k all three must be proportional to &/ and hence to each other." But unfortunately, since /3 ;J -=f(\kj\) 9 with/(0)=0, satisfies Szigeti's requirements, his argument also implicitly assumes analyticity at wave number k 5 =0. Though the rather numerous assumptions and approximations are otherwise clearly discussed by Szigeti, it is this single, unproven assumption that puts the secondorder dipole terms on the same footing as the anharmonicity, described by b ajj .Although, of course, we have not proved the validity of the molecular-field assumption (1) , 5 we have demonstrated that Szigeti's assumption conflicts with it. On the other hand, if the secondorder dipole terms are set equal to zero in his work from the outset, then his paper gives a correct, microscopic derivation of the rigid-ion result, II x OCT 3^ which was obtained by different arguments in our earlier work. 3 1 B. Szigeti, Phys. Rev. Lett. 35, 1532Lett. 35, (1975). 2 M. Born, Rev. Mod. Phys. JL7, 245 (1945). 3a P. J. Grout and N. H. March, Phys" Lett. 47A, 288 (1974). 3b P. J. Grout, N. H. March, and T. L 0 Thorp, J. Phys. C 8, 2167 (1975). 4 A conceivable way in which this could come about, suggested by Eq. (2.16) of Ref. 3b, is that the coefficient u 2 in Eq.(1) involves the difference ey-e^ between the static and high-frequency dielectric constants in such a manner tha.tu 2^~c o as Co-e^--0, the last result obviously holding in the rigid-ion limit. Such a divergence of u 2 would then herald the presence of a lower-order term in Eq. (1), proportional to \Ap\, in this singular, rigid-ion limit. 5 Indeed, we have shown elsewhere that for a ferromagnet Eq. (1) must be replaced by AU<* (Ap) m where m> 1, Thus, for a ferromagnet we have the rigorous result thatd(AU)/d(Ap) 9 at Ap = 0, is zero. This last result follows also for a pyroelectric if (1) is assumed, whereas if a term ^IA^I is included in (1), as it must be for rigid ions, then this zero derivative condition is violated.The objections by Grout and March against the author's previous paper on pyroelectricity are answered. The earlier statement concerning the behavior of certain expansion coefficients near k = 0 is proved. The conclusions are compared with the results of Born and Huang.In the preceding Comment, Grout and March 1 criticize my recent Letter 2 on pyroelectricity. They say that the molecular-field theory predicts that the pyroelectric coefficient should be proportional to T and not to T 3 at very low temperatures and that my statement concerning the wavenumber dependence of the coefficient fi 5j near k j = 0 is incorrect. It will be shown that both criticisms are erroneous.Since we are discussing the primary pyroelectric coefficient n^ we consider the solid at constant strain. Denote by p(0) the expectation value of the macroscopic dipole moment at T = 0; and by p, the same quantity at temperature T. p(0) contains the contribution of the zero-point vibrations. We may normalize £ and/>(0) to 1 mole of substance. I further define Ap by p =p(0) + Ap. These three quantities have t...