Our analysis shows that the emission line spectrum of KUG 1031+398 can be satisfactorily decomposed in a set of components which have line ratios characteristics of Hii regions or conventional NLR or BLR clouds, whithout the need to invoke the presence of an ILR characterized by 5007/H 1.
There are two main reasons why our analysis yields different results from those published by Mason et al. (1996). First, KUG 1031+398 having a redshift of 0.043, the N II 6583 line coincides with the atmospheric B band. When correcting for this absorption feature, the N II true intensity is recovered (Fig. b) and our red spectrum appears different from the published one; different line-ratios and widths are therefore not unexpected.
Second, the line-profile analysis of Mason et al. differs from ours in that, while we force each Balmer component to be associated with forbidden lines having the same velocity and width, Mason et al. allow these parameters to have different values for the Balmer and forbidden line components. As a result, they found three H components (a narrow, an intermediate and a broad one), as well as two O III components (a narrow and an intermediate one); they also detected three H components (again a narrow, an intermediate and a broad one), but only a single N II component (narrow). The measured width of the narrow H component is 150 20 kms-1 FWHM, while the width of the narrow O III lines is 265 10 kms-1; this last value, significantly larger than the narrow H line width, suggests that the O III lines may have a complex profile. Moreover, the width of the N II lines is found to be significantly larger (400 60 kms-1) than that of the narrow H component (190 40 kms-1); this could be due to an inaccurate correction of the atmospheric B band, as we have seen above.
Although our spectra have a lower resolution than those obtained by Mason et al. (3.4 Å compared to 2 Å FWHM), this does not affect the analysis; the narrow core components being identified and subtracted, all the discussion is centered on the broader components, well resolved even with our lower resolution. Similarly, the larger slit width used in our observations (21 compared to 15 for Mason et al.) does not affect the study of these broader components, since only the contribution from the extended emitting region (the Hii region, Fig. ), removed with the core, changes with the slit width.
Boller et al. (1996) and Wang et al. (1996) have suggested that the small width of the broad Balmer lines and the soft X-ray excess characteristic of NLS1 galaxies could be the effect of a high accretion rate on an abnormally small mass black hole. Mason et al. (1996) have argued that, although the emission line spectrum in KUG 1031+398 is dominated by the ILR, a weak broad component is present with line-widths of the order of 2500 kms-1 FWHM and that, therefore, at least in this object, such a model is not required.
Our analysis of the spectra of KUG 1031+398 has shown that, in the BLR, the Balmer lines are well fitted by a Lorentzian profile with 1060 kms-1 FWHM; this value is much narrower than the value found by Mason et al. ( 2500 kms-1). This is due to the fact that we used a Lorentzian, rather than a Gaussian profile to fit the broad Balmer lines; the Lorentzian profile was required by the presence of broad wings, fitted with a Gaussian by Mason et al. (1996).
We have shown (Gonçalves et al. in preparation) that in NLS1s the broad component of the Balmer lines is generally better fitted by a Lorentzian than by a Gaussian; the Lorentzian Balmer lines (component 4), without any measurable associated forbidden line, would qualify this object as a NLS1 with, in fact, very narrow lines. So, in this respect, KUG 1031+398 is a normal NLS1 and could be explained by the same small black hole mass model as the other objects of this class.