Although only a complete dynamical model can address the question of the mass distribution of a galaxy, it is possible to derive some hints about its structure directly from the analysis of the interplay between the kinematics of its gas and stars. In our sample we can identify two classes of galaxies according to their kinematics, assuming that gas and stars are coplanars:

(i) Galaxies in which ionized gas rotates faster than stars and has a lower velocity dispersion than the stars (i.e., $V_{\rm g}$ $\;>\;$ $V_\star$ and $\sigma _{\rm g}$ $\;<\;$ σ_*): NGC 772, NGC 3200, NGC 3898, NGC 4419, NGC 5064 and NGC 7782. All these galaxies are classified as early-to-intermediate type spirals, except for the Sc NGC 3200. The different kinematic behaviour of the gaseous and stellar components can be easily explained by a model where the gas is confined in the disc and supported by rotation while the stars mostly belong to the bulge and are supported by random motions (i.e. dynamical pressure). In the case of NGC 772 and NGC 7782, this simple hypothesis is confirmed by the self-consistent Jeans models of Pignatelli et al. (2001). In these galaxies the ionized gas is tracing the gravitational equilibrium circular speed. This is not true in the innermost region ( $\pm0.7$ kpc) of NGC 3898, where the ionized gas is rotating more slowly than the circular velocity predicted from dynamical modelling, unveiling a more complex behaviour (see Corsini et al. 1999; Cinzano et al. 1999).

(ii) Galaxies for which $V_{\rm g}$ $\;\simeq\;$ $V_\star$ and $\sigma _{\rm g}$ $\;\simeq\;$ σ_*; over an extended radial range. This is the case of the intermediate-to-late type spirals NGC 470, NGC 949, NGC 1160, and NGC 2541, NGC 3810, and of the Sab NGC 3705. In these disc-dominated galaxies the motions of the ionized gas and stars are dominated by rotation as we can infer from their low $$\sigma\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil $\displayst... ...{\offinterlineskip\halign{\hfil$\scriptscriptstyle ...$ $\rm km\;s^{-1}$ and large $$(V/\sigma)_{\rm max}\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hf... ...r{\offinterlineskip\halign{\hfil$\scriptscriptstyle ...$ .

The Sab spiral NGC 3368 has intermediate properties (i.e., $V_{\rm g}$ $\;>\;$ $V_\star$ and $\sigma _{\rm g}$ $\;\simeq\;$ σ_*) between the two classes even though the gas rotation is quite asymmetric. The edge-on S0 NGC 980 has a very peculiar rotation curve with $V_{\rm g}$ $$\;\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil $\displaystyle ...$ $V_\star$ for $$\vert r\vert\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil $\dis... ...\offinterlineskip\halign{\hfil$\scriptscriptstyle ...$ and $V_{\rm g}$ $\;<\;$ $V_\star$ elsewhere. These cases can be explained if the gas disc is warped and not aligned with the plane of the stellar disc. Further observations on different position angles are needed to derive a detailed modelling. For NGC 224 and NGC 5854 we have no gas kinematics to perform a comparison with stars.

In the remaining galaxies of our sample gas and stellar kinematics suffers from the presence of kinematically decoupled components. Two counter-rotating stellar components have been found by Pompei & Terndrup (1998) in the edge-on Sb NGC 2683. In the nuclear region of NGC 2841 the ionized gas is rotating perpendicularly with respect to the stars and a fraction of bulge stars are counter-rotating with respect to the rest of the galaxy (Sil'Chenko et al. 1997). In the centre of NGC 3031 our gaseous kinematic data suggest the presence of a circumnuclear Keplerian disc of ionized gas (e.g. Bertola et al. 1998) which is consistent with the gaseous disc observed by Devereux et al. (1997) and rotating around a supermassive black hole (Bower et al. 1996). In NGC 7331 the possible presence of a counter-rotating bulge has been discussed by Prada et al. (1996) and ruled out by Bottema (1999).

$\begin{figure} {\psfig{figure=MS10597f4.eps,width=11cm,angle=0} }% \end{figure}$

Figure 4: Distribution of the ionized gas ( left panel) and stellar velocity dispersions ( right panel) measured in the centres of the galaxies plotted in Fig. 3. Galaxies have been sorted according to their RC3 morphological type. The velocity-dispersion bins are 33 $\rm km\;s^{-1}$ wide.

$\begin{figure} { \psfig{figure=MS10597f5.eps,width=11cm,angle=0} }% \end{figure}$

Figure 5: As in Fig. 4 but for ionized gas and stellar velocity dispersions measured at $R_{\rm e}/4$ .

The recent results implying a tight $M_\bullet-\sigma$ relation of its host spheroids (Ferrarese & Merritt 2000) have made us look for possible relationships between the velocity dispersion of gas and stars. With this aim we compiled data from a sample of about 40 disc galaxies for which the major-axis velocity curve and velocity dispersion profiles of both ionized gas and stars are available, by adding the early-type disc galaxies of Bertola et al. (1995), Fisher (1997), and Corsini et al. (1999) to the spirals of our present sample. Although stellar and/or ionized gas kinematics have been studied in a larger number of S0's and spirals (e.g. Héraudeau & Simien 1998; Héraudeau et al. 1999) we selected these few authors since only they provide the radial trend of the gas velocity dispersion. For each object we derived the values of σ_* and $\sigma _{\rm g}$ in the centre and at $R_{\rm e}/4$ , where $R_{\rm e}$ is the half-surface brightness radius of the galaxy listed in the RC3 (Fig. 3).

The central values of σ_* and $\sigma _{\rm g}$ seem to be correlated, since galaxies with higher σ_* tend to show also higher values of $\sigma _{\rm g}$ . Moreover σ_* and $\sigma _{\rm g}$ cover the same range of values and can reach values higher than 300 $\rm km\;s^{-1}$ with $\sigma _{\rm g}$ $$\;\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil $\displaystyle ...$ σ_*. However there is no clear dependence on the morphological type as seen in Fig. 4. The high central values of $\sigma _{\rm g}$ may be partially due to the smearing effect of the seeing, which are more noticeable on the gas kinematics of the early-type disc galaxies on account of their large central velocity gradients (e.g. Rubin et al. 1985). On the other hand, a high central $\sigma _{\rm g}$ could be also due to intrinsic properties of the galaxy. This is the case for the broad emission lines which are the signature of an unresolved Keplerian velocity field due to a gaseous disc rotating around a supermassive black hole (e.g. Bertola et al. 1998; Maciejewski & Binney 2000).

At $R_{\rm e}/4$ we find $\sigma _{\rm g}$ $$\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil $\displaystyle ...$ $\rm km\;s^{-1}$ in almost all the objects, while σ_* ranges between 40 and 240 $\rm km\;s^{-1}$ . This correlates with the galaxy type as seen in Fig. 5. The low value of $\sigma _{\rm g}$ indicates that we are observing dynamically cold gas, which is rotating in the disc component. In addition, there are a few S0's and early-type spirals in which $\sigma _{\rm g}$ is much higher ( $\sigma _{\rm g}$ $\;\simeq120$ $\rm km\;s^{-1}$ ) than expected from thermal motions or small-scale turbulence. Such a high $\sigma _{\rm g}$ cannot be explained as the result of seeing smearing of velocity gradients since it is measured at a distance $R_{\rm e}/4$ which is larger than 5 seeing FWHM's for all the sample objects. We suggest that the high- $\sigma _{\rm g}$ galaxies are good candidates to host dynamically hot ionized gas as in the case of the S0 NGC 4036 (Bertola et al. 1995; Cinzano et al. 1999), even if the question whether pressure-supported gas is related to the dynamics of the bulge stars is still open (Pignatelli et al. 2001). The Hubble Type - σ_* relation observed at $R_{\rm e}/4$ is an indication that at this radius the stellar kinematics of early and late-type disc galaxies dominated by bulge and disc component, respectively. In late-type spirals, which host low or negligible bulges, σ_* $\;\simeq\;$ $\sigma _{\rm g}$ $\;\simeq\;$ 50 $\rm km\;s^{-1}$ .

4 Discussion and conclusions