived again at Wien's radiation formula 
 by a more rigorous method. Starting from Kirchhoff's Law 
 of Cavity Radiation, according to which the presence of any 
 emitting or radiating substance whatsoever in a uniformly 
 heated enclosure produces and ensures the maintenance of 
 the condition of black-body radiation, Planck chose as the 
 simplest schematic model of such a substance a system 
 of linear electromagnetic oscillators, and investigated the 
 equilibrium of the radiation set up between them and the 
 radiation of the enclosure. This is to be understood as fol- 
 lows : Each of the Planck oscillators as such we may, for 
 example, assume bound electrons capable of vibration pos- 
 sesses a fixed natural frequency v and responds, on account 
 of its weak damping, only to those waves of the radiation in 
 the enclosure whose frequencies lie in the immediate neigh- 
 bourhood of v, while all other waves pass over it without 
 effect. The oscillator thus acts selectively, as a resonator, in 
 just the same way as a tuning-fork of definite pitch com- 
 mences to sound only when its own " proper" tone, or one 
 very near it, is contained in the volume of sound which strikes 
 it. In this process of resonance, however, the oscillator ex- 
 changes energy with the radiation inasmuch as, on the one 
 hand, it acts as a resonator in abstracting energy from the 
 external radiation, and, on the other, it acts as an oscillator 
 and radiates energy by its own vibration. Hence a dynamic 
 equilibrium is set up between the oscillator and the radiation 
 of the enclosure, and, indeed, between just those waves of 
 the radiation which have the frequency v. In this state of 
 equilibrium the radiation of frequency v acquires an intensity 
 K t . which, according to Kirchhoff's Law, is equal to the intensity
 
 THE QUANTUM HYPOTHESIS 7 
 
 of black-body radiation at this temperature. Secondly, the 
 energy U of the oscillator passes in the course of time through 
 all possible values, the mean value 19 U of which is found to 
 be proportional to the intensity K v , a result which seems im- 
 mediately plausible since the excitation of the oscillator will 
 be greater the more intense the radiation that falls on it. 
 The exact calculation of this relationship between S v and U 
 on the basis of classical electrodynamics this is the first 
 part of Planck's calculations leads to the fundamental 
 formula : 
 
 *,=*? - - ' (7) 
 
 In the second part Planck * determined U, although by a 
 method that is not free from ambiguity, as a function of v and 
 T on the basis of the second law of thermodynamics. He 
 obtained 
 
 ff-*-4 .... (8) 
 
 The combination of (7) and (8) gives us Wien's Law of Radia- 
 tion (6). 
 
 4. The Quantum Hypothesis. Planck's Law of Radiation 
 
 Lummer and Pringsheim, however, refused to surrender. 
 In a fresh investigation 21 in 1900 they showed that in the 
 region of long waves Wien's radiation formula undoubtedly did 
 not agree with the results of observation. As a result of this, 
 Planck, in an important paper 22 which must be regarded as 
 marking the creation of the quantum theory, decided to 
 modify his method of deducing the law of radiation, namely, 
 by altering the expression (8) which gives the mean energy 
 of the oscillator, but which is not unique. He proceeded as 
 follows. 23 In order to distribute the whole available energy 
 among the oscillators, he imagined this energy divided into 
 a discrete number of finite " elements of energy " (energy 
 quanta) of magnitude e, and supposed these energy quanta 
 to be distributed at random among the individual oscillators 
 exactly as a given number of balls, say 5, may be distributed 
 among a certain number of boxes, say 3. Each such distri- 
 bution (of 5 balls among 3 boxes) may obviously be carried
 
 8 THE QUANTUM THEORY 
 
 out in a number of different ways, whereby, however, we are 
 not concerned with which particular balls lie in which par- 
 ticular boxes, but with the number contained in each. 2 * Now 
 since each such " distribution " corresponds to a definite state 
 of the system, it follows from what has just been said that 
 each condition may be realised in a number of different 
 ways, that is, each condition is characterised by a certain 
 number of possibilities of realisation. This number is called 
 by Planck the thermodynamic probability W of the condition 
 in question. For it is obvious that the probability of a con- 
 dition or state is the greater, i.e. it will occur the more fre- 
 quently, the greater the number of ways in which it may 
 be realised. By means of the usual formulae of permuta- 
 tions and combinations, of which the latter alone come into 
 consideration here, it was possible to calculate the probability 
 of any given distribution of the elements of energy among 
 the oscillators, and thus also the probability of a given 
 energetic condition of the system of oscillators as a func- 
 tion of the mean energy U of an oscillator and of the energy 
 quantum. Now, L. BoUzmann has given an extremely 
 fertile rule, which connects the probability of state W of a 
 system with its entropy S, a magnitude which, as is well 
 known, plays a similar role in the second law of thermodyna- 
 mics to that played by energy in the first. Thus S was ob- 
 tained as a function of U and c. If now, on the other hand, 
 one applied the second law itself, which expresses the en- 
 tropy S as a function of the mean energy U and the absolute 
 temperature T, the following result was obtained by this cir- 
 cuitous process : the entropy, as an auxiliary magnitude, was 
 eliminated, and a relation between U, T, and e was gained. 
 This fundamental result, first obtained by Planck, is as 
 follows : 
 
 U = _! - (k being a constant) . . (9) 
 
 But from (7) and TPtenVDisplacement Law (4) it follows 
 that for the mean energy U of an oscillator, a relationship of 
 the following form exists :
 
 THE QUANTUM HYPOTHESIS 9 
 
 A comparison of (9) and (10) shows that U assumes the 
 form required by (10) only when e is set proportional to v, the 
 frequency. This is an essential point of Planck's Theory : if 
 we are to remain in agreement with Wien's Displacement Law, 
 the energy element c must be set equal to hv 
 
 t = hv . . . . (11) 
 
 The constant h, which, on account of its dimensions (energy 
 x time), is called Planck's Quantum of Action, has played, 
 as we shall see, a r61e of undreamed-of importance in the 
 further development of the quantum theory. 
 
 By combining the formulae (7), (9), and (11) the renowned 
 radiation law of Planck follows at once : 
 
 - (12) 
 - 1 
 
 which Planck first deduced in the year 1900 in the manner 
 above described, that is, by the hypothesis of energy quanta. 
 In the same year as well as in the following year this Law of 
 Kadiation was confirmed very satisfactorily by H. Rubens and 
 F. Kurlbaum 2 for long waves, and by F. Paschen Z1 for short 
 waves. The later measurements of radiation emitted by 
 black bodies,28 particularly the exact work carried out by E, 
 Warburg and his collaborators at the Eeichsanstalt, have also 
 demonstrated the validity of Planck's formula. In opposition 
 to this, W. Nernst and Th. Wulf& as the result of a critical 
 review of the whole experimental material available up to that 
 date, have recently shown the existence of deviations (up to 
 7 per cent) between the measured and the calculated values 
 according to Planck's formula, and hence feel themselves 
 constrained to decide against the exact validity of Planck's 
 formula. Whatever view is taken of this criticism, it is at any 
 rate a powerful incentive to take up anew the measurement 
 of the radiation emitted by black bodies with all the finesse 
 and precautions of modern experimental science, and thereby 
 to decide finally the important question whether Planck's 
 Law is exactly valid or not. 
 
 For short wave-lengths, i.e. high frequencies (more exactly,
 
 10 THE QUANTUM THEORY 
 
 for high values of =-^\, Planck's formula assumes the form 
 
 and thus passes over into Wien's Law (cf. formula (6), 
 which, as we have seen, was confirmed by experiment for 
 these frequencies). In the other limiting case, i.e. for long 
 
 waves, low frequencies (more exactly for small values of ?-=, 
 Planck's formula assumes the form 
 
 *.- .... (H) 
 
 as is easily found by developing the exponential function 
 
 efk as a series. This limiting law, which has been confirmed 
 in the region of long wave-lengths, had been given pre- 
 viously by Lord Rayleigh. 30 Planck's formula thus contains 
 Wien's Law and Eayleigh's Law as limiting cases. 
 
 If we use the wave-length X instead of the frequency v, 
 Planck's Law takes the form 
 
 . . . (15) 
 
 To make this clear, the intensity of radiation E^ is plotted 
 in Pig. 1 as a function of A. for various values of T. The 
 curves which exhibit K. v as a function of v have a quite 
 similar appearance. The maximum of the ^-curves lies at 
 
 the point at which _A has the value 4-9651. 
 It follows that 
 
 Xinax ' T = TaSr-T = 6>042 x 1 9 ? = 8 (16) 
 
 4'yoOl . K K 
 
 a relation, which is identical in form with Wien's Displace- 
 ment Law (5a). 
 
 For the total radiation we get from (12) or (15)
 
 CONSEQUENCES OF PLANCK'S THEORY 11 
 
 an equation which gives expression to the Stefan- Boltzmann 
 
 Law 31 (3). 
 
 From (16) and (17) we recognise that the measurement (a) 
 of the total radiation (K) and (6) of the 
 wave-length of the maximum (A. max ), at 
 a fixed known temperature, allows us to 
 calculate the two constants h and k of the 
 radiation formula. 32 From Kurlbaum's 
 measurements of the Stefan-Boltzmann 
 constant y, which were available at that 
 time, and from the constant 8 of Wien's 
 Displacement Law (measured by Lummer 
 and Pringsheim) Planck 33 found the fol- 
 lowing values : 
 
 h = 6-548 x 10- 27 [erg.sec.] 
 
 k = 1-346. 10 -"fl^l 
 Ldeg.J 
 
 (18) 
 
 Corresponding to the varying values 
 which have been found in the course of 
 time for the constants y and 8, the values 
 h and k have undergone changes which 
 are not worth while 
 recording here. 
 For particularly 
 
 _/l the measurement of 
 
 p IG i_ the total radiation 
 
 as we see from 
 
 the strongly varying values given in note 15 has not yet 
 reached a sufficient degree of certainty, to allow a very ac- 
 curate calculation of the two radiation constants h and k to be 
 based on the Stefan-Boltzmann constant. Methods which 
 allow h to be determined with undoubtedly much greater 
 accuracy will be described later. 
 
 5. Consequences of Planck's Theory 
 
 The deduction of the radiation formula and the determina- 
 tion of its constants did not, however, exhaust the successes 
 of Planck's new theory ; on the contrary, important relation- 
 ships of this theory to other departments of physics became
 
 12 THE QUANTUM THEORY 
 
 immediately revealed. For it was found 3* that the constant 
 k of the radiation formula is nothing other than the quotient 
 of the absolute gas constant B (which appears in the equa- 
 tion of state of an ideal gas) and the so-called Avogadro 
 number N, i.e. the number of molecules in a grammolecule. 
 
 *-* '. . . . (19) 
 
 As the value of R is sufficiently accurately known from 
 thermodynamics 
 
 Planck,** by making use of the radiation measurements, was 
 able to calculate the value of N. By using (18) he found 
 
 N = 6-175 x 10 23 . \. . (20) 
 
 The agreement of this value with the values deduced by 
 quite different methods is very striking. 86 Avogadro s Law 
 forms the bridge to the electron theory. For it is known 
 that the electric charge which travels in electrolysis with 
 1 gramme-ion, that is, with N-ions, is a fundamental con- 
 stant of nature, which is called the Faraday. Its value was, 
 according to the position of measurements at that time, 
 9658 . 3 . 10 10 electrostatic units (the value nowadays ac- 
 cepted 37 is 9649-4 . 2-999 . 10 10 ). If now each monovalent- 
 ion carries the charge e, of the electron, the equation 
 
 Ne = 9658 . 3 . 10 10 . . . (21) 
 must hold. From this, by using (20), we get 
 
 e = 4-69 x 10- 10 electrostatic units .. (22) 
 
 The value of the electron charge thus calculated by Planck 
 from the theory of radiation differs only by about 2 per cent 
 from the latest and most exact measurements of R. A. 
 Millikan** who found the value 
 
 e = 4-774 . 10 - 10 electrostatic units. . (23) 
 A truly astonishing result.
 
 CHAPTEE II 
 The Failure of Classical Statistics 
 
 I. The Equipartition Law and Rayleigh's Law of Radiation 
 
 IF these great successes had justified faith in Planck's 
 Theory, it was also soon recognised as had already been 
 emphasised by Planck in his first papers that the central 
 point of the theory lay in the Quantum Hypothesis, i.e. in the 
 novel and repulsive conception, that the energy of the oscilla- 
 tors of natural period v was not a continuously variable 
 magnitude, but always an integral multiple of the element of 
 energy, that is e = hv. The recognition of the necessity of 
 this hypothesis has forced itself upon us more and more in the 
 course of time, and has become established, more especially 
 through indirect evidence, inasmuch as every attempt to work 
 with the classical theory has led logically to a false law of 
 radiation. For when Planck turned the radiation problem 
 into a problem of probability for a definite amount of energy 
 was to be divided among the oscillators according to chance, 
 and the mean value U of the energy of an oscillator was to 
 be calculated it became possible to apply the methods of 
 the statistical mechanics founded by Clerk Maxwell, L. 
 Boltzmann, and Willard Gibbs. And the application of these 
 methods to the case in question appeared to be demanded 
 from the start, if the standpoint, self-evident in classical 
 physics, that the energy of the oscillator could assume in 
 continuous sequence all values between and CD were 
 adopted. What, then, did statistical mechanics require? 
 One of its chief laws is the law of the equipartition of kinetic 
 energy, * according to which in a state of statistical equilibrium 
 at absolute temperature T every degree of freedom of a mechan- 
 ical system, however complicated, possesses the mean kinetic 
 13
 
 14 THE QUANTUM THEORY 
 
 energy ^TcT. In this expression the constant k is defined by 
 (19), and is thus the same constant as that which appears 
 in the Law of Eadiation. A system of/ degrees of freedom, 
 therefore, possesses at a temperature Ta mean kinetic energy 
 / . $kT. For example, the atom of a monatomic gas is a 
 configuration which possesses three degrees of freedom, if we 
 regard it from the point of view of mechanics as a mass- 
 point. Its kinetic energy at the temperature T has therefore 
 a mean value* %kT, independent of its mass, a result which 
 has been known in the kinetic theory of gases since the time 
 of Maxwell, and which is deduced as a consequence of his 
 law of distribution of velocities. 
 
 Planck's linear oscillator, which is essentially identical 
 with an electron vibrating in a straight line, possesses one 
 degree of freedom ; its kinetic energy at the temperature T 
 has therefore the mean value -^kT. Now the mean potential 
 energy of the oscillator is equal to its mean kinetic energy.* 1 
 As a result, its mean total energy (kinetic plus potential) has 
 the value 
 
 C7= kT . . . . (24) 
 
 This result of classical statistics, when combined with the 
 relation (7) deduced from classical electrodynamics, gives 
 Rayleigh's Law of Radiation 
 
 K v = ^kT . . ... (25) 
 
 which, as we saw (cf. (14)), is contained in Planck's Law of 
 Radiation as a limiting case for small values of pL that is, 
 
 for long waves or high temperatures. 
 
 This Law of Radiation of Bayleigh which, deduced as it 
 is from the fundamental principles of classical statistics and 
 electrodynamics, should be able to claim general validity for 
 all frequencies and all temperatures, stands none the less in 
 glaring contradiction to observation. For while all observed 
 curves of distribution of energy of a black body (i.e. K v plotted 
 as a function of v, T being constant) always show a maximum, 
 the curve expressed by (25) rises without limit for rising 
 values of v, and therefore gives for the sum K 2 / Kvdv an 
 infinitely large value.
 
 FRUITLESS ATTEMPTS AT IMPROVEMENT 15 
 
 2. Fruitless Attempts at Improvement 
 
 From very different quarters and in the most varied ways 
 attempts were made, as time went on, to escape from 
 Rayleigh's Law without discarding classical statistical 
 mechanics. All in vain. Thus 7. H. Jeans,* 2 without 
 making use of a "material" oscillator, considered only the 
 radiation as such in an enclosure, and distributed the whole 
 energy of radiation according to the Law of Equipartition 
 over the individual " degrees of freedom of radiation " (which 
 are here the individual vibrations that are possible in an en- 
 closure). Further, H. A. Lorentz** deduced in a penetrating 
 investigation the thermal radiation of the metals, starting from 
 the conception that the free " conduction electrons," which 
 carry the current, produce the radiation by their collisions 
 with the atoms, and applying the Law of Equipartition to 
 the motion of these electrons. The problem was attacked 
 in a somewhat different fashion by A. Einstein and L. Hop/.** 
 They imagined the Planck oscillator firmly attached to a 
 molecule, and then considered this complex exposed to the 
 radiation and the impacts of other molecules. The Law of 
 Radiation could then be deduced from the condition that the 
 impulse, which the impacts of the molecules give to the com- 
 plex, must not on the average be changed by the impulses, 
 which the radiation gives to the oscillator. We may also 
 mention a paper of A. D. Fokker** which was supplemented 
 by M. Planck.* 6 In this, by the aid of a general law due to 
 Einstein, the statistical equilibrium between the radiation 
 and a large number of oscillators was examined on the basis 
 of the classical theories. All these different ways ended, 
 however, at the same point ; they all led to Rayleigh's Law. 
 And finally, at the Solvay Congress in Brussels in 1911, 
 H. A. Lorentz showed, in the most general manner 
 imaginable, that we arrive of necessity at this wrong law, 
 if we assume the validity of Hamilton's Principle and of 
 the Principle of Equipartition for the totality of the pheno- 
 mena (of mechanical and electromagnetic nature) which 
 take place in an enclosure containing radiation, matter, and 
 electrons. Only in the limiting case of high temperatures or 
 small frequencies do the results of the classical theory agree 
 with the results of observation.* 8
 
 CHAPTEE III 
 
 The Development and the Ramifications of the 
 Quantum Theory 
 
 i. The Absorption and Emission of Quanta 
 
 A S stated above, the conviction was bound to establish 
 Xl^tself that every attempt to deduce the laws of radiation 
 on the basis of classical statistics and electrodynamics was 
 doomed from the outset to failure, and it was necessary to 
 introduce a hitherto unknown discontinuity into the theory. 
 It was, of course, clear that this " atomising of energy " would 
 conflict sharply with existing and apparently well-founded 
 theories. For if the energy of the Planck oscillator was only 
 to amount to integral multiples of e = hv, and therefore was 
 only to be able to have the values 0, e, 2e, 3e . . . then, since 
 the oscillator only changes its energy by emission and ab- 
 sorption, the conclusion was inevitable that oscillators cannot 
 absorb and emit amounts of energy of any magnitude but only 
 whole multiples of e. (Quantum emission and quantum 
 absorption.} This conclusion is in absolute contradiction to 
 classical electrodynamics. For, according to the electron 
 theory, an electromagnetic oscillator, for instance a vibrat- 
 ing electron, emits and absorbs in a field of radiation perfectly 
 continuously, that is to say, in sufficiently short times it emits 
 or absorbs indefinitely small amounts of energy. 
 
 2. Einstein's Light-quanta ; Phenomena of Fluctuation in a Field 
 of Radiation 
 
 Thus at the very entrance into the new country there 
 yawned a gulf, which had either, in view of the previous 
 success of the classical theory, to be bridged over by a com- 
 promise ; or, failing this, tradition would have to be discarded 
 and the gap would be relentlessly enlarged. Einstein felt him- 
 16
 
 EINSTEIN'S LIGHT-QUANTA 17 
 
 self compelled to take the latter radical course. On the basis 
 of very original considerations, 49 he set up the hypothesis that 
 the energy quanta not only played a part, as Planck held, in 
 the interaction between radiation and matter (resonators or 
 oscillators), but that radiation, when propagated through a 
 vacuum or any medium, possesses a quantum-like structure 
 (Light- quantum hypothesis). Accordingly, all radiation was 
 to consist of indivisible " radiation quanta " ; when the energy 
 is being propagated from the exciting centre, it is not divided 
 evenly in the form of spherical waves over ever-increasing 
 volumes of space, but remains concentrated in a finite numbe 
 of energy complexes, which move like material structures, 
 and can only be emitted and absorbed as whole individuals. 
 Einstein believed himself forced to this strange conception, 
 which breaks with all the observations that appear to 
 support the undulatory theory, by several investigations, 
 all of which led to the same conclusion. He was per- 
 suaded to this view by the result of calculations dealing 
 with certain phenomena of fluctuation, phenomena which 
 are familiar to us in statistics and particularly in the kinetic 
 theory of gases. It is well known that in a gas which 
 contains n molecules in a volume v , the spatial distribution 
 of these molecules is far from constant, being subject to vari- 
 ation on account of the motion of the molecules. Indeed, in 
 principle, extreme cases are possible as that, for example, in 
 which all n molecules are collected at a given moment in a 
 fractional part v(<v ) of the volume. The probability of 
 this rare constellation is known to be 
 
 . . . . (26) 
 
 an extraordinarily small number when n is great ; that is to 
 say, the event in question occurs extremely rarely. 
 
 Now, the spatial density of the radiation enclosed within a 
 volume v (} is subject to quite analogous variations. If E is 
 the total energy of the radiation (supposed to be monochro- 
 matic) and if its frequency v is so great, or its temperature 
 so low, that Wien's Law of Radiation holds for it, then the 
 probability that the whole radiation occupies the partial 
 volume -u V Q ) is, according to Einstein,*
 
 18 THE QUANTUM THEORY 
 
 io-^\C . (27) 
 
 A comparison with (26) shows that the radiation, within 
 the limits of validity of Wien's Law, behaves as if it were made 
 
 wp of n ( = r- j independent complexes of energy, each of mag- 
 nitude hv. 
 
 Two other investigations si of Einstein led to the same 
 conclusion. In the first, a very large volume filled with 
 black-body radiation is considered, which communicates with 
 a small volume v. If E is the momentary energy of the 
 radiation of frequency v in the volume v, this energy varies, 
 as is known, irregularly with the time about a mean value 
 E ; the magnitude e = E - E is called the fluctuation of the 
 energy. Now, the general theory of statistics leads to the 
 following value fl2 for the mean square, that is, for e 2 , 
 
 P-fcT'.g;. . . .- (28) 
 
 If we replace E by the value obtained from Planck's Law of 
 Eadiation, we obtain for the mean square of fluctuation an 
 expression with two terms, 93 in which only one term can be 
 calculated on the basis of the classical undulatory theory; 
 the second, which greatly exceeds the first in magnitude 
 when the density of radiant energy is low (that is, at high 
 frequencies or at low temperatures, in short, when Wien's 
 Law is valid), can only be understood when we again picture 
 the radiation as composed of indivisible energy-quanta. 
 
 The second of Einstein's two investigations, to which we 
 referred above, deals with the fluctuations of impulse which 
 a freely movable reflecting plate is subjected to in a field of 
 black-body radiation on account of the irregular fluctuations 
 of the pressure of radiation. If, in addition, the plate is sub- 
 jected to the irregular blows of gas-molecules, under the 
 influence of which it executes Brownian movements, there 
 must be equilibrium between the impulses which the mole- 
 cules on the one hand, and the radiation on the other, im- 
 part to the plate. If, now, we assume Planck's Law to hold 
 for the radiation, there again follows for the mean square of 
 the variations in impulse due to the radiation an expression
 
 TRANSFORMATION OF LIGHT-QUANTA 19 
 
 in two terms, only one of which is explained by the un- 
 dulatory theory of light. The other term points to a 
 quantum-like structure of the radiation, and this suggests the 
 introduction of the light-quantum hypothesis. 
 
 3. Transformation of Light-quanta into other Light-quanta or 
 Electronic Energy 
 
 However strange this hypothesis appeared, it was not to be 
 denied that it was capable of explaining simply and naturally 
 a number of phenomena which completely baffled the un- 
 dulatory theory. A very striking example of this is afforded 
 by the laws of phosphorescence, investigated by P. Lenard 
 and his co-workers, and especially by Stokes' Law. For if 
 v p is the frequency of the phosphorescent light emitted, 
 and v e the frequency of the light exciting phosphorescence, 
 then, according to Einstein's conception, 34 one quantum hv e 
 of the exciting radiation is changed through absorption by 
 the atom of the phosphorescent substance into one quantum 
 hv p of the light of phosphorescence. According to the prin- 
 ciple of energy, we must have hv e !> hv p , i.e. v e > v p . And 
 this is Stokes' Law. 
 
 Further, another fact in the realm of phosphorescence 
 phenomena speaks against the undulation hypothesis and in 
 favour of that of light-quanta. According to the classical 
 undulatory theory, all molecules of a phosphorescent body 
 on which a light- wave impinges, should absorb energy from 
 the wave, and thus all simultaneously become able to emit 
 phosphorescent light. In reality, relatively only very few 
 molecules are excited to phosphorescence at the same time, 
 and only gradually, in the course of time, does the number of 
 molecules excited increase. It would thus appear as if the 
 light-wave falling on the phosphorescent body has not equal 
 intensity along its whole front as the classical theory 
 assumes but rather as if it consists of single energy-com- 
 plexes thrown out by the source of light, so that the wave-point 
 possesses, as it were, a "beady" structure, in which active 
 portions (light-quanta) alternate with inactive gaps. 
 
 This conception of the " beady " wave-front had played a 
 part before the advent of Einstein's hypothesis of light-quanta. 
 J". /. Thomson 8B had tried to make use of it to explain the
 
 20 THE QUANTUM THEORY 
 
 fact that, when a gas is ionised by ultra-violet light or Kontgen 
 rays, only a relatively extremely small number of gas-mole- 
 cules are ionised. This is a phenomenon which is quite 
 analogous to the above-named phenomenon of phosphor- 
 escence ; for these, too, according to Lenard's view, the exci- 
 tation consists in the disjunction, through the agency of the 
 radiation, of electrons from the molecules of the phosphor- 
 escent body, and these electrons attach themselves to " storage 
 atoms." On the return of these electrons to the parent 
 molecules, energy is set free and sent out as phosphorescent 
 light. The ionisation of gases by ultra-violet light or Ront- 
 gen rays M is also capable of being explained naturally by the 
 light-quantum hypothesis. If we suppose with Einstein, 
 that one light-quantum hv is used up in ionising one mole- 
 cule, then hv > /, where / is the work required to ionise 
 one molecule, that is to say, to remove an electron from it. 
 We have under consideration here a phenomenon which be- 
 longs to the great branch of photo-electric phenomena,* 1 i.e. 
 the liberation of electrons from gases, metals, and other sub- 
 stances by the action of light. According to the hypothesis 
 of light-quanta, in all these processes light-quanta are changed 
 into kinetic energy of the electrons hurled off from the body. 
 If we again adopt Einstein 's standpoint, according to which 
 one light-quantum hv is transformed into the kinetic energy 
 of one projected electron, we must have the following re- 
 lation M for the energy of emission of the emitted electrons, 
 each having a mass ra : 
 
 fyn V 2 = hv - P . . . . (29) 
 
 This is called Einstein's Law of the Photo-electric Effect. In 
 this, P is the work that has to be done to tear the electron 
 away from the atom, and to project it from the point at 
 which it is torn from the atom up to the point at which it 
 leaves the surface of the body. For the energy of the emitted 
 electrons we thus obtain a linear increase with the periodicity 
 of the light which releases them. This law, which many in- 
 vestigators have attempted to prove, with varying success, 
 has recently been verified by B. A. Millikan* 9 for the normal 
 photo-electric effect w of the metals Na and Li with such a 
 degree of accuracy that we can actually use this method for
 
 ELECTRONIC ENERGY 21 
 
 the exact determination of h. The value found by Millikan, 
 h = 6-57 x 10 ~ 27 , is in good agreement with the value 
 h = 6'548 x 10' 27 found by Planck from radiation measure- 
 ments. 
 
 In an entirely similar manner as was used for the 
 phenomena of phosphorescence, the phenomena of fluores- 
 cence in the regions of the Rontgen and visible radiations may 
 be explained by the hypothesis of light-quanta. The in- 
 vestigations of Ch. Barkla, Sadler, M. de Broglie, and 
 E. Wagner 61 have shown the following : if a body is inun- 
 dated with Eontgen rays, and if the absorption of these rays by 
 the body is measured whilst the hardness (i.e. the frequency 
 v e ) of the rays is varied, the absorption, as we pass from 
 lower to higher v e , suddenly increases to a high value for a 
 certain value of v e . At the same moment the body begins, 
 at the expense of the energy absorbed, to emit a secondary 
 Rontgen radiation characteristic of the body itself in the form 
 of a line spectrum. It further appears that all lines emitted 
 have a lower v than that of the exciting radiation. As a 
 matter of fact, the hypothesis of light-quanta requires that the 
 radiation-quantum hv of all rays emitted as secondary radia- 
 tion should be smaller than the quantum hv of the primary 
 exciting rays. For example, the region of frequencies which 
 serves to excite the " ^-series " stretches from a sharply 
 defined limit v k (the so called " edge of the absorption band ") 
 upwards towards higher frequencies; whereby v k is some- 
 what larger than the hardest known line (y) of the ^-series. 
 In other words, the excitation of secondary Rontgen radiation 
 by primary Rontgen rays also obeys Stokes' Law. 
 
 4. The Transformation of Electronic Energy into Light-quanta 
 
 It is very significant, that the transformation of light- 
 quanta into kinetic enei'gy of electrons is also, as it were, 
 "reversible," that is, the opposite process also occurs in 
 nature, by which light-quanta result from the kinetic energy 
 of charged particles. A good example of processes of this 
 kind is afforded by the generation of Rontgen rays by the 
 impact of quickly-moving electrons (cathode rays) on matter. 
 If, say, the characteristic X"-series of a certain element is to
 
 22 THE QUANTUM THEORY 
 
 be generated by the impact of cathode rays upon an anti- 
 cathode formed of the said element, then the kinetic energy 
 E of an impinging electron must exceed a critical value E K . 
 For if we imagine E changed into a light-quantum hv e , then 
 v e must fall within the region of excitation of the JT-series, and 
 must thus be ^ V K (V K being the frequency of the edge of the 
 absorption band). It follows that E ^> hv K ( = E K ). From this 
 there follows an important relation between the frequency V K 
 of the edge of the absorption band and the critical value EK 
 of the electronic energy, i.e. the smallest value of the energy 
 at which the electron is just able to generate the required 
 secondary radiation. This quantum-relation E K = hv K has 
 proved quite correct according to measurements carried out 
 by D. L. Webster & and E. Wagner w and conversely presents, 
 when E K and V K are sufficiently accurately known, a method 
 for the determination of /i. 6 * 
 
 Now, it is known that the cathode rays, on striking the 
 anti-cathode, do not merely excite the characteristic Eontgen 
 radiation, that is a line spectrum, but excite a continuous 
 spectrum as well, the so-called " impulse radiation " (Brems- 
 strahlung). If we therefore select any frequency v of this 
 continuous spectrum, the ideas of the hypothesis of light- 
 quanta immediately suggest the conclusion that a definite 
 minimum energy E m of the impinging electrons is necessary 
 to excite this frequency v, and that we must have E m = hv. 
 The investigations of D. L. Webster & W. Duane and F. L. 
 Hunt**, A. W. Hull and M. Bice^E. Wagner, M F. Dessauer 
 and E. Back 66 have confirmed these formulae with the 
 greatest accuracy, and thus form the foundation of one of 
 the most trustworthy methods for the precise measurement 
 of the magnitude h. The following values were obtained : 
 h = 6-50 x 10-27 (Duane-Hunt) ; h = 6-53 x 10 - 2 ~ (Webster) ; 
 h = 6-49 x 10 ~ 27 (Wagner). 
 
 We also meet with similar phenomena in the visible and 
 neighbouring regions of the spectrum. Thus /. Franck 
 and Cr. Hertz m showed that the impact of electrons upon 
 mercury vapour molecules can be used to excite a definite 
 characteristic fluorescence line of mercury of wave-length 
 A = 25364 (i.e. v = 1-183 . 10 16 ), if the kinetic energy of the
 
 HYPOTHESIS OF LIGHT-QUANTA 28 
 
 electron exceeds a certain critical value E Q . In this con- 
 nexion they found that the relation E = hv was again 
 fulfilled with great accuracy. 70 We shall return to these 
 experiments and others connected with them later, since they 
 play an important part in confirming the most recent model 
 of the atom. 
 
 5. Other Applications of the Hypothesis of Light-quanta 
 
 In a considerable number of other cases, which shall only 
 be noticed shortly at this point, the hypothesis of light-quanta 
 has proved of value, especially in the hands of /. Stark 71 and 
 Einstein. Thus Stark ra has made use of this hypothesis to 
 interpret the fact that the canal-ray particles emit their 
 " kinetic radiation " only when their speed exceeds a certain 
 value. He has also propounded general laws for the position 
 of band-spectra of chemical compounds by arguing on the basis 
 of the hypothesis of light-quanta. 78 Finally, Einstein 7 * and 
 Stark 78 have considered photo-chemical reactions from the 
 standpoint of the hypothesis of light-quanta and have enun- 
 ciated a fundamental law, which has been verified, at least 
 partially, by the detailed investigations of E. Warburg. 
 
 6. Planck's Second Theory 
 
 In spite of all the successes which the quantum hypothesis 
 of light is able to show, we must not leave out of consideration 
 that this radical view, at least in its existing form, is very 
 difficult to bring into agreement with the classical undulatory 
 theory. Since on the one hand the phenomena of interference 
 and diffraction, in all their observed minutiae, are excellently 
 described by the wave-theory, but offer almost insuperable 
 difficulties to the quantum theory of light, it is easy to under- 
 stand that few scientists could make up their minds to ap- 
 prove of such a far-reaching change in the old and well-tested 
 conception of the propagation of light, a change that entailed 
 perhaps its complete abandonment. This more cautious and 
 conservative standpoint was taken up by M. Planck, who 
 retains it to this day, inasmuch as he preferred to locate the 
 quantum property in matter (the oscillators) or at least to 
 confine it to the process of interaction between matter and
 
 24 THE QUANTUM THEORY 
 
 radiation while endeavouring to retain the classical wave- 
 theory for the propagation of radiation in space. None the 
 less, serious hindrances had already intruded themselves in 
 the development of his first quantum hypothesis (quantum 
 emission and quantum absorption). For H. A. Lorentz 1 ? 1 
 pointed out quite rightly that the conception, especially of 
 quantum absorption, leads to peculiar difficulties. He showed 
 that the time which an oscillator requires for the absorption 
 of a quantum of energy turns out to belong to an improbable 
 degree when the external field of radiation is sufficiently weak. 
 Moreover, it would be possible to interrupt the radiation at 
 will before the oscillator had absorbed a whole quantum. As 
 a result of these objections Planck determined to modify the 
 quantum hypothesis as follows. 78 Absorption proceeds con- 
 tinuously and according to the laws of classical electrodynamics : 
 the energy of the oscillators is therefore continuously variable, and 
 can assume any value between and oo . On the other hand, 
 emission occurs in quanta, and the oscillator can emit only 
 when its energy amounts to just a whole multiple of = hv. 
 Whether it then emits or not is determined by a law of prob- 
 ability. But if it does emit, then it loses its whole momentary 
 energy, and therefore emits quanta. Between two emissions its 
 energy -content grows by absorption continuously and in pro- 
 portion to the time. 
 
 According to this second theory of Planck, which is called 
 the theory of quantum emission, the mean energy U of a 
 
 linear oscillator is ~ greater than in the first theory. 79 While 
 
 in the former case the mean energy of the oscillator at abso- 
 lute zero was equal to zero (see equation (9) from which, 
 when T = 0, U 0), in the case of this second theory it is 
 
 equal to . The oscillators retain therefore at the zero- 
 point a zero-point energy of value - as a mean, inasmuch 
 
 2 
 
 as they assume, when T = 0, all possible energies between 
 and hv. Nevertheless, this theory also, when the relation 
 (7) is correspondingly modified, leads to Planck's Law of 
 Kadiation. 
 
 In the course of time Planck has made several further
 
 ZERO-POINT ENERGY 25 
 
 attempts 80 to enlarge and modify this second theory too. For 
 example, he has temporarily assumed the emission also to be 
 continuous, and relegated the quantum element to the excita- 
 tion of the oscillators by molecular or electronic impacts. He 
 has, however, repeatedly returned in essentials to the second 
 form of his theory (continuous absorption, quantum emission). 
 
 7. Zero-point Energy 
 
 In more than one direction, this theory has had further 
 results. The appearance of the mean zero-point energy, 
 which is peculiar to this second theory of Planck, became 
 the starting-point of a series of researches, in which certain 
 physicists, going beyond Planck, postulated the existence of 
 a true (not mean] zero-point energy equal for all oscillators. 
 On this basis, Einstein and 0. Stern 81 have given a deduction 
 of Planck's Law which avoids all discontinuities other than 
 the existence of this zero-point energy. 
 
 In the year 1916, Nernst 82 took a still more radical step in 
 postulating the existence of a " zero-point radiation " which 
 was also to be present at the absolute zero of temperature 
 and was to exist independently of heat radiation, filling the 
 whole of space, and such that the oscillators, as well as all 
 molecular structures, set themselves in equilibrium with it by 
 taking up the zero-point energy. Even if we regard these 
 views more or less sceptically, one thing cannot be ignored : 
 many facts undoubtedly support the conception that at the 
 absolute zero by no means all motion has ceased. We need 
 only draw attention to the fact, that, according to the view of 
 F. Richarz*z P. Langevin& and according to the experiments 
 of Einstein, W. J. de Haas * and E. Beck** Para- and Dia- 
 magnetism are produced by rotating electrons and that this 
 magnetism remains in existence down to the lowest tempera- 
 tures. 
 
 8. Theory of the Quantum of Action 
 
 In yet another respect has Planck's theory proved stimu- 
 lating, in virtue of a special formulation which Planck gave 
 it 87 at the Solvay Congress in Brussels during 1911. For 
 here Planck gave expression for the first time to the idea 
 that the appearance of energy-quanta is only a secondary
 
 26 
 
 THE QUANTUM THEORY 
 
 matter, being only the consequence of a deeper and more 
 general law. This law, which is to be regarded as the pre- 
 cursor of the latest development of the doctrine of quanta, 
 may be formulated as follows : Suppose the momentary state 
 of a Planck oscillator, say a linearly vibrating electron, to be 
 defined according to Gibb's method by its displacement q from 
 its position of rest and by its impulse or momentum p, and 
 suppose it to be represented in a q-p plane (the state- or 
 phase-plane). Every point of the q-p plane, that is, every 
 phase-point, corresponds to a definite momentary condition of 
 the oscillator. The postulate is then made that not all points 
 of this plane of states are equivalent. On the contrary, there 
 
 FIG. 2. 
 
 are certain states of the oscillator which are distinguished by 
 a peculiarity. The totality of the phase-points that cor- 
 respond to these peculiar states form a family of discrete 
 curves which surround one another. In the case of the 
 Planck oscillator these curves are concentric ellipses (see 
 Fig. 2) which divide the phase-plane into ring-like strips. 
 The postulate of the quantum theory now consists in this, 
 that these ring strips all possess the same area h. If we 
 calculate on this basis the energy possessed by an oscillator 
 in one of these unique states, we find 80 that it is a whole 
 multiple of hv. These special states (represented in the 
 phase-plane by the points of the discrete ellipses) are, there-
 
 THEORY OF THE QUANTUM OF ACTION 27 
 
 fore, according to Planck's first theory, the only dynamically 
 possible and stable states of the oscillator. If an oscillator 
 emits or absorbs, its phase-point jumps from one ellipse to 
 another. The state of affairs is different if we accept Planck's 
 second theory. According to this, all conditions of the oscil- 
 lator, that is all points on the phase-plane, are dynamically 
 possible. On the other hand, emission takes place only in 
 the states specially distinguished by the ellipses. Seen from 
 this new point of view, the energy-quanta are, therefore, only 
 a result of the partitioning of the phase-plane. Mathe- 
 matically, we may express this " structure of the phase-plane " 
 thus : the ?ith unique curve encloses a surface of area nh, or, 
 in symbolic language, 
 
 I" [dqdp = \ 
 
 nh . '. (30) 
 
 The double integral is taken over the surface ; the single 
 integral is taken around the boundary curve of the nth 
 ellipse. 
 
 On this basis for systems of one degree of freedom, which 
 is called Planck's theory of " the action-quantum " for h has 
 the dimensions of an action the modern extension of the 
 quantum theory for several degrees of freedom has, as we 
 shall see, been erected. 
 
 Further, a line of argument proposed and developed by 
 A. Sommerfeld takes its origin here. Starting from the fact 
 just mentioned, that Planck's constant h possesses the dimen- 
 sions of action (energy-time), Sommerfeld set up the hypo- 
 thesis w that for every purely molecular process, say the release 
 of an electron in the photo-electric effect, or the stopping of 
 an electron by the anticathode in the generation of Eontgen 
 
 rays, the quantity called action (L - V)dt, known to us 
 V J 
 
 from Hamilton's Principle, has the value - . Here L and V 
 
 Air 
 
 are the kinetic and potential energies of the electron respec- 
 tively, T is the duration of the molecular process 4 ! say, for 
 example, the time which is required for the release of the 
 electron from the atomic complex during the photo-electric 
 effect, or the stopping of the electron by the anti-cathode.
 
 28 THE QUANTUM THEORY 
 
 This formulation of the quantum hypothesis is, as it were, 
 an expression of the well-known fact that large amounts of 
 energy are absorbed or given up in short times, whereas small 
 amounts are absorbed or emitted in longer times by the 
 molecules, so that on the whole the product of the energy 
 transferred and the duration of the time of exchange is a 
 constant. In fact, fast cathode rays, for example, are stopped 
 by matter in a shorter time and therefore generate harder 
 Kontgen rays than slow cathode rays. Sommerfeld has 
 applied his theory successfully to the mechanism of the 
 generation of Rontgen rays and y-rays. 90 Sommerfeld and 
 P. Debye w have worked out on the same basis a theory of 
 the photo-electric effect, which, like the hypothesis of light- 
 quanta, also leads to Einstein's Law (29).
 
 CHAPTEK IV 
 
 The Extension of the Doctrine of Quanta to the 
 Molecular Theory of Solid Bodies 92 
 
 i. Dulong and Petit's Law 
 
 IT was a particularly fortunate circumstance for the con- 
 solidation of the doctrine of quanta that the failure of 
 classical statistics was not confined to the theory of radiation, 
 but, as appears later, extended to the molecular theory of solid 
 bodies. Thus there arose in quite another field a strong sup- 
 port for the quantum hypothesis, namely, in the field of Atomic 
 Heats. The Atomic Heat of a substance (in the case of poly- 
 atomic bodies we say the " Molecular Heat ") is defined as the 
 product of its specific heat and its atomic weight (or molec- 
 ular weight) ; or, otherwise expressed, it is that amount of 
 heat which must be communicated to a " gramme-atom " m 
 (or gramme-molecule) of the body, in order that its tempera- 
 ture may be raised by one degree. According to our present 
 conceptions, the thermal content of a monatomic solid, say 
 a crystal, is nothing more than the energy of the elastic 
 vibrations of its atoms, which are arranged in the form of 
 a space-lattice, about their positions of equilibrium. If we 
 apply classical statistics to these vibrations, and particularly 
 the law of equipartition of kinetic energy, we arrive at the 
 following conclusion : The mean kinetic energy of an atom 
 
 01L/TT 
 
 vibrating in space, i.e. with three degrees of freedom, is , and 
 
 its mean potential energy is equal to the same amount, 94 so 
 that its total energy is therefore 3kT. If we now consider 
 1 gramme-atom of the body, that is, a system of N atoms 
 (where N is the Avogadro number, approximately 6 x 10 23 ), 
 we get for the mean energy of the body, remembering (19), 
 
 E = 3kTN = 3RT . . . (31) 
 29
 
 30 THE QUANTUM THEORY 
 
 where R is the absolute gas-constant. It follows that the 
 atomic heat of the body at constant volume becomes : 
 
 0,.^-3*-fr94[] . . (32) 
 
 This is the law of Dulong and Petit, 9 * according to which 
 the atomic heat (at constant volume) of monatomic solid bodies 
 
 has the value 5'94 jj , independently of the temperature. 96 
 
 This law is actually obeyed by many elements more or less 
 closely. 97 On the other hand, elements have long been known 
 which are far from following this rule, and which show 
 systematic differences, especially at low temperatures. 
 
 Thus, as early as the year 1875, F. H. Weber & found that 
 
 the atomic heat of diamond at - 50 C. is about 0'75 ?^. The 
 
 deg. 
 
 atomic heats of other elements as well (boron, beryllium, 
 silicon) have also been shown to be much too small at 
 ordinary temperatures. And altogether it appeared that the 
 defect from Dulong and Petit' 's normal value occurs quite 
 generally at low temperatures, and becomes the more pro- 
 nounced, the lower the temperature. The classical theory 
 offered no solution of these low values of the atomic heat. 99 
 
 2. Einstein's Theory of Atomic Heats 
 
 Einstein was the first to recognise 10 that in this case, too, 
 the quantum theory was destined to solve the difficulty. 
 Precisely as in the theory of radiation, the method of 
 classical statistics leads of necessity to a wrong law in the 
 field of atomic heats. Hence, here also, we must abandon 
 the laiv of the equipartilion of energy. In fact, we need only 
 imagine electric charges distributed among the atoms 101 and 
 then we see that, exactly like the Planck oscillators, they must 
 set themselves in equilibrium with the heat-radiation which 
 is always present in the body. This means, however, that 
 
 the relation (7), according to which U = ~a K,., must be set 
 
 up between the mean energy U of an atom vibrating linearly 
 with frequency v, and the intensity of radiation K,,. If we 
 now take Planck's radiation formula (12) as empirically
 
 EINSTEIN'S THEORY OF ATOMIC HEATS 31 
 
 given, it follows immediately that the mean energy U of the 
 linearly vibrating atom must possess, not the value kT given 
 by classical statistics, but the value given by the quantum 
 
 theory, namely, U = 
 
 hv 
 
 For the atom which vibrates 
 
 in space we get, therefore by an obvious generalisation in 
 place of the classical value 3kT, the quantum value : 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 ^j 
 
 
 
 
 
 
 ^- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 p-- 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 .. 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 X 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 / 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 / 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 I 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 / 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 1 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 / 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 . 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 o 4 
 
 1 
 
 j 0.3 a, 
 
 * <?* fl 
 
 s 0,7 0,8 0,9 io 1,5 i.tf zja 
 x THr 
 
 FIG. 3. 
 
 The heat-content of the gramme-atom will therefore be 
 3Nhv 
 
 . . . . (33) 
 
 w - 1 
 
 from which we get for the atomic heat at constant volume 
 Einstein's formula 
 
 According to this, the atomic heat of monatomic solid bodies 
 is not a constant which is independent of the temperature, as 
 
 Dulong and Petit' a Law requires, but is a function of JL and
 
 32 THE QUANTUM THEORY 
 
 is therefore in the case of a definite body (i.e. with v fixed) 
 a funct'ion of the temperature. Its form is such (see Fig. 3), 
 that for T (i.e. x co ) the atomic heat itself is 
 zero, and then increases gradually with increasing tem- 
 perature, approaching asymptotically at high temperatures 
 (i.e. with small x) the classical value 3R. Dulony and 
 Petit's Law is therefore only true in the limit for small 
 
 values of-=-jL that is, for low frequencies of atomic vibration, 
 
 fCJ. 
 
 or high temperatures, exactly as is the case with Rayleigh's 
 Law of Radiation. The departures from Dulong and Petit's 
 Law, in passing from high to low temperatures, become marked 
 the sooner the greater the frequency of the atoms. 
 
 3. Methods of Determining the Frequency 
 
 This frequency v the only unknown magnitude in Einstein's 
 formula (34) may be determined by several independent and 
 very noteworthy methods. One way that is always possible 
 is of course the following : For a given substance we choose 
 an experimentally well-known value of the atomic heat C* 
 which corresponds to a definite temperature T*. From (34) 
 
 x 2 e x C* 
 
 it follows then that = ~, an equation from which 
 
 (e* 1) oH 
 
 x = ^ can be determined, and thence v. From the v thus 
 
 found the course of the whole G v curve can be calculated for 
 all temperatures, and compared with experiment. 
 
 Besides this "empirical" method of determining v, there 
 are a number of other more " theoretical " methods which 
 do not require the use of the values of the atomic heat. 
 Einstein, 102 as far back as 1911, discovered an important 
 connection between the frequency v and the elastic properties 
 of the body. That such a connexion must exist is easily 
 recognised from the following considerations : imagine the 
 atoms of the body arranged upon a space-lattice, as in a 
 crystal, and suppose a certain definite atom arbitrarily dis- 
 turbed from its position of rest, then this atom, when released, 
 will execute vibrations about its position of equilibrium. If 
 we suppose these vibrations to be simply periodic (" mono- 
 chromatic") we shall, however, soon recognise that this
 
 DETERMINING THE FREQUENCY 33 
 
 supposition is an inadmissible approximation we see that 
 the frequency v is the greater the smaller the atomic mass, 
 and therefore also the atomic weight of the body, and the 
 greater on the other hand the force which restores the 
 atom to its position of equilibrium. This restoring force is, 
 however, for its part the stronger, the less extensible and 
 therefore compressible the body is. Hence v must turn out 
 the greater, the smaller the atomic weight and the compres- 
 sibility of the substance. The exact working out of this idea 
 led Einstein to the formula 103 
 
 2-8. 10 7 ,.. 
 
 .... (35) 
 
 Where A is the atomic weight, p the density, and * the 
 compressibility of the body. 
 
 A further interesting relation, which connects v with ther- 
 mal data, namely, the melting-point, was found by F. A. 
 Lindemann 1M by working out the conception that the ampli- 
 tude of vibration of the atom at the melting-point is of the 
 order of magnitude of the distances between the atoms. If 
 T s is the absolute melting-point, then it follows that 
 
 . . . (36) 
 
 Another formula deduced by E. Gruneisen may also be 
 given here : 
 
 v = 2-91.10".^-^[cj.a-i.pi] . . (37) 
 
 Here C v is the atomic heat at constant volume, and a is the 
 coefficient of thermal expansion ; the index means that the 
 value of C\a.-*pk at absolute zero is to be used. 
 
 From formulae (35) and (36) we recognise at once the 
 abnormal behaviour of diamond, for example, in respect to its 
 atomic heat. For it is known that diamond has a high melt- 
 ing-point and very low compressibility accompanied by a low 
 atomic weight. Its v is therefore comparatively large, and 
 it follows therefore, according to the above considerations, 
 that its atomic heat falls below Dulong and Petit' s value of 
 
 3B = 5-94 at comparatively high temperatures. In fact 
 3
 
 34 THE QUANTUM THEORY 
 
 the atomic heat of diamond at 284 abs. is only T35 -? '-, at 
 
 413 abs. it is 3'64 ~~ ' and even at 1169 abs. it reaches only 
 deg.' 
 
 the value 5'24 
 deg. 
 
 Finally, particular importance attaches to a relation, first 
 discovered by E. Madelung 106 and W. Sutherland, between 
 the frequency v of the atoms and the optical properties 
 of bodies. The two investigators started in this case from 
 the following conception : Crystals of diatomic compounds 
 (binary salts), such as rock-salt (NaCl), sylvin (KC1), 
 potassium bromide (KBr), and others, are known to be 
 cubical space-lattices, in which the single atoms carry electric 
 charges, and therefore appear as ions. In fact, the points of 
 the space-lattice are occupied alternately by the positively 
 charged Na + (or K+) atoms, and the negatively charged Cl~ 
 (or Br~) atoms. If an electromagnetic light- wave of frequency 
 v falls upon this crystal, the two ions are thrown into forced 
 oscillations relatively to one another, and further, on account 
 of " resonance," the more strongly, the more exactly the fre- 
 quency v of the impinging wave agrees with the natural fre- 
 quency v r , which lies in the infra-red, of the ions themselves. 
 Since the ionic vibrations are set up at the cost of the energy 
 of the impinging wave, this energy will be weakened (ab- 
 sorbed) the more during its passage through the body, the 
 nearer v lies to v r . On the other hand, the vibrating ions 
 radiate back waves of frequency v since they are compelled 
 to execute these vibrations, when set into forced vibration, 
 doing so the more strongly, the more pronounced the reson- 
 ance is, again, therefore, the nearer v lies to v r . Hence a 
 region of maximum absorption and strongest (metallic) re- 
 flection will lie in the neighbourhood 108 of v = v r . These 
 regions of metallic reflection of a given substance may be 
 detected by the method of " Eeststrahlen " (residual rays) 
 worked out by H. Rubens and E. F. Nichols.' 109 For this 
 purpose we only require to reflect radiation of a considerable 
 range of frequency about v repeatedly from the substance. 
 In this way all waves will be gradually absorbed except those 
 most strongly reflected. These are, however, just those of
 
 NERNST'S HEAT THEOREM 35 
 
 frequency v r . They are thus "residual." The ultra-red fre- 
 quency v f of the ions therefore agrees with the frequency of 
 the residual rays. 110 On the other hand, this vibration of the 
 charged atoms is dependent on the elastic properties of the 
 substance, as we recognised in considering the formula (35). 
 We thus conclude that the " elastic" frequency of the atoms 
 of binary salts agrees to a close approximation with the 
 "optical" frequency of their residual rays. But since the 
 " elastic " frequency of the atoms determines the behaviour 
 of their atomic heat, the ring is thereby closed, and W. 
 Nernst m was thus justified in propounding the fundamental 
 law, that in calculating the atomic heat of binary salts, we 
 may simply insert for the atomic frequencies v the frequencies 
 of the residual rays. 
 
 In this way a number of independent ways were opened 
 up for determining the atomic frequencies required for the 
 calculation of the atomic heat. A comparison of the various 
 values of v determined by these different methods shows in 
 general satisfactory agreement, at any rate in order of magni- 
 tude. 112 One could hardly expect more, as we shall soon see, 
 in view of the many idealised conditions that were used in 
 the theory. 
 
 4. Nernst's Heat Theorem 
 
 With a view to discovering experimentally the general law 
 for the decrease of the atomic heat when approaching low tem- 
 peratures W. Nernst 113 began in 1910, in co-operation with 
 his research students, a series of masterly and widely planned 
 researches. For, by an entirely different route from Einstein 
 namely, by way of thermodynamics he also had become 
 convinced that the atomic heat of solid bodies must become 
 vanishingly small on approaching absolute zero. In his 
 opinion this result was only one of several consequences of a 
 general principle, namely, a new law of heat. 11 * This Heat 
 Theorem of Nernst often called the Third Law of Thermo- 
 dynamics states, in its original form, the following fact : If 
 we regard a system of condensed (i.e. liquid or solid) bodies, 
 which passes at temperature T by means of an isothermal 
 reaction from one state to another, and if A is the maximum 
 work which can be gained from this reaction, then
 
 36 THE QUANTUM THEORY 
 
 gy-0 for the limit T = . . (38) 
 
 tliat is to say, in the immediate neighbourhood of absolute zero, 
 the maximum work which can be gained is independent of the 
 temperature But it follows immediately from this, if we 
 apply the two laws of thermodynamics, 119 that for any 
 reaction which changes the system from the initial condition 
 with energy U^ to the final condition with energy U. 2 , the 
 relation holds that 
 
 ^-^for the limit !T0 . . (39) 
 
 Now, since -5^, if we take a gramme-atom of the substance, 
 
 gives the atomic heat, we are led to enunciate the following 
 rule : in the immediate neighbourhood of absolute zero, the 
 atomic heat of condensed systems remains unchanged during 
 any transformation. 
 
 Planck m has given Nernst's Theorem a still more general 
 form : Not only the difference of the atomic heats (before and 
 after the reaction) is to assume the value at absolute zero, but 
 also each atomic heat itself is to do the same. Thus it follows 
 from the extended Nernst Theorem, in agreement with the 
 demands of the quantum theory, that the atomic heats of 
 solid bodies disappear at absolute zero. 
 
 5. The Improvement on Einstein's Theory of Atomic Heats 
 
 The experiments of Nernst and his collaborators proved 
 quite convincingly that the atomic heat of all solid bodies 
 tends towards a zero value as the temperature falls. In 
 the main, the courses of these decreasing values showed a 
 notable agreement with Einstein's formula (34). At low 
 temperatures, however, systematic discrepancies were found 
 in all cases, in the sense that the observed atomic heats fell 
 off much more slowly than Einstein's formula demanded. 1 " 
 W. Nernst and F. A. Lindemann "8 tried to take these dis- 
 crepancies into account by constructing an empirical formula, 
 and this actually expressed the observations much more 
 accurately than did the Einstein formula. This Nernst-
 
 IMPROVEMENT ON EINSTEIN'S THEORY 37 
 
 Lindemann formula, which is now only of historical interest, 
 is as follows : 
 
 hv 
 where x . (40) 
 
 It receives a meaning if we suppose that one half of all the 
 atoms vibrate with the frequency v, the other half with the 
 
 frequency n While this supposition is untenable in this 
 
 raw form, it contains a kernel of truth, namely, recognition 
 of the fact that the " monochromatic " theory of atomic 
 heats, which assumes only a single fixed frequency v for all 
 atoms, goes too far, being an idealisation of the real state of 
 affairs. Einstein, who at first, for the sake of simplicity, 
 reckoned with only one frequency, had himself already 
 recognised how matters stood, and drawn attention to the 
 need for amending his theory. 119 Nowadays, in fact, we think 
 of a solid body, say a crystal, as built up of atoms regularly 
 arranged upon a space-lattice, according to Bravais' concep- 
 tion ; and this hypothesis has been verified as a certainty 
 through Lane's discovery of the interference of Rontgen rays. 
 In such a complicated mechanical system, however, the 
 single atoms do not vibrate independently of one another 
 with a single, frequency v. But the position of equilibrium 
 of each atom, and thereby the type of its oscillations about 
 that position, is determined rather by the forces which all the 
 other atoms of the body exert upon the atom in question. 
 We are confronted with a structure which is comparable to 
 the one-dimensional case of a vibrating string, and which 
 thus possesses a whole spectrum of natural frequencies, 
 corresponding to the overtones of the string. If the body 
 consists of N atoms, it possesses in general 3N natural 
 frequencies, 120 of which the slowest are sound waves, while 
 the quickest fall in the infra-red. The most general possible 
 movement of each atom then consists in a super-position 
 of all these natural frequencies. Now, since each natural 
 frequency represents a linear, i.e. simple periodic, motion, 
 exactly like the motion of a Planck oscillator, the idea 
 
 66038
 
 88 THE QUANTUM THEORY 
 
 naturally suggested itself, in calculating the energy-content 
 of the body, to allot to each natural frequency of period v the 
 
 hv 
 theoretical quantum amount -^ as if the natural period 
 
 e Ff -l 
 
 were identical with a linear oscillator. The total mean 
 energy of the body then becomes 
 
 in which the summation is carried over all 3N natural 
 frequencies v v v 2 , v 3 , . . . v 3 jv, that is, over the whole elastic 
 spectrum of the substance. By differentiation with respect 
 to T we obtain the atomic heat 
 
 6. Debye's Theory of Atomic Heats 
 
 The kernel of the problem thus consists in calculating the 
 " elastic spectrum " of a given body, that is, in determining 
 for any body the position of its natural periods. In this sense, 
 the theory has been worked out from two different sides ; on 
 the one hand by P. Debye, nl who took an elastic continuum 
 as an approximation to the actual atomically constructed 
 body, and on the other by M. Born and v. Kdrmdn, who 
 replaced the crystal of limited size by one of infinite di- 
 mensions. The difference between these two methods of 
 approximation causes the main problem, namely, the working- 
 out of the elastic spectrum, to be solved quite differently in 
 the two cases. The Debye theory, which from the outset 
 leaves out of consideration the crystalline, and even the 
 atomic, structure of the body, rests upon the classical theory 
 of elasticity, which, of course, treats bodies as structureless 
 continua. From it follows the important law : the number 
 Z(v)dv of all those natural periods, the frequency of which 
 falls within the interval v, v + dv, amounts to 123
 
 DEBYE'S THEORY OF ATOMIC HEATS 39 
 
 Here V is the volume of the body, GI and c t are the velocities 
 with which longitudinal and transverse waves, respectively, 
 are propagated within the body. In this case, however, the 
 following difficulty occurs in replacing the body, which in 
 reality consists of N atoms, by a continuum, namely, the 
 elastic spectrum extends to infinity, that is, the number of 
 natural frequencies becomes infinitely great. For example, 
 the number of natural frequencies (fundamental tone and 
 over-tones) of a linear string of length L are 
 
 Vi = c t ^f and vi = GI ^j. respectively (i = 1, 2, . . . x> ) 
 
 2-Lv aJ 
 
 according as to whether we are considering transverse or 
 longitudinal frequencies. The series of overtones therefore 
 extends without limit to infinity. In reality, however, as the 
 body consists of N atoms (mass-points), it may not possess 
 more than 3N natural frequencies. In order to attain this, 
 Debye helps himself out by means of the following bold 
 supposition. Instead of calculating strictly the elastic spec- 
 trum of the real body consisting of N atoms, he replaces it by 
 that of the continuum as an approximation, but breaks it off 
 arbitrarily at the 3Nth natural period. Debye thus gets the 
 greatest frequency v m which occurs, that is, the upper limit 
 of the elastic spectrum, from the condition : 
 
 
 
 therefore 
 
 [9JV I 
 "(H) 
 
 . (44) 
 
 The atomic heat of the body, which follows from (42), is
 
 40 THE QUANTUM THEORY 
 
 a result which can easily be brought into the following more 
 simple form : 1M 
 
 The atomic heat is therefore only a function of the magnitude 
 x m , that is, it depends only on the ratio ^ : here ^ = ^ 
 
 This result may be expressed in Debye's terms thus : reckon- 
 ing the temperature T as a multiple of a temperature which 
 is characteristic of the particular body, then the atomic heat is 
 represented for all monatomic bodies by the same curve. Hence 
 we must be able to bring the C curves of all monatomic 
 bodies into coincidence, if only the scale of temperature be 
 suitably chosen for each substance. 128 For high tempera- 
 tures, the Debye formula passes over, as it must do, into the 
 classical value of Dulong and Petit, C v = 3.R, 128 just as do 
 the Einstein and Nernst-Lindemann formulae. On the other 
 hand, it differs from these latter in falling much more slowly 
 at low temperatures. For while the atomic heats, according 
 to both Einstein and Nernst-Lindemann, fall exponentially 
 
 / 1 constN 
 
 (with 7- e ~~T~) at low temperatures, Debye's formula leads 
 
 to the fundamental law, 127 that the atomic heats of all bodies at 
 low temperatures are proportional to the third power of the 
 absolute temperature. 
 
 It is further remarkable, that we may write formula (44) 
 for the maximum natural frequency in a form such that only 
 measurable magnitudes occur in it. For if we express the two 
 velocities of sound Ct and c\ in terms of the elastic constants of 
 the body, and replace the volume Fof the gramme-atom by 
 
 ,. . atomic weight (A) . 
 the quotient -- density g (p) ( >, * follows that 
 
 ,5-28.10*.^) 
 where " '
 
 DEBYE'S THEORY OF ATOMIC HEATS 41 
 
 In it K is again the compressibility of the body, <r the 
 Poisson ratio, that is, the ratio of the transverse contraction 
 to the extension. The similarity of this formula with the 
 Einstein relation (35) strikes one immediately. But in this 
 case the second elastic constant of the isotropic body, cr, 
 enters into the equation as well. Altogether, the upper limit 
 v m of the elastic spectrum, at which, as one can show, 129 the 
 natural frequencies always crowd together closely, plays in 
 the stricter theory an analogous role to that played by the 
 single natural frequency v in the " monochromatic " theory. 
 
 Comparison with experiment shows 130 that the Debye 
 formula, at any rate for the monatomic elements such as 
 aluminium, copper, silver, lead, mercury, zinc, diamond, de- 
 scribes the course of values of the measured atomic heats 
 very accurately. Particularly at low temperatures, the pro- 
 portionality between the atomic heat and the third power of 
 the absolute temperature receives fair confirmation. 131 In 
 view of the fact that the idealised view (replacement of the 
 actually atomic body by a continuum) is carried very far, we 
 must not regard the agreement between theory and experi- 
 ment as self-evident. At low temperatures, Debye's idealis- 
 ation will justify itself. For then ^ is large, and hence the 
 
 amount of energy v is small, excepting when v itself as- 
 
 ekf - 1 
 
 sumes small values. At low temperatures, there/ore, only long 
 waves will contribute sensibly to the energy of a body, and hence 
 to its atomic heat. For long waves, however, that is, for 
 waves, the length of which is great compared with the dis- 
 tance between the atoms, the specific atomistic construction of 
 the body plays no part ; for them the substance is almost a 
 continuum. The position is quite different at high tempera- 
 tures, at which the longer frequencies up to the maximum v m 
 (that is, the shorter waves down to the smallest) furnish con- 
 tributions of energy. For the waves which correspond to the 
 highest frequencies possess lengths, as can easily be shown, 132 
 which are comparable with the distances between the atoms, 
 and for these shorter waves the medium cannot fail to betray 
 its atomic structure. Here, therefore, its replacement by a
 
 42 THE QUANTUM THEORY 
 
 continuum becomes questionable since the approximation is 
 only very rough. 
 
 7. The Lattice Theory of Atomic Heats according to Born and 
 Karman. The Elastic Spectrum of the most general Crystal 
 
 At this point the above-mentioned investigations of Born 
 and Kdrmdn intervene, which, going beyond Debye, take 
 account of the real crystalline structure of the body, that is 
 to say, the space-lattice arrangement of the atoms. In order 
 to overcome the great mathematical difficulties involved, they 
 imagined, as has already been said, the actual limited crystal 
 replaced by one extended indefinitely. Thus the disturbing 
 effect of the surface on the interior could be eliminated, so 
 that now all atoms were exposed to the same conditions. 
 Here also the main problem is again to determine the elastic 
 spectrum, or if we dispense with the exact calculation of 
 the proper frequencies at least to discover the law, accord- 
 ing to which the proper (or natural) frequencies are distributed 
 among the different regions of frequency. This problem was 
 first solved by Born and Kdrmdn for regular crystals. The 
 laws thus obtained were then extended to the case of simple 
 point-lattices of arbitrary symmetry, and finally, Born de- 
 duced them, in his "Dynamics of the Crystal Lattice," for 
 the most general form of space-lattice. 133 
 
 These most general space-lattices arise from the periodic 
 repetition in space of a definite group of atoms and electrons 
 (basic group) which on the whole is electrically neutral, and 
 is enclosed in a parallelepiped of space, the "elementary 
 parallelepiped." In Fig. 4 such a lattice, in this case, how- 
 ever, plane, is illustrated, in which the basic group consists 
 of three particles (-ox). All particles form together a 
 simple lattice, as do the o and x particles. We have in this 
 way three interlocked simple lattices. 
 
 Thus, for example, the halogen compounds of the alkalies 
 (NaCl, LiCl, KC1, KBr, KI, BbCl, EbBr, Ebl, and so forth) 
 form cubic space-lattices, in which the lattice points are 
 alternately occupied by the positive alkali ion and the negative 
 halogen ion (see Fig. 5). If we regard the whole cube here 
 pictured as the " elementary cube," then the basic group would
 
 LATTICE THEORY OF ATOMIC HEATS 43 
 
 contain eight particles, namely, four ions of each sort (they are 
 numbered here). We have thus eight interpenetrating simple 
 lattices. Every four of them would, however, consist of the 
 same kind of particle. Hence it is advisable to select in this 
 case in place of the cube the rhombohedron (double-lined in 
 
 FIG. 4. 
 
 the figure) as the elementary parallelepiped. Then the basic 
 group consists only of the two different particles 1 and 8, of 
 which the one lies in a corner, the other in the middle of the 
 parallelepiped. In fact we can get the whole lattice by displac- 
 ing the basic group in the direction of the three rhombohedral 
 edges, a distance equal to a whole multiple of the length of
 
 44 THE QUANTUM THEORY 
 
 the edge. The lattice consists therefore, according to this 
 view, of two interlaced simple cubical atomic lattices. Further- 
 more, they are " surface-centred " lattices, that is to say, such 
 that not only the corners of the cubes, but also the middle 
 points of the cube-surfaces, are occupied. If in the most 
 general case the basic group contains s different particles, the 
 lattice consists of s interlaced simple lattices. 
 
 FIG. 5. 
 
 In order now to get a general view of the laws which 
 govern the elastic spectrum of such a most general crystal, 
 we proceed according to Born and Kdrmdn as follows : We 
 imagine an elastic wave of definite wave-length and definite 
 direction (the normal to the wave front) passing through the 
 crystal. For each wave thus defined there are 3s natural
 
 LATTICE THEORY OF ATOMIC HEATS 45 
 
 frequencies with periodicities v l v 2 v 3 . . . v 3 .. The first three 
 frequencies v lt v.,, v 3 correspond to those natural frequencies 
 of the crystal, by which the single interpenetrating simple 
 lattices are similarly distorted to a first approximation with- 
 out being compelled to move relatively to one another. 
 These are the three ordinary acoustic natural periods (one 
 longitudinal, two transverse). The remaining 3(s - 1) 
 frequencies, on the other hand, correspond to another type 
 of motion of the crystal, namely, to those natural frequencies 
 with which the single simple lattices oscillate with respect to 
 one another without distortion. If the basic group contains 
 only one particle (s = 1), i.e. if the crystal consists of only a 
 simple lattice, this second type of motion disappears alto- 
 gether, and we are left with only the three acoustic natural 
 frequencies v v v. 2 , v 3 . If, on the other hand, we are dealing 
 with a crystal, say of the halogen compounds of an alkali, 
 for example, rock-salt (NaCl), s = 2, there exist, as we have 
 seen, besides the three acoustic oscillations, three further 
 natural frequencies of the second type. In consequence 
 of the regular crystal character of the alkaline halides, these 
 three natural frequencies exactly coincide, at any rate for 
 long waves, and give rise to that motion in which the sodium 
 lattice vibrates approximately as a rigid structure against the 
 likewise rigid chlorine lattice. We see at once that it is 
 just the natural frequency last considered that will play the 
 chief part in the optics of these crystals. For when an 
 electromagnetic wave meets the crystal, the sodium ions 
 are driven by the electric force of the wave to the one side, 
 and the oppositely-charged chlorine atoms are drawn to the 
 opposite side. It is thus just the type of vibration described 
 above that is brought about. If the frequency of the 
 external wave approaches closely to that of the natural 
 period, resonance occurs. These infra-red vibrations, there- 
 fore, are what determine the course of the refractive index, 
 especially in the infra-red. They are the so-called " infra- 
 red dispersion frequencies." It is also in their neighbourhood 
 that the places of metallic reflection lie which are detected 
 by the method of residual-rays. 
 
 What has just been stated for the special case s = 2 
 (alkaline halides) may, of course, be immediately generalised.
 
 46 THE QUANTUM THEORY 
 
 For if the basic groups consists of s different particles, it 
 is just the 3(s - 1) natural frequencies that determine the 
 dispersion of the crystal. Among them are those, in the 
 neighbourhood of which the regions of metallic reflection 
 (residual rays) lie. If the basic group contains p positive 
 atomic residues and s - p electrons, the frequencies v 4 . . . v 3 , 
 fall correspondingly into two classes : the first class consists 
 of 3(p - 1) infra-red frequencies, which arise from the 
 atomic residues ; the second consists of 3(s - p) ultra-violet 
 frequencies, which are to be ascribed to the influence of the 
 electrons. The infra-red natural frequencies decide the 
 course of the refractive index in the infra-red, the position 
 of the residual rays, and, as we shall see, the atomic heats ; 
 the ultra-violet natural frequencies, on the other hand, deter- 
 mine chiefly the refractive indices in the visible and ultra- 
 violet. Incidentally, the general lattice-theory of Born 134 
 confirms the law previously enunciated by Haber 13S that the 
 frequencies of the first class (infra-red) bear the same ratio to 
 the second (ultra-violet) class, as regards order of magnitude, 
 as the square root of the mass of the electron bears to the 
 square root of the mass of the atom. 
 
 After this digression let us now return to our starting-point. 
 Up to the present we have always considered a wave of 
 definite length A. and with a definite normal direction n, and 
 we have seen that corresponding to it there are, in the most 
 general case, 3s natural frequencies i/j . . . v 3S . Let us now 
 allow the wave-length X to vary continuously, keeping the 
 wave-direction constant, by going from infinitely long waves 
 to the smallest. Then each of the 3s natural frequencies will 
 also vary continuously, and will pass through a continuous 
 range of values. In other words, the 3s natural frequencies 
 are certain functions of the wave-length X : 
 
 v< -/<(*). 
 
 From this, however, we learn the fundamental fact that all 
 these ranges of values of the single natural frequencies are 
 only finite in extent and that, therefore, each of the 3s continua 
 of frequencies automatically breaks off at a higJiest limiting 
 frequency. " Automatically," i.e. without our arbitrary assist- 
 ance (as in Debye's case), solely on account of the analytical
 
 LATTICE THEORY OF ATOMIC HEATS 47 
 
 form of the function /,-. This is explained by the fact that 
 the wave-length X of possible waves in the crystal has a 
 lower limit set to it : waves of length below a certain lowest 
 value cannot exist. This is most simply recognised from the 
 following instructive example. If we consider a simple 
 cubical lattice having the atomic distance a, and examine, 
 for example, longitudinal waves, which are being propagated 
 along an edge of the cube so that all atoms on an edge at 
 light angles to this side oscillate in the same phase in the 
 direction of the edge then we see at once that the smallest 
 wave that is possible here has the length X m j n = 2a. For 
 this wave, namely, successive planes of the cube swing 
 in opposite phase, that is, " against " one another. The 
 functional relation between v and X assumes the special 
 form : 138 
 
 For infinitely long waves (X = co ), v = ; if we pass on 
 to shorter waves, v increases continuously, until, for X = 2a, 
 it reaches its maximum value v m . At this limiting frequency 
 v m the range of possible v's breaks off automatically. 
 
 Up to the present we have given the wave-direction (n, the 
 direction of the normal) a certain fixed value, and have 
 allowed the wave-length X to vary. We now give the wave- 
 direction by degrees other values, and at each step we allow 
 the wave-length to vary from the value GO to the least 
 possible value. Then the nature of the functional dependence 
 of the magnitude vt or X, and the position of the limiting 
 frequencies also change continuously with the wave-direction, 
 so that we may say : the 3s natural frequencies are, in 
 general, continuous functions of the wave-length X and of 
 the wave-direction n : 
 
 vi =/<(X, n), (i = 1, 2, 3, . . . 3s) . (48) 
 
 In it, each of the functions / t - breaks off automatically for a 
 minimum value of the wave-length at an upper limit 
 (vi)max, which itself still depends on the wave-direction. 
 These equations express the law of dispersion of waves in 
 crystals, for they determine for each wave the 3s frequencies
 
 48 THE QUANTUM THEORY 
 
 vf and hence also tell us how the rates of propagation 
 q f = vi'X depend on the wave-length and the wave-direction. 
 The dispersion law becomes particularly simple in the region 
 of long waves : for the three acoustic vibrations the re- 
 lations 137 
 
 A A. A. 
 
 hold. In them the three magnitudes q^(n), q. 2 (n), and q s (n) 
 are three, in general different, functions of the wave- 
 direction. And further, these are the three velocities of 
 propagation of the three acoustic vibrations. In the region 
 of long waves, therefore, the three velocities of propagation 
 of the three slow acoustic vibrations are independent of the 
 wave-length to a first approximation. 
 
 The dispersion law (for long waves) assumes a very 
 different appearance for the 3(s - 1) rapid vibrations 
 v., V K . . v,.. It assumes the form 
 
 (i = 4, 5, . . . 3s) . (50) 
 
 A 
 
 here the v9's are constants, the p(w)'s are again certain 
 functions of the wave-direction. The velocities of propaga- 
 tion here assume the values 
 
 q t = Vi \ = v\ + p { (n) . . . (51) 
 
 and would thus be linear functions of the wave-length. 
 
 We may summarise thus : the elastic spectrum of the most 
 general crystal, the basic group of which contains s particles, 
 consists of 3s separate parts (" branches "). Each part consists 
 of a finitely extended continuum of frequencies. The three first 
 parts contain the totality of all sloiv, acoustic natural fre- 
 quencies (sometimes called "characteristic"). The remaining 
 3(s - 1) parts include the rapid (infra-red and ultra-violet) 
 natural frequencies, which play the chief part in determining 
 the optical dispersion and the positions of metallic reflection. 
 
 % 8. Continuation. The Law of Distribution of the Natural 
 Frequencies 
 
 While this knowledge of the general character of the elastic 
 spectrum is, as we shall soon see, of great value, it is none
 
 LATTICE THEORY OF ATOMIC HEATS 49 
 
 the less insufficient for the question of the energy-content and 
 molecular heat of the crystal, inasmuch as, even for the 
 simplest crystal, a strict calculation of the elastic spectrum 
 is not possible at the present time. We know, however, on 
 the other hand, that we need not know the whole details of 
 the elastic spectrum to calculate the energy-content and the 
 molecular heat, but that it suffices to know the law according 
 to which the natural frequencies are distributed over the 
 elastic spectrum (or its individual "branches"). This is 
 the more true, the closer together the natural frequencies 
 lie. Now, in reality the finite crystal possesses, if it con- 
 sists of the basic group (of s particles) N times repeated, 
 SNs natural frequencies, which are distributed so that N fre- 
 quencies fall to each of the 3s branches of the spectrum. If N 
 becomes infinite, the N individual natural frequencies of each 
 branch merge into one another to form a continuum, and we 
 get exactly the elastic spectrum that we have just been con- 
 sidering. We see from this, that the more we are justified 
 in replacing the finite crystal by one of infinite extent the 
 better our results if we know only the distribution law of the 
 natural frequencies (without knowing their position exactly). 
 The law of distribution of the natural frequencies, which 
 was discovered by Born and Kdrmdn and extended by Born 
 in his " Dynamics of the Crystal Lattice " to the most general 
 type of crystal, may be formulated thus : Select from the 
 totality of all elastic loaves the small group, whose lengths lie 
 between A. and \ + d\, and ivhose normal direction lies in the 
 elementary solid angle 138 dO. Each of the 3s branches of the 
 
 Y 
 spectrum then contribute dXdto natural frequencies to this 
 
 group. Here V denotes the volume of the finite crystal. 
 
 9. Continuation. The Atomic Heats at Low, very Low, and 
 High Temperatures 
 
 The knowledge of this law of distribution allows us to 
 write down at once the thermal capacity of the crystal con- 
 sisting of Ns particles. From (42) it is :
 
 50 
 
 THE QUANTUM THEORY 
 
 . (52) 
 
 This formula is to be interpreted as follows : the natural fre- 
 quencies vi are, by (48), to be expressed as functions of the 
 wave-length X and the wave-direction n : then the integra- 
 tion is to be performed with respect to X from the smallest 
 wave-length X m (w), which itself depends upon the wave- 
 direction n, up to the maximum X = oo . The result of this 
 integration still depends on the wave-direction and the index 
 i. Finally, integration is to be performed over all directions 
 (that is, over all elementary solid angles between and 4?r) 
 and summation over all 3s branches of the spectrum. But 
 we have seen that the 3s branches of the spectrum fall into 
 two groups. The first 3 branches (i = 1, 2, 3) contain the 
 totality of slow acoustic natural frequencies; for these 
 branches we have the dispersion law (49) which is valid for 
 long waves. The remaining 3(s - 1) branches contain the 
 totality of the quick (infra-red and ultra-violet) natural fre- 
 quencies, with the entirely different type of dispersion law 
 (50), which also holds for long waves. Hence the sug- 
 
 3t 
 
 gestion naturally occurs of dividing the sum J of (52) into 
 
 i = l 
 
 two parts, corresponding to the two different groups of fre- 
 quencies and of writing 
 
 r p = r? + r<*> 
 
 where 
 
 JL i (53) 
 
 r ( > = kv ; r<*> = kv 
 
 These still very complicated formulae may, according to 
 Born, be brought into a very simple and comprehensive form
 
 LATTICE THEORY OF ATOMIC HEATS 51 
 
 by limiting our considerations to low temperatures and in- 
 troducing certain approximations. As we have already re- 
 cognised, at low temperatures only the long waves contribute 
 to the energy-content. Hence we shall apply in formula (53) 
 all those approximations which are introduced by confining 
 ourselves to long waves. Let us consider first r ( ;p. Here we 
 set in place of the v{S of (50) the constant values v., which 
 are independent of the wave-length A. and of the wave- 
 direction. If we do this, we can place the constant factors 
 
 in front of both integration signs, and write 
 
 !-*> V^,. FU>3 , where,, 
 
 The factor in square brackets has, however, a simple meaning. 
 From the law of distribution of the natural periods we see, 
 namely, that this factor gives the sum-total of all natural 
 frequencies that occur in one of the 3s branches of the 
 spectrum ; it therefore has the value N, which as has already 
 been said, is the number of basic groups which go to make 
 up the crystal. If we choose the piece of crystal under con- 
 sideration such that its size is so that N is equal to the 
 Avogadro number, then if we remember that Nk = R for 
 I* 2 ), the expression 
 3 
 
 follows. If we compare this result with (34) we see that rop 
 excepting for the missing factor 3 consists of 3(s - 1) 
 Einstein functions. We write the expression in the form 
 
 
 
 M where ^ =_^ . (55) 
 in which the abbreviation is obvious. The fact that, in using
 
 52 THE QUANTUM THEORY 
 
 these approximations, we come across Einstein factors, i.e. 
 that we encounter the " monochromatic " theory, might have 
 been anticipated. For since we treated the v/s here as con- 
 stants that are quite independent of wave-length and wave- 
 direction, these vibrations represent processes which have 
 nothing to do with the propagation of elastic waves in the 
 crystal as a whole : and this means that the individual 
 particles, uncoupled as it were, perform 3(s -- 1) mono- 
 chromatic vibrations. 
 
 The approximate evaluation of the first part rW is quite 
 different. For here we have to use for the frequencies 
 v i V 2> V 3 * ne relations (49), which connect the three acoustic 
 natural frequencies with wave-length and wave-direction. 
 Here we have therefore to deal with three real elastic oscilla- 
 tions, which are propagated in the crystal with the three 
 different acoustic velocities q-^ri), q*(ii), q s (n), each of which 
 depends on the direction (n). The crystal acts here as a 
 dynamic whole, exactly as in Debye's point of view. Hence 
 we may conjecture that r ( J> allows itself to be brought 
 into the form of three Debye functions (45). The more 
 exact calculation confirms this supposition, and gives us 1M 
 
 (56) 
 
 which, taking Debye's formula (45) into consideration, we 
 may write in the following immediately intelligible form : 
 
 I^-IDft . . (57) 
 
 i=l 
 
 The three magnitudes Xi here play the part of three upper 
 limits of frequency. Their values are 
 
 where the three magnitudes qt represent certain mean direc- 
 tions of the acoustic velocities, which therefore no longer
 
 LATTICE THEORY OF ATOMIC HEATS 53 
 
 depend on the wave-direction. From (55) and (57) we get for 
 the thermal capacity of the piece of crystal considered 
 
 - (59) 
 
 Now, since JV particles of each of the s different kinds of 
 particles are present, that is one gramme-atom of each kind 
 of particle exactly for N is the Avogadro number the piece 
 of crystal contains s gramme-atoms of different sorts of 
 particles. If, therefore, we cut the crystal into s equal 
 
 N 
 pieces in such a manner, that each piece comprises only - 
 
 basic groups, then each of these pieces contains a so-called 
 
 "mean" gramme-atom. Hence if we now consider only a 
 
 p 
 single one of these pieces, its thermal capacity is ; we call 
 
 S 
 
 it the " mean atomic heat " C v , and we may write 
 3 
 D + (* . . (60) 
 
 Here the #/s have the same meaning as in (58). For the 
 
 N 
 
 piece of crystal now under consideration consists of basic 
 
 s 
 
 groups, and has therefore the volume . Formula (58), 
 however, obviously remains unchanged when we replace in 
 
 it N and V by and . The quantity -, the volume of a 
 s s s 
 
 mean gramme-atom, is also called the mean atomic volume. 
 
 In the case of chemical compounds, in which several sorts 
 of atoms occur in the basic group, and also in the case of 
 polyatomic elements, in which the basic group contains several 
 particles of a like sort, we frequently speak of the molecular 
 heat. In doing so, we follow the usual chemical conception, 
 inasmuch as we imagine the s particles of the basic group 
 divided into one or several sub-groups, and regard each sub- 
 group, taken alone, as a molecule. If then the molecule
 
 54 THE QUANTUM THEORY 
 
 contains q atoms, then qC,, is the mean molecular heat; for 
 example, the basic group of rock-salt (NaCl) contains one 
 sodium ion and one chlorine ion. The whole piece of crystal, 
 
 which, by definition, contains - = basic groups, comprises 
 
 S A 
 
 therefore sodium ions and the same number of chlorine 
 ions, that is to say - " NaCl-molecules." q is in this special 
 
 case equal to 2. Hence 2C,, represents the thermal capacity 
 of N " NaCl-molecules,'' that is, the mean molecular heat of 
 rock-salt. 
 
 If among the s particles of the basic group there are p 
 atomic residues and s p electrons, the number of Einstein 
 factors in (59) reduces to 3(p - 1), since the 3(s - p) ultra- 
 violet frequencies arising from the s - p electrons contribute 
 only in a vanishingly small degree to the atomic heat as com- 
 pared with the infra-red. We thus arrive at the law : the 
 mean molecular heat of a crystal wJiose basic group includes p 
 (similar or different) atomic residues, is made up, at a suf- 
 ficiently low temperature, of three Debye terms (with, in general, 
 three different upper limits of frequency) and 3{p - 1) Einstein 
 terms (in which the 3(p-I) infra-red natural frequencies for 
 long waves appear as frequency numbers). 
 
 When we descend to the lowest temperatures, the Einstein 
 terms disappear exponentially, and only the three Debye terms 
 remain, for these, as we know, decrease much more slowly. 
 In them we can further replace all the upper limits of the 
 three integrals (see (56)) by GO , so that the integrals thereby 
 become numerical constants. Eemembering (58) we get the 
 fundamental law, that the molecular heat of every crystal at 
 the loivest temperatures is proportional to the third power of 
 the absolute temperature. So the general lattice theory con- 
 firms Debye's result. The formula obtained has the following 
 simple form : 141 
 
 where VA is the " mean atomic volume " 
 
 mean atomic weights 
 
 - _ - . - i* - 1
 
 TESTS OF THE BORN-KARMAN THEORY 55 
 
 and q represents a quantity which, if suitably defined, may be 
 called the mean acoustic velocity, introduced in place of the 
 three different acoustic velocities q lt q 2 , q s . 
 
 Also in the other extreme case, for high temperatures, a 
 very useful formula can be obtained, as H. Thirring iK 
 showed. He started from (52) and developed the exponential 
 functions in series. The following value is then obtained for 
 the mean atomic heat : 
 
 where the coefficients J v J 2 , J 3 , . . . depend in a complicated 
 manner on the elastic constants of the crystals, the atomic 
 masses, and the atomic distances. 
 
 10. Tests of the Born-Karman Theory 
 
 How do matters stand with regard to the testing of the 
 Born-Kdrmdn Theory? We see at once that it is incom- 
 parably more difficult than in the case of Debye's Theory : for 
 even in simple cases, the calculation of the mean atomic 
 heat of a crystal is very complicated, and requires above all 
 a more exact knowledge of its elastic behaviour than we at 
 present possess. Only by restricting our attention to low 
 and very low temperatures on the one hand, where the 
 formulae (60) and (61) may be applied, and, on the other, 
 to the region of high temperatures, within the limits of 
 applicability of Thirring's formula (62), are we enabled to 
 carry our calculations for a number of simple substances to 
 the point of comparison with experimental results. Born 
 and Kdrmdn themselves, in one of their first publications 14a 
 tested the formula (61), valid for the lowest temperatures 
 (Debye's TMaw), by comparing its results with those of 
 experiment. They limited themselves in this case to metals 
 (Al, Cu, Ag, Pb) which, however at any rate in the usual 
 form are not proper crystals, but irregular crystalline 
 aggregates. For this reason, they proceeded as if the metal 
 were an isotropic body, and obtained the mean acoustic 
 velocity the only quantity in (61) which in general requires
 
 56 THE QUANTUM THEORY 
 
 extensive calculation from the following relation which 
 holds for isotropic bodies : 1H 
 
 Q 1 O 
 
 N+! < 68) 
 
 Here q l and q t are the velocities of propagation of the 
 longitudinal and transverse elastic waves, magnitudes, there- 
 fore, which may be simply calculated from the two elastic 
 constants of the isotropic body and its density. 1 * 9 The 
 agreement of the values of C v thus found with the experi- 
 mental data is, especially in the case of Al and Cu (and also Pb), 
 quite good. A. Eucken w has, however, pointed out rightly, 
 that no weight should be attached to this agreement. For 
 the values of the elastic constants which Born and Kdrmdn 
 used for calculating q t and q t are those which are correct at 
 the ordinary room temperature. If we take their dependence 
 on temperature into account, the good agreement between 
 theory and experiment disappears. Metals are, indeed, not 
 isotropic bodies, and hence it is not permissible to use the 
 observable elastic constants, which depend upon temperature, 
 in calculating q. 
 
 Matters are much more favourable in the case of real 
 crystals, in which, as experiments by E. Madelung i show, 
 the elastic constants vary very little with temperature. But 
 here the calculation of the mean acoustic velocity q gives rise 
 in general to notable difficulties, 1 * 8 which may, however, 
 be cleared away in simple cases by a very practical method 
 due to L. Hopf and G. Lechner. 9 Eopf and Leclmer were 
 thus enabled successfully to carry out the calculations for 
 sylvin (KC1), rock-salt (NaCl) fluor-spar (CaF 2 ) and pyrites 
 (FeS 2 ). They proceeded to calculate the quantity q from the 
 observed values of C v , assuming the correctness of formula 
 (61), and they then compared these with the value of q 
 calculated from elastic data. The result showed very satisfac- 
 tory agreement. 1 o 
 
 It is of particular interest to test the very clear formula 
 (60) which gives the mean atomic heat as a sum of three 
 Debi/e, functions and 3(s - 1) Einstein functions. Here the 
 three infra-red natural frequencies v, vj?, v coincide, and the
 
 TESTS OF THE BORN-KlRMAN THEORY 57 
 
 three Einstein functions become equal to one another. If we 
 introduce the further approximation of replacing the three 
 different quantities Xi in the Debye formula by a mean value 
 x, it follows that 
 
 C, = \{D(x) + E(x)} . . . (64) 
 
 In this we use the value of x deduced from formula (58) 
 by merely replacing <?, in it by a mean value q, which can be 
 calculated by the method of Hopf and Lechner just mentioned. 
 
 j, o 
 x, on the other hand, according to (54), = ^ , where v is the 
 
 infra-red natural frequency of the crystal (for long waves), 
 which may be determined from the dispersion in the infra- 
 red or by the method of residual rays. 
 
 Formula (64) had already been given, previously to Born, 
 by W. Nernst, in who, however, based his argument on a 
 supposition which is no longer tenable. Nernst started 
 from the conception that, for example, in the case of rock- 
 salt, the NaCl-molecules are located upon the points of the 
 space -lattice, and that the most general state of oscillation 
 of the lattice arises from the superposition of two modes of 
 motion, firstly the oscillation of the whole molecules in the 
 lattice-structure, which give a Debye term, and secondly the 
 intra-molecular oscillations of the two atoms, which, being 
 almost monochromatic, lead to an Einstein term. The 
 agreement of the Born-Nernst formula (64) with the ex- 
 perimental data is not very satisfactory in the case of NaCl 
 and KC1, but much better in the case of AgCl, which belongs 
 to the same crystal type. 182 The reason for this is believed 
 by E. Schrodinger *** to lie in the excessively rough ap- 
 proximation inherent in formula (64). 
 
 Finally, Thirring's formula (62) has also been tested, by 
 Thirring himself, IM for NaCl, KC1, and, by neglecting certain 
 factors, for CaF 2 and FeS 2 . Taking into account the variation 
 of the elastic constants with temperature (which, however, is 
 to be regarded as uncertain and provisional since the values 
 are only obtained by interpolation) he found good agreement 
 between theory and experiment. In connection with the 
 Thirring formula, Born 1 ** has also calculated the atomic 
 heat of diamond and compared it with experiment. Since in
 
 58 THE QUANTUM THEORY 
 
 this case, however, the elastic constants were unknown, Born 
 proceeded to evaluate the curves of atomic heat for various 
 possible values, and to select from them that curve which 
 conformed most closely to the results of observations. Thus, 
 
 for example, the value O63 x 10 ~ 12 , m was obtained for 
 
 LdyneJ 
 
 the compressibility ; this is in satisfactory agreement with the 
 value, probably too small, measured by W. Bichards, viz. 
 
 0-5x10-4^1. 
 LdyneJ 
 
 From all this we see that the possibilities of testing the 
 Born-Kdrmdn Theory of Atomic Heats, partly on account of 
 the great difficulties of calculation, partly on account of our 
 insufficient knowledge of the elastic behaviour of crystals, are 
 exceedingly sparse, so that for the present Debye's much more 
 tractable formula (if necessary, with the addition of Einstein 
 terms) appears more useful. If, in spite of this fact, so much 
 space has been devoted here to the Born-Kdrmdn Theory, the 
 reason is to be sought in the conviction that this theory has 
 gone much further than that of Debye into the kernel of the 
 matter. For, without a more exact treatment of the structure 
 of the space-lattice and its dynamics, our knowledge of the 
 nature of the solid state must without doubt remain faulty. 
 
 ii. The Equation of State of a Solid Body 
 Linking up with this new development of the theory 
 of atomic heats, a number of investigators, chiefly E. 
 Gruneisen, 1 * 6 S. Ratnoivski, 161 and P. Debye,*** have worked 
 out a theory of the solid state with the object of creating as 
 a counterpart to the Kinetic Theory of Gases a Kinetic 
 Theory of Solids. One of the main problems in this con- 
 nexion is to formulate an " Equation of State," that is, a 
 relation between pressure (p), volume (V), and temperature 
 (T), a problem, which, according to the doctrine of thermo- 
 dynamics, is to be regarded as solved as soon as the " free 
 energy " F of the body is known as a function of the tempera- 
 ture and the volume. 189 Then the pressure, for example, will 
 follow from the simple equation
 
 EQUATION OF STATE OF A SOLID BODY 59 
 
 which, as a relation between p, V, and T, gives the equation 
 of state at once. If this is known, we have mastered quanti- 
 tatively the behaviour of the body for all changes of state. 
 For example, the coefficients of expansion a, and the com- 
 pressibility K, result from the well-known formulae 
 
 * 
 
 (F is the volume at the zero-point.) 
 
 P. Debye 16 was the first to draw attention to the fact that 
 the model of the solid body which forms the basis of the 
 atomic heat theories of Einstein, Debye, and JBorn-Kdrmdn, 
 is necessarily too highly idealised ; for this idealised solid 
 body has, as is easily seen, a zero coefficient of expansion. In 
 fact, if, as has always been assumed hitherto, the forces 
 which pull the atoms back into their position of equilibrium 
 are proportional to the first power of their relative displace- 
 ments (assumption of quasi-elasticity, Hooke's Law), then 
 the atoms will execute symmetrical oscillations about this 
 position of rest. If this supposition, viz. Hooke's Law, be 
 valid for all temperatures, then the mean volume of the 
 body that is, the volume that it possesses when all atoms are 
 exactly in their positions of rest must be just as often over- 
 shot as undershot, however great the amplitude of the heat- 
 vibrations may be. Hence, if we warm the body from zero 
 until it possesses the volume VQ, and if we assume that all 
 atoms are at rest at zero, then its mean volume at any tem- 
 perature will also be equal to F . The body, therefore, does 
 not change its mean, observable volume with rise of tempera- 
 ture ; its coefficient of expansion is therefore 0. If we desire to 
 represent the actual behaviour of the solid body, namely, its 
 expansion when heated, as known to us from thousandfold ex- 
 perience, we are necessarily obliged, according to Debye, to re- 
 place Hooke's Law of Force by an expression involving higher 
 powers of the variation of atomic distance. Then the oscilla- 
 tions of the atoms become unsymmetrical, and there occurs a 
 displacement of their position of rest as the energy of vibration 
 increases. If we arrange the generalisation of Hooke's Law 
 so that a greater force is necessary to bring the atoms nearer 
 together than to separate them, then the change in the
 
 60 THE QUANTUM THEORY 
 
 position of rest occurs in such a manner that for increasing 
 energy of vibration, that is, for rise of temperature, the 
 relative distances of the atoms increase, and hence the body 
 increases in volume. Debye has extended the theory in this 
 sense. Among other things this gives us the law previously de- 
 duced by Griineisen m that at sufficiently low temperatures the 
 thermal coefficient of expansion a is proportional to the specific 
 heat. Moreover, the very small change in compressibility 
 with temperature is well accounted for on Debye 's Theory. 
 
 12. The Thermal Conductivity of Solid Bodies according to Debye 
 
 The importance of Debye' s Theory is by no means confined 
 to thermal expansion. On the contrary, it became manifest 
 that another important group of phenomena require this 
 generalisation of Hooke's Law. In the idealised solid body, 
 in which the elastic forces obey Hooke's Law, the elastic 
 waves will become superposed without disturbance, and will 
 penetrate the whole body without becoming weakened. If 
 we imagine the idealised body as a horizontal, infinitely 
 extended plate of finite thickness, and if we transmit a 
 powerful motion (high temperature) to the upper layer of 
 atoms, while we keep the lower layer at rest (i.e. at zero 
 temperature), then an elastic energy current (heat current) 
 will pass continually from above to below. An energy 
 gradient (temperature gradient) does not, however, exist 
 in the body, since, on account of the undamped character 
 of the wave, the mean density of energy is everywhere the 
 same. Since, in general, the conductivity for heat is equal 
 to the flux of heat divided by the gradient of temperature, 
 it follows that the idealised solid' body possesses an infinite 
 thermal conductivity. The case becomes different, however, 
 if we extend Hooke's Law in the manner described, and thus 
 pass over to the " real " solid body. The waves in the body 
 will then, on account of the departure of the equations of 
 motion from linearity, no longer pass over one another un- 
 disturbed. On the contrary, an oscillation already present 
 will, in consequence of the fluctuations in density caused by 
 it, disturb the oscillations superimposed upon it, with the 
 effect that a scattering, and therefore a weakening of the 
 waves in the body results, in precisely the same way as a
 
 THE ELECTRON THEORY OF METALS 61 
 
 " cloudy " medium scatters and weakens light passing through 
 it. Hence, in the case taken, a temperature gradient is set 
 up in the plate from the top to the bottom. In the case of 
 the real body we thus arrive at a finite thermal conductivity. 
 The mathematical development of this conception led Debye to 
 the law 162 that the thermal conductivity of crystals is inversely 
 proportional to the, absolute temperature (if we confine ourselves 
 to temperatures which are so high that classical statistics are 
 applicable). This deduction seems to be in excellent agree- 
 ment with experimental results obtained by A. Eucken. 1 * 3 
 
 13. The Electron Theory of Metals and its Modification by the 
 Quantum Theory 
 
 If matters are already complicated in the intrinsically 
 clear case of crystals, the position becomes still more 
 difficult when we turn to metals which, in general, con- 
 sist of an irregular conglomerate of crystallites. In this 
 case the conductivities, namely, of heat and electricity, are 
 particularly deceptive. According to the classical theories of 
 P. Drudew E. Riecke, and H. A. Lorentz, these pheno- 
 mena are brought about by the free conductivity-electrons, 
 which, like gas-molecules, fly about in the space between the 
 fixed atomic residues, exchange energy with these upon 
 collision, and so take part in the establishment of thermal 
 equilibrium. Thus the conduction of electricity is explained 
 as follows : in a piece of metal of uniform temperature, an 
 equal number of electrons fly, on the average, in each 
 direction through an element of surface. Hence, on the 
 average, there is no transport of electrical charges through 
 this element of surface, that is, no electric current is flowing 
 in the piece of metal. If now we apply a potential difference 
 to the ends of the metal, an electric field exists in the metal, 
 and this field impresses upon the electrons during their " free 
 paths " (i.e. their paths between two encounters with atoms) 
 a certain one-sided additional velocity which is super- 
 imposed upon the irregular heat-motion. Now, therefore, 
 more electrons will pass per second through the element of 
 surface in one direction than in the other, and since the 
 electrons carry a negative charge, and so move against the 
 field, i.e. in a direction opposite to the field, we have now an
 
 62 THE QUANTUM THEORY 
 
 electric current in the metal. The mathematical calculation 
 of this simple conception gives for the electrical conductivity 
 o- of the metal 167 
 
 N^ 
 '-2S5 ' ' ' ' (67) 
 
 Here N is the number of electrons per unit of volume, e and 
 ra charge and mass of the electrons, q their average velocity, 
 and I their free path. If we write the expression (67) in the 
 form 
 
 we may, according to the assumptions of the classical theory, 
 replace the mean kinetic energy \m(f of the electrons by %kT. 
 For since, as we assumed, the electrons take part in 
 establishing heat-equilibruim, the law of equipartition of 
 kinetic energy applies to their motion, and there is thus 
 allocated to each of the three degrees of freedom of the 
 electrons the energy ^kT. In this way we arrive at the 
 formula 
 
 . - (676) 
 
 Analogously, we get from Drude's Theory the coefficient of 
 thermal conductivity 168 
 
 . . . . (68) 
 
 so that a combination of the two formulae leads to the 
 fundamental relation 
 
 I-f -T .... (69) 
 
 which is the law of Wiedemann- Franz and of Lorenz. m It 
 states that the ratio of the thermal to the electrical conductivity 
 has the same value for all pure metals and is proportional to 
 the absolute temperature. 
 
 Thus all appeared in the best of order. The classical 
 theory appeared here also to have worked successfully and 
 the law of equipartition celebrated a triumph. But upon 
 closer inspection, gaps appeared in the apparently solid 
 theoretical structure, and serious doubts arose. For if the
 
 THE ELECTRON THEORY OF METALS 63 
 
 free electrons really took part in the thermal equilibrium, 
 and hence claimed their full share, fNfcT (per unit of 
 volume), in the equal division of kinetic energy, then this 
 share of energy should be plainly noticeable in the atomic 
 heat of the body, namely, to the extent of f N*&, where N* 
 denotes the number of electrons in a gramme-atom. Such 
 an increase in the atomic heat of the metals as compared 
 with the non-metals (which contain no, or vanishingly few, 
 free electrons) has never been observed. This difficulty could 
 be avoided by assuming that the number of electrons is small 
 compared with the number of atoms per unit volume, and 
 then their contribution to the atomic heat would be relatively 
 small. But then we should expect from (676) much smaller 
 conductivities than experiment has disclosed, unless we were 
 to assume high values, that are improbable, for the mean free 
 path, "a 
 
 Further, H. A. Lorentz^ 1 has shown, as we have seen, 
 that, if the law of equipartition for the motion of the electrons 
 is assumed, the metals would radiate in the region of long 
 waves according to Bayleigh's Law, whereas we have un- 
 questionably to expect, especially at low temperatures, the 
 radiation to take place according to Planck's Law. 
 
 The calculated dependence of the conductivity on tempera- 
 ture can only be made to agree with experience by making 
 particular assumptions at high temperatures, whereas no 
 assumptions seem to be able to make calculation and obser- 
 vation agree for low temperatures. At high temperatures 
 the resistance of the metals increases proportionally to 
 
 the temperature, that is, o- decreases with . This can 
 
 only be reconciled with (676), if the product Nlq is inde- 
 pendent of the temperature. If we assume with /. J. 
 Thomson 112 that N increases proportionately to *JT, then, 
 since q is likewise proportional to >JT, I must decrease with 
 =f a hypothesis which, as we shall see, has latterly been 
 
 upheld by several investigators. 
 
 Now, although the agreement between theory and experi- 
 ment could thus be compelled by special assumptions at high 
 temperatures, the region of low temperatures revealed itself
 
 64 THE QUANTUM THEORY 
 
 as the vulnerable point of the theory. For experiments by 
 H. Kamerlingli-Onnes m in the laboratory for low tempera- 
 tures at Leyden had shown that the resistance of metals 
 at very low temperatures (the experiments extended as far 
 down as 1'6 abs.) falls away to a quite extraordinary degree, 
 and practically disappears before the zero-point is reached. 
 At any rate, the resistance cannot, as follows in view of what 
 has just been said from formula (67&), sink proportionately 
 to only the first power of the temperature ; on the contrary 
 the fall is without doubt proportional to a higher power. 
 That the Wiedemann- Franz Law also ceases to be valid in 
 this region, has been proved by experiments of C. H. Lees 174 
 and W. Meissner. 
 
 In order to escape from all these difficulties the quantum 
 theory was appealed to, and attempts were made, in the most 
 varied ways, to make it harmonise with the existing theory. 
 A first attack was ventured by W. Nernst 176 and Kamerlingh- 
 Onnes, 111 who gave for the resistance of the metals empirical 
 formulae which linked up directly with the form of Planck's 
 energy equation (9) and which gave the change in the resist- 
 ance with temperature satisfactorily. F. A. Lindemann m 
 and W. Wien 119 conceived more detailed theories. Linde- 
 mann accepts in his first paper J. J. Thomson's hypothesis, 
 according to which N is proportional to >JT, and retains the 
 equipartition law for the motion of the electrons, so that q 
 also becomes proportional to JT. Then, according to (676), 
 
 the variation of the resistance - with the temperature depend* 
 <r 
 
 entirely on the mean free path I. But this is, according to 
 well-known considerations of the theory of gases, the greater 
 the smaller the " radius of action " of the metallic atoms ; for 
 the electrons can pursue greater paths freely, i.e. without 
 collisions, the smaller the hindrances set in their path. The 
 novel part of Lindemann's Theory is the fact that he brings 
 the radius of action of the atom into relation with its ampli- 
 tude of swing in its heat-motion. For it is at once obvious 
 that the atoms in this heat-motion will cover a greater space 
 in a given time, and their sphere of action will be the greater, 
 the larger their amplitude of oscillation, i.e. the higher the 
 temperature. Thus the mean free path also becomes a
 
 THE ELECTRON THEORY OF METALS 65 
 
 function of the temperature, inasmuch as it is brought into 
 relation with the energy of vibration of the atoms. But, for 
 the latter term, Lindemann inserts the value given by the 
 quantum theory, and finds for the resistance the formula 18 
 
 hv ' l~fc, ' ~ "" ' * ('") 
 
 F-I V^-i 
 
 where v denotes the frequency of the atoms (again the mono- 
 chromatic theory) ; A and B are constants. For high tem- 
 peratures W then becomes proportional to the temperature 
 T; for low temperatures W decreases exponentially with 
 
 hv 
 
 e 2kT to a constant value B 2 . With the help of this formula, 
 Lindemann succeeds in representing the observations quite 
 well (the formula contains two constants which can be mani- 
 pulated) ; but, since the law of equipartition has been retained 
 for the electrons, the difficulties of the excessive atomic heat 
 and of Rayleigh's radiation formula remain. Moreover, this 
 theory is unable to explain the departures from the Wiede- 
 mann-Franz Law at low temperatures; for the mean free 
 path I the only quantity dependent on T which occurs in a- 
 disappears entirely from the formula (69). 
 
 W. Wien attacked the question much more radically than 
 Lindemann. In order once and for all to get rid of the 
 contribution of the electrons to the atomic heat this is the 
 weak point of all theories which make use of the law of 
 equipartition he assumed that the electrons do not take 
 part in the thermal equilibrium, but possess a velocity q 
 which is independent of the temperature. Moreover, he 
 makes the number N of the electrons per unit volume equal 
 for all temperatures. Then, according to (67), the variation 
 
 of - with temperature is again determined only by the 
 
 dependence of the mean free path I on the temperature. 
 Wien, in a manner similar to that of Lindemann, connects I 
 with the energy of vibration of the metallic atoms, taking, 
 however, the complete elastic spectrum into account accord- 
 ing to Debye. He thus gets for the resistance the value 
 5
 
 THE QUANTUM THEORY 
 
 v m 
 
 vdv 
 W = const. I A, ... (71) 
 
 ( vdv 
 . I A* 
 
 3f--\ 
 
 For high temperatures this formula gives W = const. T, i.e. 
 proportionality with the temperature. For low temperatures 
 it follows that W =- const. T 2 , i.e. a parabolic decrease. 
 The observations are very well represented by Wien's formula. 
 But, above all, the unsatisfactory fact remains that this 
 method does not lead us on to a theory of heat-conduction, 
 unless we make new assumptions, nor to the Wiedemann- 
 Franz Law. For, by the condition that the motion of the 
 electrons takes place quite independently of the temperature, 
 Wien has taken away the possibility of also ascribing the 
 transport of heat to the electrons. 
 
 This difficulty arises again in a more recent paper of 
 F. A. Lindemann in which, in continuation of the con- 
 ceptions of Born and Kdrmdn, the supposition is introduced 
 that just as the atoms in a crystal the electrons in a metal 
 form a lattice. F. Haber 181a has also adopted a similar hypo- 
 thesis. The conduction of electricity is then explained by 
 supposing this electron lattice to move practically as a rigid 
 structure relatively to the atomic lattice and so through the 
 metal. This model has many advantages. Since, in the 
 heat-motion, in which the electron lattice naturally takes 
 part, the electronic vibrations, on account of their mass, are 
 extremely rapid (high frequency), these vibrations of the 
 electron, according to Planck's formula for the energy, make 
 no appreciable contribution to the atomic heat. In addition 
 the abnormal conductivity (supra-conductivity) which has 
 been observed at very low temperatures may, if we use 
 earlier considerations by /. Stark, 1 ** be explained without 
 difficulty by the conception that at these very low tempera- 
 tures at which the atomic space-lattice is practically at rest, 
 the electronic lattice glides almost unimpeded through the 
 gaps of the atomic lattice. 
 
 G. Borelius, 1 ** in a sketch which was recently published, 
 uses ideas similar to those of Lindemann. 
 Finally, we may refer to a paper by K. Herzfeld 1M which,
 
 THE ELECTRON THEORY OF METALS 67 
 
 in contrast to the preceding investigations, attacks the ques- 
 tion from a more phenomenological point of view without 
 making use of a particular model. For if, in the formulae 
 for o- and y, (67) and (68), we bring into evidence the energy 
 E = $mq 2 of the electrons by writing the equations thus : 183 
 
 . . . (72) 
 
 we get 
 
 as the expression which represents the Wiedemann- Franz 
 Law. Herzfeld then shows that, if we compare the result 
 with observation, the formula (73) can be made to agree well 
 with the actual measurements if we set Planck's expression 
 
 s -g~ for E. (The factor % has been introduced because 
 
 P -\ 
 
 the energy of the electrons is solely kinetic.) The values for 
 v which have to be used stand in no recognisable relation to 
 the atomic frequencies. A paper by F. v. flawer 186 works 
 along similar lines. 
 
 If we survey the whole field of the conduction of heat and 
 electricity in metals we recognise that here the last word 
 has not been spoken, and that a great deal of hard work will 
 be necessary to clear up finally the extraordinarily com- 
 plicated relationshipsi But much would doubtless be gained 
 for the theory if in future the observations, as far as possible, 
 are no longer made on crystal aggregates, but on metal 
 crystals that are pure and homogeneous.
 
 CHAPTEE V 
 
 The Intrusion of Quanta into the Theory of Gases 
 
 i. The Heat of Rotation of Diatomic Gases according to the 
 Quantum Theory 
 
 WHILE the molecular theory of the solid state thus 
 gained new nourishment from the doctrine of quanta, 
 the kinetic theory of gases could no longer be preserved from 
 the influx of the new views. W. Nernst 187 had pointed out 
 quite early that quantum-effects are to be expected in the 
 rotation of di- and polyatomic gas-molecules, and also in the 
 
 oscillation of atoms in the 
 molecule. Let us take as 
 an example the diatomic 
 gas hydrogen, the mole- 
 cule of which we may 
 picture provisionally as a 
 rigid " dumb-bell " (Fig. 6). 
 The knobs of the dumb-bell 
 are the hydrogen atoms, 
 the grip represents their 
 chemical bond. Such a 
 molecule is known to 
 possess, besides its transla- 
 tory motion (three degrees 
 of freedom), the possibility 
 of rotating about an axis at right angles to the line joining the 
 atoms (two degrees of freedom, corresponding to the two axes 
 dotted in the figure). Eotation about the line joining the 
 atoms does not if we accept Boltzmann's conception of the 
 absolutely rigid smooth atom come into play in the ex- 
 change of energy by collision and hence in the distribution 
 of energy among the separate degrees of freedom : for this 
 68 
 
 FIG. 6.
 
 HEAT OF ROTATION OF DIATOMIC GASES 
 
 rotation cannot be changed by collision. Considered fror 
 the new point of view of quanta, which rejects " rigidity " and 
 " smoothness " as an unjustified idealisation, the position is as 
 follows : The moment of inertia of the molecule relatively to 
 the line joining the atoms is extremely small compared with 
 the moment of inertia about either of the axes at right angles 
 to this line. Buk it is known that rotations which take place 
 about axes with small moments of inertia occur with much 
 greater rapidity than those about axes with large moments of 
 inertia (the same energy having been imparted in each case). 
 If, therefore, we identify the revolutions per second with fre- 
 quencies, and use the Planck energy expression for the energy 
 of rotation (which is not strictly correct quantitatively), a line 
 of argument which has already been frequently applied shows 
 us that the rotation about the line joining the atoms possesses 
 only a vanishingly small share of the energy. For the same 
 reason (high frequencies) the degrees of freedom, which 
 correspond to the vibration of the atoms in the molecule, 
 become of importance only at high temperatures. As a result 
 
 JcT 
 of all this, classical statistics gives us the value 2 = TcT 
 
 for the mean energy of rotation of the hydrogen molecule ; 
 per gram me- molecule it therefore becomes NkT = RT. 
 Hence that part of the molecular heat which arises from 
 
 rotation is equal to R, that is, about T98 ^-l , and it is in- 
 
 deg.' 
 
 dependent of the temperature. In crass contradiction to this, 
 A . Eucken 188 found experimentally that the rotation part of 
 the molecular heat of hydrogen has the value R demanded by 
 the classical theory only at high temperatures. On the other 
 hand, it gradually decreases as we pass to lower temperatures, 
 and approaches asymptotically the value zero at the abso- 
 lute zero. In the immediate neighbourhood of absolute zero, 
 hydrogen behaves as a monatomic gas. Eucken's result was 
 confirmed by experiments conducted by K. Scheel and W. 
 Heuse,w who, however, measured the values of the molec- 
 ular heat only for three temperatures (92, 197, and 239 
 on the absolute scale). This falling off of the rotational heat 
 is without doubt a quantum-effect, similar to the decrease in 
 the atomic heat of solid bodies.
 
 70 THE QUANTUM THEORY 
 
 The first attempt to calculate this phenomenon theoretically, 
 is due to A. Einstein and 0. Stern, 190 who proceeded as 
 follows : If / and v are the moment of inertia and the 
 number of revolutions per second of the molecule respectively, 
 then its rotational energy is 
 
 JSr-.(arv) . . . (74) 
 
 If, to simplify matters, we now suppose that all molecules 
 rotate with the same mean number of revolutions v per second, 
 then we can introduce for the corresponding mean energy of 
 rotation 
 
 E r = ?(2*W .... (75) 
 the theoretical quantum value 191 
 
 E r - -j^ (according to Planck's first theory) (76) 
 
 E r = A - V + - (Planck's second theory) . (77) 
 
 From (76) or (77), by combining with (75), we obtain v as a 
 
 dP 
 function of T. If finally we form -5, and multiply by the 
 
 Avogadro number N, then we get the share of the energy of 
 rotation in the molecular heat, and we see how it depends on 
 the temperature. It thus appeared that only by using the 
 expression (77) for E r would we be enabled to get a satis- 
 factory connexion agreeing with Eucken's measurements, a 
 fact which Einstein and Stern used at the time as an argu- 
 ment for the existence of a zero-point energy. It must, how- 
 ever, be emphasised that this theory can only be regarded as 
 a first attempt to find general bearings and that it does not 
 fulfil more rigorous requirements. For the Planck energy 
 formulae used, (76) or (77) are valid, as is shown by their 
 genesis, only for configurations whose frequency v is a con- 
 stant quantity independent of the temperature. Here, on
 
 HEAT OF ROTATION OF DIATOMIC GASES 71 
 
 the contrary, we have made use of a mean speed of rotation 
 v, dependent on the temperature. 
 
 P. Ehrenfest i n 1913 built up a theory of the heat due to 
 rotation on a stricter basis. He had, however, to confine 
 himself to configurations with one degree of freedom, that is, 
 to rotations of the molecule around a fixed axis, as at that 
 time an extension of the quantum theory to several degrees 
 of freedom had not yet been worked out. The expression 
 thus obtained for the heat of rotation was then, in order to 
 take into account both degrees of freedom, simply multiplied 
 by 2, a method which readily suggests itself, but is not justi- 
 fiable. Ehrenfest started, in his calculation, from the original 
 form of the quantum hypothesis, according to which the 
 energy of linear oscillators may only be whole multiples of hv, 
 and accordingly made the condition that the rotational energy 
 of a configuration with one degree of freedom (fixed axis) may 
 
 only consist of whole multiples of 75-. The factor - appears, 
 
 because the energy of rotation in contrast with the vibra- 
 tional energy of the oscillator is solely kinetic by nature. 
 The Ehrenfest condition is, therefore, according to (74) : 
 
 (n-0,1,2,3 . . .) . (78) 
 hence 
 
 vn = ^j (n = 0, 1, 2, 3 . . .) . (79) 
 
 and by substitution in (78) 
 
 (n = 0, 1, 2, 3 . . .) . (80) 
 
 Hence tJie molecules can only rotate with quite definite, discrete 
 speeds v ll} and correspondingly acquire only a series of discrete 
 rotational energies E^\ quite in agreement with the sense of 
 Planck's quantum theory. It is noteworthy that these dis- 
 crete rotational energies are related to one another as the 
 squares of the whole numbers, whereas the energies of the 
 Planck oscillators are proportional to the whole numbers 
 themselves. With the discovery of the discrete values (80) 
 of the energy, the dynamical part of the problem was solved.
 
 72 THE QUANTUM THEORY 
 
 But we require the mean energy E r of a totality of N similar 
 molecules. It is here, then, that the second, statistical part 
 of the calculation begins. If w n denotes the probability that 
 a molecule possesses the rotational energy E (n J at the tempera- 
 ture T (w n is therefore the "distribution-function " which has 
 been extended in accordance with the quantum theory), then 
 the mean rotational energy of a molecule is known to be equal 
 
 ^ ' w n . Multiplication by N and differentiation with 
 
 n = 
 
 respect to T give us immediately the heat due to rotation. 193 
 Ehrenfest thus obtained for the relationship between the 
 rotational heat and the"Hemperature a curve which could, it is 
 true, be made to agree well with the measurements obtained 
 at low temperatures by choosing the arbitrary constant / (the 
 moment of inertia) suitably, but at high temperatures it showed, 
 before reaching the classical value B, a maximum and a sub- 
 sequent minimum, which did not correspond with the existing 
 observations. 
 
 We may note here an important consequence of equation 
 (79), since it has played a noteworthy r6le in the further 
 development of the quantum theory. If, namely, we write 
 down the angular momentum (the moment of momentum 194 ) 
 of the molecule, that is, the quantity p = J x I 2irv, then it 
 follows from (79) that only the special quantum values 
 
 Pn=^ (n-0,1,2,3 . . .) . (81) 
 
 of the turning-moment exist. This relation may also be 
 deduced directly from the theory of the quantum of action as 
 formulated in (30). For, if we select as our general co- 
 ordinate, in this case q, the angle of rotation <f>, then the 
 corresponding momentum or impulse p is known to be 
 none other than the moment of momentum. 193 It follows 
 from this, since p is independent of </>, that 
 
 2irp* = rih . . . (82) 
 in agreement with (81).
 
 HEAT OF ROTATION OF DIATOMIC GASES IS 
 
 In the same way, on the basis of Planck's first theory, 
 namely, the conception that the special quantum rates of 
 rotation v n are the only possible ones, and using the dumb-bell 
 model, the author 196 has recently carried out the strict calcula- 
 tion for structures with two degrees of freedom (free axes of 
 rotation), making use of the later ideas of the quantum theory. 
 This stricter method likewise gives us curves for the rotational 
 heat which are useless, for they also have a maximum and 
 a subsequent minimum, as in Ehrenfest's case. Only by 
 making special subsidiary assumptions, such as excluding 
 certain quantum states, can we get curves which rise steadily 
 with increasing temperature, and which agree, at least to a 
 certain extent, with observation. 197 . "* . 
 
 f-r E, E r E' r E r 
 
 FIG. 7. 
 
 Not much more satisfactory results were obtained in those 
 investigations which, again with the use of the dumb-bell 
 model, were based on Planck's second theory. According to 
 this theory, the discrete values v n of the rotational speeds are 
 not the only possible ones ; on the contrary, the molecule can 
 rotate with all rotational speeds between and oo , and hence 
 can assume all values of rotational energy between and oo, 
 exactly like the Planck oscillators in Planck's second theory. 
 The peculiarity of the special quantum values (80) for the 
 energy here consists in the following : imagine the energies 
 E r plotted as abscissae (Fig. 7) and the corresponding prob- 
 abilities w as ordinates ; then a step-ladder curve results,
 
 74 THE QUANTUM THEORY 
 
 the steps of which lie exactly at the values Ety. The prob- 
 ability that a given value E r of the rotational energy will 
 appear is therefore constant within the range of energy between 
 E (r ? and JB,"* 1 * but changes suddenly at the ends of this range. 
 According to Planck's first theory, which allows only she 
 quantum values E%\ the encircled points alone have a 
 meaning. Only at those points is the probability other taan 
 zero, while all intermediate values of the energy possess the 
 probability zero, that is, do not occur. 
 
 In this case, too, the problem was first solved for one 
 degree of freedom (fixed axis of rotation). E. Holm lK and 
 J. v. Weyssenhoff 199 found, in agreement with one another, a 
 steadily rising curve for the rotational heat, which fitted the 
 observations well at low temperatures, but undoubtedly went 
 too high at higher temperatures (from about 140 abs. up- 
 wards). 
 
 But when the modern development of the quantum hypo- 
 thesis for several degrees of freedom, to which we shall be 
 introduced later, was available, a stricter calculation for 
 free axes of rotation, i.e. for two degrees of freedom, could be 
 carried out. This problem was attacked on the one hand by 
 M. Planck,* 10 on the other by Frau S. Rotszayn but was 
 treated differently in each case. Planck started with the 
 premise that this problem belongs to the category of so-called 
 " degenerate " problems. This term is to convey the follow- 
 ing: the molecule rotates, when no external forces act 
 on it, according to the doctrines of mechanics, with con- 
 stant speed in a spatially-fixed plane. The position of 
 this plane in space must, so Planck argues, be of no im- 
 portance for the statistical state of the molecule. Hence the 
 condition of rotation of the molecule in the sense of the 
 quantum theory is determined by a single quantity, namely, 
 the rotational energy. In spite of the fact, therefore, that 
 the problem is originally and naturally a problem of two 
 degrees of freedom for the position of the molecule in space 
 is determined by two angles we must, according to Planck, 
 treat it in the quantum theory as a problem of only one 
 degree of freedom. The two degrees of freedom coalesce, as 
 it were ; they are " coherent" 
 
 In contrast to this, Frau Rotszayn proceeds to turn the
 
 HEAT OF ROTATION OF DIATOMIC GASES 75 
 
 problem into a non-degenerate one by the addition of an ex- 
 ternal field, and after solving this problem, reduces the field 
 of force till it vanishes. This method which was also used 
 by the author in the paper above cited, appears to be par- 
 ticularly advantageous, when the calculation is based on 
 Planck's first theory, for peculiar difficulties arise in " degen- 
 erate " cases. Success here decides in favour of the second 
 method. For while Planck finds a curve 2 02 which rises above 
 the classical value to a maximum, and then descends asymp- 
 totically towards the value R and is therefore of no use 
 the calculation of Frau Rotszayn gives a steadily rising curve, 
 which agrees well with the measurements for lower and higher 
 temperatures ; only the value observed at T = 197 abs. lies 
 about 10 per cent too low. 203 
 
 While all the above-mentioned investigations are based on 
 the dumb-bell model, which can only be regarded as a pro- 
 visional, schematic construction, P. S. Epstein* in 1916 
 carried out the corresponding calculations for another mole- 
 cular model proposed by N. Bohr.w* This model of the 
 hydrogen molecule, to which we shall return later, is built up 
 of two positive hydrogen atoms, each of which carry a single 
 positive charge, and around the connecting line of which two 
 electrons, diametrically opposite, rotate in a fixed circle at a 
 fixed rate (see Fig. 8). Since the equilibrium in this purely 
 electrical system is determined by the play of the Coulomb 
 attractions and the centrifugal forces, and since the radius of 
 the electron is determined by a quantum condition, this model 
 possesses the advantage that all its dimensions are completely
 
 76 THE QUANTUM THEORY 
 
 fixed, so that there is no longer any question of the arbitrari- 
 ness of the moment of inertia. The " dumb-bell knobs " are 
 represented here by the two positively-charged hydrogen 
 atoms; the rotations of the molecule hitherto considered 
 would therefore correspond to those motions in which the 
 molecule rotates with a moment of inertia J" about an axis 
 at right angles to the line joining the atoms. But to this 
 there must very plainly be added the rotation of the system 
 about the axis of symmetry (i.e. the line joining the atoms), 
 which results from the extremely rapid rotation of the elec- 
 trons. The moment of inertia corresponding to this axis is, 
 in consequence of the extremely small mass of the electrons, 
 very small compared with J. The whole system obviously 
 possesses, if we regard it approximately as rigid, the properties 
 of a symmetrical top. Its motion is therefore, in consequence 
 of its own rotation about the axis of symmetry, not a rotation, 
 but instead the well-known motion, "regular precession." 206 
 Epstein treated the problem from this point of view but could 
 not obtain agreement at low temperatures with the moment 
 of inertia calculated from the model itself, 207 namely, / = 2 -82 
 x 10 ~ 41 . Presumably, this failure depends on the fact that 
 the model does not correspond with reality, and in fact we 
 shall see later, that well-founded doubts have arisen as to the 
 correctness of the Bohr model. We must therefore admit, 
 unfortunately as one of a number of instances in the quantum 
 theory, that the important problem of the rotational heat of 
 hydrogen still awaits solution. 
 
 2. The Bjerrum Infra-red Rotation-spectrum 
 
 N. BjerrumM* has applied the relation (79) in a very 
 interesting manner to the infra-red absorption of polyatomic 
 gaseous compounds. These gases (for example HC1, HBr, 
 CO, H 2 in the form of steam, but on the other hand not 
 the elementary gases H 2 , 2 , N 2 , C1 2 ) show, according to the 
 investigations of S. P. Langley, F, Paschen, H. Eubens, 211 
 H. Bubens and E. Aschkinass, H. Eubens and G. Hettner, 
 W. Burmeisterpu Eva v. Bahr,* u extensive absorption bands 
 in the short- and long- wave infra-red. While in the long- 
 wave infra-red we account for the absorption by the rotating 
 molecule, which, composed of positively and negatively
 
 INFRA-RED ROTATION-SPECTRUM 77 
 
 charged atoms, act like electric double poles and hence in 
 turning emit and absorb radiation, Bjerrum was the first to 
 point out that the molecular rotation must also make itself 
 noticeable in the short-wave infra-red. For if there exists 
 in this region a linear vibration V Q of the ions in the molecule 
 relatively to one another and hence an absorption at this 
 point and if, in addition, the whole molecule rotates at the 
 speed v r , then it is known that there will be produced as a 
 result of the composition of the vibration with the rotation 218 
 two new vibrations (and, correspondingly, two new regions 
 of absorption) having the periods v e + v r and VQ - v r , sym- 
 metrically disposed on both sides of the ionic vibration VQ. 
 On the whole, then, we have three points of absorption : 
 v r , VQ ~ VT, vo + v r , to which we must add the non-rotational 
 state VQ as a fourth. But if now, according to Planck's first 
 theory, the molecule can only rotate with discrete speeds of 
 rotation v n [see (79)], we get symmetrically to the original 
 position of absorption v = VQ and, on both sides of it, a series 
 of further discrete equidistant positions of absorption : 
 
 v = i/o + v n = v + n~-} 
 
 * \(n - I, 2, 3 . . .) . (83) 
 v = v - v n = v - n 
 
 These discrete equidistant positions of absorption have 
 actually been found by Eva v. Bahr in the case of water 
 vapour and gaseous hydrochloric acid, and were measured 
 later with still greater accuracy by H. Rubens and G. Hettner 
 for water vapour. In an examination carried out on an 
 extensive scale E. S. Imes zl1 has once more thoroughly 
 investigated the hydrogen halides (HC1, HBr, HF) and con- 
 firmed the law (83) for the position of the absorption lines. 
 It was thereby found that the middle line VQ was always 
 missing. From the standpoint of the theory here described 
 this would mean that the non-rotational state does not exist, 
 that is, that the molecules always rotate (zero-point rotation). 
 A. Eucken? 1 * who discussed the results of E, v. Bahr, 
 which were at that time the only ones known, deduced from 
 the good agreement between observation and calculation that 
 Planck's second theory is not valid, for the experiments
 
 78 THE QUANTUM THEORY 
 
 seemed so obviously to prove that the molecule can actually 
 only rotate with the discrete speeds v n . This conclusion, 
 however, is not inevitable, as Planck 219 showed in a pene- 
 trating investigation. On the contrary, the observations 
 may after all be explained, surprising as it may seem, on the 
 basis of his second theory (continuous " classical " absorp- 
 tion ; all speeds of rotation possible). This curious result is 
 explained as follows : Let w(E r )dE r be the probability that 
 a molecule possesses exactly the rotational energy E r ; hence 
 for N molecules Nw(E r )dE r will be the number that will 
 possess exactly the rotational energy E r . These molecules 
 rotate therefore according to (78) with the speed 
 
 _ 1 l2E r 
 
 "" \ </ 
 
 The quantity w(E r ) is here, according to Planck's second 
 theory, the step-ladder curve pictured in Fig. 7. Planck's 
 calculation then leads to the following result : the absorption 
 of an external radiation of frequency v* is not as one should 
 expect -proportional to the number of molecules having a 
 rotational speed v r = v*, that is, to the quantity w(E r ) but to 
 
 its differential coefficient . This differential coefficient 
 
 dE r 
 
 is, however, as Fig. 7 shows, everywhere equal to zero, 
 excepting at the special quantum energy-values E^\ that is, 
 at the rotational speeds v n . It follows from this, that here 
 also, from the standpoint of Planck's second theory, absorption 
 takes place only at the special quantum rates of rotation v. 
 It thus comes about that, at present at any rate, the infra- 
 red absorption spectra of the polyatomic gases, contrary to 
 all expectation, do not decide one of the most fundamental 
 questions of the whole quantum theory, whether, namely, 
 Planck's first or second theory is correct. An important 
 remark must be added here. The deductions of the relation 
 which gives the position of the infra-red absorption bands is 
 half in accordance with the classical and half in accordance 
 with the quantum theory. For although the rotational 
 speeds v n are determined by the quantum theory, the resolu- 
 tion of the oscillation v into the two components VQ v n 
 are determined by the classical methods. How to attack this
 
 THE DEGENERATION OF GASES 79 
 
 problem from a point of view entirely consistent with the 
 quantum theory will be seen later in Chapter VIII. 
 
 3. The Degeneration of Gases 
 
 The phenomena described above which were observed in 
 the case of polyatomic gases (falling-off of the molecular heat, 
 and infra-red absorption) justify fully the application of the 
 quantum theory to motions of rotation. On the other hand, 
 the attempts to go a step farther and to apply it to the 
 translational energy of gases rest upon a much more insecure 
 basis. If this step is taken, the hitherto exceptional position 
 occupied by the monatomic gases, whose molecules contain 
 only translational energy, becomes destroyed, for then they, 
 too, must succumb to the quantum law. This problem 
 has been attacked from various quarters [0. Sackur, H, 
 Tetrode m W. H. Keesom, W. Lenz and A. Sommerfeld,*** 
 P. Scherrer , M. Planck.&S] Thus, for example, Tetrode, 
 Keesom, Lenz and Sommerfeld imagine the thermal motion 
 of the gas split up into a spectrum of natural frequencies, and 
 they then distribute the energy in quanta, that is, according 
 to formula (9), over the individual natural frequencies, quite 
 analogously to the manner of Debye and Born-Kdrmdn in 
 the case of solid bodies. Scherrer and Planck, on the other 
 hand, apply the quantum hypothesis directly to the motion 
 of the individual gas-atoms, basing their argument on the 
 modern formulation of the quantum conditions for several 
 degrees of freedom. How such a quantum resolution of the 
 translator^ motion is effected, is perhaps most easily seen 
 by the following simple example (Scherrer) : Let a gas-atom 
 of mass ra fly to and fro in a cube-shaped space of side a 
 with the speed v parallel to one of the edges. It then 
 
 executes a sort of oscillation with the period " = o~- If we 
 set its kinetic energy, E = %mv 2 , according to Planck's first 
 
 theory = n- (n = 0, 1, 2, 3 . . .) then it follows that 
 
 h v
 
 80 THE QUANTUM THEORY 
 
 hence 
 
 Hence the velocity of the atom and its translatory energy 
 can acquire only discrete, quantum-determined values. 
 
 The calculations of the above-named investigators lead to 
 two important main results, at least in qualitative agreement ; 
 in the first place, there results an alteration in the gas laws 
 at very low temperatures. The necessity for this " degenera- 
 tion " of the monatomic gases had already been recognised by 
 Nernst, who deduced it on the basis of his new heat theorem. 226 
 For if the equation of state of ideal gases 
 
 = W 
 
 p = pressure 
 
 V = volume of a gramme-atom 
 
 IT- -prril ' .v^^-no vi u, Qiuimmv-aiw^ i , Q _, 
 
 J \R = absolute gas-constant 
 temperature 
 
 were exactly true for all temperatures down to the lowest, 
 then the maximum work A, which could be gained from 
 the isothermal expansion of the gas from the volume V l to 
 the volume F 2 , would have, as we know, for all temperatures 
 the value 
 
 v v 
 
 ' ' 
 
 = RT log, (pj). 
 
 A = 
 
 For all temperatures down to absolute zero, -Tm = R log ( -^ 
 
 would differ from zero, in direct contradiction to the condition 
 (38) of Nernst' s Theorem. Hence it follows that in the region 
 of the lowest temperatures, the equation of state (85) must 
 undergo modification. In fact, experiments of 0. SacJcurW 
 on hydrogen and helium appear to speak in favour of the 
 existence of this " degeneration." 
 
 4. The Chemical Constants of Monatomic Gases 
 
 The second main result given by the application of the 
 quantum theory to monatomic gases, is an extremely in- 
 teresting relation of the Planck constant h to the so-called
 
 CONSTANTS OF MONATOMIC GASES 81 
 
 "chemical constant" of the gas, a quantity which plays 
 an important part in changes of the state of aggregation 
 (vaporisation, sublimation) and in chemical states of equi- 
 librium. But it is here specially emphasised that the re- 
 lationship just mentioned is not bound to the undeniably 
 hypothetical resolution of the translatory energy into quanta. 
 On the contrary, 0. Stern 228 has succeeded in deducing it un- 
 objectionably, without applying the quantum tJieory to the gas. 
 The original method, which Stern adopts, may be shortly 
 sketched here. Consider the process of sublimation, i.e. the 
 passage from the solid into the vapour state. Let the 
 vapour obey the gas laws, and let its density be negligible 
 compared with that of the condensed solid. Then classical 
 thermodynamics gives for the pressure p of the saturated 
 vapour as a function of the temperature the following 
 equation : 
 
 ~ + f log T ~ T + c (86) 
 
 Here X is the heat of vaporisation (per gramme-atom) at 
 absolute zero, E^ is the energy of the condensed solid (per 
 gramme-atom) at the temperature T; the constant C, which 
 is the chemical constant of the vapourising substance, 
 remains undetermined, according to thermodynamics. On 
 the other hand, the integral on the right-hand side of 
 equation (86), which contains the energy of the solid 
 material, may be completely calculated upon the basis of 
 our assured knowledge of the energy- content of the solids. 
 We only require to assume the solid to be a Born-Kdrmdn 
 crystal, and hence to use the quantum-theoretical value (41) 
 for E ( T\ If we now restrict ourselves to high temperatures, 
 to a region, therefore, in which the classical theory is valid, 
 (86) assumes the form 
 
 log p = ----- S r- - i log T + 3 log^|) + C (87) 
 6
 
 82 THE QUANTUM THEORY 
 
 (The BN quantities v; here form the elastic spectrum of the 
 solid body ; v is their geometric mean.) The formulation of 
 this equation constitutes the first step of Stern's deduction. 
 It gives the result of thermodynamics, extended by the 
 application of the quantum theory to the condensed sub- 
 stance. The second step is the formulation, in accordance 
 with molecular theory, of a vapour-pressure formula for high 
 temperatures, in the region therefore of classical statistics. 
 Here, also, the Born-Kdrmdn solid model is used for the con- 
 densed substance, and, on the basis of known laws of 
 probability, the number of the atoms is calculated which are 
 in statistical equilibrium in the vapour phase. In this way 
 the density of the vapour, and hence, as a result of the gas 
 laws, its pressure, are given. So Stern finds 
 
 log,? - - ^ - i log T + log 
 
 Here m denotes the mass of an atom, and X' is the work 
 which is necessary to bring N atoms (N is the Avogadro 
 number) from complete rest to the gaseous state. An un- 
 determined constant naturally does not appear in this formula 
 deduced from pure molecular theory. For the molecular 
 model is completely determinate, and hence gives the absolute 
 value of the vapour pressure, not only its temperature co- 
 efficient, as in the case of thermodynamics. A comparison 
 of (87) with (88) shows, firstly, that 
 
 jp . (89) 
 i 
 and secondly, that the chemical constant C has the value 
 
 . . . (90) 
 
 Eelation (89) may be interpreted by making the supposition 
 that the solid body already possesses an energy amounting to 
 3tr 
 
 } -FT at the absolute zero, that is, a "zero-point energy," to 
 i 
 which the latent heat of vaporisation A must be added, i n
 
 CONSTANTS OF MONATOMIC GASES 88 
 
 order to set the atoms completely free from their union in the 
 crystal. Equation (90) gives us the solution of the problem 
 before us. It gives the chemical constant of the monatomic 
 gases as a function of the atomic mass and the universal 
 constants h and k. Nowhere, however, in the whole 
 deduction this should be emphasised once more has the 
 quantum hypothesis been applied to the gas itself. 
 
 In order to make formula (90) available for comparison 
 with experiment 229 we may expediently introduce the molec- 
 
 7~> 
 
 ular weight M = mN, and set k >: then 
 
 C = C + a log M, where C = log ^7 = 10'17 
 
 If we finally use the base 10 instead of the natural base e for 
 our logarithms, and measure the vapour pressure not in 
 absolute measure but in atmospheres, we get the chemical 
 constant C' used by Nernst, which is related to Sterns vai-ie 
 for C thus : 
 
 C' = xJL* = 6-0057 
 
 For it we finally get the simple expression 
 
 C' = C' + 4 Iogi M, where C' = - 1-59 . (91) 
 This formula has been brilliantly verified by experiment. The 
 hitherto most trustworthy measurements of vapour pressure 
 and chemical states of equilibrium give in the case of hydro- 
 gen, argon, and mercury the values 
 
 - 1-69 0-15, - 1-65 0-06, - 1-62 0'03 
 We are therefore justified in saying with Stern that the 
 expression (90) for the chemical constant of the monatomic 
 gases is theoretically and experimentally one of the best 
 founded results of the Quantum Theory.
 
 CHAPTEE VI 
 
 The Quantum Theory of the Optical Series. The 
 Development of the Quantum Theory for several 
 Degrees of Freedom 23 
 
 i. The Thomson and the Rutherford Atomic Models 
 
 THE greatest advance since M. v. Lane's discovery of the 
 method of Eontgen-spectroscopy for determining crystal 
 structure was made in the realm of atomic theory in 1913, 
 when the Danish physicist Niels Bohr placed the atomic 
 models in the service of the quantum theory. Bohr's labours 
 have in their turn reacted on the quantum theory and fertil- 
 ised it, and thus a marvellous abundance of notable successes 
 have been achieved in recent years through the interaction be- 
 tween the dynamics of the atom and "the quantum hypothesis. 
 Among serviceable atomic models, the one proposed by 
 J. J. Thomson long occupied a much favoured position ; accord- 
 ing to it, the electropositive part of an atom, which constitutes 
 the most important part of its mass, is supposed to be a 
 sphere of "atomic dimensions" (radius about 10 ~ 8 cms.) 
 filled with a positive space charge in the interior of which the 
 negative parts, the electrons, rest in a stable position of equi- 
 librium. This model has the great advantage of explaining 
 on purely electrical grounds the possibility of " quasi-elastic- 
 ally bound " electrons, i.e. such electrons as, being displaced 
 <from their position of rest, are drawn back into it by a force 
 ivhich is proportional to the displacement.^ 1 And it was just 
 with the help of such electrons that, as is well known, P. 
 Driide? W. Voigt** M. Planck,** and H. A. Lorentz 
 succeeded in building up large regions of theoretical optics, 
 namely, the theory of dispersion and absorption, and the 
 magneto-optical effects (magneto-rotation and Zeeman effect). 
 84
 
 THOMSON AND RUTHERFORD MODELS 85 
 
 Moreover, the Thomson atomic model was able, by following 
 the classical doctrine of the theory of electrons, to do what 
 must be demanded of every serviceable atomic model, viz. to 
 explain the emission, as a result of the oscillation of its 
 electrons, of sharp, i.e. essentially monochromatic " spectral 
 lines," the position of which, on account of the quasi-elastic 
 restoring force, 236 was independent of the intensity of the 
 excitation, that is, of the energy of the oscillations. 
 
 In three important points, on the other hand, the model 
 failed completely. In the first place no success at all, unless 
 with complicated and artificial hypotheses invented ad hoc, 
 attended efforts to deduce from Tfwmson's model the formulae 
 for the optical series, for example, the simple formula for the 
 Balmer series of hydrogen. 237 Secondly, the model could not 
 account for the division of the spectral lines in an electric 
 field as observed and closely studied by J. Stark* (Stark 
 effect), in spite of the fact that it had been found most 
 valuable, in the hands of H, A. Lorentz, for explaining and 
 calculating the Zeeman effect. 239 Thirdly, it was not in a 
 position to explain the large individual deflections, sometimes 
 exceeding 90, which, according to H. Geiger and Marsden, 
 a-particles undergo in passing through thin metallic foils. 
 For on their way through the metallic foil, the a-particles, 
 which are known to be doubly charged helium atoms, come 
 into the neighbourhood of the metallic atoms and are more 
 or less deflected from their straight paths by the electric fields 
 of the atoms. If, now, the metallic atoms were Thomson 
 atoms, the electric field of these atoms would attain its 
 greatest value at the surface of the positive sphere, at a 
 distance therefore of about 10 8 cms: from the centre of the 
 atom. For from the surface outwards the field decreases, 
 
 according to Coulomb's Law, with - while it grows from the 
 
 centre to the surface proportionately to r. Those a-particles, 
 therefore, which pass close to the surface of the positive 
 sphere, must undergo the greatest deflection. An easy ap- 
 proximate calculation shows, however, that the field at this 
 distance from the centre is far from being strong enough to 
 explain the great deflections which Geiger and Marsden have 
 observed. This weighty reason led E. Rutherford 2* 1 to set up,
 
 86 THE QUANTUM THEORY 
 
 instead of the Thomson model, a new one, which was able to 
 explain the large deflections of the a-rays. According to the 
 Rutherford atomic picture, the electropositive part of the 
 atom is compressed into an extremely small space M2 the so- 
 called nucleus. Its charge E consists in general of z positive 
 elementary charges e, so that E = ze. Here z is, according to 
 a hypothesis of van den Broek 3 the atomic number of the 
 element, i.e. the number which gives the position of the 
 element in the series of the periodic table. Thus, for example, 
 z = 1 for hydrogen, 2 for helium, 3 for lithium, and so on. 
 About this nucleus the electrons describe planetary paths, 
 that is, circles or Kepler ellipses with the nucleus as focus, 
 since the electrons are attracted by it in accordance with 
 Coulomb's Law (inversely proportional to the square of the 
 distance). 
 
 In the electrically neutral atom having the atomic number 
 z, e electrons circle round the nucleus. For example, the 
 neutral hydrogen atom consists of a singly charged nucleus 
 (E = e) around which one electron revolves in a circular or 
 elliptic path. 
 
 That this Rutherford model is actually able to explain the 
 cause of large deflections of the a-particles is seen at once ; 
 for the field-strength of the nucleus, in contrast to Thomson's 
 model, increases strongly up to the immediate neighbourhood 
 of the nucleus, in accordance with Coulomb's Law; hence, 
 if the positively charged a-particles come very close to the 
 nucleus that is, much nearer than 10 ~ 8 cms. then they are 
 exposed to the extremely powerful repulsion of the nucleus. 
 
 On closer examination, the Rutherford atomic model dis- 
 appoints us seriously : for the revolutions per second, v, of 
 the electrons depend on the energy of the system. 2 ** If, 
 therefore, we suppose, according to the classical electron 
 theory, that an electron revolving at v revolutions sends out 
 an electromagnetic radiation of frequency v, then, since the 
 system loses energy by this radiation, v must diminish cor- 
 respondingly. But this means that the atom is unable to emit 
 a sJiarp, homogeneous spectral line. 
 
 2. Bohr's Model of the Atom 
 
 It thus appears that we are obliged to reject this model at 
 the very outset. But the history of physics has decided
 
 BOHR'S MODEL OF THE ATOM 
 
 87 
 
 otherwise. With deep-sighted intuition, Niels Bohr saw the 
 possibilities of Rutherford's model and brought it under the 
 quantum theory by making three bold hypotheses.^ I n the 
 first place, he assumed that the electron (or electrons) cannot 
 revolve around the nucleus in all paths possible according to 
 the view of mechanics, but only in certain discrete orbits 
 determined by the quantum theory. If we restrict ourselves, 
 as Bohr did initially, to circular paths, then only those paths 
 of an electron are allowable from the view of the quantum 
 theory for which the moment of momentum (angular mo- 
 mentum) of the revolving electron is a whole multiple of 
 
 - , in exact agreement with the quantum condition (81) or 
 
 (82) for the rotating molecule. 
 This gives, in the simplest 
 case for the quantum paths of 
 the electrons, a discrete family 
 of concentric circles around 
 the nucleus, with radii, which 
 are related to one another as 
 the squares of the whole 
 numbers (1 : 4 : 9 : 16 : 
 ). See Fig. 9. 
 Secondly, these " allow- 
 able " orbits are stationary ; p I(Jj 9. 
 they are in a certain sense 
 
 stable states of motion. This stability is gained by making 
 the radical condition that the electron in striking contrast 
 with everything that the classical theory has taught us 
 shall not radiate when in the stationary paths. Since by 
 this " decree " the loss of energy is abolished, the electron 
 can continually revolve in such a "quantum path." That 
 there are such "non-radiating" paths in the atom, is be- 
 yond doubt. Among other things, the constancy, in time, 
 of the para- and ferro-magnetism of bodies, which is 
 generated by revolving electrons, speaks in favour of this 
 view. But how electrodynamics must be altered in order to 
 guarantee the non-radiation of the quantum paths, and only 
 of these, is a question which as yet remains unanswered. As 
 we have now abolished the "classical" radiation of the
 
 88 THE QUANTUM THEORY 
 
 atom, the actually observed emission must be accounted for 
 by a new hypothesis. Here, again in direct connexion 
 with Planck's original quantum rule, Bohr's third condition 
 takes effect : when the electron passes from one allowable 
 quantum orbit, in which the energy is W 2 , into another 
 allowable quantum path of energy W lt energy amounting to 
 Wk - W\ is radiated in the form of an energy-quantum hv of 
 homogeneous, monochromatic radiation. The frequency of 
 the radiation emitted is determined by " Bohr's Frequency 
 Condition : " 
 
 We can follow Einstein 2 * 6 in imagining the passage from 
 the state of higher energy to the state of lower energy as a 
 sort of radio-active disintegration, the occurrence of which in 
 time is determined by chance. The details of this passage and 
 the release of energy accompanying it are, however, entirely 
 obscure up to the present. 
 
 3. The Hydrogen Type of Series according to Bohr's Atomic 
 Model 
 
 However bold and unorthodox Bohr's three hypotheses 
 may have appeared, their success was surprising. If we apply 
 them to a "hydrogen type" of Rut}ierford-a,tom in which 
 a single electron revolves around a positive nucleus with a 
 z-fold charge, we get 247 for the frequencies of the spectral 
 lines, which the electron emits in passing from the ?ith to the 
 sth quantum path, the following values : 
 
 fe, m charge and mass~| 
 J of electrons I (93) 
 
 ~ 2 j \s, n whole numbers J 
 
 Nz* 
 
 If we here set z = 1 (hydrogen), s = 2, n = 3, 4, 5 . . . we 
 get in exactly the same form the empirical expression for the 
 Balmer series of glowing hydrogen M8 
 
 (n = 3, 4, 5 . . .) . (94)
 
 THE HYDROGEN TYPE OF SERIES 89 
 
 For the constant N which appears in the empirical formula, 
 the so-called Eydberg number, Bohr's Theory therefore gives 
 the expression 
 
 N=*^ .... (95) 
 
 If we use here the well-known values 
 e = 4-774 x 10 - 10 (Millikan) h = 6'55 x 10 ~ 27 (Planck) 
 
 = 1-77 . 10 7 
 me 
 
 then it follows from (95) that 
 
 N = 3-27 . 10 1S 
 
 while the empirical Eydberg number has the value 3-29 . 10 15 . 
 This striking agreement and the resolution of the Bydberg 
 number into universal constants is one of the main achieve- 
 ments of Bohr's Theory, 249 and forms a strong argument for 
 its innate power. We may say that, according to Bohr's 
 original theory, the individual lines of the Balmer series (H a , 
 Hp, H y , . . .) are emitted when the electron jumps from the 
 3rd, 4th, 5th ... orbit into the 2nd. 
 
 With this statement, however, the achievements of formula 
 (93) are not exhausted. For it includes, as we easily see, 
 further spectral series of hydrogen. Namely, if we set s = 1, 
 n = 2, 3, 4 . . . we get the ultra-violet series that was found 
 and measured by Lyman. 1 If on the other hand we set 
 s = 3, 71 = 4, 5, 6 . . . we get the infra-red Bergmann series, 
 the first two lines of which were measured by F. Paschen.** 1 
 
 The element which follows hydrogen in the Periodic System 
 is helium (atomic number z = 2). While, however, the con- 
 stitution of the neutral helium atom with its two electrons 
 is already more complicated according to the latest investi- 
 gations, the two electrons circle around the nucleus in two 
 different orbits the simply ionised helium atom, which has 
 therefore a single positive charge, is entirely " of the hydrogen 
 type ; " for it consists of a doubly-charged positive nucleus 
 around which an electron rotates. The sole difference, as 
 compared with the hydrogen atom, thus consists in the 
 doubling of the nuclear^charge, = 2. The series emitted
 
 90 THE QUANTUM THEORY 
 
 from the positive helium atom may therefore, according to 
 (93), be comprised in the formula 
 
 >-*(?-*} < 96 > 
 
 where N is again the Rydberg number as defined in (98). If 
 we here set s = 3, n = 4, 5, 6 . . . we get the so-called 
 " principal series of hydrogen " which was observed by 
 Foivler 2fl2 and very recently measured with great care by F. 
 Paschen. 3 For s = 4, n = 5, 6, 7 . . . we get the so-called 
 " second subsidiary series of hydrogen," which was observed 
 by Pickering 28* and Evans. m Both series were, before the 
 advent of Bohr's Theory, falsely ascribed to hydrogen. 
 
 A new and extremely noteworthy result of Bohr's Theory is 
 revealed, if we allow for the movement of the nucleus in our 
 calculations. For, in reality, the nucleus is not stationary, 
 but nucleus and electron revolve about their common centre 
 of gravity. By taking this fact into account we are led to 
 a slightly altered expression for the Rydberg constant. In 
 place of (95) we get the formula 
 
 in which M denotes the mass of the nucleus. It follows 
 from this that for different elements, for instance, hydrogen 
 and helium, the Rydberg constant differs somewhat and is 
 smaller for hydrogen than for helium (since M H < Mff e ). In 
 general, the value of the Rydberg constant increases with 
 increase of atomic weight tending towards a limiting value. 
 All this is in perfect agreement with the results of many 
 years of spectroscopic research. 
 
 In the same way as emission, absorption has a quantum- 
 like character, according to Bohr's model. If light, say of 
 the first Balmer line (E a ), falls upon a hydrogen atom, a 
 quantum hv of this external H a radiation is used to "raise" 
 the electron into the third quantum orbit. An amount of 
 energy hvff a is taken from the external radiation, that is, light 
 from the line H* is absorbed.
 
 STRUCTURE OF THE PERIODIC SYSTEM 91 
 
 4. The Structure of the Periodic System 
 
 Even in his earliest papers Bohr endeavoured to construct 
 for the higher elements as well (Li, Be, B, C, etc.), in con- 
 nexion with the Periodic System, suitable atomic models 
 with several rings of electrons, each occupied by several 
 electrons, in which, for example, the well-known octaves of 
 the system are reproduced by a regular arrangement of the 
 external electrons which recurs at every eighth element, 
 while the number of the electrons revolving in the outermost 
 ring is equal to the valency of the element in question. 
 
 W. Kossel 286 arrived at a similar structure of the atoms as 
 a result of a profound investigation of the formation of mole- 
 cules from atoms. Also, L. Vegard, 251 A. Sommerfeld* and 
 B. Ladenbnrg 2 s9 have constructed analogous atomic models, 
 particularly taking into account the well-known up-and-down 
 curve of atomic volumes, and using them to explain other 
 periodically varying properties (paramagnetism, ionic colour). 
 These considerations, although they are tending indisputably 
 along the right lines as far as the general principles are con- 
 cerned, are not yet firmly established in detail. 
 
 5. The Quantum Hypothesis for Several Degrees of Freedom 
 
 While the quantum hypothesis in its most primitive form 
 demonstrated in this way its innate power by entering the 
 field of atomic dynamics, it had, in doing so, gained little as 
 far as its own development was concerned. But the fruits of 
 Bohr's Theory ripened more rapidly than could have been 
 divined. Already the year 1915 brought a decisive develop- 
 ment : almost simultaneously, Planck and Sommerfeld inde- 
 pendently found the solution of a problem that had long been 
 a burning question, namely, the extension of tlie quantum 
 theory to several degrees of freedom. Sommerfeld^ retained a 
 close connexion with Bohr's Theory in attacking the problem. 
 The first main condition of this theory related to the choice 
 of " allowable " stationary orbits among all those mechanically 
 possible. According to this, as we saw, only those orbits 
 were allowed for which the moment of momentum (Impuls- 
 
 moment) p is a whole multiple of -. This may also be
 
 92 THE QUANTUM THEORY 
 
 expressed according to (81) and (82) thus : among all mechan- 
 ically possible paths, only those are allowable and stationary 
 for which the pJuise-integral fulfils the condition : 
 
 nh . (98) 
 
 In this quantum condition we are to replace according to 
 (82) the general co-ordinate q by the angle of rotation (the 
 " azimuth ") <f>, the impulse ^> by the " impulse (or momentum) 
 corresponding to <f>," namely, p^ (the moment of momentum). 
 The integration is thereby to be extended over the whole range 
 of values of the variable q, that is, in the present case, from 
 to 2*-. 
 
 In the case of the original Bohr Theory, which considers 
 only circular orbits, there naturally exists only a single 
 quantum condition, namely, that for the case q = <, since 
 the angle of rotation <f> is the only variable of the path. 
 Matters are otherwise, when we reject the limitation to 
 circular orbits, and hence take .STe^/cr-ellipses into account. 
 Then each point of the path is determined by two variables, 
 namely, by the distance r of the electron from the nucleus, 
 which is at the focus of the ellipse, and by the angle </> (the 
 " azimuth ") which r makes with a fixed direction (say with the 
 straight line, which joins the nucleus to the perihelion). In 
 this case we are presented with a problem of two degrees of 
 freedom, with two generalised co-ordinates, r and $ (polar 
 co-ordinates). The simple extension of the quantum hypo- 
 thesis by Sommerfeld now consists in setting up in this case' 
 two quantum conditions of the form (98), one for the co- 
 ordinate <, which agrees with the single quantum condition of 
 Bohr's Theory, and a new one for the co-ordinate r, so that 
 the selection of the stationary orbits is here determined by 
 the two following equations : 
 
 nh. . . (99) 
 
 n'h . . . . (100) 
 
 n and n are here whole numbers, p$ and p,. are the impulses 
 (momenta) corresponding to the co-ordinates <f> and r. 261 The
 
 THE QUANTUM HYPOTHESIS 93 
 
 integration in (100) is to be taken over the full range of 
 values of r, that is, from the smallest value r m i n (perihelion) 
 to the greatest value r max (aphelion) and back to the smallest 
 fmin. (99) is called the azimuthal quantum condition, n being 
 the azimuthal quantum number ; (100) is the radial quantum 
 condition, ri the radial quantum number. 
 
 In a corresponding manner the extension may be carried 
 out for more than two degrees of freedom. If the system has 
 / degrees of freedom, and if it is therefore characterised by 
 the / generalised co-ordinates q v q.^, q s . . . and the corre- 
 sponding impulses p v p. 2 , p 3 . . ., then the " allowable " 
 movements of the system are limited by the / quantum 
 conditions : 
 
 \Pi d( li = n i h > p-A = nji, - - \Pfdqj- = njh . (101) 
 (n v n. 2 . . . HJ- are positive whole numbers). 
 
 In every one of the / phase-integrals the integration is to 
 be performed over the full range of values of the co-ordinate 
 in question. 
 
 A difficulty, which arose here from the outset, was the 
 question as to which co-ordinates ought to be chosen for the 
 application of the quantum rule (101), or whether the choice 
 is immaterial. In general, we may characterise a system of 
 several degrees of freedom by various types of co-ordinates ; 
 for instance, we may describe the Kepler movement of the 
 electron either by polar co-ordinates r and <f>, or by Cartesian 
 co-ordinates x and y. This question is the more urgent, 
 
 when one considers that the separate phase-integrals Ip^; do 
 
 not really become constants for every choice of co-ordinates, 
 as is required by the quantum rule (101) , 262 P. S. Epstein 263 
 and K. Schwarzschild 2 ^ have solved, independently of one 
 another, this problem of the " correct choice of co-ordinates " 
 to a certain extent. Incidentally, an interesting and sur- 
 prising relation of the quantum rules (101) to a long-known 
 theorem of classical dynamics was revealed, which had been 
 propounded by Jacobi and Hamilton, and had hitherto been 
 successfully applied in celestial mechanics. Finally, quite 
 lately, A. Einstein, 26 * by modifying the expression (101), has
 
 94 THE QUANTUM THEORY 
 
 put forward a quantum hypothesis which has the advantage 
 of being independent of the choice of co-ordinates. But a 
 closer discussion of these abstract investigations would lead 
 us too far here. 
 
 The second formulation of the quantum hypothesis for 
 several degrees of freedom is due, as already mentioned, to 
 M. Planck.* It is, as it were, more cautious in its nature 
 than the more radical attack of Sommerfeld. Planck, con- 
 tinuing directly from the division of the phase-plane of linear 
 oscillators already discussed, starts from the so-called Gibbs 
 phase-space to deal with more complicated systems. For a 
 system of /degrees of freedom, which is characterised by the 
 co-ordinates q v q. 2 . . . qy and the impulses p lt p. 2 . . . pf, 
 the Gibbs phase-space is that 2/ dimensional space, the points 
 of which possess the 2/ 1 co-ordinates q 1 . . . p/. Each point 
 of the phase-space (phase-points) represents, therefore, a 
 definite momentary state of the system in question. Planck 
 now gives this phase-space, in exact analogy to the phase- 
 plane, a cellular structure, by bringing into prominence 
 certain specially distinguished boundary surfaces. At the 
 same time the size of the cells is proportional to h f . The 
 points of intersection of those boundary surfaces then repre- 
 sent the distinctive quantum states or phases of the system 
 (that is, according to Planck's first theory the only possible, 
 the "allowable" conditions). In contrast with Sommerfeld' s 
 Theory, in which the motion of a system of / degrees of 
 freedom is always determined by / quantum conditions, in 
 Planck's, under certain circumstances, the case may occur 
 that fewer quantum conditions than degrees of freedom exist, 
 so that several (" coherent ") degrees of freedom are limited 
 by a single quantum condition. 
 
 6. Sommerfeld's Theory of Relativistic Fine-structure 
 
 That these theories had found the kernel of the matter was 
 soon to be shown by applying them to Bohr's atomic model. 
 According to them from among all the mechanically possible 
 paths, which the electron can describe about the -fold 
 positively charged nucleus, the allowable, stationary paths 
 must be determined by the two quantum conditions (99) and 
 (100). This gives, in place of the discrete, quantised circles
 
 RELATIVISTIC FINE-STRUCTURE 95 
 
 of Bohr, discretely quantised Kepler ellipses, among which 
 also the Bohr circles are included, as special cases. And 
 further, the ellipses are quantum-determined, both with re- 
 ference to their sizes (i.e. to their major axes), and to their 
 form (i.e. the relation of the axes to one another), so that here 
 every orbit, as compared with Bohr, is characterised by two 
 quantum numbers n and w'. 267 In place of formula (93) for 
 the hydrogen type of series, we get the general formula : a68 
 
 v = Nzf, _ J_ _ 1 "I (102) 
 
 L(s + s'Y (n + nj] 
 
 Here again N, the Rydberg constant, is given by (95), or 
 more exactly (the motion of the nucleus being taken into 
 account) by (97) ; s and s' are the two quantum numbers 
 (azimuthal and radial) of the final orbit of the electron ; n 
 and n' are the quantum numbers of its initial orbit. Since 
 also, as a result of this more complete view of Sommerfeld, 
 the number of allowable orbits is greatly increased, as com- 
 pared with those arising from Bohr's Theory (owing to the 
 addition of the ellipses), the electrons have a great many 
 more possibilities in passing from one orbit to another, that is, 
 the chances of generating spectral lines are multiplied. But 
 we easily recognise the following fact : if we choose as the 
 final orbit of the electron any one of those orbits, for which 
 the sum of the quantum numbers s + s' has a definite value, 
 say s + s' = 2, and as initial orbit, any one of those paths, 
 for which n + n' has a definite value, say n + n' = 3, then 
 all the different transitions of the electrons from any one of 
 these initial orbits to any one of these final orbits generate 
 always the same line (in the case of the figures above chosen 
 it will be the first Balmer line) ; for according to (102) the 
 frequency of the line emitted depends only upon the sum 
 s + s', and the sum n + n', and on the other hand not on the 
 separate values of s, s', n, n'. It would thus appear as if 
 nothing is gained physically by Sommerfeld' s elaboration of 
 the theory as compared with Bohr's original theory. How- 
 ever, as Bohr had already pointed out, the calculations are 
 incomplete in one important respect, which become of funda- 
 mental importance when consistently taken into account, 
 and which represents the main achievement of Sommerf eld's
 
 96 THE QUANTUM THEORY 
 
 theory of spectral lines. Namely, the velocities of the 
 electrons, which appear in these problems, cannot be con- 
 sidered negligibly small compared with the velocity of light. 
 In this case, however, we cannot, as we know, calculate by 
 the methods of classical mechanics, which regards the mass 
 of the electron as constant, but must take our stand upon the 
 theory of relativity, and hence take into account the variations 
 of the mass of the electron with its speed. Sommerfeld com- 
 pleted the calculation in this respect. The paths of the 
 electron and the nucleus differ, in this refinement of the 
 theory, from the ordinary Kepler ellipse in that the perihelion 
 of the orbit advances in the course of time, and that the path 
 loses its closed character. This has the effect that the energy 
 of the electron in the stationary quantum-chosen orbits which 
 here also are determined by (99) and (100) are no longer 
 solely dependent on the sum of the quantum numbers as in 
 the case of the non-relativistic Kepler motion, but that the 
 quantum numbers n and n also enter, separately, into the 
 expression for the energy. Only as a first approximation, 
 therefore, i.e. when the relativity correction is neglected, 
 will the frequency v of the spectral line emitted depend on 
 the quantum sums s + s' and n + n' alone, as (102) shows. 
 If we take into account the relativistic change of mass of the 
 electron, on the other hand, v will also depend on the 
 individual values of s, s', n, w'. 269 It follows, therefore, that 
 the various possibilities, above considered, of the generation of a 
 definite spectral line, that is, the passage of an electron from 
 any one of the initial orbits s + s' = constant to any one of the 
 final orbits n + n' = constant, no longer produce exactly the 
 same line, but give rise to slightly different lines, which, how- 
 ever, on account of the smallness of the relativity effect, lie 
 very close together. This is Sommerfeld' s explanation of the 
 fine-structure of the spectral lines in the case of the hydrogen 
 type of spectra. For example, according to Sommerfeld, the 
 first line of the Balmer series (the red hydrogen line H a ) must 
 consist of five components, which are arranged in two chief 
 groups (of two and three each). The mean distance of these 
 two groups from one another should amount, according to the 
 theory, 270 to about 0'126A ; the best measurements of the 
 hydrogen doublet gave the value 0'124A (Paschen, Mciasner).
 
 HIGHER ELEMENTS 97 
 
 If this agreement already speaks strongly in favour of Sommer- 
 f eld's Theory, the exact measurements, by F. Paschen, of the 
 fine-structure of the lines of positive helium (Fowler series) have 
 given a still more convincing proof of its correctness ; almost 
 without an exception, all the components required by the 
 theory of the fine-structure appeared on the photographic plate, 
 and thus proved strikingly the existence of the stationary paths 
 of the electron and its relativistic change of mass. 
 
 Two interesting consequences may yet be mentioned here ; 
 they are directly connected with Sommerfeld's Theory and 
 Paschen's observations. First of all they have rendered 
 possible the use of the fine-structure measurements for a 
 direct " spectroscopic " determination of the three funda- 
 mental constants e, m (mass of the electron at infinitely low 
 speeds), and h.^ 2 Secondly, K. Glitscher was able to 
 show that we only find the spectroscopic observations, for 
 example, the size of the hydrogen doublet, in agreement with 
 the theory, when we use for the variation in the mass of the 
 electron the formula given by the theory of relativity. On 
 the other hand, Abraham's Theory of the rigid electron leads 
 to formulae which do not agree with experiment. 
 
 7. Higher Elements 
 
 We thus see that Rutherford's atomic model as further 
 developed by Bohr and Sommerfield far exceeded the ex- 
 pectations which it could reasonably be expected to fulfil. At 
 any rate, it has revealed to us the optical series of hydrogen 
 and helium with undreamed-of precision as far as the finest 
 details. But beyond these primary gains, it has undertaken 
 a further series of successful attacks. Thus Landew* was 
 successful in calculating the two series-systems of neutral 
 helium (helium and parhelium) by taking, in contra- 
 distinction to Bohr, a model of the neutral helium atom in 
 which the two electrons circle around the double positive 
 nucleus in two different orbits, either co-planar or else 
 inclined at an angle to one another. In this case then, the 
 external electron, the leaps of which generate the radiation, 
 moves in a field in which the simple Coulomb Law no longer 
 holds, on account of the disturbing influence of the inner 
 electron. Examples of this type which differ from that of 
 7
 
 98 THE QUANTUM THEORY 
 
 hydrogen have been generally investigated by Sommp.rfeld, 
 who has shown 273 that by giving up the Coulomb field we 
 arrive, to a first and second approximation, at the Bydberg 
 and Eitz forms of the series laws. A very promising 
 beginning in setting up a quantum theory of the spectral 
 lines was thus made. 
 
 8. The Stark Effect and the Zeeman Effect in Bohr's Theory 
 of the Atom 
 
 Under the circumstances the question forces itself upon us, 
 whether the atomic model in its present state of development 
 is able to account for the Stark effect, that is, the splitting up of 
 the spectral lines as a result of the action of an external electric 
 field on the electrons emitting the lines. For, as we may 
 remember, the original TJwmson model had completely failed 
 just at this point. And how do matters stand as regards the 
 Zeeman effect, the splitting up of spectral lines as a result of 
 an external magnetic field? Could the new model explain 
 these phenomena as well as the old ? Both questions have 
 fortunately been answered in the affirmative. As regards the 
 Stark effect, P. S. Epstein, in an important paper, succeeded 
 in demonstrating the following : if we calculate the motion of 
 the electron under the influence of the nucleus and the 
 external field, according to the methods usual in celestial 
 mechanics, and then choose from among all mechanically 
 possible motions the allowable stationary orbits by applying 
 the modern quantum rules for several degrees of freedom, and 
 if, thirdly, we allow the electron to leap from one of these 
 stationary paths into another (whereby we limit the infinite 
 number of possible passages by a "principle of selection" 
 presently to be discussed), then the Bohr frequency formula '(92) 
 gives with the most admirable accuracy and completeness, both 
 as regards position and number, all the components of the 
 resolved lines as observed by Stark in the cases of hydrogen 
 and positive helium. This astonishing result must be re- 
 garded as a further strong support of the correctness of 
 Bohr's model and its system of quanta. The theory of the 
 explanation of the Zeeman effect has up to the present not 
 been quite so successful, It is true that Debyc and
 
 SELECTION OF RUBINOWICZ AND BOHR 99 
 
 Sommerfield 278 have been able to derive the normal Zeeman 
 effect (division of the original line into a triplet when the 
 line of observation is perpendicular to the lines of force) by 
 calculation from the model. The explanation, however, of 
 two important phenomena in this field has not yet been 
 accomplished : firstly, the anomalous Zeeman effect and its 
 laws (Runge-Preston rule), and secondly, the fact, discovered 
 by Paschen and Back, 1219 that even in the case of lines with a 
 complicated fine-structure, the normal triplet is formed as 
 the magnetic field grows. Further investigation will, it may 
 be hoped, unravel those difficulties. 
 
 9. The Principles of Selection of Rubinowicz and Bohr 
 
 Inasmuch as the foregoing considerations deal only with 
 the position of lines in the spectrum, i.e. with their frequency, 
 we are still confronted with the problem of their form of 
 vibration, i.e. their intensity and polarisation. Moreover, the 
 important question had yet to be answered, whether all leaps 
 of the electron from any one stationary path to any other 
 are possible, or whether the number of allowable transitions 
 must be limited by some " principle of selection." This 
 also is, fundamentally, a question of intensity, for the position 
 may be regarded as follows : the forbidden transitions corre- 
 spond to zero intensity. The solution of this whole complex 
 of problems has been greatly advanced quite recently. In 
 the first place, A. Rubinowicz, & by applying the law of the 
 conservation of the moment of momentum (impuls-moment) 
 to the system atom + radiated wave, arrived at a principle 
 of selection and a rule of polarisation of the following form : 
 in atoms of the hydrogen type, which are removed from the 
 influence of external fields of force, the azimuthal quantum 
 number n of the electron [see formula (99)] can only alter by 
 0, +1, or - 1, when emission takes place. In the first case, 
 the light radiated is linearly polarised, in the two other 
 cases circularly. The position of the plane of the orbit 
 remains unchanged during the process of emission. In 
 the case of atoms differing from the hydrogen type, and 
 of more complicated structure, the position is less simple; 
 if we set the total moment of momentum of all the 
 masses forming part of the system (we know that this
 
 100 THE QUANTUM THEORY 
 
 impulse remains constant during the motion), equal to a 
 whole number, n*, times ^, it is just the changes in this 
 
 number n* during the emission which must be limited by the 
 principle of selection in the same manner, as, in the case 
 above, the alterations in the azimuthal quantum number of 
 the individual electron in its leaps were limited. Here also, 
 zero change in the azimuthal quantum number gives linear 
 polarisation, changes by + 1, on the other hand, lead to 
 circular polarisation. In place of the orbital plane we get 
 the " invariable plane " (at right angles to the total moments 
 of momentum or impulse-moments), the position of which in 
 space remains unaltered. If, finally, the atom is exposed to 
 an external field, say a homogeneous electric field (Stark 
 effect) or a homogeneous magnetic field (Zeeman effect), then, 
 as we know, only that component of the total turning 
 impulse remains constant during the motion of the masses 
 forming parts of the atom which is parallel to the external 
 
 field. If we set these components of impulse = n^, then 
 
 only the alteration of this number n will be limited by the 
 principle of selection (that is, the alterations must be Q l 1). 
 The principle of selection is thus clearly weakened in its 
 action by the external field, and can, if fields of irregular 
 strength and direction act on the atom, become completely 
 illusory, as, for example, in the case of electric discharges. 
 
 By means of entirely different considerations, N. Bohr 281 
 arrived at results which coincide, in essentials, with those of 
 Eubinowicz, but exceed them greatly in range. Bohr started 
 from the fact that in the limit for large quantum numbers, 
 when the successive stationary states of the atom differ very 
 little in the energy they involve, the frequency that the 
 electron emits in its passage between neighbouring states 
 becomes identical with the rate of revolution in the stationary 
 orbit. 282 The electron therefore emits, according to Bohr's 
 frequency condition, the same line that it sends out accord- 
 ing to the classical theory of electrons. In other words, for 
 very high quantum numbers, the quantum theory passes over 
 into the classical theory. (Bohr's "Principle of Correspon- 
 dence or Analogy.") Arguing from this principle, Bohr pro-
 
 SELECTION OF RUBINOWICZ AND BOHR 101 
 
 ceeds as follows : according to classical mechanics, the motion 
 of the electron in Bohr's atom may be represented as the super- 
 position of component harmonic vibrations of the frequency : 
 
 "kl = Tl<l + T 2 0> 2 + . . . + TjWf . . (103) 
 
 Here, T I . . . T/ are whole numbers which in general may 
 have all values between oo and + oo ; the o^ . . . ay are 
 certain constants which depend on the character of the 
 motion : / is the number of degrees of freedom. Let the 
 amplitude of the partial vibration characterised by the 
 numbers TJ to T/ be A T i . . . A r f. Then, according to classical 
 electrodynamics, vki is the frequency of the radiated partial 
 wave (T! . . . T,) and A^ . . . A*f is a measure of its in- 
 tensity. On the other hand, the following result is derived 
 from the quantum theory (Bohr's frequency formula) for high 
 quantum numbers : in the transition from an initial state 
 characterised by the quantum numbers m v m 2 . . . w/ into a 
 final state corresponding to the quantum numbers n^ . . . Hf, 
 a line of frequency 
 
 VQU = (% - n^ + (ra 2 - w 2 )w 2 + . . . + (m f - n f )u>f . . . (104) 
 
 is emitted. Here the quantities <o 1 . . . <D/ are the same 
 constants as in (103). But, according to Bohrs Principle of 
 Analogy, for high quantum numbers vki = VQ U . Hence there 
 follows from a comparison of (103) with (104) 
 
 T i = m i ~ n i> T 2 = m 2 ~ n 2 > . . . T/ = ra/ - w/ . . . (105) 
 i.e. the " classical " partial vibration (r { . . . T/) corresponds to 
 that quantum transition, in which the quantum numbers alter 
 by exactly TJ . . . T/. The polarisation and intensity of the 
 wave emitted during this q^lantum transition may be calculated 
 from the form of vibration and amplitude of the " corresponding 
 classical" partial oscillation. This principle which has been 
 derived for high quantum numbers is extrapolated by Bohr 
 with great boldness over the region of all quantum numbers. 
 Thus the important " principle of correspondence " is obtained. 
 If in the development of the electronic motion in terms of 
 partial vibrations the term (r lt r 2 . . . ?/) is missing, then 
 the corresponding transition
 
 102 THE QUANTUM THEORY 
 
 is not present. Hence there follows, for example, for atoms 
 of the hydrogen type in a field free from force, the law that 
 the azimuthal quantum number can in all emissions only change 
 by + 1 or 1, both of which lead to circularly polarised 
 radiation. This law is somewhat more limited in form than 
 that of Rubinowicz. 
 
 Both the principles of selection and the rules for the 
 polarisation and the intensity have stood the test of compari- 
 son with experiment. Bubinowicz himself showed that his 
 principle of selection and the rule of polarisation are in agree- 
 ment with Paschen's measurements of the fine-structure of 
 the helium lines, and further with the observations of the 
 Stark effect and the normal Zeeman effect. P. S. Epstein 283 
 and H. A. Kramers* went still further, and were able to 
 prove by profound investigations, based on Bohr's Theory, that 
 the calculations of intensity along the lines sketched above 
 were also in surprising agreement with observation. Finally, 
 Sommerfeld and Kossel 283 in an interesting study have applied 
 the Rubinowicz principle of selection to spectra differing from 
 the hydrogen type as well, and have shown that it is able to 
 explain why certain series appear more readily and are more 
 favoured than others, as it were, and that, by the selection of 
 the " possible " transitions, it sets a limit to the multiplicity 
 of possible combinations in a manner which, so it appears, 
 entirely agrees with experience. 
 
 10. Collision of Electrons on the Basis of the Bohr Atom 
 
 While in this way, through the interpretation and unravell- 
 ing of the universe and the almost bewildering abundance of 
 spectroscopic observations, the conviction of the correctness 
 of Bohr's atomic model deepened more and more, a series of 
 observations of quite another kind became known and contri- 
 buted considerably to the consolidation of Bohr's Theory. 
 These were the investigations already mentioned earlier in 
 connexion with the light-quantum hypothesis, which dealt 
 with the collision of free electrons with gas molecules and 
 atoms. These researches were conducted particularly by 
 J. Franck and G. Hertz 28fl and, in succession, by a considerable 
 number of American investigators in a systematic manner. 
 The manifold results of these interesting researches may be
 
 COLLISION OF ELECTRONS 
 
 103 
 
 sketched here schematically by a simple example. What 
 have we to expect when electrons collide with a Bohr Atom ? 
 As a simple type of Bohr atom, let us choose a model in which 
 z electrons revolve around a 2-fold positively charged nucleus 
 in stationary quantum paths. The nature and spatial arrange- 
 ment of these paths, as well as the distribution of the electrons 
 among the individual paths will be left open, and we shall 
 
 FIG. 10. 
 
 make only the simplifying assumption that one electron the 
 so-called valency electron revolves alone in the outermost 
 orbit (1) (see Fig. 10). Let this be the " normal," unexcited 
 state of the atom. The hydrogen atom (z = 1) is, as we know, 
 constituted in this way, and, of the neutral complicated atoms, 
 the atoms of the vapours of the alkali metals (Li, Na, K, Kb, 
 Cs) very probably also fall under this scheme. If by any 
 addition of energy the electron is " raised " from its normal
 
 104 THE QUANTUM THEORY 
 
 orbit (1) to a higher orbit (that is, one having more energy), 
 say into the orbit (2), (3), (4) and so forth, and if it "falls" 
 from these back into orbit (1), then the 1, 2, 3 . . . line of the 
 so-called " Absorption-series of the unexcited atom " (principal 
 series) is emitted. The frequencies of the lines emitted are 
 regulated by Bohr's frequency condition (92), i.e. that the loss 
 of energy W n W l incurred in passing from the nib. to the 
 first orbit is equal to a quantum hv n>l of the line emitted : 
 
 Wn-W^hv^. . . . (106) 
 
 The additional energy required to " raise " the electron to the 
 higher energy level can be obtained in two ways : firstly by 
 absorption of external radiation ; secondly (and that is the case 
 we are dealing with here) by electronic impact. If external 
 radiation of frequency v n<l falls upon the atom, a quantum 
 hv n ,i of this radiation is absorbed and is used to raise the 
 electron from the energy level W l to the higher level W n 
 = W l + hv n ,i- ^ n foiling from this to the original level, the 
 electron then emits the light corresponding to the line absorbed. 
 The circumstance is further noteworthy, that the electron, 
 when it is raised to the level (2), has no other choice than to 
 return to the initial level, whereas from orbit (4) it can make 
 one of three possible transitions to (3), (2), and (1). If, 
 therefore, the atom has absorbed light corresponding to the 
 line j/ 2a from the external radiation, it will re-emit this line 
 with its full complement of energy. The first line of the 
 absorption series is, therefore, in contrast with all other lines, 
 a so-called resonance line. 
 
 If the energy required -to raise the electron is furnished by 
 tfce impact of an outside electron, then as Franck and Hertz 
 were the first to prove the intruding foreign electron will be 
 reflected from the atom perfectly elastically (according to the 
 mechanical laws of elastic impact), as long as its energy 
 remains below a certain critical value ER. If this energy 
 value is reached, the impinging electron loses all its energy, 
 and gives it up to the electron of the atom which has been 
 struck (" inelastic impact "). What does this mean according 
 to Bohr's view of the atom ? Obviously ER is nothing other 
 than W 2 - W v that is, the energy which is necessary to raise 
 the electron from its normal state in the atom to the orbit (2).
 
 COLLISION OF ELECTRONS 105 
 
 The result of this electronic impact, which adds energy of 
 amount ER to the atom must therefore be the emission of 
 the resonance line. If this view represents the kernel of the 
 matter, then the energy ER must be connected with the 
 frequency i/ 2n of the resonance line by the quantum relation 
 
 E R = hv^ .... (107) 
 
 This relationship has been excellently verified by experiment. 
 Thus Tate and Footed for example, find in the case of 
 sodium, that the first inelastic electronic impact takes place 
 when the impinging electron is accelerated by a potential 
 of VR = 2-2 volts, the so-called resonance potential. The 
 energy communicated by this potential to the impinging 
 electron is 
 
 On the other hand, the resonance line that is under con- 
 sideration here is the D-line, hence 
 
 We thus see that the relation (107) is fulfilled with great 
 accuracy. The same holds for potassium (Vp. = T55 volts, 
 .-. E R = 2-47 - 10- 12 , X 21 = 7-685 10~ 5 /. hv zl = 2-55 - 1Q- 12 ). 
 In the case of the inert gases (helium, neon, etc.) and the 
 vapours of mercury, zinc and cadmium, similar qualitative and 
 quantitative relations with some modifications occur. The 
 excitation, by electronic impact, of the mercury resonance line 
 A. = 2-536 10 - 5 , that is 2-536.4, discovered by Franck and 
 Hertz, and already referred to, presents a charactei-istic 
 example. The observed resonance potential is here 4-9 volts, 
 while from the relation 
 
 v 300 p 300, SQQhc 
 
 VR = - &R = - AT.* = 
 
 e e eAj,! 
 
 the value VR = 4-86 volts is deduced. 
 
 If the energy of the impinging electron is increased beyond 
 ER, then an "inelastic" impact, accompanied by complete 
 loss of the energy, is to be expected every time as soon as E
 
 106 THE QUANTUM THEORY 
 
 has become equal to W n - W l (n = 3, 4, 5 . . .). By these 
 various additions of energy the electron attached to the atom 
 is raised successively to the 3rd, 4th, 5th . . . level of energy. 
 If, finally, E = E& = W& - W v then the energy of the 
 impinging electron is just sufficient to remove the electron 
 attached to the atom to infinity, i.e. to ionise the atom. E^ 
 is thus the ionisation energy, and the voltage corresponding 
 
 SOOT' 1 
 to it, FOQ = 9P_ ( i s called the ionisation potential. From 
 
 the relation (106) we get immediately the important equation 
 
 E * = e m =/lv ' 1 (108) 
 
 That is to say, the ionisation energy is equal to the quantum 
 which corresponds to the last line of tJie absorption series, that is, 
 to the " series limit." This quantum relation has also been 
 excellently confirmed in all cases. For sodium, for example, 
 Tate and Foote found : FQQ = 5-13 volts, which gives an 
 ionisation energy of the value E^ = 8'17 10~ 12 . On the 
 other hand, the limit of the principal series has the wave- 
 length AQO x = 2-413 10- 5 , from which h v<x > 1 = 8-14 - 10' 12 , 
 in striking agreement with the value of E^ . 
 
 For mercury vapour, the limit in question of the principal 
 series A.^ = 1-188 10 - 5 . From this follows, according to 
 (108), FOQ = 10-4 volts while the measurements of various 
 workers gave the value 10*2 to 10'3 volts (Tate, Bergen, 
 Davis and Goucher ; Hughes and Dixon ; Bishop 6 ). From 
 all these examples, which could be considerably multiplied, 
 the conclusion may be drawn with convincing clearness that 
 the Bohr conceptions have laid bare the nature of the con- 
 struction and the mode of action of the atom with un- 
 precedented lucidity. 
 
 11. Einstein's Deduction of Planck's Law of Radiation on the 
 Basis of the Bohr Atom 
 
 Under these circumstances the suggestion naturally arises to 
 refound the law of black-body radiation by taking as the ele- 
 mentary absorbing and emitting structure Bohr's model in 
 place of the linear oscillator used by Planck. Einstein 289 has 
 taken this step. In a highly important study he investigated
 
 PLANCK'S LAW OF RADIATION 107 
 
 the equilibrium of energy and momentum between black-body 
 radiation and a generalised Bohr model, which, stripped of all 
 special properties, has only to fulfil the quantum condition of 
 being able to assume a discrete series of different states. For 
 the interaction between the radiation and the atom absorption 
 (Einstrahhing) and emission (Ausstrahlung) Einstein intro- 
 duces the following simple hypotheses : the frequency of the 
 emissions, i.e. the transitions, accompanied by loss of energy, 
 of the atom from a condition (2) of higher energy, E%, to 
 a condition (1) of lower energy, E v shall follow the same 
 statistical law as that which governs the disintegration of 
 radioactive bodies, i.e. the number of transitions 2 -> 1 in the 
 time dt, or, as we may say, the number of atoms (2) that " dis- 
 integrate " in this time is proportional to dt N v where N% 
 denotes the number of atoms momentarily in the state (2). 
 
 But, according to Einstein, a different law regulates the 
 processes called into existence by the effect of external radi- 
 ation. Under the influence of external radiation two things 
 may happen : either an atom may pass from state (1) to state 
 (2) by taking up energy, this is the " proper positive absorp- 
 tion." Or the case may also occur, that, as a result of the 
 phase-relation between the field of the external radiation and 
 the atom, the atom loses energy through the action of the im- 
 pinging radiation, and hence passes from state (2) to state (1) 
 ("negative absorption"). The rate at which both kinds of 
 transition are repeated is then proportional to the intensity 
 K v of the external radiation : the number of transitions 1 -> 2 
 associated with positive absorption in the time dt is therefore 
 proportional to N-^dtK.,, ; the number of transitions 2 - 1 as- 
 sociated with negative absorption is proportional to N 2 dtK. v . 
 Here N^ is the number of atoms momentarily in the state (1). 
 Nj_ and N 2 are determined by the laws of distribution known 
 from the theory of gases and statistical mathematics and en- 
 larged in conformity with the quantum theory. There follows 
 from the energy equilibrium between in-coming and out-going 
 radiation at the temperature T 
 
 . . . (109)
 
 108 THE QUANTUM THEORY 
 
 where k is Boltzmann's constant, and A is a constant inde- 
 pendent of the temperature. From Wieris Displacement Law 
 (4) it follows, firstly that A is proportional to v 3 and secondly 
 that E 2 - E! is proportional to v. If, therefore, we write 
 
 E 2 - E l = hv . . . . (110) 
 
 we recognise in this expression Bohr's frequency condition (92). 
 In this way K,, assumes the form of Planck's Law of Eadia- 
 tion, arising in a surprisingly simple and elegant manner from 
 a minimum of hypotheses of a general character. Einstein, 
 in pursuing and deepening these conceptions by writing down 
 the expression for the equilibrium of the momenta in addition 
 to the energies of the in-coming and out-going radiation, was led 
 to the remarkable conclusion that the radiation of Bohr atoms 
 cannot take place in spherical waves, as the classical theory 
 of electrons requires, but that the process of emission must 
 have a particular direction like the shot from a cannon. We 
 cannot fail to recognise that this brings the conception that 
 radiation has a quantum-like structure (light-quantum hypo- 
 thesis) within realisable bounds.
 
 CHAPTEK VII 
 The Quantum Theory of Rontgen Spectra 
 
 i. The Analysis of Rontgen Spectra 
 
 T) AEALLEL with the development of the science of optical 
 [ spectra, a theory of Eontgen spectra has been developed of 
 late years upon the same basis. This theory has already shed 
 much light on the structure of atoms and thus forms a 
 desirable extension of the theory of optical spectra. The 
 investigations of Ch. Barkla, W. H. and W. L. Bragg, Moseley 
 and Darwin, Siegbahn and Friman,' 290 among others, have 
 shown that by the impact of cathode rays upon the anti- 
 cathode of a Eontgen tube two kinds of Eontgen rays arise : 
 first, the so-called " impact radiation " (Bremsstrahlung) con- 
 sisting of an extensive and continuous range of wave-lengths 
 (similar to the continuous background of visible spectra) ; 
 secondly, the " characteristic radiation," a typical line- spectrum, 
 the structure of which depends so essentially on the material 
 of the anti-cathode that a glance at this spectrum suffices us 
 to deduce immediately and unmistakably the nature of the 
 material of which the anti-cathode is composed. Thus along- 
 side the optical spectrum analysis of Bunsen and Kirchhoff a 
 Eontgen- or X-ray analysis presents itself. It has further 
 been shown that the characteristic X-ray spectrum is a 
 purely atomic property, and, indeed, an additive one. If we 
 examine, for example, the X-ray spectrum, which is emitted 
 by an anti-cathode of brass (copper + zinc), we find the 
 lines of both copper and zinc unaltered and occupying the 
 same positions as if only one metal were present in turn. No 
 new lines appear. Accordingly we are led to suppose that 
 the line-spectrum arises in the atoms of the anti-cathode, and 
 is generated there by the impinging electrons of the cathode 
 109
 
 110 THE QUANTUM THEORY 
 
 rays. The further important fact appeared that the lines of 
 the characteristic spectrum may be arranged in series, just 
 like those of the optical spectrum. Thus we have discovered 
 up to the present a short-wave ^-series, a long- wave .L-series, 
 and a still longer-wave M-series. 
 
 The most curious feature of these spectra is their connexion, 
 by a definite law, with the atomic number of their element in 
 the periodic system. If we plot the position of a certain line 
 (say the first line K a of the ^-series) for the successive 
 elements of the periodic system, a perfectly regular progres- 
 sive shift is revealed : the line advances with increasing 
 atomic number steadily towards the shorter waves. The re- 
 gularity of this advance is such that we can recognise gaps or 
 false positions of elements in the periodic system immediately 
 by an excessive jump. Now, according to the hypothesis, 
 already mentioned, of Eiitherford, v. d. Broek, and Bohr, the 
 atomic number of an element is nothing other than the 
 number of its nuclear charge, that is, the number of elemen- 
 tary positive charges of its nucleus. If to this we add the 
 phenomenon just discussed, according to which the steady 
 advance of the nuclear charge in the series of the elements is 
 reflected in the steady displacement of the X-ray lines, then 
 we are forced to the view that the origin of tlie X-ray spectra 
 must be localised in the immediate neighbourhood of the nucleus, 
 that is, in the inmost part of the atom. For in this region the 
 nucleus clearly has the greatest power and is least disturbed 
 by external electrons, and hence it is here, too, that the growth 
 of the nuclear charge will make itself most felt. 
 
 The connexion between the position of the X-ray lines and 
 the atomic number z was first formulated by G. Moseley. 
 He found for the frequency of K a (first line of the JfT-series) 
 and L a (first line of the Z/-series) the empirical relation 
 
 (Ill) 
 
 where N is the Eydberg number. 
 
 The similarity of these relations, which are only approxj-
 
 THE ANALYSIS OF RONTGEN SPECTRA 111 
 
 mately valid, with Bohr's formula (93) for the series of the 
 hydrogen type is so striking, that it was an obvious step to 
 seek to find the explanation of the Eontgen series by arguing 
 on the basis of Bohr's model. 
 
 This problem was attacked chiefly by W, Kossel, 2 A. 
 Sommerfeld, L. Vegard, P. Debye, J. Kroo,&* and A. 
 Smekal. 1 And thus, in addition to the theory of the optical 
 spectra which take their origin at the periphery of the atom, 
 a theory of the Kontgen spectra has arisen which leads us 
 
 FIG. 11. 
 
 into the inmost regions of the atom. According to this theory 
 we may picture to ourselves, in general terms, the emission 
 of the Rontgen spectra as follows : we consider a neutral 
 Bohr atom, consisting of a 2-fold nucleus, around which 
 z electrons revolve. These z electrons may be arranged in 
 different rings. The innermost, single-quantum ring, the so- 
 called -ST-ring, carries, let us say, p l electrons in its normal 
 state ; let the second ring, the ir-ring, be a two-quantum ring 
 occupied by p. 2 electrons, the third, three-quantum, the .M-ring 
 with^ 3 electrons, and so on (Fig. 11). The question whether
 
 112 THE QUANTUM THEORY 
 
 we can reach our goal with this conception of the ring by 
 assuming the quantum numbers to increase as we go outwards, 
 and whether we are to take the rings as co-planar or inclined 
 to one another will be left open. The preparation for the 
 emission of the -BT-series consists in this, that by the addition 
 of energy whether by absorption of external radiation or by 
 electronic impact an electron of the K-ring is removed to in- 
 finity, that is, the atom is, so to speak, ionised " inside," i.e. 
 in the .BT-ring. If the energy of the atom before this inner 
 ionisation = W , and after the ionisation = W K , then the 
 amount W K - W of energy must be provided. Hence every 
 radiation, the energy quantum of which satisfies the condition 
 hv ^ W K - W , can on being absorbed effect the tearing of the 
 electron out of the K-r'mg. If we allow the v of the external 
 radiation to grow slowly from small values, then, at the point 
 
 "TT7" \JU 
 
 V K = *_ 9, a sudden increase of the absorption occurs, 
 
 because from this point onwards the external radiant energy 
 is used for the " ionisation of the K-ring." Thus an absorp- 
 tion-band extends from v = V K towards higher frequencies, the 
 edge of the band lying at V K . This phenomenon of the " edge 
 of the absorption-band " has already been interpreted above 
 in the sense of the hypothesis of light-quanta. If the addition 
 of energy is provided by the impact of a strange electron, 
 coming from without, then its energy must be E> W K - W , 
 that is, E ^> ~hv K , a relation, which we have already deduced 
 earlier from the standpoint of the quantum hypothesis of light. 
 By ionisation of the .ST-ring the atom is now prepared for 
 ^-emission. If now an electron falls from the 2-quantum 
 Zv-ring into the 1-quantum J5"-ring, filling up, so to speak, the 
 gap produced there, then the first line of the K- series, K a , 
 will be emitted. If on the other hand the gap in the .ST-ring 
 is filled by an electron of the 3 -quantum If -ring, or the 
 4-quantum .W-ring, Kp or K y result respectively. The position 
 is quite analogous as regards the L- and .M-series. If, by the 
 addition of energy (absorption or electron-impact), an electron 
 of the Zi-ring is battered off, that is if the L-ring is ionised, 
 then the atom is prepared for the emission of the ZJ-series. 
 If, now, the gap in the 2-quantum L-ring is filled by an 
 electron of the 3-quantum M-ring, the first line of the Z/-series,
 
 FINE-STRUCTURE OF RONTGEN LINES 118 
 
 L a , results; if it is filled by an electron of the N-riug, the 
 second line of the .L-series, Ly, results (the notation is not 
 quite consistent but will serve the present purpose), and so 
 forth. 
 
 The converse phenomenon to line emission, viz. line absorp- 
 tion, with which we are acquainted in visible spectra, appears 
 at first sight to be missing here. That is, however, as W. 
 Kossel 298 recently showed, an error. It is true that the ejected 
 electron of the .ST-ring, for example, cannot in general be 
 caught upon the L-, M-, or .N-ring, because all places on them 
 are already occupied. An absorption of the lines K a , Kp, K y , 
 is therefore in this case impossible. But the electron of the 
 K-ring can certainly come to rest on an unoccupied quantum 
 orbit outside the occupied rings, that is, outside the surface 
 of the atom. In this process a " line " is actually absorbed, 
 namely, that line of which the hv is equal to the energy- 
 difference between the K-rmg and the final orbit of the ejected 
 electron. This refinement of our considerations shows, then, 
 that the electron from the .fiT-ring does not need to be raised 
 immediately to infinity, but that line absorptions may occur 
 before the edge of the band of absorption is reached. 
 
 2. The Fine-structure of Rontgen Lines 
 
 It is particularly noteworthy that Sommer/eld succeeded 
 also in the field of X-ray spectra in explaining the fine- 
 structure of the lines by calling in the aid of the theory of 
 relativity. Thus, for example, the 2-quantum .L-orbit is 
 "double"; it can occur as a circle (n = 0, n = 2) or as an 
 ellipse 2" (n = 1, n = 1). Hence the line which is emitted 
 by the electron of which the L-rmg is the initial orbit, namely, 
 K a , is a doublet (K a and K a -). In just the same way, those 
 lines for which the Iv-orbit is the final orbit of the electron 
 are doublets, namely, the line L a (more exactly L a >) to which 
 Lp is added to make a doublet ; further, L y which forms 
 a doublet with L&, and so forth. The distance between the 
 components of the doublets (expressed in frequencies) comes 
 out, according to Sommerfeld's Theory, as approximately pro- 
 portional to the fourth power of the atomic number z. Hence 
 here, in the X-ray region, where we are dealing for the most 
 part with elements having fairly high atomic numbers, the 
 8
 
 114 THE QUANTUM THEORY 
 
 doublets appear microscopically enlarged as compared with 
 the microscopic hydrogen-doublet (z = 1). During the emis- 
 sion of X-rays the electron approaches very near to the 
 highly-charged nucleus, and hence the relativistic effects of 
 the resolution of the lines are much greater than in the case 
 of the optical spectra, in which the electron is moving at the 
 surface of the atom, where it is almost entirely screened from 
 the action of the strong nucleus by the remaining electrons. 
 With the help of the following relation deduced theoretically 
 and adapted to experimental evidence, 
 
 < - fr")' -.< 112 > 
 
 Sommerfeld was able to calculate the hydrogen-doublet from 
 the observed L-doublets, and compare it with the results of 
 experiment. The agreement is very satisfactory. 
 
 3. The Distribution of Electrons among the Rings. Objections 
 to the Ring-arrangement of Electrons 
 
 The quantitative calculation of the simplest case, namely, 
 the emission of K a , led Debye to the conclusion that the 
 .ST-ring in the normal state consists of three electrons. To 
 this Kroo, by elaborating the calculation, adds the con- 
 clusion that the L-r'mg contains in its normal state nine 
 electrons. With these two distribution numbers, p l = 3, 
 p 2 = 9, the position of K a could be represented as a function 
 of the atomic number z for all elements. The emission of K a 
 takes place according to the following obvious scheme : 
 
 I K-ring L-ring 
 
 Normal state | 3 9 . _ . . , , _ . 
 
 Initial state [ 2 | 9 > I ni8atlon of * Z-nng. 
 KEafBtate | 8 | 8 > EmiSS1On f *" 
 
 The two distribution numbers (Besetzungszahleri) thus found 
 for the two innermost rings excite our attention. For on 
 the basis of the Periodic System with its periods of eight 
 we ought to expect, according to Kossel, the numbers 2 and 8.
 
 DISTRIBUTION OF ELECTRONS 115 
 
 The strange occurrence of the numbers 3 and 9 becomes 
 an objection, when we consider the case of sodium (z = 11). 
 Here, according to Kossel, we should expect the numbers 2, 
 8, 1, since in all probability an electron (the valency electron) 
 revolves alone, as in the case of all alkali metals, around the 
 outside quantum orbit (M-ring). In any case it is impossible 
 that the two innermost rings together should, in the normal 
 state, contain 12 (= 3 + 9) electrons. If we attempt to go 
 a step further still on the basis of Kroo's numbers 3 and 9, 
 and to set up a formula which represents for all 2*8 the 
 position of L a in conformity with observation, and thereby 
 to determine the number of electrons p 3 on the .M-ring, we 
 find, as A. Smekal 300 showed, that this mode of representation 
 is impossible with any combination 3, 9, p y Nor do we fare 
 better if we incline the various rings to one another, and take 
 their interaction into account. The suspicion is forced upon 
 us, that perhaps the whole conception of the arrangement 
 into plane rings does not correspond with fact, but that, rather, 
 the electrons in the atom form spatially symmetrical figures. 
 This suspicion is very much strengthened by a series of pro- 
 found investigations carried out by M.. Born and A. Landd. 901 
 Following on M. Bom's investigations of the dynamics of 
 the crystal-lattice, which we discussed in detail earlier hi 
 connection with the atomic heat of solids, the two in- 
 vestigators asked themselves the question, whether it is 
 possible to build up the cubic crystal-lattice of the alkaline 
 halides (NaCl, NaBr, Nal; KC1, KBr, KI, etc.) from ions of 
 Bohr atoms, by taking into account only the mutual electro- 
 static forces; and whether this method, if possible, would 
 enable them to prophesy the crystal properties (lattice-con- 
 stant, compressibility) from the atomic models of the two 
 constituent ions. The answer to this question has been, on 
 the whole, in the affirmative. But when the calculation of 
 the compressibility of these crystals was carried out, the 
 remarkable result manifested itself that crystals are found to 
 be too soft, that is, insufficiently rigid, if the conception of the 
 ring-arrangement of electrons in the atom is maintained. On 
 the other hand, we get good agreement with the observations 
 if, following Born, we introduce the hypothesis that the 
 electrons are arranged spatially. A complex of eight electrons,
 
 116 THE QUANTUM THEORY 
 
 as occurs in sodium, potassium, etc., does not therefore occupy 
 a plane 8-ring ; the eight electrons describe paths of ciibical 
 symmetry. Into the still obscure region of these " spatial " 
 electron paths, A. Lande 302 has made some successful in- 
 cursions. 
 
 From all that has been said it would appear to be certain 
 that in dealing with Rontgen spectra, too, we can no longer be 
 content with the arrangement of the electron rings in planes, 
 and that the whole quantitative theory of the Eontgen series, 
 including Sommerfeld's fine-structure of the K- and the L- 
 doublets, must be built up on a fresh foundation.
 
 CHAPTEB VIII 
 
 Phenomena of Molecular Models 
 i. Dispersion and Magneto-rotation of the H 2 Molecule 
 
 WHILE the X-ray spectra and the spectra of the optical 
 series arise from the atoms of the elements (and hence 
 their theory links up with the atomic models), there is a series 
 of phenomena which, in the case of polyatomic substances, 
 are peculiar to the molecules, and the theory of which, 
 therefore, is founded on the molecular models. Chief among 
 these are the normal dispersion, the rotation of the plane of 
 polarisation in the magnetic field (magneto-rotation), and, 
 further, the great and complicated subject of band-spectra. 
 Up till a few years ago, dispersion and magneto-rotation had 
 been exclusively treated from the standpoint of the Thomson 
 model, that is, with the help of quasi-elastically bound 
 electrons, and this explanation had served in turn as a 
 powerful support for this model. Nevertheless, discrepancies 
 in these theories had long been known. For example, 
 measurements calculated upon the basis of the dispersion 
 theories of Drude, Voigi, or Planck led to values for the ratio 
 
 of the charge to the mass of the electron ( ) which, in com- 
 parison with the direct measurements of this quantity (based 
 upon the deflection of the cathode- or /3-rays in the electric 
 and magnetic fields) which were much too small. When, 
 however, the Thomson model became displaced by the 
 Rutherford-Bohr model, and the successes of the Bohr atomic 
 model increased at an undreamed-of rate, the question arose 
 whether an unobjectionable theory of dispersion and magneto- 
 rotation could not be founded upon these new views. The 
 difficult position, into which we are brought by this problem, 
 117
 
 118 THE QUANTUM THEORY 
 
 arises from the fact that we do not actually know a single 
 instance of the exact manner in which a polyatomic Bohr 
 molecule is built up from its nuclei and electrons. The 
 exact knowledge of this structure, and the motion of all the 
 electrons is absolutely necessary, if we desire to know how 
 the molecule reacts upon external waves (dispersion). It is 
 true that W. Kosseiw* has, in a detailed study already 
 referred to above, pointed out the general guiding lines along 
 which, from the chemical point of view, the building-up of 
 the atom from molecules must be carried out, but the details 
 of this construction remain open. Only in a few of the 
 simplest cases have detailed molecular pictures been con- 
 structed and closely tested. Thus Bohr, as we remarked in 
 discussing the atomic heat of gases, has already proposed a 
 model of the diatomic hydrogen molecule. It has the follow- 
 ing construction (see Fig. 8) : two singly-positive nuclei (that 
 is, each consisting of only a single positive charge) are 
 separated by the distance 26. In the vertical plane which 
 bisects the line joining the nuclei, two electrons rotate, 
 diametrally opposite one another, on a circle of diameter 2a. 
 The equilibrium of the Coulomb and the centrifugal forces 
 requires that a = b^/3. By means of this relation, and by the 
 quantum condition that each electron must have the moment 
 
 of momentum , the model is completely determined in all 
 
 2ir 
 
 its dimensions and speeds. It was this model which was 
 the first to be proposed : it was examined by P. Debye 3* 
 with reference to its dispersion. On account of its sym- 
 metrical structure the molecule possesses no electrical mo- 
 ment in its normal state. If, on the other hand, it is struck 
 by an external light wave, the motion of its electrons is 
 periodically disturbed ; they depart from the normal quantum 
 path, fall into forced vibration, and thus generate an electric 
 moment which changes periodically in step with the external 
 wave. Thus the original motion of the primary wave is 
 changed, and dispersion results. We may conceive this as 
 follows : Let c be the velocity of the primary wave in vacuo. 
 The oscillations of the electrons generate a secondary wave 
 which spreads out from the molecules. All these secondary 
 waves combine with the primary wave to a form new wave
 
 OBJECTIONS TO BOHR'S MODEL 119 
 
 which moves with the altered velocity q, the value of which 
 depends on the frequency of the primary wave. But just 
 this is the phenomenon of dispersion. The electronic vibra- 
 tions which occur here are not oscillations about positions of 
 equilibrium, as in the case of the quasi-elastic model, but 
 oscillations about stationary paths. Moreover, here, the force 
 holding the electrons, as opposed to the usual classical 
 theories of dispersion, is anisotropic (that is, the electron is 
 held by different forces in different directions) ; above all, by 
 means of this anisotropy, it was possible to explain away the 
 
 disagreement in the value of , which had previously been 
 
 WIC 
 
 found to be too small ; and Debye succeeded, on the basis of 
 the normal value of , in deducing from the theory the 
 
 observed dispersion curve of hydrogen, that is, the curve 
 which shows how its coefficient of refraction depends on the 
 wave-length. It should be noted that in the formula for the 
 coefficient of refraction, no single constant is arbitrary, but 
 that the dispersion formula is made up entirely of universal 
 constants. 
 
 Using the same method (calculus of disturbances), P. 
 Scherrer 30 * has calculated the rotation of the plane of 
 polarisation which linearly polarised light undergoes in its 
 passage through hydrogen under the influence of a magnetic 
 field. His efforts were equally successful. 
 
 2. Objections to Bohr's Model of the Hydrogen Molecule 
 
 In spite of the successes which the Bohr model of the 
 hydrogen molecule has won, a list of weighty objections to 
 it has accumulated in the course of time. That the con- 
 tribution which the rotation (more accurately, the regular 
 precession) of this molecule makes to the molecular heat at 
 low temperatures, does not correspond with the observations 
 of Eucken, has been shown by P. S. Epstein, as we have 
 already mentioned. Also at high temperatures, when the 
 oscillations of the two nuclei relatively to one .another con- 
 tribute to the molecular heat, no agreement between theory 
 and observation has been found in the case of the Bohr model, 
 as G. Laski 306 recently showed,
 
 120 THE QUANTUM THEORY 
 
 Further, the model must possess, in consequence of the 
 revolving electrons, an almost fixed magnetic moment parallel 
 to the axis of the nucleus, that is to say, it must be equivalent 
 to a molecular elementary magnet, which endeavours to set 
 itself, in an external magnetic field, parallel to the lines 
 of force. Hydrogen ought, therefore, to be paramagnetic, 
 whereas it is diamagnetic. 
 
 Another very important objection, to which Nernst in 
 particular drew attention, is the following : if we calculate 
 the work which is necessary to separate the molecule into 
 its two atoms, the so-called heat of dissociation, we get s 07 the 
 value, 61,000 calories. On the other hand, Langmuir 3 08 found 
 84,000 cals., Isnardi** 95,000 cals., /. FrancJc, P. Knipping 
 and Thea Kriiger 81,000 ( 5700) cals. In any case, the 
 calculated heat of dissociation comes out 25 per cent, too 
 small.sioa 
 
 Finally, W. Lenz 311 has recently increased the objections 
 to the hydrogen model by an important one based on a 
 theory of band-spectra, which we shall discuss below. He 
 proved that the band-lines of hydrogen and nitrogen can 
 exhibit the observed Zeeman effect, only if these molecules 
 possess no moment of momentum around the nuclear axis. 
 The fact that the two electrons in Bohr's molecular model 
 revolve in the same sense, however, endows it with just such 
 a moment of momentum. On the whole, the Bohr model does 
 not seem to correspond to reality ; the arrangement of the two 
 nuclei and electrons must plainly be quite different. No 
 satisfactory model, however, has yet been found. 
 
 3. Models of Higher Molecules 
 
 Matters are no better in the case of models of the more 
 complicated molecules. It is true that Sommerfeld 312 and 
 F. Pawer 313 have also worked out the theories of dispersion 
 and magneto-rotation in the case of the more general Bohr 
 models (N 2 and O 2 ) which are constructed on the lines of the 
 hydrogen model. According to Sommerfeld, four electrons 
 revolve about the line joining the two nuclei in the case of 
 oxygen, each of which acts with an effective charge + 2e ; in 
 the case of nitrogen, a ring of six electrons rotates about the 
 nuclear axis, while the nuclei carry triple effective charges.
 
 QUANTUM THEORY OF BAND-SPECTRA 121 
 
 Sommerfeld was able to obtain agreement with observation only 
 by setting up for each electron of a valency ring of 2s-electrons 
 the unaccountably strange quantum condition : moment of 
 
 momentum = ^ ^ s > undoubtedly a most unsatisfactory 
 
 result. Gerda Laski 31 * obtained better results with some- 
 what different models, which she chose in such a way that 
 the specific heat of the two gases at high temperatures agreed 
 with the observations of Pier. 318 According to her ideas, the 
 nitrogen molecule must consist of two seven-fold positive nuclei, 
 each of which is closely surrounded by a 1-quantum ring of 
 two (or three) electrons. The " valency ring " in the central 
 vertical plane is 2-quantum and contains ten (or eight) 
 electrons. Analogously, the oxygen molecule consists of two 
 eight-fold positive nuclei, each encircled by a 1-quantum ring 
 of two (or three) electrons, whereas the 2-quantum valency 
 ring contains twelve (or ten) electrons. The same objections 
 apply to some extent to these models of Sommerfeld and Laski 
 as to the hydrogen model. For example, they give no account 
 of why oxygen should be paramagnetic, and nitrogen, on the 
 other hand, diamagnetic. Moreover, the above-mentioned 
 objection of Lenz applies in full force to these models ; for 
 they all possess moments of momentum around the nuclear 
 axis. In conclusion, we feel bound to admit that the exact 
 constitution of even the simplest models is at present unknown 
 to us. 
 
 4. The Quantum Theory of Band-spectra 
 
 To conclude this chapter, we shall turn our attention to the 
 band- spectra, and collect together shortly what the quantum 
 theory has been able to assert about them up to the present 
 time. That they belong to molecules and compounds may 
 nowadays be regarded as certain. The first attempt to con- 
 struct a logical quantum theory of band-spectra was under- 
 taken by K. Schwarzschild 316 who clearly recognised the 
 importance of the rotation of the molecule in the production 
 of these spectra. His conceptions may be defined as follows : 
 a system of electrons revolves at a definite quantum distance 
 around a molecule which itself rotates according to quantum 
 conditions, the assumption being made for the sake of
 
 122 THE QUANTUM THEORY 
 
 simplicity that the motion of the electrons is not influenced 
 by the motion of the molecule. If E is the quantum energy 
 of the electrons, E r the quantised rotational energy of the 
 molecule, then E + E r = E is the total energy of the system. 
 If the three chief moments of inertia of the molecule / are 
 equal to one another, then it follows, just as in (80), that 
 
 where n denotes the rotational quantum number. Therefore 
 
 If, now, the system passes from one quantum state having 
 the electronic energy E Q and the rotational quantum number 
 n into another quantum state having the electronic energy E' 
 and the rotational quantum number n', then it follows from 
 Bohr's frequency formula (92) that the frequency of the line 
 radiated is given by 
 
 _ E -E' (n - n> 
 
 7 T o 2 T ' ' V A / 
 
 If we keep all the quantum numbers which occur here, except- 
 ing n, constant, and allow n to vary, then we get a series of 
 lines progressing towards the violet and having the frequencies 
 
 v = a + bn 2 (a and b are constants) . (115) 
 
 This is a formula which had already been given empirically 
 by Deslandres,* 11 and which is approximately true for the lines 
 of many bands. 
 
 Following Schwarzschild, T. Heurlinger 318 and W. Lenz* 19 
 in particular, have further developed and refined the quantum 
 theory of band-spectra. For example, Lenz has pictured the 
 molecule as a symmetrical top having two moments of inertia 
 and a rotational rigidity (moment of momentum) around the 
 axis of the figure, and hence deals from the outset with a regular 
 precession of the molecule in place of a rotation. Using 
 Bohr's frequency formula, and applying the principles of 
 selection, he obtained the following general foi-mula for the 
 lines of a band : 
 
 v a + bn + en 2 (a, b t c are constants) . (116)
 
 QUANTUM THEORY OF BAND-SPECTRA 123 
 
 which is obeyed, according to Heurlinger, in the case of the 
 so-called " cyanogen " lines of nitrogen, for example. In 
 addition to the lines given by (116), Lenz's Theory requires the 
 occurrence of the series given by the formula 
 
 + + 2 . . . (117) 
 
 for the case that the molecule really possesses a finite moment 
 of momentum about its axis of figure. A series which follows 
 this law does not, however, exist in the cyanogen bands, ac- 
 cording to Heurlinger. Lenz deduces from this the conclusion 
 already mentioned, that the nitrogen model does not possess 
 a rotational rigidity about its axis. By calculating the 
 Zeeman effects of the band lines, and comparing them with 
 observation, Lenz was able to confirm this, and to extend it to 
 the hydrogen molecule. 
 
 The infra-red Bjerrum absorption bands of the diatomic and 
 polyatomic gas compounds, which we had discussed at length 
 in Chapter Y, belong to the general type of band-spectra. If 
 we are to deduce them from a theory consistently founded on 
 quanta and not, as we did earlier, half according to the 
 quantum, half according to the classical theory we must 
 follow closely the course pursued above, with the difference 
 that, in place of the energy of the electronic system there will 
 appear the energy of the atoms, 920 with which the rotational 
 energy of the molecule is combined, as a first approximation, 
 additively. The logical carrying out of this calculation (in 
 which Bohr's frequency formula and the principle of corre- 
 spondence are applied), which was undertaken by Heurlinger & l 
 and the author, 322 gives for the structure of the " fluted " ab- 
 sorption bands an arrangement of lines which at first sight 
 does not appear to agree with the beautiful and exact measure- 
 ments of Imes. 323 The theory gives for the position of the 
 absorption lines a formula 
 
 l/ = Vo(n + i) JL, (n=l,2,3...) (118) 
 
 and therefore requires that all neighbouring lines be equi- 
 distant, including the two in the middle (n = 0). On the other 
 hand, Imes' observations show with indubitable clearness that 
 the interval between the two middle lines is twice as great as
 
 124 THE QUANTUM THEORY 
 
 the interval between all neighbouring lines. This apparent 
 contradiction is explained, as A. Kratzer 32 * recently showed, 
 in a surprising fashion, if we take into account the intensity 
 of the absorption lines according to Bohr's Principle of Analogy. 
 For it then appears that the first absorption line to the right 
 of the middle v -line, namely, the line 
 
 h 
 
 " = "o + SW 
 
 (which is derived from formula (118) by setting n = and 
 using the positive sign for the second term) is of vanishingly 
 small intensity. This line is generated when the molecule 
 passes over from an initial rotationless and vibrationless state 
 into the final state in which the two ions oscillate relatively 
 to one another with one quantum, and in which, at the same 
 time, the molecule rotates as a whole with one quantum. The 
 rotationless and vibrationless state has, however, a vanishingly 
 small probability ; the number of transitions from this initial 
 state per second, and therefore the intensity of the correspond- 
 ing absorption line, is hence vanishingly small. By the dis- 
 appearance of the first line to the right of the middle position 
 v , the structure of the lines as observed by Imes is actually 
 reproduced, as one may easily recognise ; in the formula, the 
 " middle" of the line structure is displaced from the point v 
 
 to the right by the amount oZFr The absorption lines group 
 themselves equidistantly and symmetrically on both sides of 
 the missing " middle," v = v + Q-^J- This state of affairs 
 
 O7T J 
 
 may be expressed by writing, in formal agreement with (83), 
 
 " = "''4^7 (*-:i,8,8...)| 
 where . (119) 
 
 From the constant interval between neighbouring lines, namely 
 
 . (120) 
 
 the moment of inertia of the rotating molecule can be cal- 
 culated with great accuracy. 328
 
 CHAPTEE IX 
 
 The Future 
 
 IN the preceding pages the author has attempted to give 
 in broad outline the most important features of the 
 doctrine of quanta, its origin, its development, and its 
 ramifications. If we now survey the whole structure, as 
 it stands before us, from its foundations to the highest story, 
 we cannot avoid a feeling of admiration ; admiration for the 
 few who clear-sightedly recognised the necessity for the pew 
 doctrine and fought against tradition, thus laying the founda- 
 tions for the astonishing successes which have sprung from 
 the quantum theory in so short a time. 
 
 None the less, no one who studies the quantum theory 
 will be spared bitter disappointment. For we must admit 
 that, in spite of a comprehensive formulation of quantum 
 rules, we have not come one step nearer to understanding 
 the heart of the matter. That there are discrete mechanical 
 and electrical systems, characterised by quantum conditions 
 and marked out from the infinite continuity of " classically " 
 possible states, appears certain. But where does the deeper 
 cause lie, which brings about this discontinuity in nature? 
 Will a knowledge of the nature of electricity and of the con- 
 stitution of the electromagnetic field serve to read the riddle ? 
 And even if we do not set ourselves so distant a goal, there 
 remains an abundance of unanswered questions. The 
 decision has not yet been made, as to whether, as Planck's 
 first theory requires, only quantum-allowed states exist (or 
 are stable), or whether, according to Planck's second formula- 
 tion, the intermediate states are also possible. We are still 
 completely in the dark about the details of the absorption 
 and emission process, and do not in the least understand 
 125
 
 126 THE QUANTUM THEORY 
 
 why the energy quanta ejected explosively as radiation 
 should form themselves into the trains of waves which we 
 observe far away from the atom. Is radiation really pro- 
 pagated in the manner claimed by the classical theory, or 
 has it also a quantum character ? 
 
 Over all these problems there hovers at the present time 
 a mysterious obscurity. In spite of the enormous empirical 
 and theoretical material which lies before us, the flame of 
 thought which shall illumine the obscurity is still wanting. 
 Let us hope that the day is not far distant when the mighty 
 labours of our generation will be brought to a successful 
 conclusion.
 
 Mathematical Notes and References 
 
 1 0. Lummer and E. Pringsheim, Wiedem. Ann. 63, 395 (1897) ; 
 Verhandl. d. deutsch. physikal. Ges. 1899, pp. 23, 215; ibid., 1900, p. 163. 
 Of. also O. Lummer and E. JahnJee, Drudes Ann. 3, 283 (1900), and O. 
 Lummer, E. Jahnke and E. Pringsheim, Drudes Ann. 4, 225 (1901). 
 
 2 Of. M. Planck, Vorlesungen iiber die Theorie der Warmestrahlung 
 (Leipzig 1906), 10. 
 
 3 Frequency () = velocity oflight in vacuo (c) ^ 
 
 wave-length in vacuo (A.) 
 
 iCf., for example, M. Planck, Vorlesungen iiber Warmestrahlung 
 (1906), 17. 
 
 3 O. Kirchhoff, Gesammelte Abhandlungen (J. A. Earth, Leipzig 1882), 
 pp. 573 et seq. ; Berliner Akademieberichte, 1859, p. 216 ; Poggend. Ann. 
 109, 275 (1860). 
 
 6O. Lummer and W. Wien, Wiedem. Ann. 56, 451 (1895). Cf. also 
 O. Lummer and F. Kurlbaum, Verhandl. d. deutsch. physikal. Ges. 17, 
 106 (1898). 
 
 7 Of. Note 5. 
 
 8 L. Boltzmann, Wiedem. Ann. 22, 291 (1884). 
 9J. Stefan, Wiener Ber. 79, 391 (1879). 
 
 10 The Stefan-Boltemann Law is deduced as follows : Let the energy 
 of black-body radiation at the temperature T, which is enclosed in a 
 space of volume V having a movable piston, be U = Vu, where u is 
 the " spatial " density of the radiant energy. The pressure, equal in 
 all directions, which the radiation exerts upon the piston and walls is, 
 according to electrodynamics, p = $u. If we supply to this system at 
 the temperature T (that is, isothermally) an amount of heat d'Q, then 
 its energy increases by dU, and the radiation does work pdV in push- 
 ing back the piston. Therefore, according to the first law of thermo- 
 dynamics, and owing to the two relations above : 
 
 d'Q = dU + pdV = udV + Vdu + dV = | udV + Vdu. 
 According to the second law of thermodynamics, -Q must be a com- 
 plete differential. Hence the following relation holds : 
 \_ 3 m 4 d u\dT _1 
 
 ,'ldt* u)dT_l 
 
 4 fl dw _ u] 
 3\TdT Z^J 
 
 127
 
 128 THE QUANTUM THEORY 
 
 i.e. -^ = 4^, which, integrated, gives u = a!" 
 
 where a is a constant. Now, as we can easily see, the total radiation 
 
 00 
 
 K 2 / TH v dv is distinguished from the density of radiation u only by 
 
 
 
 a constant factor (see M. Planck, Lectures in Radiation (1906), 22), 
 hence the total radiation is proportional to the fourth power of the 
 absolute temperature and this is the Stefan-Boltzmann Law. 
 
 11 W. Wien, Sitzungsber. d. Akad. d. Wissensch. Berlin, 9 Feb. 1893, 
 p. 55 ; Wiedem. Ann. 52, 132 (1894). Of. also Max Abraham, Theorie 
 der Elektrizitat II, 43 (1914); M. Planck, Vorlesungen iiber die 
 Theorie der Warmestrahlung (Leipzig 1906), pp. 68 etseq. ; W. Westphal, 
 Verhandl. d. deutsch. physikal. Ges. 1914, p. 93; H. A. Lorentz, Akad. 
 d. Wissensch. Amsterdam, 18 May 1901, p. 607. 
 
 12 Formula (4) of the text (Wieris Law of Displacement) may be 
 obtained by means of a simple dimensional calculation, as L. Hopf 
 recently showed in the " Naturwissenschaften " (8, 109, 110 (1920)). 
 We assume that Kx depends only on v, T, and the velocity of light c. 
 The dimensions of Kx are obtained from the fact that, according to (1), 
 energy 
 
 surface x time 
 From this it follows that 
 
 [K,] - [-']. 
 If we set 
 
 Kx = const. v x T y c z 
 
 then, remembering that T has the dimensions of energy, we get 
 [m- 2 ] = const. [-* m y - V* r *" f t~ Z 1 
 
 = const. \m y ?V+* r*~ -*] 
 Hence x = 2; # = 1 ; 2 = -2 
 
 which gives us, Kx = const. . ?1 . T. 
 
 This relation is not, however, as we shall see, generally valid. In fact, 
 
 oo 
 
 it would give no finite value for K = 2 \"K. v dv. But, according to the 
 
 6 
 
 Stefan-Boltzmann Law (3), K = y T 4 . Hence the constant of Kx may 
 still depend on a dimensionless combination of the four variables 
 y, v, T, c. If, therefore, we set const. =f(y^T n c^y' a ) then the argument 
 of the function / must have the dimension 0. If, further, we remember 
 that 
 
 [ BfKf* surface x time 
 
 H = .
 
 NOTES AND REFERENCES 129 
 
 it then follows that 
 
 = [t-t .mri.fr, t-*i .K.t -<T. m- 3 . l-**> . t ] 
 = [< -*--?+. 
 
 Hence const. = /[(?)". c- - 7 ,] = < 
 
 Therefore K , = Jr.* (j) = *. J. 
 
 or, finally, Kl> = 3 F (j} 
 
 13 If we plot K, - *(} function of v, keeping T constant, the 
 
 maximum of this curve if one is present lies at that point at which 
 = 0. This gives 
 
 where F 1 is the differential coefficient of F with respect to the argument. 
 This equation, in which only occurs as unknown, gives a definite value 
 
 for ^. In other words, for v = max, it follows that ^^ = const. 
 
 II W. Wien, Wied. Ann 58, 662 (1896). 
 
 13 0. Lummer and E. Pringslieim, Wied. Ann. 63, 395 (197) ; Drude's 
 Ann. 3, 159 (1900) : Veih. d. deutsch. phys. Ges. 1, 23 and *15 (1899). 
 
 The total radiation emitted per second from 1 cm. 2 in one direction is, 
 by formula (1) ( 
 
 s = 
 
 According to the Stefan-Boltzmann Law, S is nroportional to T 4 , there- 
 fore S = ffT*. (The constant of proportionality a is related to the 
 constant y occurring in (3) by the equation v = iry.) The absolute 
 measurement of S gave the following values for a, in chronological order : 
 
 <r - 5-45 10' 12 r V ! a !' t ,"| according to F. Eurlbaum [Wiedem. Ann. 
 Lcm/ deg.*J 
 
 65, 746 (1898); Verhandl. d. deutsch. 
 
 physikal. Ges. 14, 576, 792 (1912)]. 
 = 5-58. 10 ~ 12 ti according to S. Valentiner [Ann. d. Phys. 
 
 31, 255 (1910) ; 39, 489 (1912)]. 
 = 5-90 . 10 " 12 .1 according to W. Gerlach [Ann. d. Phys. 
 
 38, 1 (1912)].
 
 180 THE QUANTUM THEORY 
 
 a- - 5-30 . 10 ~ 12 T cm Tde 1 according to E. Bauer and M. Moulin 
 
 [Soc. Franc, de Phys. Nr. 301, 2-3 
 
 (1909)]. 
 = 6*30 10~ 12 > according to Ch. Ffry [Bull. Soc. Franc. 
 
 Phys. 4 (1909)]. 
 = 6-51. 10 ~ 12 > according to Ch. F&ry and M. Drecq 
 
 [Journ. de Phys. (5) 1, 551 (1911)]. 
 = 5*67 . 10 ~ 12 i> according to G. A. Shakespear [Proc. Roy. 
 
 Soc. (A) 86, 180 (1911)]. 
 ==5-54. 10 ~ 12 M according to W. H. Westphal [Verhandl. 
 
 d. deutech. physikal. Ges. 14, 987 (1912)]. 
 - 6-05. 10 " 12 >, according to L. Puccianti [Cim. (6) 4, 31 
 
 (1912)]. 
 = 5-89. 10 ~ 12 ,, according to Keene [Proc. Roy. Soc. (A) 
 
 88, 49 (1913)]. 
 = 5-57.10-12 according to W. H. Westphal [Verhandl, 
 
 d. deutsch. physikal. Ges. 15, 897 (1913)]. 
 -= 5-85 . 10 ~ 12 , according to W. Gerlach [Phys. Zeitschr. 
 
 17, 150 (1916)]. 
 
 As regards Wieris Law of Displacement, the relation (5a) was tested 
 and found to be confirmed. From Fig. 1, in which E*. is plotted as a 
 function of A. for different values of A., we see clearly how the maximum 
 of the curve becomes displaced towards shorter wave-lengths as the 
 temperature rises. 
 
 For the constant on the right-hand side of relation (5o) the measure- 
 ments gave the following values : 
 const. = 0-294 [cm. deg.] according to O, Lummer and E. Pringsfoim 
 
 [Verhandl. d. deutsch. physikal. Ges. 1, 23 
 
 and 215 (1899)]. 
 = 0-292 according to F. Paschen [Drude's Ann. 6, G57 
 
 (1901)]. 
 = 0-2911 according to Coblentz [Bull. Bur. of Stand. 10, 
 
 1 (1914)]. 
 
 16 O. Lummer and E. Pringsheim, Verhandl. d. deutsch. physikal. 
 Ges. 1, 215 (1899). 
 
 17 F. Paschen, Berliner Ber. 1899, pp. 405, 959. 
 
 18 M. Planck, Absorption und Emission elektr. Wellen durch Resonant. 
 Sitzungsber. d. Berl. Akad. d. Wiss. 21 March 1895, pp. 289-301 ; Wiedem. 
 Ann. 57, 1-14 (1896). tJber elektr. Schwingungen, welche durch Re- 
 sonanz erregt und durch Strahlung gedampft werden. Sitzungsber. d. 
 Berl. Akad. d. Wiss. 20 Febr. 1896, pp. 151-170 ; Wiedem. Ann. 60, 577-599 
 (1897). tTber irreversible Strahlungsvorgange. (1. Mitteilung.) Sitzungs- 
 ber. d. Berl. Akad. d. Wiss., 4 Febr. 1897, pp. 57-68. (2. Mitteilung) ibid., 
 8 July 1897, pp. 715-717. (3. Mitteilung) ibid., 16 Dec. 1897, pp. 1122- 
 1145. (4 Mitteilung) ibid., 1 July 1898, pp. 449-476. (5. Mitteilung) ibid., 
 18 May 1899, pp. 440-480. (Supplement.) ibid., 9 May 1901, pp. 544-555 ;
 
 NOTES AND REFERENCES 181 
 
 Drudes Ann. 1, 69-122 (1900). (Supplement.) Drudes Ann. 6, 818-831 
 (1901). Entropie und Temperatur strahlender Warme. Drudes Ann. 1, 
 719-737 (1900). 
 
 19 In place of the mean value, with respect to time, of the energy of a 
 single oscillator, we may use the spatial mean value of the momentary 
 energy of a whole system consisting of very many oscillators. 
 
 20 In this second, more difficult part of the calculation, Planck takes his 
 stand upon the second law of thermodynamics, and seeks, from this view, 
 to determine a phase-quantity S of the oscillator, which possesses the 
 well-known property of the entropy, that it increases in all irreversible 
 processes. He arrived at the solution : 
 
 This function possessed, as Planck showed, the required property of en- 
 tropy, but it was not the only function with this property. And in fact 
 it appeared later, that in the deduction of the above expression, a readily 
 suggested but unjustified supposition had been made. The expression 
 given in the text, formula (8), for the mean energy U follows from S by 
 applying the second law in the form : 
 
 MO. Lummer and E. Pringsheim, Verhandl. d. deutach. physikal. 
 Ges. 1900, p. 163. 
 
 22 M . Planck, Verb. d. deutsch. phys. Ges. 1900, p. 237. It is of 
 historic interest to note that Planck had already, in a somewhat earb'er 
 paper (Verh. d. deutsch. phys. Ges. 1900, p. 202), arrived at the true 
 law of radiation by a purely formal alteration of Wien's formula, which 
 was not further explained. Cf. also Ann. d. Phys. 4, 553 (1901) ; 4, 561 
 (1901) ; 6, 818 (1901) ; 9, 629 (1902). 
 
 23 Let N oscillators be present. Let the total energy to be divided 
 among them be UN = NU. The " state " or phase of the oscillator- 
 system, the probability of which is to be calculated, is then defined by the 
 fact that N oscillators possess the energy Ujt. We divide UN into P 
 energy elements , so that 
 
 UN = N . U = P*. 
 
 The number of possible ways of distributing P balls among N boxes is, 
 however, 
 
 (N+P- 1)1 
 (N - 1) 1 P ! ' 
 
 This is therefore the probability of the state, which corresponds to the 
 distribution of P energy elements among N oscillators. P. Ehrenfest aud 
 H. Kamerlingh-Onnes give a very simple deduction of this formula in 
 Ann. d. Phys. 46, 1021 (1915). 
 The rule mentioned in the text, which is due to Boltzmann, states
 
 182 THE QUANTUM THEORY 
 
 that the entropy Sjt of the oscillator system is connected with the prob- 
 ability TFby the fundamental relation 
 
 Sir = k log W 
 where k is a constant. 
 
 In this theorem of Boltzmann the following law of the growth of en- 
 tropy (second law of thermodynamics) is contained : if a system passes 
 from an improbable condition into a more probable one, then by this 
 transition W, and therefore the entropy S, increases. If we here insert the 
 value of W, and, since N and P are very large numbers, use Stirling's 
 approximation formula 
 
 \oge(Nl)=N(logeN- 1) 
 
 then, if we set for P, N , we get by an easy calculation 
 
 and hence the entropy S of one oscillator becomes : 
 
 But according to the Second Law (see note 20) 
 dS 1 
 
 If we carry out the differentiation on the left-hand side, and solve the re- 
 sulting relation between U, T, and e, with respect to U, we get the ex- 
 pression (9) of the text. 
 
 2< Of. the paper by Ehrenfest and Katnerlingh-Onnes cited in the 
 previous note. 
 
 2S This law is essentially identical with Boltzmann' 's H- Theorem. Of. 
 L. Boltztnann, Vorlesungen iiber Gastheorie Bd. I, p. 38 (1896) ; Sit- 
 zungsber. d. Wiener Akad. d. Wiss. (II) 76, 373 (1877). Of. also P. 
 Ehrenfest, Phys. Zeitschr. 15, 657 (1914). 
 
 36 H. Rubens and F. Kurlhaum, Sitzungsber. d. Berl. Akad. d. Wiss. 
 1900, p. 929 ; Ann. d. Phys. 4, 649 (1901). 
 
 27 F. Paschen, Ann. d. Phys. 4, 277 (1901). 
 
 28 L. Holborn and 8. Valentiner, Ann. d. Phys. 22, 1 (1907) ; Coblentz, 
 Physical Review, 31, 317 (1910) ; E. Baisch, Ann. d. Phys. 35, 543 (1911) ; 
 E. Warburg, O. Leithttuser, E. Hupka and C. Milller, Ann. d. Phys. 40, 
 609 (1913) ; E. Warburg and C. Milller, Ann. d. Phys. 48, 410 (1915). 
 
 W. Nernst and Th. Wulf, Ber. d. deutsch. phys. Ges. 21, 294 (1919). 
 80 Lord RayUigh, Phil. Mag. 49, 539 (1900). 
 
 31 The " Stefan-Boltzmann constant of total radiation" <r, introduced 
 in note 15, has therefore the value
 
 NOTES AND REFERENCES 133 
 
 32 In order to determine the constants h and k which occur in the 
 radiation formula, we can, instead of using the equation : \max . T = 
 const., compare other relations with the measurement of the total 
 radiation. For example, we can proceed as follows : At a constant 
 temperature T we measure the ratio of the intensity of radiation for two 
 different wave-lengths \j and \^ (isothermal method). Now this ratio 
 is, according to (15) 
 
 &* \ A 1/ _C_ K 
 
 e* lT - 1 
 
 From this relation, since everything excepting C is known, C, that is, 
 - may be calculated. Another method is the following : we measure for 
 
 a fixed wave-length x the ratio of the intensity of radiation at two 
 different temperatures T l and T 2 (isochromatic method). Then it follows 
 that 
 
 This is a relation from which C, that is,- can again be calculated. 
 
 With the help of these methods, the researches, for example, of 
 Warburg and his co-workers cited in note 28 have yielded values for 
 
 C = ^ which lie in close proximity to C = 1-430. This value was taken 
 
 by Nernst and Wulf (see note 29) for their critical investigation. 
 
 For the constant of Wien's Law of Displacement in the form \inax . T 
 = b we would accordingly get from (16) : 
 
 4-9651 
 
 a value smaller, therefore, than that given by direct measurement (see 
 no'e 15). Whether Warburg's value, C = 1-430, or the measured values 
 of &(> 0-29) or both, are seriously affected by experimental error, or 
 whether after all as Nernst and Wulf maintain Planck's formula is 
 not right, must be left for the future to decide. 
 
 38 M. Planck, Ann. d. Phys. 4, 553 (1901). 
 
 84 If we apply Boltzmann's relation S = k log W (quoted in note 15), 
 which connects the entropy S with the probability of state W, to one 
 gramme-molecule of an ideal gas, then by calculating the probability of 
 a certain state, i.e. a certain distribution of velocities among the 
 molecules, we arrive at the following value for the entropy of the gas 
 
 S kN(l log* U + log V) + const.
 
 184 THE QUANTUM THEORY 
 
 (Of., for example, M. Planck, Lectures on the Theory of Radiation 
 (1906), 143.) Here N is the number of molecules in a gramme- 
 molecule (Avogadro's number), U the energy, V the volume of the gas. 
 Now, according to the Second Law of Thermodynamics, 
 
 jo dU + 
 
 must be a complete differential, where p and T denote pressure and 
 temperature of the gas. Hence the relation 
 
 \dVju~ T 
 
 must hold. This gives 
 
 kV = P ie p _kNT 
 
 If we compare this with the equation of state of an ideal gas in thermo- 
 dynamics, p = ~, we get for the absolute gas constant R the value 
 B = kN 
 
 from which formula (19) of the text follows. 
 
 88 M. Planck, Ann. d. Phys. 4, 564-566 (1901). 
 
 38 Compare, for example, the table of the values of Avogadro's number 
 given in the report of J. Perrin at the Solvay Congress in Brussels 
 (1911). [A. Eucken, Die Theorie der Strahlung und der Quanten. 
 Abhandlungen der Bunsen-Gesellschaft Nr. 7, Wilh. Knapp, Halle 1914.] 
 
 37 R. A. Millikan, Phil. Mag. (6) 34, 13 (1917). 
 
 Mlbid., from the values given by Millikan for the electronic charge 
 e = 4-774 x 10- 10 (electrostatic units) and from the electrochemical 
 constant F = 969-4 . 2'999 . 1010 electrostatic units, there follows for 
 Avogadro's number the value N = 6-0617 . 1023. 
 
 39 Of., for example, W. Gibbs' Elements of Statistical Mechanics, 
 Chapter V. 
 
 WThe term "mean value" may be taken as referring to time or to 
 space. If we select a definite atom, and follow it a long time upon its 
 zig-zag path, and from the mean of the values which its kinetic energy 
 assumes in the course of time, we get the " time-mean." If, on the 
 other hand, we select a large number of identical atoms of the gas at a 
 particular instant and again form the mean of the values of the kinetic 
 energies which these atoms possess at the instant in question, we get 
 the " space-mean." 
 
 *1 If x is the elongation of the oscillator (electron) vibrating with the 
 natural frequency, then x = A sin (2mrf), where A is the amplitude and 
 t the time ; the mean kinetic energy becomes
 
 NOTES AND REFERENCES 135 
 
 The mean potential energy is : 
 V 
 
 Hence, as stated, L = V: i.e. the mean kinetic energy = the mean 
 potential energy. 
 
 42 /. H. Jeans, Phil. Mag. 10, 91 (1905). 
 
 MH. A. Lorentz, Proc. Kon. Akad. v. Wet., Amsterdam 1903, p. 666. 
 The theory of electrons (Teubner, Leipzig 1909), Oh. II. 
 
 444. Einstein and L. Hopf, Ann. d. Phys. 33, 1105 (1910). 
 
 UA. D. Fokker, Ann. d. Phys. 43, 810 (1914). 
 
 46 M. Planck, Ber. d. Berl. Akad. d. Wiss., 8 July 1915, p. 512. 
 
 47 H. A. Lorentz. Die Theorie d. Strahlung u. d. Quanten ; Abhand- 
 lungen der Deutschen Bunsen-Gesellschaft. Nr. 7. v. A. Eitcken. 
 Halle, W. Knapp 1914 pp. 10 et seq. 
 
 48 By a suitable modification of classical statistics in the sense of the 
 quantum theory, we can obtain the expression (9) for the mean energy 
 of an oscillator in the following manner which is worthy of notice. 
 Let a number N of similar oscillators with the most varied values for 
 the energy be given. We require to find how great is the probability 
 w, that an oscillator possess a certain energy value U; or, otherwise 
 expressed, how many of the N oscillators possess the energy U. In 
 order to answer this question, we find it best to take first of all the 
 standpoint of Oibbs' statistical mechanics, that is, of " classical " 
 statistics. In place of the special case in question, namely, that of the 
 linear oscillator, let us consider at once quite generally a system of / 
 degrees of freedom, and characterise it by / generalised co-ordinates 
 2i2z ^ and by the corresponding impulses or momenta 2*1 .Pa . . Pf. 
 (Here, the impulse pi is thus defined : form the kinetic energy of 
 
 the system as a function of the generalised velocities qi J-, then 
 
 OT- . <** 
 
 pi = -~r. \ In particular, the linear oscillator (vibrating electron) will 
 
 qi' 
 
 be described by a co-ordinate q, namely, the elongation of the electron, 
 and the impulse p = m -%. In general, therefore, 2/ quantities are 
 
 necessary in order to define completely the momentary state of a 
 system. Hence we can represent this momentary state by a point 
 (" phase-point ") in the 2/- dimensional space in which 9i . . . Pf (of the 
 " phase-space ") are co-ordinates. 
 
 We now consider a number N of similar systems of this kind, 
 which are in thermodynamic equilibrium with a very large reservoir 
 at the temperature T. Then the probability that the co-ordinatea 
 and impulses lie in the small intervals q 1 . . . g t + dq v etc., and 
 Pi . . . Pi + rfpj, etc., that is, that the "phase-point" of the system lie 
 in the element dn = dq^dq^ . . . dqj, dp^p t . . . dpf of the phase- 
 space is, according to Oibbs,
 
 136 THE QUANTUM THEORY 
 
 Here E is the energy of the system, and k is the constant defined in 
 (19). The integration in the denominator is to be taken over all 
 possible values of the 2/ quantities q 1 . . . pf, or, as we may say, over 
 all possible " phases," or over the whole region of the phase-space 
 concerned. 
 
 Among the N systems there are then Nw, whose phase-points lie in 
 the element dfi of the phase-space. This is therefore a " distribution " 
 of the N systems over the phase-space. This distribution is called 
 Canonical ; it represents a generalisation of Maxwell's familiar law of 
 distribution of velocities which may be deduced from it by special- 
 ising it for the case of the gas atom, that is, by setting/ =3. 
 
 The sum of all probabilities is naturally 1. Indeed, it is at once 
 clear that 
 
 L-> 
 
 For the mean value of the energy E we get 
 
 J-M. -/*'**. 
 /."*> 
 
 If we apply this equation to the linear oscillator we get 
 
 / f - 
 TT JjUe Wgqdp 
 
 ffe-^dp 
 N W ' <7 -!' + -0MV 
 
 If we introduce the auxiliary variables { and n, defined by 
 j = TW2m 
 
 j n = ,JL, and hence dqdp = ddn 
 * fj'2m TV 
 
 we get 
 
 U = f 2 + T? 
 and, therefore, it suggests itself to us to write 
 
 f|= 
 
 \rj= 
 
 where <f> is a parametric angle. If we interpret { and rj as Cartesian co- 
 ordinates of a point in the plane, then JU and <p are the polar co-ordinates
 
 NOTES AND REFERENCES 137 
 
 of this point. The element of surface ddrj is written in polar co-ordinates, 
 as we know, thus 
 
 hence 
 
 Hence 
 
 dqdp = 
 
 r r - 
 J Jta M 
 
 tr=0*> = 
 
 in agreement with (24). This is the standpoint of classical statistics. 
 
 The quantum statistics of the oscillator may be immediately deduced 
 from this, if we elaborate the canonical law of distribution 
 
 i - 
 
 Je 
 
 
 
 kT dqdp 
 
 in a suitable manner 
 
 If we here again introduce dqdp = -dUdf, and integrate with respect 
 to $, we get 
 
 _ 
 
 wu= e * r dU 
 
 Je'^dU 
 
 as the probability that the energy of the oscillators lies between U and 
 U H- dU. 
 
 Now the quantum theory demands that the energy U shall assume only 
 the discrete values U , U lt U , . . . Un- The transition may best be 
 effected by laying down the condition : E shall only be able to assume the 
 values contained in the narrow intervals between U and U + a, U^ and 
 I7j + o, and generally Un and U n + o. Then dU = a, and the integral in 
 the denominator changes into a sum. Thus it follows that 
 
 n n 
 
 thus a is eliminated ; if we now proceed to the limit a = 0, w remains
 
 138 THE QUANTUM THEORY 
 
 unaltered. Hence w n is the canonical distribution function generalised 
 for quantum conditions, and hence, among N oscillators, Nw n have an 
 energy of the value U n . 
 
 We now get for the mean energy 
 
 V * 
 
 2ft* v 
 
 Now, according to the first form of the quantum theory, 
 
 U n = tie = nh v (n = 0, 1, 2, 8 . . . oo )". 
 Therefore 
 
 u = 
 
 2 
 
 V 
 
 o S, 
 
 o 
 If we set fc5>, for convenience, = x, then 
 
 Further, 
 
 from which we get 
 
 in agreement with (9) 
 
 The canonical distribution may be still further generalised by the intro- 
 duction of certain "weight factors," which are intended to express the 
 fact that the individual quantum states of the system considered have, 
 a priori, different probabilities. This happens, for example, if each quantum 
 state may be realised in different ways, and if the number of these possi- 
 bilities of realisation is different for the different quantum states. Then, 
 
 the different states will have different "weights," and a "weight factor" 
 
 v n 
 
 p n has to be included in the exper mental function e ~*r so that the can- 
 onical distribution function assumes the form 
 
 - = C . f^ff' 
 Here C depends on the temperature ; p n , on the other hand, doe not.
 
 NOTES AND REFERENCES 189 
 
 194. Einstein, Ann. d. Phys. 17, 132 (1905) ; 20, 199 (1906) ; Verhandl. 
 d. deutsch. physikal. Ges. 11, 482 (1909) ; Bericht Einstein auf dem 
 Solvay-Kongress in Brussels 1911 ; cf. A. Eucken, Die Theorie der 
 Strahlung und der Quanten ; Abhandl. d. deutsch. Bunsen-Gesellschaft, 
 Nr. 7 (Halle, W. Knapp 1914), pp. 330 et seq. Cf. also W. Wien, 
 Vorlesungen iiber neuere Probleme der theoretischen Physik (Teubner, 
 Leipzig and Berlin 1913), 4. Vorlesung. H. A. Lorentz, Les theories 
 statistiques en thermodynamique (Teubner, Leipzig and Berlin 1916), 
 42 et seq. 
 
 SO A. Einstein, Ann. d. Phys. 17, 132 (1905). 
 
 31 A. Einstein, Phys. Zeitschr. 10, 185 (1909). 
 
 52 This formula may be deduced as follows : Firstly, from e = E - E 
 the frequently used relation 
 
 ? = W - 2E . IS + (E)* = W - (IT) 2 
 
 follows. In order now to calculate the two quantities E* (mean of the 
 squares of the energy) and (E)* (square of the mean energy), which 
 are known to differ from each other in general, we do best to take the 
 standpoint of Gibbs' statistical mechanics (see note 48). According to 
 this, the probability that the co-ordinates and impulses lie in the small 
 intervals q l . . . q l + dq lt etc., p^ . . . p 1 + dp lt etc, that is, that the 
 "phase-point" lies in the element dq t dq 2 . . . dqfdp^dp^ . . . dp, = dn 
 of the " phase-space " : 
 
 Then the mean of the energy follows in the usual way : 
 
 Likewise, 
 We then form 
 
 dE
 
 140 THE QUANTUM THEORY 
 
 Therefore, 
 
 dE 
 F-Wjy 
 
 We also arrive at the same formula, if instead of the classical 
 canonical distribution function, we start from the quantum distribution 
 function 
 
 68 The mean energy of radiation of frequency v in the volume v is 
 = vu v dv, where the monochromatic density of radiation is 
 
 if Planck's Law is taken as the basis. (Of., for example, M. Planck, 
 Lectures on the Theory of Radiation, Engl. Transl.) 
 
 According to formula (28) deduced in the previous note, it therefore 
 follows that 
 
 If we eliminate T on the right-hand side by substituting for e** its 
 value 1 + 8vhvS , it follows that 
 
 C'Uy 
 
 u v vdv , 
 
 The second term on the right is required by the Undulatory Theory 
 for at each point of the volume v the most varied trains of waves of 
 radiation cross one another's paths with every possible amplitude and 
 phase. The interference of all these waves thus generates at the point 
 considered an intensity, which varies continually, and hence the energy 
 of the volume v also varies. If we calculate the mean of the square 
 of the energy, i.e. a , wo find precisely the second term of the above 
 formula. (Of., for example, H. A. Lorents, Les theories statistiques en 
 thermodynamique (Teubner, Leipzig and Berlin), 1916, pp. 114 et seq,) 
 
 The first term is not, however, explained by the classical undulatory 
 theory. On the other hand, it becomes endowed with meaning if we 
 suppose that the radiant energy consists of a certain whole number
 
 NOTES AND REFERENCES 141 
 
 (n) of finite energy complexes of the value hv. For then E = n hy, 
 and therefore E = n hv, where n is the mean about which the number 
 n varies. If 5 = n -_ n be the variation of the number n, then it 
 follows that e = E - E = Shy^ where 2 = S 2 feV. But, according to 
 a well-known law of statistics, S 2 = n. (Cf., for example, H. A, Lorentz, 
 loc. cit., 26 and 27.) Hence ? = nfeV = ~E'hv. This is exactly the 
 first term in the above formula. 
 
 MA. Einstein, Ann. d. Phys. 17, 144 (1905). 
 
 95 J. J. Thomson, Conduction of Electricity through Gases. 
 
 96.4. Einstein, Ann. d. Phys. 17, 147 (1905). 
 
 97 Of. R. Pohl and P. Pringsheim, Die lichtelektrischen Erschein- 
 ungen. Sammlung Vieweg Heft 1 (Braunschweig 1914). 
 
 98 A. Einstein, Ann. d. Phys. 17, 145 (1905). 
 
 99 R. A. Millikan, Phys. Zeitschr. 17, 217 (1916). 
 
 60 According to Pohl and Pringsheim, we have to distinguish between 
 the normal and the selective photo-effect : in the case of the normal 
 effect the number of electrons torn off (per calorie of the Jight-energy 
 absorbed) is independent of the orientation of the electrical vector of the 
 light-wave, and increases, starting from an upper limit of the wave- 
 length, in general uniformly as the wave-length decreases. In the case 
 of the selective effect, on the other hand, which only appears when the 
 electrical vector of the light-wave possesses a component vertical to the 
 metallic surface, the number of electrons torn off (per calorie of light- 
 energy absorbed) shows a decided maximum at a definite wave-length. 
 
 61 Oh. Barkla, Phil. Mag. 7, 543, 812; 15, 218. Jahrb. d. Radioak- 
 tivitat u. Elektronik, 5, p. 239, 1908. Ch. Barkla and Sadler, Phil. Mag. 
 17, 739. Ch. Barkla, Jahrb. d. Radioaktivitat u. Elektronik, 1910, p. 
 12. if. de Broglie, G. R. 25 May and 15 June 1914, p. 1785. Ch. Barkla, 
 Phil. Mag. 16, 550. E. Wagner. Ann. d. Phys. 46, 868 (1915); Sit- 
 zungsber. d. bayer. Akad. 1916, p. 39. 
 
 62 D. L. Webster, Proc. Americ. Acad. 2, 90(1916); Physic. Review, 7, 
 587 (1916). 
 
 63 E. Wagner, Ann. d. Phys. 46, 868 (1915). 
 
 64 Of., for example, E. Wagner, Phys. Zeitschr. 18, 443 (1917). The 
 value that Wagner calculates for h is : h = 6-62 .10-27. 
 
 65 W. Duane and F. L. Hunt, Physic. Review, 6, 166 (1915). 
 86.4. W. Hull and If. Rice, Proc. Americ. Acad. 2, 265 (1916). 
 
 67 E. Wagner, Phys. Zeitschr. 18, 440 et seq. (1917) ; Ann. d. Phys. 57, 
 401 (1918). 
 
 68 F. Dessauer and E. Back, Ber. d. deutsch. physikal. Qes. 21, 168 
 (1919). 
 
 69 J. Franck and G. Hertz, Verhandl. d. deutsch. physikal. Ges. 16, 
 512 (1914). 
 
 70 The critical potential measured by Franck and Hertz amounted to 
 
 V= 4-9 volts = electrostatic units, and therefore the critical energy 
 
 300 
 of the electron is 
 
 v _ 4-774. 10 -10. 4-9 
 300
 
 142 THE QUANTUM THEORY 
 
 The wave-length A of the mercury line emitted is 
 \ = 25364 = 2-536.10-5. 
 Hence we must get 
 
 17 j.e i. V* 4-774. 10 -10. 4-9. 2-536. 10 -5 
 ,7= fc-, ,.e. h = = - 3 . 102-3 ^r 
 
 = 6-59.10-27 
 
 and this is in good agreement with the results of other measurements. 
 
 71 Cf., for example, J. Stark, Prinzipien der Atomdynamik II. (S. 
 Hirzel, Leipzig 1911), Chs. IV and V. 
 
 72 J. Stark, Ber. d. deutsch. phys. Ges. 10, 713 (1908) ; Phys. Zeitschr. 
 8, 913 (1907) ; 9, 767 (1908). 
 
 Canal-rays are positively charged particles of matter, which move in a 
 vacuum tube in the direction : anode to cathode ; the latter is pierced 
 with holes through which the canal-rays pass into the space behind the 
 cathode. If we generate such canal-rays in a vacuum tube rilled with 
 hydrogen, we find that the series lines of hydrogen are emitted. Now, 
 if we observe this emission spectroscopically " from the front," that is, so 
 that the canal-rays are moving towards the observer, we see, firstly, at 
 its usual place in the spectrum, the sharp series line (line of rest, "in- 
 tensity of rest ") ; secondly, we see displaced towards the violet, a 
 broadened strip (line of motion, "intensity of motion" or "dynamic 
 intensity"). These lines represent the series line emitted by the 
 moving canal-ray particles, which is displaced towards the region of 
 higher frequencies on account of the Doppler effect. Since the canal-rays 
 do not possess a single uniform velocity, and since particles with all 
 possible velocities occur, the displaced strip is not sharp, but softened 
 and broadened. The " intensity at rest " is therefore emitted when the 
 quickly moving canal particles strike " resting " molecules, i.e. gas- 
 molecules which are moving comparatively slowly and irregularly, and 
 excite these to emit the series lines. The " intensity of motion," on the 
 other hand, is excited by the unidirectionally moving canal particles 
 themselves, when they hit gas-molecules. 
 
 Now, it is very remarkable that the interval between the intensity of 
 rest and that of motion is not filled in, but that the emission of the in- 
 tensity of motion becomes observable only above a certain velocity. 
 Stark interpreted this fact in terms of the light-quantum hypothesis 
 thus : If %mv z is the kinetic energy of a canal-ray particle, and if the 
 fraction arau 2 (a > 1) is transformed into a light-quantum hv upon 
 collision with a gas-molecule, then we must have h < -me 2 ; that is, the 
 spectral line of frequency v can be generated only by canal-rays, the 
 
 Telocity of which >A/?5 
 \ can 
 
 The proportionality between the critical velocity and *Jv has been 
 fairly well borne out. 
 
 It should be remarked here that J. Stark has lately abandoned the 
 theory of light-quanta. (Cf. J. Stark, Verh. d. deutsch. physik. Ges. 16, 
 304 (1904) ; 18, 42 (1916).)
 
 NOTES AND REFERENCES 148 
 
 73 J. Stark, Phys. Zeitschr. 9, 85, 356 (1908). J. Stark and W. 
 Steubing, Phys. Zeitschr. 9, 481 (1908). J. Stark, Phys. Zeitschr. 9, 
 889 (1908). 
 
 In these papers J. Stark defends the view that the band-spectra are 
 emitted when a "valency electron" belonging to the atom or molecule 
 is pushed out of its normal position and then returns again to its initial 
 position, counterbalancing the work done in displacement. If the 
 energy of deformation (valency energy) E is changed into a light- 
 
 quantum, then we must have hv = E, i.e. v ^- All lines of the 
 
 ^ 7t 
 
 ET 
 
 band must therefore lie below the edge v = _ . If the valency energy 
 
 E is changed by chemical processes, the band-spectrum must be dis- 
 placed accordingly. 
 
 74,1 Einstein, Ann. d. Phys. 17, 148 (1905). 
 
 75 J. Stark, Phys. Zeitschr. 9, 889 (1908) ; Ann. d. Phys. 38, 467 (1912). 
 
 The fundamental law of photochemical decomposition enunciated by 
 Stark and Einstein states : If a molecule dissociates at all owing to j,he 
 absorption of radiation of frequency v, then it will absorb an amount of 
 energy hv when it dissociates. This energy, therefore, represents the 
 heat of reaction, which will be set free upon recombination of the 
 products of decomposition. 
 
 This law was later deduced by A. Einstein for the range of validity of 
 Wierfs Law of Radiation without the assistance of the light-quantum 
 hypothesis, by purely thermodynamical methods. (Of. Ann. d. Phys. 
 37, 832 (1912), and 38, 881 (1912).) 
 
 nE. Warburg, Ber. d. Berl. Akad. d. Wiss. 1911, p. 746; 1913, p. 
 644 ; 1914, p. 872 ; 1915, p. 230 ; 1916, p. 314 ; 1918, pp. 300, 1228. Cf. 
 also " Naturwissenschaften," 5, 489 (1917). 
 
 77 H. A. Lorentz, Phys. Zeitschr. 11, 1250 (1910). 
 
 78 M. Planck, Ber. d. deutsch. physikal. Ges. 13, 138 (1911); Ann. d. 
 Phys. 37, 6i2 (1912). 
 
 79 On account of the continuous (classical) absorption, all energy values 
 of the oscillator in an elementary region, say between n and (n + l)e, 
 are equally probable. The mean energy in the nth elementary region 
 is, therefore, 
 
 Prom the canonical law of distribution extended in the sense of the 
 quantum theory, it then follows that
 
 144 THE QUANTUM THEORY 
 
 
 (cf. note 48). If we further set e = hy it follows that 
 
 In place of relation (7) of the text we get here 
 
 and this leads to Planck's Law of Radiation. 
 
 80 M. Planck, Sitzungsber. d. Kgl. Preuss. Akad. d. Wiss. 3 April, 
 1913, p. 350; ibid., 30 July 1914, p. 918; ibid., 8 July 1915, p. 512. 
 
 81 A. Einstein and O. Stern, Ann. d. Phys. 40, 551 (1913). 
 
 82 W. Nernst, Verhandl. d. deutsch. physikal. Ges. 18, 83 (1916). 
 
 83 F. Richarz, Wiedem. Ann. 52, 410 (1894J. 
 
 84 Report by P. Langevin at the Solvay Congress in Brussels, 1911. 
 Cf. A. Eucken, Die Theorie der Strahlung und der Quanten. Abhandl. d. 
 deutsch. Bunsen-Ges., Nr. 7 (W. Knapp, Halle 1914), pp. 318 et seq. 
 
 8M. Einstein and W. J. de Haas, Verhandl. d. deutsch. physikal. Ges. 
 17, 152, 203, 420 (1915). A. Einstein ibid., 18, 173 (1916). W. J. de 
 Haas, ibid., 18, 423 (1916). 
 
 86 E. Beck, Ann. d. Phys. 60, 109 (1919). 
 
 87 Report by Planck at the Solvay Congress in Brussels, 1911. See 
 A. Eucken, Die Theorie der Strahlung und der Quanten. Abhandl. d. 
 deutsoh. Bunsen-Ges., Nr. 7 (W. Knapp, Halle 1914), p. 77. 
 
 88 If q is the elongation of a linearly vibrating electron of mass ra (os- 
 cillator) and v its period of oscillation, then the energy of this configur- 
 ation is 
 
 The first term represents the kinetic and the second the potential energy. 
 Now the impulse (the momentum) is p = m ~Ti- Therefore, we may write 
 
 i.e.
 
 NOTES AND REFERENCES 145 
 
 The curves U = const., that is, those curves in the phase-plane, which 
 correspond to the states of constant energy of the oscillator, are therefore 
 ellipses with the semi-axes 
 
 For a definite value of U we get a completely definite ellipse. The 
 " phase-point " of the oscillators would continually revolve in this ellipse, 
 if the electron, without emitting or absorbing, were to execute pure har- 
 monic oscillations : for then its energy would remain permanently constant. 
 If we allow U to vary continuously, i.e. if we give it other and again 
 other values in continuous succession, we get an unlimited manifold of 
 concentric ellipses. 
 
 The quantum theory, as formulated in (30) in the text, selects from this 
 infinite manifold a discrete set of ellipses, and distinguishes them as the 
 "quantised" ellipses which correspond to the " characteristic states " of 
 the oscillator. To these belong the "quantum energy- values " U , U lt 
 U a . . . Dn. 
 
 Now the nth ellipse encloses an area nh. The area of the nth ellipse 
 is, however, 
 
 hence we must have 
 
 = nh i.e. U n = 
 
 that is, in the nth quantum state the oscillator possesses an amount of 
 energy nt = nhv. 
 
 tOA. Sommerfeld, Phys. Zeitschr. 12, 1057 (1911). Report by A. 
 Sommerfeld at the Solvay Congress in Brussels, 1911. Cf. A. Eucken, 
 Die Theorie der Strahlung und der Quanten. Abhandl. d. deutsch. 
 Bunsen-Ges., Nr. 7 (W. Knapp, Halle 1914), p. 252. 
 
 80 Report by Sommerfeld at the Solvay Congress, 1911. 
 
 91 A. Sommerfeld and P. Debye, Ann. d. Phys. 41, 873 (1913). 
 
 92 Cf., for example, the recent summary by E. Schrddinger, Der 
 Energieinhalt der Festkorper im Lichte der neueren Forschung. Phys. 
 Zeitschr. 20, 420, 450, 474 (1919). A complete set of references accom- 
 panies this account. 
 
 93 One gramme-atom of a substance, the atomic weight of which is a, 
 is defined as the quantity a grammes of the substance. For example, 
 one gramme-atom of copper is equal to 63'57 grammes of copper, since 
 63-57 is the atomic weight of copper. Exactly analogous is the 
 definition of the gramme- molecule (also called " mol"). One gramme- 
 molecule of oxygen is 32 grammes of oxygen, for the molecular weight 
 of oxygen (diatomic) is 32. 
 
 If c is the specific heat of a substance of atomic weight a, it signifies 
 that one gramme of the substance requires an amount of heat c to raise 
 its temperature by 1 C. Hence we must communicate to a gramme-atom 
 10
 
 146 THE QUANTUM THEORY 
 
 of the substance, i.e. to a grammeB of it, an amount of heat C = ea in 
 order to raise its temperature by 1 C. C is then called the atomic heat. 
 
 94 The equality of the mean potential and the mean kinetic energies is 
 true here as in the case of the linear Planck oscillator (vibrating electron), 
 of. note 41. This equality is, in general, always present when the forces 
 which act upon the atoms and restore them to their positions of rest 
 (zero positions) are Ivnea/r functions of the relative displacements of the 
 atoms, that is, when the force is "quasi-elastic," that is, proportional 
 to the displacement from the zero position. Cf. in this connexion 
 L. Boltzmann, Wiener Ber. 63 (11), 731 (1871), and F. Richarz, Wied. 
 Ann. 67, 702 (1899). 
 
 WDulong and Petit, Ann. de chim. et de phys. 10, 395 (1819). 
 
 96 The quantity usually obtained by measurement is not the atomic 
 heat at constant volume C,,, but the atomic heat at constant pressure C f . 
 For this we get values which in general fluctuate about the value 6'4 
 cal./deg. The calculation of C c from C p is based on the thenno- 
 dynamically deduced formula 
 
 C p - C, = a * VT 
 
 where a is the cubical coefficient of thermal expansion, K the (isothermal) 
 
 cubical compressibility, and V the atomic volume = atomic wei g ht . 
 
 density 
 
 97 E.g. we find 
 
 for silver at C C p = 6-00 
 
 aluminium 58 C p = 5-82 
 
 copper 17 C C p = 5-79 
 
 lead 17 C p = 6'33 
 
 iodine ,, 25 C C p = 6-64 
 
 zinc 17 C C p = 6-03 
 
 98 F. E. Weber, Poggend. Ann. 147, 311 (1872) ; 154, 367, 553 (1875). 
 
 99 As a possible way out, the "agglomeration hypothesis," supported by 
 F. Richarz [Marburger Ber. 1904, p. 1], C. Benedicks [Ann. d. Phys. 42, 
 183 (1913)] and others, has been put forward. According to this, as the 
 temperature falls the number of degrees of freedom of the system 
 diminishes by " freezing-in," as it were, in that certain linkages become 
 completely rigid. According to this, however, the compressibility should 
 decrease greatly as the temperature falls, which, according to E. Orii- 
 neisen's measurements is not the case [Verb. d. deutsch. phys. Ges. 13, 
 491 (1911)]. Compare also in this connexion the report of E. Schro- 
 dinger quoted in note 92. 
 
 100.4. Einstein, Ann. d. Phys. 22, 180, 800 (1907). 
 
 101 Cf. A. Einstein, Ann. d. Phys. 35, 683 ff. (1911), also the report by 
 Einstein at the Solvay Congress in Brussels, 1911 ; see A. Eucken, Die 
 Theorie der Strahlung und der Quanten. Abhandl. d. deutsch. Bunsen-Ges., 
 Nr. 7 (W. Knapp, Halle 1914), pp. 330 et seq. 
 
 1024. Einstein, Ann. d. Phys. 34, 170, 590 (1911) ; 35, 679 (1911). 
 
 IWThe nature of the dependence of the frequency v on the three
 
 NOTES AND REFERENCES 147 
 
 quantities A, p, K may, according to Einstein (loc. cit.), be obtained by a 
 simple dimensional calculation. If we assume that v depends only on the 
 mass in of the atoms, their distance apart d, and the compressibility K of 
 the body, then an equation of the following form must hold 
 
 v = C . m x . & . K z . 
 
 C is here a numerical constant ; x, y and z are numbers which remain to 
 be determined. 
 
 The dimensions of the frequency \y] are \t - 1] ; the dimensions of m and 
 d are \ni] and [i], and the dimensions of the compressibility K follow from 
 its definition : 
 
 _ _ increase in volume _ 
 increase in pressure x original volume 
 
 K has therefore the dimensions 
 
 I! = [-surface "I 
 ->"-" ' L force J L 
 
 We thus get the following dimensional equation 
 
 t~ l 
 Hence 
 
 x - 2 = 0; y -f z = ; "2z = - 1 
 from which we get 
 
 x- -J; y- +4; 2- -4 
 
 We have therefore, 
 
 Let N be Avogadro's number, i.e. the number of atoms in the gramme- 
 atom. Then the atomic weight of the body is numerically equal to the 
 mass of the gramme-atom, i.e. 
 
 A = mN. 
 
 If we imagine the atoms arranged upon a cubical space-lattice with 
 sides d, then the density must satisfy the equation 
 
 from this it follows that 
 
 and hence 
 
 i. 
 
 from which, it follows that
 
 148 THE QUANTUM THEORY 
 
 Einstein determines the factor C by assuming simply that only the 
 twenty-six neighbouring atoms act upon the displaced atom. 
 
 IMF. A. Lindemann, Phys. Zeitschr. 11, 609 (1910). Lindemann's 
 formula may be shortly deduced thus : Let r = a sin (2-irvt) be the elonga- 
 tion of an atom which is vibrating with the amplitude a and the frequency 
 y. The mean energy of this atom is 
 
 E ' f (ar)" + f ' 
 
 At the melting-point, according to Lindemann's conception, a is of the 
 same order as d (distance apart of atoms). On the other hand, the mean 
 energy of the atoms at high temperatures = 'SkT, or, at the melting-point 
 3kT s . (The melting-point, as a rule, is high.) From this it follows that 
 
 = SkT, 
 
 But we have (see note 103) 
 
 m =^j- d = m*,-i = AlN-*f-*> 
 Hence 
 
 v = const. I 7 ,* . 4~*JSM-*V = const. T,* . 4~* . />*. 
 
 105 E. Qrilneisen, Ann. d. Phys. 39, 291 et seq. (1912). 
 
 106 E. Madelung, Nachr. d. kgl. Ges. d. Wiss. zu Qottingen, mathem.- 
 physikal. Klasse 1909, p. 100, and 1910, p. 1. 
 
 107 W. Sutherland, Phil. Mag. (6), 20, 657 (1910). 
 
 108 If n and K are the coefficients of refraction and extinction of a 
 substance respectively, then, according to Maxwell's Theory, its reflect- 
 ing power is 
 
 n - 
 
 If we require the point of maximum reflection, we have to form the 
 equation ^^ = 0, which gives after reduction the following relation : 
 
 From this we see that the position of maximum reflection does not 
 
 coincide exactly with the position of maximum absorption (^ = Oj, 
 
 \ov / 
 
 but that it lies the nearer to it, the less the coefficient of refraction 
 varies with the frequency. On the other hand, the point of maximum 
 -absorption lies, according to the dispersion theory, in the immediate 
 ^neighbourhood of the natural frequency v,.. 
 
 109 H. Rvtens and E. F. Nichuh, Wiedem. Ann. 60, 418 (1897). Also
 
 NOTES AND REFERENCES 
 
 149 
 
 77. Rubens and H. Hollnagel, Ber. d. kgl. preuss. Akad. d. Wiss. 1910, 
 p. 45; //. Hollnagel, Dissert. Berlin 1910; H. Rubens, Ber. d. kgl. 
 preuss. Akad. d. Wiss. 1913, p. 513 ; H. Rubens and H. v. Wartenberg, 
 ibid., 1914, p. 169. 
 
 As an example we give here the following small table in which \ 
 denotes the wave-length of the " residual " rays, as given by the above 
 investigators. 
 
 
 A. 
 
 
 A. 
 
 NaCl 
 KC1 
 AgCl 
 HgCl 
 
 52 M 
 
 63-4 M 
 81 -5 M 
 98-8/i 
 
 T1C1 
 KBr 
 AgBr 
 TIBr 
 
 91-6 M 
 82-6 M 
 112-7 M 
 117 M 
 
 HO Of., however, note 108. 
 
 111 W. Nernst and F. A. Lindemann, Sitzungsber. d. kgl. preuss. 
 Akad. d. Wiss. 1911, p. 494 ; W. Nernst, Ann. d. Phys. 36, 426 (1911). 
 
 112 The following short table gives the values for v which are calculated 
 from Einstein's formula (35), Lindemann's formula (36), from the 
 "residual rays" (see note 109), and from the observed atomic heat 
 according to an empirical formula (40) proposed by Nernst and 
 Lindemann. For more detailed data with, in part, corrected numerical 
 factors see C. E. Blom, Ann. d. Phys. 42, 1397 (1913). 
 
 Substance 
 
 V E 
 
 "L 
 
 ''residual rays 
 
 ''atomic heat 
 (Nerntt-Lindemann) 
 
 Al 
 
 6-7 . 10 12 
 
 7-6 . 10" 
 
 
 8-3 . 10" 
 
 Cu 
 
 5-7 . 10 12 
 
 6-8 . 10 12 
 
 
 6-7 . 10 12 
 
 Zn 
 
 
 4-4 . 10 12 
 
 
 4-8 . 10 12 
 
 Ag 
 
 4-1 . 10 12 
 
 4-4 . 10 12 
 
 
 4-5.10 12 
 
 Pb 
 
 2-2 . 10 12 
 
 1-8. 10 12 
 
 
 1-5 . 10 12 
 
 Diamond 
 
 
 32-5 . 10 12 
 
 
 40 . 10 12 
 
 NaCl 
 
 
 7-2. 10 12 
 
 5-8 . 10 12 
 
 5-9 . 10 12 
 
 KC1 
 
 
 5-6 . 10 12 
 
 4-7 . 10 12 
 
 4-5 . 10 12 
 
 113 W. Nernst, F. Koref, F. A. Lindemann, Untersuchungen iiber die 
 spezifische Warme bei tiefen Temperaturen. I. u. II. Sitzungsber. d. 
 kgl. preuss. Akad. d. Wiss. 1910, 3 March. W. Nernst, idem III., ibid., 
 1911, 9 March. F. A. Lindemann, idem IV., ibid., 1911, 9 March. W. 
 Nernst and F. A. Lindemann, idem V., ibid., 1911, 27 April. W. Nernst 
 and F. A. Lindemann, idem VI., ibid., 1912, 12 Dec. W. Nernst, idem 
 Vn., ibid., 1912, 12 Dec. W. Nernst and F. Schwers, idem VIII., loc. 
 tit., 1914. W. Nernst, Der Energieinhalt fester Stoffe. Ann. d. Phys. 
 36, 395 (1911). 
 
 HI W. Nernst, Die theoretischen und experimentellen Grundlagen des 
 neuen Warmesatzes. (W. Knapp, Halle 1918.) 
 
 115 The First Law states : If d'Q is the heat supplied to a system, d'A
 
 150 THE QUANTUM THEORY 
 
 the work done on the system from outside, then the increase of energy 
 U of the system is given by 
 
 dU = d'Q + d'A. 
 
 The Second Law states : if d'Q is supplied reversibly at the temperature 
 T, then ^ is the complete differential of the entropy S, hence 
 
 Let us follow Helmholtz and introduce the " free energy" F denned by 
 
 F= U- T- S. 
 Then it follows that 
 
 dF = dU - T.dS - S.dT=d'Q + d'A - T.dS - S.dT 
 
 i.e. dF = d'A - S . dT 
 
 for every reversible process. 
 
 If the process is isothermal (dT = 0) then it follows that dF = d'A 
 or, for a finite change of state, F. A - F l = A. If we set A' = - A, so 
 that A' is the work gained, we get 
 
 Fj - F, = A'. 
 
 That is, the work gained in the isothermal reversible process which is, 
 as may be shown, the maximum obtainable is equal to the decrease of 
 free energy. 
 
 Further, it follows, since at constant volume F the work d'A = 0, 
 that 
 
 Therefore, formulating these expressions for two states, we get 
 /pjOffii - "a)~| _ ^ _ p^ _ ^J7 } _ {/ 2 j 
 
 or, finally, if we write for short t7j - U z = U' 
 
 an equation much used in physical chemistry. 
 Since, now, according to Nernst's heat theorem, 
 
 ('),. =o 
 
 \OT Jhm T=Q 
 (A' - U') vanishes for T = 0, being above the first order. 
 
 Hence lim ?L4' ~ . ^') = Q 
 
 and hence also
 
 NOTES AND REFERENCES 151 
 
 This is equation (89) of the text. 
 
 From -= - S, it follows further that ^ F ] ' " F ^ - S 2 - 8 lt or 
 
 Q _ O _ 9-4' 
 
 and hence Nernst's Theorem may be formulated thus 
 
 lim (S 2 - S,) = 
 
 r=o 
 
 that is, in /; neighbourhood of the absolute zero all processes proceed 
 without change of entropy. 
 
 116 Cf., for example, M. Planck, Lectures on Thermodynamics. Planck 
 goes further than Nernst inasmuch as he postulates that not only the 
 difference of the entropies S 2 - S x is zero at absolute zero (see previous 
 note) but also that the individual values themselves become zero. Hence, 
 according to Planck, at the absolute zero of temperature the entropy of 
 every chemically homogeneous body is equal to zero. From this the con- 
 clusion given in the text, 
 
 lim^^,j = 
 
 may be deduced immediately. It follows from the relation (occurring 
 in the last note) 
 
 F - U = - TS 
 
 and from Planck's version of Nernst's Theorem, that F - 0" vanishes for 
 
 T = 0, being of higher order than the first. 
 
 Hence 
 
 = or limf^-H S) 
 T=o\oT ) 
 
 or, finally, 
 
 117 For low temperatures, that is, for high values of x = ^ Einstein's 
 
 formula (34) takes the following form : C v = 3Rx*e~ x . The falling-off at 
 low temperatures therefore follows an exponential law ; more exactly, it 
 varies as 
 
 i const 
 
 118 W. Nernst and F. A. Lindemann, Sitzungsber. d. kgl. preuss. Akad. 
 d. Wiss. 1911, p. 494 ; Zeitschr. f. Elektrochemie, 17, 817 (1911). 
 
 119 A. Einstein, Ann. d. Phys. 35, 679 (1911). 
 
 120 For if we regard the atoms as mass-points, then each atom hao three 
 degrees of freedom ; the whole body has therefore '3N degrees of freedom. 
 As is proved in mechanics, however (cf. R. H. Weber and R. Oans,
 
 152 THE QUANTUM THEORY 
 
 Repertoriuin der Physik Bd. I. pp. 175 et seq.), a mechanical system of 
 3N degrees of freedom has 3N natural frequencies, and the moit general 
 small motion of each atom consists in a superposition of these 3N natural 
 frequencies. 
 
 121 P. Debye, Ann. d. Phys. 39, 789 (1912). 
 
 122 M. Born and Th. v. Kdrmdn, Phys. Zeitschr. 13, 297 (1912); 14, 
 15, 65 (1913). Of. also M. Born, Ann. d. Phys. 44, 605 (1914) ; M. Born, 
 Dynamik der Kristallgitter (Teubner, Leipzig and Berlin 1915). 
 
 123 Of., for example, R. Ortvay, tjber die Abzahlung der Eigenschwin- 
 gungen fester Korper. Ann. d. Phys. 42, 745 (1913). 
 
 Ortvay considers the natural frequencies of an elastic cube, each side 
 of which has the length L. There are found to be three groups of natural 
 frequencies. The first two groups are the transversal frequencies, the 
 third group is the group of the longitudinal frequencies. That the trans- 
 versal frequencies form two groups (moreover identical) is easily seen. 
 For in the case of a transversal vibration, which is propagated in, say, the 
 direction of the aj-axis, two equal alternatives are probable, namely, that 
 the particles vibrate parallel to the y- or to the 2-axis. In the case of the 
 longitudinal oscillations, however, there is naturally only one group ; for 
 in the case of propagation along the z-axis there is only one possibility, 
 namely, that the particles vibrate parallel to the x-axis. The frequencies 
 of the first two groups are characterised by the values 
 
 the third group by 
 
 Here c t and c t are the velocities of propagation of transversal and longitu- 
 dinal waves in the body, whereas a, b, c are arbitrary positive whole 
 numbers. If therefore we give a, b, c all possible values in all possible com- 
 binations, we get all the possible transversal and longitudinal natural fre- 
 quencies, which together form the elastic spectrum of the cube. If now 
 we inquire how many transversal natural frequencies of the first group 
 fall below v, this means nothing else than inquiring how many trios of 
 values (a, b, c) fulfil the condition 
 
 Imagine a, b, c as co-ordinates of a point in space. Then all possible trios 
 (a, b, c) of values are represented by the total "lattice-points" of the 
 positive space octant, and the above question is answered by counting how 
 
 many lattice-points are at a distance less than ''from the origin (0, 0,0). 
 
 c
 
 NOTES AND REFERENCES 158 
 
 All these lattice-points lie within the positive octant of the sphere whose 
 
 radius is -. Since now one lattice-point is assigned to every volume of 
 
 c t 
 
 magnitude 1 namely, every elementary cube the required number of 
 lattice-points, provided that it is sufficiently large, is equal to the volume 
 
 of the positive spherical octant of radius ", i.e. is equal to 
 
 If V = L 3 is the volume of the given cubical body, then the number 
 of the transverse natural frequencies below v belonging to the first 
 group is 
 
 Zl = T F cJ 
 
 The number belonging to the second group is the same, that is 
 
 rr r? 4ir T r V 
 
 Finally, the number of the longitudinal frequencies corresponding to 
 these is 
 
 4ir v"' 
 
 We thus get for the total of all natural frequencies below v 
 
 The total of natural frequencies in the interval v . . . v + dv follows by 
 differentiation with respect to v 
 
 and this is just formula (43) of the text. 
 
 124 In formula (43) for Z(v)d v let us replace, according to formula (44) 
 of the text, the factor 
 
 Then it follows that
 
 154 THE QUANTUM THEORY 
 
 If we now aet - and - m , we get 
 
 128 A table showing how the Debye function C depends on x m is given 
 by Nernst (Die theoretischen und experimentellen Grundlagen des neuen 
 Warmesatzes. W. Knapp, Halle 1918, p. 201). In it the simple 
 Einstein function [formula (34) of the text] is also tabulated. 
 
 126 If T is great, then x m is small compared with 1 ; then we may 
 replace in the integral of (45) e x by 1 in the numerator, and e x - 1 by x 
 in the denominator. It then follows that 
 
 127 If Tia small, then x m is large, and we may replace the upper 
 limit of the integral as a first approximation by oo . The integral will 
 thus become a numerical constant independent of x m , and it follows that 
 
 C. - j* . const. - 9 - . T3 . const. 
 128 From the theory of elasticity it follows that 
 and c = 
 
 J HE 
 \H-o 
 
 where K is the compressibility, p the density, and a the ratio 
 
 transverse contraction 
 longitudinal dilatation* 
 
 If we insert these values in (44) and note further that V = -, formula 
 
 (46) of the text follows. 
 
 129 As the number of frequencies below v is proportional to v*, we get, 
 for example, the following picture : if we divide the interval from to 
 v m into 10 parts, and if only one natural frequency lies in the first 
 division, then in the following divisions there will be 7, 19, 37, 61, 91, 
 127, 169, 217, 271 natural frequencies ; i.e. the natural frequencies crowd 
 continually closer together. 
 
 180 P. Debye, Ann. d. Phys. 39, 789 (1912) ; W. Nernst and F. A. 
 Lind&mann, Sitzungsber. d. Berl. Akad. d. Wiss. 1912, p. 1160. 
 
 131 A. Eucken and F. Schwers, Verhandl. d. deutsch. physikal. Ges. 
 15, 578 (1913) ; W. Nernst and F. Schwers, Sitzungsber. d. Berl. Akad. d. 
 Wiss. 1914, p. 355 ; P. Grttnther, Ann. d. Phys. 51, 828 (1916) ; W. H.
 
 NOTES AND REFERENCES 155 
 
 Keesom and Kmnerlingh-Onnes, Amsterdam Proc. 17, 894 (1915). Cf. 
 also the graphic tables by E. SchrOdinger, Phys. Zeitschr. 20, 498 (1919). 
 182 If we introduce into equation (44) of the text, 
 
 a " mean acoustic velocity " c, by the obvious definition 
 
 then for the order of magnitude of the smallest wave-length \ m i n , there 
 follows 
 
 c 
 
 It now the atoms in the cubical space-lattice, for example, are arranged 
 so as to be a distance a apart, then Na* = V, and hence 
 
 VI- 
 
 Amin 
 
 \ 3 
 
 133 For references see note 122. 
 131 Of. Born, Dynamik der Kristallgitter, 19. 
 133 .F. Haber, Verb. d. deutsch. phys. Ges. 13, 1117 (1911). 
 For if the atomic residue (mass m) and the electron (mass /i) are held 
 to their zero positions by forces of the same order of magnitude, and if 
 they vibrate independently of one another (a simplifying supposition) the 
 equation of vibration of the atom is nix + a?x = 0, the solution of which is 
 
 x = A sin( ~~F^t ). The infra-red frequency of the atom is, therefore, 
 
 a 
 y r = 2a- / , and correspondingly, the ultra-violet frequency of the electron 
 
 Hence Habeas Law follows : v r : v v = <s//x : N/. The 
 
 general space-lattice theory of M. Born confirms this law and shows that 
 in the lattice, too, atomic residues and electrons appear upon an equal 
 footing, and are acted upon by forces of the same order of magnitude. 
 
 136 Cf. M. Born and Thos. v. Kdrmdn, Phys. Zeitschr. 13, 297 (1912). 
 
 We may treat this problem, which is of course one-dimensional, most 
 simply thus : If we imagine an endless chain of points of equal mass TO 
 disposed along the x-axis at a distance apart a, and if we suppose for 
 simplicity that each mass-point only acts upon its two neighbours, then 
 the equation of motion of the nth point is 
 
 mx n = a(x n+ i - X n ) - a(x lt - SC,,_i) = a(x n +i + X_i - 2 n ). 
 Here o is a constant, and n can assume all values between + o>and - <n . 
 As a solution let us set for trial 
 
 x n = A sin 2*yt - n\
 
 156 THE QUANTUM THEORY 
 
 This represents a process which is periodic in space and time, that is, a 
 wave which is propagated along the chain in the direction of increasing 
 x. The frequency of this wave is v, its length is A. Then, if after p 
 points the same displacement is to recur, pa must = A, and hence it 
 actually follows that 
 
 f 2*-a" 
 
 x n+J> = A sin 2irvt - (n + p) 
 
 In order to find the relation between v and \ (that is, the " law of dis- 
 persion"), let us insert the above formula in the equation of motion. 
 Then it follows that 
 
 - (n + 1) 1 + sin [znt - (n - 1) 1 
 
 * J L A J 
 
 -n^l\ 
 * Ji 
 
 - 2*1 sin 
 
 - *] . (l - cos *2\ 
 \ J \ A / 
 
 That is, 
 
 = * J~sin ()- , m sin (Y if we set LJt = , m . 
 W \w \A/ \A/ ir Vro 
 
 187 Of. .Bom, Dynamik der Kristallgitter, p. 51. 
 
 From the special case treated in the previous note, we also recognise 
 the truth of law (49) ; for if A is much greater than a, the dispersion law 
 
 takes the form v = n , where q = v m ita, represents the velocity of 
 
 A A 
 
 propagation of the wave, and this is independent of the wave-length. 
 
 138 The statement that a given direction lies in the element of solid 
 angle dti is intended to convey the following sense : about an arbitrary 
 origin O describe a " unit sphere," i.e. a sphere of radius 1. Now let a cone 
 of infinitely small angle be constructed of rays passing through 0, the point 
 of the cone lying at O. Let this cone cut out of the surface of the unit 
 sphere a small element of surface dn. Now let the parallel ray to the 
 "given direction" be drawn through O (here, for example, the wave- 
 normals). If this ray lies in the cone just constructed, then we say that 
 the " given direction " lies in the elementary solid angle dfl. 
 
 139 The capacity for heat of a certain finite body is that amount of heat 
 which must be imparted to the whole body in order that its temperature 
 be raised by 1 C. If M is the mass of the body, and c its specific heat, 
 then its capacity for heat is
 
 NOTES AND REFERENCES 157 
 
 From the mean energy content E of the whole body, r follows by 
 differentiation with respect to the temperature 
 
 r-* 1 
 
 140 This somewhat complicated calculation runs as follows : we start 
 from the formula 
 
 and first replace \ by g . Thus we get 
 
 and the integral with respect to A. is transformed into one with respect to 
 vi. The limits of this integral are 
 
 vi = g*'( n ) [corresponding to \ = A^n)] 
 
 and 
 
 vi = (corresponding to A. = oo ). 
 If we further set 
 
 we get 
 
 3 47T *<> 
 
 fe'T 3 ^ C dn ;'x 4 e*dx 
 
 In place of the quantities gi(i) and x(n) which still depend essenti- 
 ally on the direction, let certain mean values be introduced. Firstly, let 
 
 us set 
 
 4ir 
 
 In this way three mean acoustic velocities g lf 2 2 , g 3 , independent of the 
 direction, are defmed. We further introduce in place of Am(n) a mean 
 value independent of the direction, in the following manner. In deduc- 
 ing formula (55) we saw that 
 
 . 
 
 Am(n)
 
 158 THE QUANTUM THEORY 
 
 If we carry out the integration with respect to x, we get 
 
 to 
 
 v r da x f 
 
 o 
 
 u, 
 
 47T 
 
 dn 
 
 Now, in a way analogous to that used for the acoustic velocities g, we 
 set 
 
 Hence 
 
 Into ^ n ) = \ we introduce in place of qi(n) and \ m (n) the mean 
 
 values qi and \ m , which are independent of direction ; thereby xi(n) also 
 becomes independent of direction, and is transformed into 
 
 It follows that 
 
 '- 1 * 1 ""' 
 
 141 At the lowest temperatures 
 
 . 
 
 32^2 Ji- - } 
 i=i Xl b 
 
 3 a
 
 NOTES AND REFERENCES 159 
 
 Now the value of the integrals = ^w 4 . If we further set R = Nk, and 
 for z~J the value (59), we get 
 
 15** ' Zfe 
 
 l 
 
 If we introduce_in place of the three acoustic velocities^", 2j, 3s a mean 
 acoustic velocity q by means of the definition 
 
 it follows that 
 
 Finally for we can write V^ (mean atomic volume) and thus get the 
 formula 
 
 148 H. Thirring, Phys. Zeitsohr. U, 867 (1913) ; 15, 127, 180 (1914). 
 
 143 M. Born and Th. v. Kdrmdn, Phys. Zeitschr. 14, 15 (1913). 
 
 144 Cf. note 182. 
 148 Cf. note 128. 
 
 1484. Eucken, Verhandl. d. deutsch. physikal. Ges. 15, 571 (1913). Cf. 
 also A. E'ucken, Die Theorie der Strahlung und der Quanten (W. Knapp, 
 Halle 1914), pp. 386 et seq., Appendix. 
 
 147 Cf. A. Eucken, Die Theorie der Strahlung und der Quanten (W. 
 Knapp, Halle 1914), p. 387. 
 
 148 To calculate the mean acoustic velocity q, the relation given in note 
 141 is used 
 
 3 4ir 
 
 We have therefore to obtain from the "dispersion equation" of the 
 crystal in question (for long waves) the values of the three acoustic 
 velocities q^n), q 2 (n), q s (n) as functions of the wave-direction ; q is then 
 obtained from the above formula by integration over all directions and 
 finally summation. 
 
 149 L. Hopf and G. Lechner, Verhandl. d. deutsch. physikal. Gee. 16, 
 643 (1914). 
 
 180 The following short table is taken from the paper of Hopf and 
 Lechner cited in note 149 :
 
 160 
 
 THE QUANTUM THEORY 
 
 Crystal 
 
 2 calc. from C y 
 
 5 calc. from elastic 
 data 
 
 Sylvin . . . 
 Rock salt . 
 Fluor-spar . 
 Pyrites 
 
 2-36 . 10 B 
 2-82 . 10 8 
 4-02 . 10 8 
 5-43 . 105 
 
 2-03 . 10 s 
 2-72 . 105 
 3-82 . 10 8 
 5-12 . 105 
 
 181 W. Nernst, Vortrage iiber die kinetische Theorie der Materia und der 
 Elektrizitat. Wolfskehl-Kongress 1913 in Gottingen (Teubner, Leipzig 
 and Berlin 1914), pp. 63 et seq. 
 
 182 W. Nernst, ibid., pp. 81 et seg_. 
 
 153 E. Schrodinger, Phys. Zeitschr. 20, 503 (1919). Schrodinger correctly 
 points out that apart from the substitution of one single mean x for 
 the three quantities x> in the Debye terms the approximation above all in 
 the second part of C v (i.e. the replacement of the 3(s - 1) frequencies 
 v t . . . v 3 by the constants ^ . . . i>,) may not be permissible in many 
 cases : namely, in those cases in which the masses of the various kinds 
 of atoms are not very different from one another. If we were to allow 
 so he argues the masses of the different kinds of atoms and the forces 
 acting upon them gradually to become equal to one another, a simple 
 atomic lattice would result, and during this process the 3(s - 1) branches 
 of the spectrum, which correspond to the second type of motion, would 
 merge into the three first branches. " They cannot therefore even be 
 approximately monochromatic if the masses differ only slightly." 
 
 184 H. Thirring, Phys. Zeitschr. 15, 127, 180 (1914). 
 
 188 M. Born, Ann. d. Phys. 44, 605 (1914). 
 
 186 E. Oriineisen, Ann. d. Phys. 39, 257 (1912). 
 
 187 S. Ratnowski, Verhandl. d. deutsch. physikal. Ges. 15, 75 (1913). 
 
 188 Vortrage iiber die kinetische Theorie der Materie und der Elek- 
 trizitat. Wolfskehl-Kongresz zu Gottingen, 1913. (Teubner, Leipzig 
 and Berlin 1914), Vortrag P. Debye. 
 
 189 If U is the energy, and S the entropy of the system, then the " free 
 energy " is defined according to Helmholtz by the relation 
 
 F = U - S . T. 
 It then follows from note 115 that 
 
 dF = d'A - S . dT 
 where d'A is the work done from without. If we set in the usual way 
 
 d'A = - pdV (p = pressure, V = volume) 
 then 
 
 dF=- pdV- SdT. 
 
 From this we get immediately the equation (66) in the text
 
 NOTES AND REFERENCES 161 
 
 Similarly, 
 
 fdF\ _ _ s 
 
 and hence 
 
 160 P. Debye, loc. cit., note 158. 
 
 161 E. Oriineisen, Ann. d. Phys. 26, 211 (1908) ; 33, 65 (1910) ; 39, 285 
 (1912). 
 
 162 P. Debye, loc. cit., note 158. 
 
 1634. Eucken, Ann. d. Phys. 34, 185 (1911) ; Verhandl. d. deutsch. 
 physikal. Ges. 13, 829 (1911). 
 
 IMP. Drude, Ann. d. Phys. 1, 566 (1900). 
 
 163 #. Riecke, Wiedem. Ann. 66, 353, 545 (1898). 
 
 166 Of., for example, H. A. Lorentz, The Theory of Electrons (Teubner, 
 Leipzig, and Berlin 1909). 
 
 167 Let g be the average velocity of the electrons along the free path I. 
 
 Then the electron takes the time T = - to pass over this free path. 
 
 During this time it is exposed to the electrical force E of the external 
 field. Its increase in velocity due to this force is at the commencement 
 
 of the free path = 0, at the end of it = ?=, where e and m are the 
 
 fli 
 
 charge and mass of the electron respectively. In the mean, therefore, 
 the small additional velocity generated by the field is Ag = -^^- = |^-- 
 
 The electrons stream unidirectionally with this velocity against the field. 
 If N is the number of electrons per unit volume, then through unit 
 area of the surface there streams per second a quantity of electricity 
 
 This is, however, the "current density" I which is 
 
 known to be connected with the field E by the relation I = ffE- The 
 expression (67) for the conductivity <r therefore follows. 
 
 A more thorough treatment is due to H. A. Lorentz (see note 166). 
 He does not give the electrons a single velocity q, but introduces Max- 
 well's supposition, known from the kinetic theory of gases, that all 
 possible velocities occur, which are distributed among the electrons 
 according to a fixed law, the so-called Maxwell Law of Distribution. 
 He thus obtained a formula of the following form : 
 
 - 
 
 \3ir mq 
 
 which therefore only differs by a numerical factor from Dnide's formula 
 (67) ; here q = v 2 a , the root mean square of the velocity. 
 
 168 Let a temperature gradient along the x axis be present in the piece 
 of metal. Let a section be taken (see Pig. 12) at right angles to the 
 11
 
 162 THE QUANTUM THEORY 
 
 x axis; we shall calculate the energy transport across this section per 
 second. If we suppose that of all electrons wander in each of the three 
 directions in space, then move in the positive x direction ; and further, 
 ^_ the number of electrons which 
 
 pass through the unit of sur- 
 face in one second, will be all 
 those which are contained in 
 the small shaded cylinder with 
 the base surface area 1 and the 
 height q (velocity), namely, 
 
 FIG. 12. iNg. We also make the sup- 
 
 position, usual in the theory of 
 
 gases (although not strictly true), that the energy, which each electron 
 transports through the cross-section, has the value corresponding to that 
 which it had at the point where it last collided. 
 Now the energy in the section itself at temperature T is equal to %kT, 
 
 and hence the energy = |fcr + ^j^ . I at the points which lie at a 
 
 distance I in front of and behind the section. Here, on the average, the 
 electrons coming from the right and the left meet with their last collisions. 
 The energy transport per second through unit of cross-section is 
 therefore 
 
 Hence 7 = ^JUlqk is the coefficient of thermal conductivity. 
 
 Here also U. A. Lorentz has deepened the theory by taking the distri- 
 bution of velocity into account, and finds that 
 
 where again q = v g 2 
 
 169 G. Wiedemann and R. Franz, Poggend. Ann. 89, 497 (1853) ; L. 
 Lorenz, Wiedem. Ann. 13, 422, 582 (1881). Of. also G. Kirchho/ and 
 G. Hansemann, Wiedem. Ann. 13, 417 (1881); W. Jaeger and H. 
 Diesselhorst, Abh. d. phys. techn. Beichsanstalt 3, 269 (1900). 
 
 The following short table is taken from the paper of the two investi- 
 gators last named ; it gives the ratio "L for various metals at a temperature 
 of 18 C.
 
 NOTES AND REFERENCES 
 
 163 
 
 Metal 
 
 y . 10-10 
 
 
 a- 
 
 Al 
 
 6-36 
 
 Cu 
 
 6-65 
 
 Ag 
 
 6-86 
 
 Au 
 
 7-09 
 
 Zn 
 
 6-72 
 
 Pb 
 
 7-15 
 
 Pt 
 
 7-53 
 
 Bi 
 
 9-64 
 
 170 It follows from (67) by setting ^mq 2 
 
 , that is, a = 
 
 , that 
 ra 
 
 Now, let JY be the number of atoms per unit volume, .AT* the number 
 of atoms in a gramme-atom (Avogadro's number). If, further, A is the 
 atomic weight, M the mass of an atom, and p the density, then 
 
 (A = MN* 
 \p = MN 
 therefore 
 
 We next assume that N> the number of electrons per unit of volume, is 
 small compared with N, say 
 
 If we insert this value, then we get for the free path 
 
 200 <r 
 
 We shall make a rough calculation for copper at G. We have 
 
 ff <--> 5-4 . 10 17 (in electrostatic units) 
 A = 63-57 
 k = 1-4 . 10- 16 
 T= 273 
 m = 0-9 . 10- 27 
 N* = 6-1 . 10 23 
 P = 8-9 
 e = 4-77 . 10- 10 . 
 
 With these values we get 
 
 I is of the order 5-7 . 10-".
 
 164 THE QUANTUM THEORY 
 
 Since the atomic distance is of the order of magnitude 2 . 10~ 8 , the 
 electrons would therefore only suffer collision after passing many thou- 
 sands of atoms. This is unacceptable, since the "radius of molecular 
 action " of the atoms itself has dimensions which fall within the order of 
 magnitude of about 10 ~ 8 . 
 
 171 JET. A. Lorentz, loc. cit., note 43. 
 
 172 J. J. Thomson, The Corpuscular Theory of Matter. 
 
 173 H. Kammerlingh-Onnes, Leiden Communicat. 1913, 133. 
 171 C. H. Lees, Phil. Trans. (A) 208, 381-443 (1908). 
 
 175 W. Meissner, Ann. d. Phys. 47, 1001 (1915). 
 
 176 W. Nernst, Berl. Ber. 1911, p. 310. 
 
 177 H. Kammerlingh-Onnes, Leiden Communicat. 119, 22 (1911). 
 
 178 F. A. Lindemann, Berl. Ber. 1911, p. 316. 
 
 179 W. Wien, Berl. Ber. 1913, p. 184. Of. also Vorlesungen iiber 
 neuere Probleme der theoretischen Physik. (Teubner, Leipzig and Berlin 
 1913.) 3. Vorlesung. 
 
 180 If s is the radius of atomic action, N the number of stationary atoms 
 per unit of volume, then, according to a well-known result of the kinetic 
 theory of gases, the mean free paths of the electrons 
 
 Let us set 
 
 where s is the radius of atomic action for T = 0, that is, when the atoms 
 are at rest ; let a be the amplitude of atomic vibration. Now the mean 
 energy E of this vibration (frequency i>), on the one hand, = (2irv)*a z 
 
 (M is the atomic mass) ; on the other hand, it is, according to Planck- 
 Einstein, 
 
 From this it follows that 
 
 Now, according to formula (67) of the text, the resistance 
 
 w * 2m 2 
 r *r-I* 
 
 If we here set for q the value,*/ (cf. note 170), and for N, according 
 
 to J. J. Thomson's supposition, a^T, and for =- the value 
 *Ni* = *N(a' l + 2as + si)
 
 NOTES AND REFERENCES 165 
 
 it follows that 
 
 an expression, which contains only a and s as unknown constants. If 
 we set 
 
 then W assumes the form given in the formula (70). 
 
 181 F. A. Lindemann, Phil. Mag. 29, 127 (1915). 
 
 181a F. Haber, Berl. Akad. Ber. 1919, pp. 506 and 990. 
 
 182 J. -Stark, Jahrb. d. Radioakt. u. Elektronik 9, 188 (1912). 
 
 183 G. Borelius, Ann. d. Phys. 57, 278 (1918). 
 
 184 K. Herzfeld, Ann. d. Phys. 41, 27 (1913). 
 
 185 If we set %mq* = E, therefore q = \K, the first of the two for- 
 
 mulae (72) follows from (67). If we further take into account that in Drude's 
 
 9 /7 77 1 
 
 Theory E = f kT, that is, that & - | ^ then from (68) the second for- 
 mula (72) follows. 
 
 186 F. v. Hauer, Ann. d. Phys. 51, 189 (1916). 
 
 187 W. Nernst, Berl. Ber. 1911, p. 65. 
 1884. Eucken, Berl. Ber. 1912, p. 141. 
 
 189 K. Scheel and W. Heuse, Ann. d. Phys. 40, 473 (1913). Of. also 
 L. Holborn, K. Scheel and F. Henning, Warmetabellen der physikal.- 
 techn. Reichsanstalt (Vieweg 1919). 
 
 1904. Einstein aud O. Stem, Ann, d. Phys. 40, 551 (1918). 
 
 191 The quantum formulae (76) and (77) properly correspond to the 
 Planck oscillator, that is, to a system of one degree of freedom, while 
 here, in. the case of rotation, we have to do with two degrees of freedom. 
 But the energy of the Planck oscillator is composed of two equal parts, a 
 kinetic and a potential part, while in the case of rotation only kinetic 
 energy comes into question. This is often expressed thus : the Planck 
 oscillator possesses one potential and one kinetic degree of freedom, while 
 the rotating molecule possesses two kinetic degrees of freedom. 
 
 192 P. Ehrenfest, Verhandl. d. deutsch. physikal. Ges. 15, 451 (1918). 
 
 193 According to note 48, the quantum canonical distribution function is
 
 166 THE QUANTUM THEORY 
 
 and the mean energy is 
 
 ~kT 
 
 If we here set all p n 's = 1, and if for E n we substitute the value E (n 
 from (80), there follows for the mean rotational energy of a molecule 
 
 o 
 and for the heat of rotation of hydrogen we get the expression 
 
 19* The turning impulse (moment of momentum) of a system, the 
 mass-points of which possess the mass mi, the velocities vi, and the dis- 
 tances n from a fixed point (say the origin of co-ordinates), is a vector of 
 the value 
 
 In the present case, the system consists only of the two atoms (mass M) , 
 which rotate around a circle of radius r with the constant velocity 
 v = r 2irv. 
 Hence here 
 
 |U| =p = 2Mr a . 2 = J. 2w, 
 
 where J = 2Mr* is the moment of inertia. 
 198 The impulse (or momentum) pi corresponding to a generalised co- 
 
 (3L 
 ordinate qi is, according to note 48, defined by the relation pi = ^ . .' 
 
 where qi = -jr, and L is the kinetic energy of the system. Now here 
 
 the angle of rotation $ is chosen as a generalised co-ordinate. But the 
 kinetic energy of a body rotating about a fixed axis is known to be 
 = J ' (moment of inertia) x (angular velocity) 2 , hence 
 
 J/d<t>\*_J- 
 L= 2\-Tt) ~2*- 
 Hence 
 
 ])Q = ov = J<p = 3 ' 2irv.
 
 NOTES AND REFERENCES 167 
 
 196 F. Reiche, Ann. d. Phys. 58, 657 (1919). 
 
 197 The best curve was obtained by assigning the " weight " In to the 
 nth quantum state of rotation. The rotationless state (n = 0) thus 
 receives the weight zero, i.e. it does not exist. This amounts to the 
 same thing as the introduction of a zero-point rotation. 
 
 198 E. Holm, Ann. d. Phys. 42, 1311 (1913). 
 
 199 J. v. Weyssenhoff, Ann. d. Phys. 51, 285 (1916). 
 
 200 M. Planck, Ber. d. deutsch. physikal. Ges. 17, 407 (1915). 
 
 201 S. Rotszayn, Ann. d. Phys. 57, 81 (1918). 
 
 202 The curve is not drawn by Planck, but is discussed in the author's 
 paper cited in note 196. 
 
 203 See likewise the author's paper quoted in note 196. 
 
 204 P. S. Epstein, Ber. d. deutsch. physikal. Ges. 18, 398 (1916). Of. 
 also Phys. Zeitschr. 20, 289 (1919). 
 
 203 N. Bohr, Phil. Mag. 1913, p. 857. 
 
 206 During " regular precession " the top turns uniformly about its 
 axis of symmetry (axis of its figure), while at the same time this axis 
 describes a cone of circular section about an axis fixed in space. 
 
 207 A compilation of the moments of inercia of the hydrogen molecule 
 used by the various investigators is as follows : 
 
 .7.10. 
 
 Einstein-Stern 1-47 
 
 Ehrenfest 0'69 
 
 (2-214^ 
 Reiche \ 2-293 \ different curves. 
 
 12-095J 
 
 Holm 1-36 
 
 Weyssenhoff 0-34 
 
 Rotszayn 2-12 
 
 Epstein (Bohr's model) . . . 2-82 
 
 MSN. Bjerrum, Nernst Festschrift 1912, p. 90. Bjerrum did not, 
 by the way, start from formula (79), but calculated with the values 
 
 y n = o~Tr> s ^ uce > following a proposal of H. A. Lorentz, he set the rota- 
 tional energy E^ equal to nhv n , in contrast to Ehrenfest's formulation 
 
 (78), which rests on a sounder basis. 
 
 209 S. P. Langley, Annals of the Astrophysical Observatory of the 
 Smithsonian Institution, Vol. I, p. 127, Plate XX (1900). 
 
 210 F. Paschen, Wiedem. Ann. 51, 1 5 52, 209 ; 53, 335 (1894). 
 
 211 H. Rubens, Berl. Ber. 1913, p. 513. 
 
 212 H Rubens and E. Aschkinass, Wiedem. Ann. 64, 584 (1898). 
 
 213 If. Rubens and G. Hettner, Berl. Ber. 1916, p. 167. See also 
 G. Hettner, Ann. d. Phys. 55, 476 (1918). 
 
 214 W. Burmeister, Ber. d. deutsch. physikal. Ges. 15, 589 (1913). 
 
 215 Eva v. Bahr, Ber. d. deutsch. physikal. Ges. 15, 710, 731, 1150 
 
 (1 M6 3 <3f . Lord RayUigh, Phil. Mag. 34, 410 (1892). Let an HC1 mole- 
 cule, for example, be considered, which consists of a positively charged
 
 168 
 
 THE QUANTUM THEORY 
 
 hydrogen atom H+ and a negatively chlorine atom Cl~ (see Fig. 13). Let 
 its centre of gravity be S, and let a be the distance of the H+ atom from 
 S. Let the line joining the two atoms be the axis of x', and let this axis 
 turn in the positive direction about S at the rate of v r revolutions per 
 second with respect to the fixed x-7/-system. If, now, the two atoms 
 vibrate relatively to one another with the frequency r fl and the amplitude 
 A, then the x' co-ordinate of the H+ atoms may be represented thus 
 
 x' = a + A sin (2irv t). 
 
 If we project this vibration upon the fixed co-ordinate system, it follows 
 that 
 
 (x = x' cos (2*vJ) = o cos (2irv r t) + A sin (2trv t) cos (2*V) 
 \y = y' sin (2irv r t) = a sin (torvj) + A sin (2*V) sin (2W) 
 
 FIG. 13. 
 for which we may also write 
 
 fa = a cos (2y r O + 4 sin 2(r + , r )t + 4 sin 2*( v , - , r )t 
 |T/ = a sin (2w^) - ^cos 2 1 r(^ + , r )i + ^ cos 2(ir - Vf )t . 
 
 From the point of view of the system at rest we have thus three 
 oscillations : 
 
 (a) the left-circular oscillation 
 
 x = a cos (2^)|^ h ^ {requenc 
 T/ = oin(2x^)J 
 (6) the left-circular oscillation 
 
 with the frequency v u + v r 

 
 NOTES AND REFERENCES 
 
 (c) the right-circular oscillation 
 
 I with the frequency v n - 1 
 
 169 
 
 ~ sin 2 
 
 217 E. S. Imes, Astroph. Journ. 50, 251 (1919). 
 
 218 A. Eucken, Ber. d. deutsch. phys. Ges. 15, 1159 (1913). Eucken has 
 here, on account of the asymmetrical form of the hydrogen molecule, 
 assumed two different moments of inertia 
 
 </! = 0-96 . 10- 40 , and J t =2-21 . lO" 40 
 
 and hence obtained two different series of numbers giving the revolutions 
 Vr per second, cf . the table given there. See also the table in Rubens and 
 Hettner, loc. cit., note 213. 
 
 219 M. Planck, Ann. d. Phys. 52, 491 ; 53, 241 (1917). 
 
 220 0. Sackur, Ann. d. Phys. 36, 958 (1911) ; 40, 67 (1913). 
 
 221 H. Tetrode, Phys. Zeitschr. 14, 212 (1913) ; Ann. d. Phys. 38, 434 
 (1912). , 
 
 222 W. H. Keesom, Phys. Zeitschr. 15, 695 (1914). 
 
 2234. Sommerfeld, Vortrage liber die kinetische Theorie der Materie 
 und der Elektrizitat. Wolfskehl-Kongress in GQttingen 1913. (Teubner, 
 Leipzig and Berlin 1914), p. 125. 
 
 224 P. Scherrer, Gottinger Nachr. 8 July, 1916. 
 
 225 M. Planck, Berl. Ber. 1916, p. 653. 
 
 226 W. Nernst, Die theoretischen und experimentellen Grundlagen des 
 ueuen Warmesatzes. (W. Knapp, Halle 1918), pp. 154 et seq. 
 
 227 O. Sackur, Ber. d. deutsch. chem. Ges. 47, 1318 (1914). 
 
 228 0. Stern, Phys. Zeitschr. 14, 629 (1913) ; Zeitschr. f. Elektrochemie 
 25, 66 (1919). 
 
 229 For what follows cf. the paper by O. Stern quoted in the last note. 
 Further, W. Nernst, Die theoretischen und experimentellen Grundlagen 
 des neuen Warmesatzes. (W. Knapp, Halle 1918), Oh. XIII. 
 
 230 As regards this and the following chapter, the reader is referred for 
 more exact details to the article of P. S. Epstein in the Planck number 
 of " Naturwissenschaften " (1918, p. 230). 
 
 231 As the simplest Tiwmson atom, we are to imagine a sphere of radius 
 a, filled with the unit charge e of posi- 
 
 tive electrification, of space-density p, in 
 
 the middle of which an electron with 
 
 the charge - e rests. This structure is 
 
 externally neutral. If we draw the 
 
 electron out from the centre to a distance 
 
 r (see Fig. 14) the external (shaded) 
 
 hollow sphere exerts no force on the 
 
 electron, according to the well-known 
 
 laws of electrostatics. The inner solid 
 
 sphere of radius r, on the other hand, 
 
 acts on the electron just as if its total 
 
 charge were concentrated at the centre. 
 
 The force which draws the electron back into its position of rest is
 
 170 THE QUANTUM THEORY 
 
 therefore, 
 
 that is, it is proportional to the distance of the electron from its position 
 of equilibrium. 
 
 232 Cf. also P. Drude, Lehrbuch der Optik. 2. Aufl., Chs. V and VII 
 (Hirzel 1906). There is an English edition of this work. 
 
 233 Cf . W. Voigt, Magneto- und Elektro-optik (Teubner 1908). 
 
 234 M. Planck, Ber. d. Berl. Akad. d. Wiss. 1902, p. 470; 1903, p. 480; 
 1904, p. 740 ; 1905, p. 382. 
 
 238 H. A. Lorentz, The Theory of Electrons, Chs. Ill, IV (Teubner 1909). 
 
 236 The electron oscillates, when bound quasi-elastically, according to 
 
 the equation of motion m ? = - fx, if we restrict ourselves to linear os- 
 cillations. Here m is the mass of the electron, x is its distance from the 
 position of rest, and / is a factqr of proportionality. The solution of this 
 differential equation is represented by the pure harmonic motion 
 
 x = A cos (nt + 8) 
 where the frequency is 
 
 n = V TO' 
 
 The frequency n is therefore, as we see, independent of the amplitude and 
 therefore of the energy of vibration. 
 
 237 The frequencies v of those spectral lines of luminous hydrogen, 
 which are included under the name " Balmer series," may be represented 
 with great accuracy by the following formula given by Balmer. 
 
 n = 3, 4, 5, 6 ... oo . 
 \&' n'/ 
 
 N is here a constant, the so-called Rydberg number. If we set for the 
 current number n the values 3, 4, 5 ... we get in succession the fre- 
 quencies of the red line of hydrogen (H a ), the green line (H^), and the 
 blue line (H y ) and so forth. 
 
 238 J. Stark, Ann. d. Phys. 43, 965 (1914) ; J. Stark and G. Wendt, ibid., 
 43, 983 (1914) ; J. Stark and H. Kirschbawn, ibid., 43, 991 ; 43, 1017 
 (1914) ; J. Stark, ibid., 48, 193, 210 (1915) ; J. Stark, O. Hardtke and G. 
 Liebert, ibid., 56, 569 (1918) ; J. Stark, ibid., 56, 577 (1918) ; G. Liebert, 
 ibid., 56, 589, 610 (1918) ; J. Stark and 0. Hardtke, ibid. 58, 712 (1919) ; 
 J. Stark, ibid., 58, 723 (1919). 
 
 239 Cf. H. A. Lorentz, The Theory of Electrons (Teubner, Leipzig and 
 Berlin 1909), Ch. III. 
 
 2*0 H. Geiger and Marsden, Phil. Mag. April, 1913. 
 
 241 E. Rutherford, Phil. Mag. 21, 669 (1911). 
 
 242 According to C. G. Darwin [Phil. Mag. 27, 506 (1914)], the radius 
 of the nucleus, taken as a sphere, is in the case of gold at the most 
 = 3 . 10 - 12 cms., in the case of hydrogen at the most = 2 . 10-13 cms.
 
 NOTES AND REFERENCES 171 
 
 243,4. van den Broek, Phya. Zeitschr. 14, 32 (1913). 
 
 244 Cf. note 247. 
 
 243 N. Bohr, Phil. Mag. 26, 1, 476, 857 (1913). 
 
 2464. Einstein, Phys. Zeitschr. 18, 121 (1917). 
 
 247 The quite elementary calculation is as follows : let an electron of 
 charge e and mass m rotate around a nucleus of charge E = ez in a 
 circular orbit : then z is the atomic number (for hydrogen, in particular, 
 z = 1). If a is the radius of the circle, v the velocity, and a> the angular 
 velocity (frequency of rotation) of the electron in the circular orbit, then 
 the condition for equilibrium between the attraction of the nucleus and 
 the centrifugal force is 
 
 e _ = maw 2 or ma s w 2 = eE = e*z. 
 
 According to Bohr's second hypothesis the moment of momentum 
 p( = mva=ma?<u) is a multiple of , hence 
 
 ma*a = wA (n = 1, 2, 3 . . .). 
 
 From these two equations for a and u> we get for the discrete radii of the 
 permissible quantum orbits 
 
 and the corresponding frequencies of rotation 
 ^BvWfm 
 
 The energy (kinetic + potential) is 
 
 (aT7 t \ 2tf 
 
 -T) -*"*-*- 
 
 therefore the discrete quantum values of the energy are 
 
 * Wn? ' 
 
 If, in this expression, we set 
 
 we recognise, that W is a function of , and hence of v = -. The energy 
 
 of the electron in the Rutherford model therefore depends, as stated in 
 the text, on its frequency of rotation v. 
 
 If the electron passes from the ntb to the sth quantum path, then, ac- 
 cording to Bohr's third hypothesis, a homogeneous spectral line is emitted 
 of frequency 
 
 W n - W s 2irVw.2 2 /l 1
 
 172 THE QUANTUM THEORY 
 
 where 
 
 N = Wm 
 
 2*8 Cf . note 237. 
 
 2*9 It is of historical interest to note that, before Bohr, A. E. Hems in 
 1910 (Sitzungsber. d. Wiener Akad. 10 March, 1910) succeeded in repre- 
 senting Rydberg's number in terms of the universal constants e, h, m ; 
 his result differed from that of Bohr only by a factor 8. He deduced his 
 result as follows. Starting from /. J. Thomson's atomic model, which 
 was generally accepted at that time, he calculated the maximum oscilla- 
 tion-frequency (no. of revolutions) /max of the electron in the simplest 
 atom (hydrogen atom) for the case when this atom, provided with one 
 energy-quantum, was circling just on the surface of the positive sphere. 
 He obtained 
 
 4Vw 
 "max = p 
 
 This maximum frequency was next identified by Haas with the series 
 limit (n = oo ) in Balmer's formula 
 
 Then it follows that 
 
 which is a value 8 times greater than JVjjohr- Haas used this relation to 
 calculate from the three quantities, the Bydberg number N, Planck's 
 
 constant h, and the ratio ~ , all of which he assumed known, the charge 
 m 
 
 e of the electron. In consequence of the factor 8 he obtained the value 
 e = 3'18 . 10~ 10 , a value that is too small according to our present know- 
 ledge, but which agreed well with the measurements of J. J. Thomson and 
 H. A. Wilson, which were available at that time. 
 
 230 Th. Lyman, Phil Mag. 29, 284 (1915). 
 
 251 F. Paschen, Ann. d. Phys. 27, 565 (1908). 
 
 252.4. Fowler, Month. Not. Roy. Astron. Soc. 73, Dec. 1912. 
 
 233 F. Paschen, Ann. d. Phys. 27, 565 (1908). 
 
 23* E. C. Pickering, Astroph. Journ. 4, 369 <1896) ; 5, 92 (1897). 
 
 238 E. J. Evans, Nature, 93, 241 (1914). 
 
 236 W. Kossel, Ann. d. Phys. 49, 229 (1916) ; Die Naturwissenschaften 
 7, 339, 360 (1919). 
 
 237 L. Vegard, Verhandl. d. deutsch. physikal. Ges. 19, 344 (1917). 
 
 238 A. Sommerfeld, Atombau und Spektrallinien. (An English edition 
 translated from the 3rd German edition (1922) is being prepared by 
 Messrs. Methuen & Co., Ltd.) 
 
 259 R. Ladenburg, Die Naturwissenschaften 8, 5 (1920). 
 280 .4. Sommerfeld, Ann. d. Phys. 51, 1 (1916).
 
 NOTES AND REFERENCES 173 
 
 281 Expressed in terms of polar co-ordinates the kinetic energy L as- 
 sumes the well-known form : 
 
 L = (^ + rV). 
 
 In it, m denotes the mass of the electron, the dots represent differentia- 
 tion with respect to the time. The impulses p r and p^ are then defined as 
 follows (see note 48) : 
 
 p r =?>L = mr ; p. = !%L = mr^. 
 
 3r' ^ 
 
 262 Only when each impulse _p f depends solely on the corresponding 
 2j (or when it is a constant), and when, in addition, the limits of the 
 phase-integral are independent of the g/s, does the phase-integral work 
 out to a constant. This is by no means the case for any arbitrary choice 
 of the co-ordinate-system. 
 
 263 P. S. Epstein, Ann. d. Phys. 50, 489 ; 51, 168 (1916). 
 
 264 K. Schwarzschdd, Sitzungsber. d. Berl. Akad. d. Wiss. 4. Map 
 1916. * 
 
 263 A. Einstein, Verhandl. d. deutsch. physikal. Ges. 19, 82 (1917). 
 
 266 M. Planck, Verhandl. d. deutsch. physikal. Ges. 17, 407, 438 (1915) ; 
 Ann. d. Phys. 50, 385 (1916). 
 
 267 The semi-major axis of the ellipse, which is characterised by the 
 values n and n', here has the value 
 
 The ratio of the axis is 
 
 b _ n 
 a~ n + n'' 
 We see that n' = corresponds to the case of Bohr's circular orbits. 
 
 268 The energy of the electron moving in the Kepler ellipse (n, n') here 
 has the value 
 
 = _ 2TrVs a w _ _ Nhz* 
 
 h\n + n'Y ~ (n + n') a ' 
 
 The series formula (102) of the text then follows from Bohr's Law of 
 Frequency 
 
 269 If account is taken of the influence of relativity, the series formula 
 for the spectra of the hydrogen type become to a first approximation 
 
 V = VQ + V l 
 
 where
 
 174 THE QUANTUM THEORY 
 
 In these expressions the symbols N and a have the following meaning : 
 N= 2 f e * m o a = ^f; a 2 is of the order 5-3.10-5 
 
 m is the mass of the electron at vanishingly small velocities. 
 
 Hence whereas the first term i/ gives the old formula, which was 
 obtained by neglecting the influence of relativity, the small additional 
 term v l represents the influence of relativity. As we observe, v l does not 
 only depend on the quantum sums s + s' and n + n', but also on the 
 individual values s, s', n, n'. This member, v v is thus responsible for the 
 fine-structure. 
 
 270 If we apply the formula of the preceding note to H a , we have to set 
 z = 1, s + s' = 2, n + n' = 3. We then get 
 
 l s' 1 n'~\ 
 
 + r i + * 
 
 24 84 J 
 
 
 
 n 
 
 
 AVjj 
 
 j 
 
 1 [- 
 
 
 -i 1 
 
 r 1 1 
 
 t? *> * 
 i i 
 
 n 
 
 r c 
 
 I 
 
 1 i 
 
 1 "i 
 
 * 
 
 Pia. 15. 
 
 Corresponding to the possibilities of partition 
 
 and 
 
 2 + 01 circle 
 1 + I/ ellipse 
 
 2 ^^ orbits 
 
 n + ri = 3 = 3 + 0^ circle \ 
 
 = 2 + 1 L ellipse L 3 initial orbits 
 = 1 + 2] ellipse ) 
 
 (for dynamical reasons the azimuthal quantum number n cannot under 
 normal conditions assume a zero value), we should expect 2.3 = 6 
 possibilities of production and hence 6 components of the fine-structure 
 of Ha.. One of these components, however, namely, the one correspond- 
 ing to the transition of the electron from the circle (n = 3, n' = 0) to the 
 ellipse (s = 1, s' = 1) does not present itself under normal conditions, as 
 follows from the " Principle of Selection " enunciated by Rubinowicz
 
 NOTES AND REFERENCES 175 
 
 and Sommerfeld (see Chapter VI, 9). Hence 5 components of the 
 fine-structure remain ; their position is exhibited in Fig. 15. 
 
 As we see, the 5 components arrange themselves into two main groups, 
 containing 3 and 2 members, respectively. The ' missing " line Ila is 
 dotted in. The distance AVH between I a and II a , Ib and lib, Ic and II C is 
 called the " theoretical hydrogen doublet." 
 
 According to the above formula the frequency-number of the line 
 la (3, 0-2, 0) is 
 
 The frequency-number of the line Ha (3, >1, 1) is 
 
 Thus 
 
 A "H = */J,, - "/a = ^=1-095.1010 
 
 corresponding to AA.H = 0'157A. 
 
 The hydrogen-doublet actually observed is measured from about the 
 middle of I a and It, to the middle of lib and II C , owing to the absence of 
 IIa. This leads to the value 0'8AA. H , that is, to 0-126A. 
 
 According to a principle of correspondence enunciated by Bohr (see 
 Chapter VI, 9), as a result of which the azimuthal quantum number can 
 only vary by + 1, the components Ib and II C are also absent. 
 
 271 F. Paschen, Ann. d. Phys. 50, 901 (1916). 
 
 272 From formula (97) of the text we get for the two Rydberg constants 
 for hydrogen and helium : 
 
 Moreover, according to note 269, we get the third formula giving the 
 value of the constant for the fine-structure : 
 
 From the first two relations, by using MHO = &Ma, we get 
 
 m a , __M^ 
 
 ~
 
 176 THE QUANTUM THEORY 
 
 and hence 
 
 _L = JL N H-* N H* 
 
 m c MHC ' Nne - Na ' 
 
 The two Rydberg numbers NH and NHC have been measured by Pastfien 
 with great accuracy : 
 
 Nil = (109677-691 + 0'06) . c 
 Nue = (109722-144 0'04) . c. 
 
 Moreover, = F is the electrochemical equivalent (Faraday'* num- 
 
 MH . c 
 ber), that is, the charge which, in electrolysis, accompanies one gramme- 
 
 atom (i.e. N = - atoms). This number has the value 
 F = 9649-4 electromagnetic units. 
 
 If we insert the three values of N a , NH* and * - in the relation above 
 
 MH.C 
 deduced, we get 
 
 JL = 1-7686 . 10 7 electromagnetic units, 
 m c 
 
 a value which agrees very well with those values of this quantity which 
 were obtained by direct methods (deflection of the cathode- and /5-rays in 
 the electric and magnetic field). Let us now write 
 
 or, using the value of - given above, 
 MH 
 
 ^m^ = 3 
 h 4 
 
 The right-hand side of this equation is known. If we combine with it 
 
 
 m c 
 
 O /2 
 
 = = 7-290. 10- 3 
 
 which follows from Paschen's measurements of the fine-structure in the 
 case of helium, we have three equations in three unknowns e, m g , h. 
 From them we get 
 
 e = (4-766 0-088). 10 - 10 
 
 h = (6-526 + 0-200) . 10 - 27 . 
 
 According to Sommerfeld it is more advantageous to use Millikan's value 
 for e. We then get 
 
 fe =(4-774 0-004). 10 - 10 
 { h = (6-545 0-009) . 1Q- 87 
 U = (7-295 0-005) . 10-. 

 
 NOTES AND REFERENCES 177 
 
 273 K. Glitscher, Ann. d. Phys. 52, 608 (1917). 
 
 274 A. Lande, Phys. Zeitschr. 20, 228 (1919) ; 21, 114 (1920). 
 
 278 Cf. A. Sommerfeld, Atombau und Spektrallinien. Ch. IV, 6. 
 
 276 P. S. Epstein, Ann. d. Phys. 50, 489 (1916). 
 
 277 P. Debye, Gottinger Nachr. 3 June, 1916. 
 
 278/1. Sommerfeld, Phys. Zeitschr. 17, 491 (1916). Cf. also Atombau 
 und Spektrallinien. Ch. VI, 5. 
 
 279 F. Paschen and E. Back, Ann. d. Phys. 39, 897 (1912) ; 40, 960 
 (1913). 
 
 2804. Rubinowicz, Phys. Zeitschr. 19, 441, 465 (1918). 
 
 281 N. Bohr, On the Quantum Theory of Line-spectra. Parts I and II. 
 D. Kgl. Danske Vidensk. Seisk. Skrifter, Naturvidensk. og Mathem. Afd. 
 8, Baekke IV, 1. Kopenhagen 1918. 
 
 282 The number of revolutions of the electron per second in the sth 
 quantum circle of Bohr is, in the case of hydrogen, according to note 247 : 
 
 On the other hand, it follows from formula (93) of the text, if we take s 
 considerably greater than 1 (high quantum numbers), and n = s + 1 
 (transition between neighbouring circles), that 
 
 
 283 P. S. Epstein, Ann. d. Phys. 58, 553 (1919). 
 
 284 H. A. Kramers, Intensities of Spectral Lines. D. Kgl. Danske 
 Vidensk. Selsk. Skrifter, Naturvidensk. og Mathem. Afd. 8, Raekke III, 
 3. Kopenhagen 1919. 
 
 283.4. Sommerfeld and W. Kossel, Ber. d. deutsch. physikal. Ges. 21, 
 240 (1919). 
 
 286 J. Franck and G. Hertz, Phys. Zeitschr. 20, 132 (1919) ; in which 
 references are also given. Cf . also J. Franck and P. Knipping, Phys. 
 Zeitschr. 20, 481 (1919) ; J. Franck, P. Knipping and Thea KrOger, Ber. 
 d. deutsch. physikal. Ges. 21, 728 (1919). 
 
 287 J. Tate and Foote, Phil. Mag. July, 1918. 
 
 288 References are given in the report by J. Franck and G. Hertz, men- 
 tioned in note 286. 
 
 289 A. Einstein, Phys. Zeitschr. 18, 121 (1917). Let us consider the 
 two quantum states (1) and (2) of the atom, with the energies E l and E z 
 (.E 2 > -EJ). The number of transitions 2 - 1 which take place in the 
 time dt owing to radiation is then, according to Einstein, NiA 2l dt, in 
 which N 9 is the number of atoms in the state 2, and, therefore, accord- 
 ing to note 48 
 
 ~ &. 
 
 N,, = Nw, 
 
 N being the total number of atoms. 4 al is a factor of proportionality. 
 12
 
 178 THE QUANTUM THEORY 
 
 The introduction of external monochromatic radiation of frequency / 
 and intensity K,, firstly brings about positive absorption, that is, transi- 
 tions l->2. The number of these in the time dt is, according to 
 Einstein, JVj.B 12 K,,, in which .B 12 is a factor of proportionality, N l is the 
 number of atoms in the state 1, and hence 
 
 5 
 
 jY, = NCpje M. 
 
 Secondly, the external radiation also effects transitions 2 - 1 (nega- 
 tive absorption). The number of these that occur in the time dt 
 = -ZV-j-B^Kj/, where jB 2] is a factor of proportionality. When the energy 
 exchange is in equilibrium the number of transitions 2 > 1 must be 
 equal to the number of transitions 1 > 2, hence 
 
 ~ M' 
 
 i.e. 
 
 When the temperature increases indefinitely, K* must also increase to 
 infinitely great values ; from this it follows that 
 
 = 1. 
 
 Finally, if we set -^p- = A for shortness, we get the relation given in the 
 text: 
 
 ^ A 
 
 290 Of. the resume* by E. Wagner, Phys. Zeitgchr. 18, 405, 432, 461, 
 488 (1917). 
 
 291 O. Moseley, Phil. Mag. 26, 1024 (1913) ; 27, 703 (1914). 
 
 292 W. Kossel, Verhandl. d. deutsch. physikal. Ges. 16, 898, 953 (1914) ; 
 18, 339 (1916). 
 
 293 A. Sommerfeld, Ann. d. Phys. 51, 125 (1916) ; Phys. Zeitschr. 19, 
 297 (1918). Of. also Atombau und Spektrallinien. Oh. Ill, Ch. IV 
 4, Ch. V 5. 
 
 29* L. Vegard, Verhandl. d. deutsch. physikal. Ges. 1917, pp. 328, 344 ; 
 Phys. Zeitschr. 20, 97, 121 (1919). 
 298 P. Debye, Phys. Zeitschr. 18, 276 (1917). 
 
 296 J. Kro6, Phys. Zeitschr. 19, 307 (1918). 
 
 297 A. Smekal, Wiener Ber. Ila 127, 1229 (1918) ; 128, 639 (1919) ; 
 Verhandl. d. deutsch. physikal. Ges. 21, 149 (1919). Of. also A. Smekal 
 and F. Reiche, Ann. d. Phys. 57, 124 (1918).
 
 NOTES AND REFERENCES 179 
 
 298 W. Kossel, Zeitschr. f. Physik 1, 119 (1920). 
 
 299 Since the L-ring consists of several electrons, we take the expression 
 " elliptic motion " to mean the following type of motion : each electron 
 independently describes an elliptic path about the nucleus, whereby the 
 electrons are at each moment situated at the corners of a regular polygon 
 which shares in the motion of the electrons, alternately contracting and 
 expanding during this motion (" elliptical associates"), cf. Sommerfeld. 
 Atombau and Spektrallinien. 
 
 3004. Stnekal, Wiener Ber. Ha 128, 639 (1919). 
 
 301 M. Born and A. Lande, Berl. Akad. Ber. 1918, p. 1048 ; Verhandl, 
 d. deutsch. physikal. Ges. 20, 202, 210 (1918) ; M. Born, ibid., 20, 230 
 (1918) ; Ann. d. Phys. 61, 87 (1920). 
 
 302 A. Lande, Verhandl. d. deutsch. physikal. Ges. 21, 2, 644, 653 (1919) ; 
 Zeitschr. f. Phys. 2, 83 (1920). Cf. also A. Lande and E. Madelung, 
 Zeitschr. f. Phys. 2, 230 (1920). 
 
 303 W. Kossel, Ann. d. Phys. 49, 229 (1916). 
 3M P. Debye, Munch. Akad. Ber. 9 Jan. 1915. 
 
 305 P.. Scherrer, Die Botationsdispersion des Wasserstoffs. Dissertation, 
 Gottingen, 1916. 
 
 306 G. Laski, Phys. Zeitschr. 20, 269, 550 (1919). 
 
 307 Cf., for example, A. Sommerfeld, Atombau und Spektrallinien, Ch. 
 IV, 6. 
 
 308 Langmuir, Journ. Amer. Chem. Soc. 34, 860 (1912) ; Zeitsohr. f. 
 Electrochemie 23, 217 (1917). 
 
 309 Isnardi, Zeitschr. f. Elektrochemie 21, 405 (1915). 
 
 310 /. Franck, P. Knipping and Thea KriLger, Ber. d. deutsch. physikal. 
 Ges. 21, 728 (1919). 
 
 310a Planck has made an attempt to alter Bohr's model in such a way 
 that the right heat of dissociation results. See M. Planck, Berl. Akad. 
 Ber. 1919, p. 914. Cf. also H. Kallmann, Dissertation, Berlin 1920. 
 
 311 W. Lenz, Ber. d. deutsch. physikal. Ges. 21, 632 (1919). 
 3124. Sommerfeld, Ann. d. Phys. 53, 497 (1917). 
 
 313 F. Pauer, Ann. d. Phys. 56, 261 (1918). 
 
 314 G. Laski, see note 314. 
 
 315 M. Pier, Zeitschr. f. Elektrochemie 16, 897 (1910). 
 
 316 K. Schwarzschild, Berl. Akad. Ber. 1916, p. 548. 
 
 317 H. Deslandres, Compt. Rend. 138, 317 (1904). 
 
 318 T. Heurlinger, Phys. Zeitschr. 20, 188 (1919) ; Zeitschr. f. Physik 
 1, 82 (1920). 
 
 319 W. Lenz, see note 311. 
 
 320 A different view is upheld by /. Burgers (Versl. K. Ak. van Wet. 
 Amsterdam 26, 115, 1917), in which, also, jumping electrons produce the 
 middle line in the infra-red of band. In contrast with Schwartschild and 
 Lenz, Burger assumes that the motion of the electrons is influenced by 
 the rotation of the molecule. The energy of the system is then not com- 
 posed additively of the energy of the electrons and the rotational energy 
 of the molecule, but a third term has to be added, which is due to the 
 Coriolis force of the rotating system.
 
 180 THE QUANTUM THEORY 
 
 321 T. Heurlinger, see note 318. 
 
 322 F. Reiche, Zeitschr. f. Physik 1, Heft 4, 283 (1920). 
 
 323 E. S. Imes, Astrophys. Journ. 50, 251 (1919). 
 
 324 A. Kratzer, Dissertation, Munchen 1920. 
 
 323 In the case of the gases investigated by Imes, namely HC1, HBr, and 
 HP, the following moments of inertia were found : 
 
 /HOI = 2 ' 6 * 10- 40 ; J H Br = 3-27 . 10 - 40 ; J HF = 1'37 . 10- 40 .
 
 INDEX 
 
 Absorption band, edge of, 21, 112. 
 
 continuous, 78. 
 Acoustic vibrations, 157. 
 Action, quantum of, 9, 27. 
 Atomic heat, 29. 
 
 numbers, 86, 110. 
 Avogadro's constant, 12, 29, 163. 
 
 Law, 12. 
 
 Azimuthal quantum number, 92. 
 
 B 
 
 Back, 22. 
 
 Balmer, 90. 
 
 Barkla, 20, 109. 
 
 Beck, 25. 
 
 Benedicks, 146. 
 
 Bergen, 106. 
 
 Bergmann series, 89. 
 
 Bishop, 106. 
 
 Bjerrum, 75, 123. 
 
 Bohr, 75. 
 
 Bohr's model of the atom, 86, 119. 
 
 principle of analogy or corre- 
 
 spondence, 100. 
 
 selection, 98. 
 
 Boltzmann, 4, 13. 
 
 Born, 38, 42. 
 
 Bragg, W. H. and W. L., 109. 
 
 Bravais, 37. 
 
 Bremsstrahlung, 22, 109. 
 
 Broek, van den, 86. 
 
 Broglie, de, 21. 
 
 Bunsen, 109. 
 
 Burgers, 179. 
 
 Canal rays, 142. 
 Cavity radiation, law of, 3. 
 Characteristic Bontgen radiation, 
 109. 
 
 Chemical constant, 81. 
 Compressibility, 33. 
 Conductivity, thermal, 60. 
 Coulomb, 12. 
 
 Davis, 106. 
 
 Debye, 38, 39, 58. 
 
 Debye's formula, 40. 
 
 Degeneration of gases, 79. 
 
 Deslandres, 122. 
 
 Dessauer, 22. 
 
 Diamagnetism, 25. 
 
 Diamond, 33. 
 
 Dispersion, 47. 
 
 Displacement Law, Wien's, 4. 
 
 Distribution numbers, 114. 
 
 of velocities, Maxwell's Law of, 
 
 5. 
 
 Dix&n, 100. 
 Doppler, 4. 
 Drude, 61, 84, 17. 
 Duane, 22. 
 Dulong and Petit's Law, 29, 30. 
 
 Ehrenfest, 71. 
 
 Einstein, 1, 15, 16, 30, 107. 
 
 functions, 31, 32, 51. 
 Einstein's hypothesis of light- 
 quanta, 16, 17, 107. 
 
 Law of the photo-electric effect, 
 
 20. 
 Elastic collisions, 104. 
 
 spectrum, 42. 
 
 Electron theory of metals, 61. 
 Emissivity, 2. 
 Entropy, 8, 151. 
 Epstein, 75, 98. 
 Equipartition, 13. 
 Eucken, 56, 61, 69. 
 
 181
 
 182 
 
 THE QUANTUM THEORY 
 
 P 
 
 Fine-structure, Sommerf eld's theory 
 
 of, 94. 
 
 of the Rontgen rays, 113. 
 Fokker, 15. 
 Foote, 105. 
 Franck, 22, 102. 
 Friman, 109. 
 
 Oeiger, 85. 
 Gibbs, 13, 26. 
 Glitscher, 97. 
 Ooiicher, 106. 
 Gramme-molecule, 145. 
 Grilneisen, 58. 
 
 Haas de, 25. 
 
 Haber, 46, 66, 155. 
 
 Hamilton, 15. 
 
 Hamilton's principle, 27. 
 
 Hauer, von, 67. 
 
 Heat theorem, Nernsfs, 35. 
 
 Helium, 97. 
 
 Helmholtz, 160. 
 
 .ZTerte, 22, 102. 
 
 Herzfeld, 66. 
 
 Heurlinger, 122. 
 
 Iftmse, 69. 
 
 floo&e's Law, 59. 
 
 flop/, 15, 56. 
 
 Hughes, 106. 
 
 fltiK, 22. 
 
 Hunt, 22. 
 
 Hydrogen type of series, 88. 
 
 Imes, 123. 
 
 Impulse or momentum, 26. 
 
 radiation, 22, 109. 
 
 Infra-red dispersion frequencies, 4i 
 
 Intensity, 2. 
 
 Isnardi, 120. 
 
 Jeans, 15. 
 
 K 
 
 Kammerlingh-Onnes, 64. 
 Karmdn, 38. 
 Kepler ellipses, 95. 
 Kinetic radiation, 23. 
 Kirchhoff, 3. 
 .KbsseZ, 91, 102, 105. 
 Kramers, 102. 
 Kratzer, 124. 
 K-rings, 111. 
 roo, 111, 115. 
 K-series, 21. 
 Kurlbaum, 9. 
 
 Ladenburg, 91. 
 
 Langmuir, 120. 
 
 Losfci, 119. 
 
 Lattice theory of atomic heats, 42. 
 
 Lowe, 35 
 
 Law, Avogadro's, 12. 
 
 equipartition, 13. 
 
 of cavity radiation, 3. 
 
 of radiation, Planck's, 10. 
 --- Bayleigh's, 10, 14. 
 --- Wieris, 5, 7. 
 
 Stefan and Boltzmann, 5. 
 L-doublets, 114. 
 
 Lechner, 56. 
 Lees, 64. 
 Lenard, 19. 
 Lews, 120, 122. 
 Light-quanta, Einstein's, 16. 
 Lindemann, 33, 64, 66. 
 Lorentz, 15, 24, 61, 161. 
 Lorenz, 62. 
 L-rings, 111. 
 Lummer, 2, 3, 4. 
 
 M 
 
 , 33. 
 Marsden, 85. 
 Maxwell, 5, 13. 
 Mean atomic volume, 53. 
 Meissner, 64. 
 
 Metals, electron theory of, 61. 
 Michelson-Morley, 1. 
 Millikan, 12, 20. 
 Molecular heat, 53. 
 Momentum, 26, 166. 
 moment of, 72, 166. 
 Moseley, 110.
 
 INDEX 
 
 188 
 
 N 
 
 Nernst, 9. 
 
 and Lindemann's formula, 37. 
 Nernst's heat theorem, 35. 
 Nichols, 33. 
 
 Orbits, allowable, 87. 
 Ortvay, 152. 
 
 Paramagnetism, 25. 
 
 Parhelium, 97. 
 
 Paschen, 6, 9, 89. 
 
 Pauer, 120. 
 
 Phase-integral, 92. 
 
 Phase-space, 26, 92, 94, 136. 
 
 Photo -electrons, 20, 141. 
 
 Pier, 121. 
 
 Planck, 6, 15, 24, 27, 91, 117. 
 
 Poisson's ratio, 41. 
 
 Polarisation, 168. 
 
 Precession, 167. 
 
 Pringsheim, 2, 4. 
 
 Probability, thermodynamic, 8. 
 
 Q 
 
 Quanta, energy-, 7. 
 Quantum of action, 9. 
 
 Radial quantum number, 92. 
 Ratnowski, 58. 
 Bayleigh, 10. 
 Reflection, metallic, 45. 
 Relativity, 95, 96. 
 Residual rays, 35, 149. 
 Resonance lines, 104. 
 potentials, 105. 
 Resonators, Planck's, 6. 
 Bice, 22. 
 Richards, 58. 
 Richarz, 25. 
 Riecke, 61. 
 
 Rontgen radiation, 21. 
 Rotation spectra, 76. 
 Rotszayn, 75. 
 Rubens, 9, 33. 
 
 Rubinowicz' principle of selection, 
 98, 99, 174. 
 
 Rutherford, 84, 171. 
 Rydberg, 90, 110. 
 
 Sackur, 80. 
 Sadler, 21. 
 Scheel, 69. 
 Scherrer, 79, 119. 
 Schrodinger, 57. 
 Schwarzschild, 93, 121. 
 Selection, principle of, 98. 
 Sommerfeld, 27, 91, 94, 102, 120. 
 Stark, 23, 66. 
 effect, 98. 
 Stefan, 4. 
 
 Stefan-Boltzmann, Law, 5. 
 Stern, 25, 70, 81. 
 Stokes, 19. 
 Sutherland, 33. 
 
 Thirring, 55. 
 
 Thomson, J. J., 19, 63, 84. 
 
 atom, 169. 
 
 Vegard, 91. 
 
 117. 
 
 w 
 
 Wagner, 21, 22. 
 Warburg, 9, 23. 
 Weber, 30. 
 Webster, 22. 
 Weyssenhof, von, 74. 
 
 Wien, 3, 4. 
 
 Wien's displacement Law, 4. 
 Law of radiation, 5, 7. 
 WW/, 9. 
 
 Zeeman, 84. 
 
 effect, 98, 120.
 
 PRINTED IN GBBAT BRITAIN AT THE UNIVERSITY PRESS, ABERDEEN
 
 This book is DUE on the last date stamped below 
 
 301945 
 MAR 19 1946 
 
 aim 
 
 MAR 2 7 1953 
 JUL 27 !95b 
 
 Form L-9-35m-8,'28
 
 ,2s* 
 
 
 UC SOUTHERN REGIONAL LIBRARY FACIL 
 
 t,