Accretion occurs from a disc or possibly from polar caps in the case of magnetic novae: it is doubtful whether a spherically symmetric and homogeneous envelope configuration is eventually reached. A number of works (see in particular Kippenhahn & Thomas, 1978, King & Shaviv, 1978, Livio & Truran, 1988) adressed to this problem. Even if the matter can be almost homogeneously distributed - at least for high accretion rates - it is very likely that the chemical composition would not be uniform because of the shear mixing in the equatorial belt. Moreover, there can be temperature gradients. These and several non spherically symmetric factors (see the discussion of Tutokov & Yungelson, 1974, on the effects of rotation and magnetic fields) could lead to runaway only in a portion of the envelope. Instead of having a spherically symmetric explosion there would be a local thermonuclear runaway (LTNR). Estimates of Shara (1982) indicate that, with certain sets of parameters, the time for thermalization of the envelope would be much longer than the time to reach the conditions for the runaway and the outburst in such a case would occur only in a certain portion of the envelope. A non-spherically symmetric outburst could be the cause for the peculiar equatorial belts - polar caps geometry of the shells of many novae. There are other explanations (for instance interaction of the secondary with the expanding shell, like suggested by Livio and by Shankar et al. in this colloquium), but bidimensional spectroscopy seems to favour also difference in chemical composition in the envelope and in the polar caps (Duerbeck 1987) and high resolution Hα spectroscopy of Nova Cen 1986 hints that the blobs in the shell have formed at the time of the outburst (Bandiera & Focardi 1989) and cannot be explained by a spherically symmetric TNR model.