Having these results in mind, we optimized this last fit by using a Lorentzian profile for the H line, with no associated O III emission, and a set of three Gaussians for the remaining contribution coming from the ``narrow'' components (this is not the first time Lorentzian profiles are used to fit AGN emission lines; for example Stüwe et al. (1992) found that, in the case of NGC 4258, the narrow lines were better fitted by Lorentzians, rather than Gaussians).
The best fit is presented in Fig. e: in this model, the flux of the ``narrow'' (Gaussian) H component is only 9% of the ``broad'' (Lorentzian) H component. The H Lorentzian component is blueshifted by 150 kms-1 with a width of 915 kms-1. The Gaussian components are blueshifted by 395 kms-1 and their width is 1115 kms-1. Mason et al. found a FWHM = 1030 150 kms-1 for this component which is blueshifted by 240 30 kms-1. This blueshift, however, is measured with respect to the O III lines core which is dominated by the Seyfert 2 cloud, itself blueshifted by 95 kms-1 with respect to the Hii region; the blueshift of Mason et al.'s intermediate component is, therefore, 240 + 95 = 335 kms-1, in agreement with our value of 395 kms-1.
In Seyfert 1 galaxies, the Balmer decrement of the broad component is never smaller than that of the narrow component; in the present case, we therefore expect the ``narrow'' H component flux to be less than 9% of the ``broad'' H component flux. Moreover, the N II6583 line flux is, in Seyfert galaxies, equal or smaller than the narrow H component flux. So, in KUG 1031+398, we expect the narrow lines to be quite weak compared to the ``broad'' H component, and we fitted the ``original data minus core'' red spectrum with a single Lorentzian profile of 1205 kms-1 FWHM, blueshifted by 65 kms-1 with respect to the Hii region. This fit is shown in Fig. f.
Another model allowing, in addition, for a set of H and N II Gaussian components was also tested, resulting in a fit of similar quality; the very small 6583/H ratio observed for this solution (0.2), shows that the nitrogen lines may be considered as undetectable.