Volume 12, number 2 By vp Fig. 2 ‘well into account the radiation correction to the p-decay constant {ound by Berman 8) and Kino~ Shila and Sirlin 4) we obtain for the muon life time “ah Be2 | AZ Be2) Ag a M2 In= +5 a4, yo Lm Maa BoB 8 ye whero 7,0 is the muon life time calculated by means of universal theory of four fermion inter- action with a constant taken from s-decay without any corrections, Aj is the cut off momentum due PHYSICS LETTERS 16 September 1864 to the strong interactions, Ag ~ M, £ is the en- ergy of f-transition. According to experimental data 7), /7 4° = 0.988 + 0.004. Substituting the numbers into (1) we obtain £,,/7,2 = 1.008 and the disagreement between the theory and experiment will be in our case 1.5 + 0.4%, When discussing this result one should take into consideration that n (1) only the terms ~ e@ tn e~2 were correctly taken into account but the terms ~ 2 were discarded. Tt seoms to us that the conclusion that in the theory of weak interaction with intermediate W- meson 3~ and 1-conslants must be with good ac~ curaey the same (taking into account the correc tions due to the electromagnetic and weak inter- actions), is in favour of the weak interaction the- ory with W-meson unlike the four-fermion theory, More detailed paper will be published else~ where. ‘The author is indebted to B. V. Geshkenbein, 1. Yu. Kobsarev, L. B. Okun, A. M. Perelomov, 1.Ya, Pomeranchuk, V.S. Popov, A. P. Rudik and V. Terentyev for valuable discussions. References 1) B.-L. foffe, M.V-Terentyev (in print). 2) T_D. Lee, Phys. Rov. 128 (1962) 899. 3) S.M, Berman, Phys.Rev. 112 (1958) 267. '§ T-Rinoehita, 'A.Sirlin, Phys.Rev.113 (1959) 1682. rrr BROKEN SYMMETRIES, MASSLESS PARTICLES AND GAUGE FIELDS P.W. HIGGS ‘Tait Institute of Mathematical Physics, University of Edinburgh, Scotland Received 27 July 1964 Recently a number of pgople have discussed the Goldstone theorem 1,2): that any solution of a Lorente-invarlant theory which violates an inter nal symmetry operation of that theory must con, tain a masatess scalar pasticle, Klein and Loe 9 showed that this theorem does not necessarily ap- ply in non-relativistic theories and implied that their considerations would apply equally well to Lorentz-invariant field theorles. Gilbert 4), how- 132 ever, gave a proof that the failure of the Goldstone theorem in the nonrelativistic case is of a type which cannot exist when Lorentz invariance 1 it posed on a theory. The purpose of Unis note is to show that Gilbert's argument fails for an impor~ tant class of field theories, that in which the con- Served eurents are coupled to gauge fields. Following the procedure used by Guibert #), let ‘us consider a theory of two hermitian scalar fieldsVolume 12, number 2 1G), @g(e) which is invariant under the phase transformation 917 91 cosa +92 sina, a ©4 94 sin a + ggco8 a Then there is a conserved current j,, such that iL a8 59%, 109] = 920). @) We assume that the Lagrangian is such that sym- metry is broken by the nonvanishing of the vacuum expectation value of #9. Goldstone’s theoxem is proved by showing that the Fourier transform of {0 26), 4 (0)]) contains a term 31(2)¢ Wo) 6062), where hk, is the momentum, as a consequence of Lorentz-Covariance, the con- servation law and eq. (2). Klein and Lee 9) avoided this result in the non- relativistic case by showing that the most general form of this Fourier transform is now, in Gilbert's notation, Fis = dey 04(82, nk) +m), p9(0, mk) + Cg), oa where m,, which may be taken as (1, 0, 0, 0), picks out a special Lorentz frame. "he conver- sation law then reduces oq. (3) to the less general form BLT. = hy, 0(2)og (nk) +[Pn,,- de, (nk)]og (22, nt) + Cyn, 642). (ay Itturns out, on applying eq. (2), that all three terms in eq. (4) ean contribute to (ey). ‘Thus the Goldstone theorem fails if py = 0, which is pos- sible only if the other terms exist, Gilbert's re- nark that no special timelike vector m,, is avail- able in a Lorente-covariant theory appears to rule out this possibility in such a theory. There is however a class of relativistic field theories in which a vector ny does indeed play a Part. This is the class of gauge theories, where & auxiliary unit timelike veetor m,, must be i= PHYSICS LETTERS 19 September 1964 troduced in order to define a radiation gauge in which the vector gauge fields are well defined operators. Such theories are nevertheless Lo- rentz-covariant, as has been shown by Schwinger ®). (This has, of course, long been known of the simplest such theory, quantum elec- trodynamics.) There seems to be no reason why the vector m, should not appear in the Fourier transform under consideration. It is characteristic of gauge theories that the conservation laws hold in the strong sense, as a consequence of field equations of the form jh = ayriw, me! = 8p Ay - a, Ay! ©) Except in the case of abelian gauge theories, the fields Ay", Fy" are not simply the gauge field variables Ay, F),,, but contain additional terms with combinations of the structure constants of the group as coefficients, Now the structure of the Fourier transform of (fA ,"¢0), 0 40)]) must be given by eq. (3). Applying eq. (5) td this com- mutator gives us as the Fourier transform of ido), 910) the single term [heen ), ~ kuelak)] p(b2, nk). We have thus exorcised both Goldstone's zezo-mass bosons and the "spurion” state (at 2,, = 0) proposed by Klein and Lee. In a subsequent note it will be shown, by con- sidering some classical field theories which dis- play broken symmetries, that the introduction of gauge fields may be expected to produce qualita- tive changes in the nature of the particles de- scribed by such theories after quantization, References 2) J.Goldstone, Nuovo Cimento 19 (1961) 154, 2) J.Goldstone, A.Salam and 8, Weinberg, Phys.Rev. 137 (1962) 965, 9) A Klein and .W, Lee, Phys.Rev. Letters 12 (1964) 266, 4) W. Gilbert, Phys.Rev. Letters 12 (1964) 713. 5) J. Schwinger, Phys.Rev. 127 (962) 324. rrr 133,