We shall now try to obtain an idea about the temperature of the shell using the presence of the elements observed in emission. We have already applied this procedure in the other papers of this series, because in the other stars we did not see any part of the underlying spectrum. Judging from the the ionization stage of the elements present, we can say that the spectum corresponds to a late A or early F type object, a fact which would place the temperature of the shell at about 7000 K. Assuming the photospheric temperature to be 22 000 K (Israelian et al. 1996), and a black-body model for the two entities one can calculate the radius of a star which would dominate at wavelengths beyond 4200 Å, which is what one observes. One comes out with a radius of about ten times the radius of a 22 000 degree K star, which would be the distance of the shell. This seems to be a reasonable order of magnitude. Observe however that we have supposed implicitely that the shell is spherical. Since probably the shell is restricted to the equatorial region, the distance of the shell should be less, in order to cover a sizable fraction of the stellar disc. Otherwise we should see the absorption line spectrum over all wavelengths.