The classical problem of diffraction of radio waves around a spherical earth is briefly reviewed, with the purpose of pointing out a mistake in the mathematical derivation of the theory which has been generally accepted. The correct way of analysis is indicated.