Abstract. The problem of low-frequency sound propagation in slowly varying ducts is systematically analyzed as a perturbation problem of slow variation. Webster's horn equation and variants in bent ducts, in ducts with nonuniform soundspeed, and in ducts with irrotational mean flow, with and without lining, are derived, and the entrance/exit plane boundary layer is given. It is shown why a varying lined duct in general does not have an (acoustic) solution. The usual derivation is based on the assumption of a crosswise uniform acoustic pressure field, such that, by averaging over a duct cross section, the spatial dimensions of the problem are reduced from three to one.Although it shows a remarkable evidence of ingenuity and physical insight, this derivation is mathematically unsatisfying. It is not clear (i) what exactly is the small parameter underlying the approximation, (ii) why the pressure may be assumed to be uniform, (iii) what the error is of the approximation, (iv) what the conditions are on the duct geometry and on the frequency of the field, (v) how to generalize to similar problems, (vi) how to generate higher order corrections, and (vii) what happens near the source or duct entrance or exit plane.An asymptotically systematic derivation of the three-dimensional (3D) classic problem was given by Lesser and Crighton [4], extending the derivation of Lesser and Lewis in [5,6]. They also showed for a number of 2D configurations how abrupt changes of the geometry (open end, slit in the wall) can be incorporated as boundary layer regions in a setting of matched asymptotic expansion. Their approach, based on introducing different longitudinal and lateral scales, is a special case of the method of slow variation put forward by Van Dyke [7]. Although only an asymptotically sound derivation is able to indicate the range of validity and the order of the error of the approximation, we found in the literature no variants of this problem (e.g., with mean flow [8,9,10,11,12]) that strictly follow that approach.Particularly interesting would be an investigation of the related problems of lined ducts without and with flow, as this would form a natural long wavelength closure of the multiple scales theory of sound propagation in slowly varying ducts [13,14,15,16].