$\begin{figure} \par\includegraphics[width=7cm,clip]{h2765f1.ps} \par\end{figure}$

Figure 1: Image of the total H I line emission of NGC1808 (upper left) and NGC1792 (lower right), the two dominant members of the NGC1808 group of galaxies, and their surroundings. The cross in the northeastern spiral arm of NGC1808 marks the position of SN1993af. The angular resolution of the data is indicated in the lower left; the linear grey scale ranges from 0 to the maximum observed surface brightness of 1.912 Jy beam^-1.

Our map of the integral H I line emission from NGC1792, NGC1808 and the intergalactic space in between is displayed in Fig. 1. This map was created using the AIPS routine MOMNT to calculate three "moment maps'' (total intensity, velocity field and velocity dispersion) from the original data cube, before applying the primary beam correction. It shows clearly resolved H I emission from both galaxies. The 1- $\sigma$ rms noise of the data at the positions of both galaxies is of order 10 mJy beam^-1 per channel; at the centre of the FOV, where all six pointings overlap within the FWHM of the primary beam, it is approximately 5 mJy beam^-1 per channel.

$\begin{figure} \par\psfig{figure=h2765f2.ps,width=8.8cm} \end{figure}$

Figure 2: Integral H I line spectra of NGC1792 (dotted line) and NGC1808 (solid line) superimposed onto each other to illustrate the overlap in velocity space and the relative intensities.

Table 2: Integral properties of NGC1792 and NGC1808.
Property NGC1792 NGC1808

W₂₀ (km s^-1) $326\pm10$ $342\pm10$

W₅₀ (km s^-1) $295\pm10$ $266\pm10$

$v_{\rm hel}$ (km s^-1) $1207\pm10$ $1001\pm10$

$f_{\rm HI}$ (Jy km s^-1) $39.5\pm5$ $63.4\pm6$

$f_{\rm 1.34}$ (mJy) $325\pm10$ $616\pm15$

log( $L_{\rm 1.34}$ ) (W Hz^-1) 21.66 21.94

**Table 2:** Integral properties of NGC1792 and NGC1808.
Property	NGC1792	NGC1808
W₂₀ (km s^-1)	$326\pm10$	$342\pm10$
W₅₀ (km s^-1)	$295\pm10$	$266\pm10$
$v_{\rm hel}$ (km s^-1)	$1207\pm10$	$1001\pm10$
$f_{\rm HI}$ (Jy km s^-1)	$39.5\pm5$	$63.4\pm6$
$f_{\rm 1.34}$ (mJy)	$325\pm10$	$616\pm15$
log( $L_{\rm 1.34}$ ) (W Hz^-1)	21.66	21.94

Assuming optically thin H I line emission ( $\tau\ \ll\ 1$ ), the observed flux densities can be converted into H I column densities following the relation (Eq. (12.2) by Giovanelli & Haynes 1988)

Table 3: Global properties $^{\rm a}$ of NGC1792 and NGC1808.
Property Symbol NGC1792 NGC1808

Absolute blue magnitude $^{\rm b}$ $M_{\rm B}$ (mag) $-20.15\pm 0.10$ $-20.24\pm 0.10$

Total blue luminosity $L_{\rm B}$ ( $L_\odot$ ) $1.6\times 10^{10}$ $1.7\times 10^{10}$

Total H I gas mass $M_{\rm HI}$ ( $M_\odot$ ) $1.1\times10^{9}$ $1.8\times10^{9~~{\rm c}}$

Total virial mass $M_{\rm T}$ ( $M_\odot$ ) $0.54\times 10^{11~~{\rm d}}$ $1.1\times10^{11~~{\rm e}}$

H I gas to total mass ratio $M_{\rm HI}/M_{\rm T}$ 0.021 0.016 $^{\rm c}$

H I gas to blue light ratio ${M_{\rm HI}\over L_{\rm B}}$ ( ${M_\odot \over L_\odot}$ ) 0.07 0.10 $^{\rm c,e}$

Total mass-to-light ratio ${M_{\rm T}\over L_{\rm B}}$ ( ${M_\odot \over L_\odot}$ ) 3.4 6.5

Optical colour index $^{\rm b}$ B-V (mag) $0.68\pm 0.14$ $0.82\pm 0.14$

Total FIR luminosity $L_{\rm FIR}$ ( $L_\odot$ ) $0.63\times10^{10~~{\rm d}}$ $2.4\times 10^{10~~{\rm f}}$

Supernova Rate $\nu_{\rm SN}$ (yr^-1) 0.45 0.87

Star Formation Rate $^{\rm g}$ SFR ( $M_\odot$ yr^-1) 11.0 21.2

Total Normalised SFR ${{\it SFR}\over{A_{\rm SF}}}$ ( ${{M_\odot}\over{{\rm yr}\ {\rm kpc}^{2}}}$ ) 0.072 34.1 $^{\rm h}$

**Table 3:** Global properties $^{\rm a}$ of NGC1792 and NGC1808.
Property	Symbol	NGC1792	NGC1808
Absolute blue magnitude $^{\rm b}$	$M_{\rm B}$ (mag)	$-20.15\pm 0.10$	$-20.24\pm 0.10$
Total blue luminosity	$L_{\rm B}$ ( $L_\odot$ )	$1.6\times 10^{10}$	$1.7\times 10^{10}$
Total H I gas mass	$M_{\rm HI}$ ( $M_\odot$ )	$1.1\times10^{9}$	$1.8\times10^{9~~{\rm c}}$
Total virial mass	$M_{\rm T}$ ( $M_\odot$ )	$0.54\times 10^{11~~{\rm d}}$	$1.1\times10^{11~~{\rm e}}$
H I gas to total mass ratio	$M_{\rm HI}/M_{\rm T}$	0.021	0.016 $^{\rm c}$
H I gas to blue light ratio	${M_{\rm HI}\over L_{\rm B}}$ ( ${M_\odot \over L_\odot}$ )	0.07	0.10 $^{\rm c,e}$
Total mass-to-light ratio	${M_{\rm T}\over L_{\rm B}}$ ( ${M_\odot \over L_\odot}$ )	3.4	6.5
Optical colour index $^{\rm b}$	B-V (mag)	$0.68\pm 0.14$	$0.82\pm 0.14$
Total FIR luminosity	$L_{\rm FIR}$ ( $L_\odot$ )	$0.63\times10^{10~~{\rm d}}$	$2.4\times 10^{10~~{\rm f}}$
Supernova Rate	$\nu_{\rm SN}$ (yr^-1)	0.45	0.87
Star Formation Rate $^{\rm g}$	SFR ( $M_\odot$ yr^-1)	11.0	21.2
Total Normalised SFR	${{\it SFR}\over{A_{\rm SF}}}$ ( ${{M_\odot}\over{{\rm yr}\ {\rm kpc}^{2}}}$ )	0.072	34.1 $^{\rm h}$

We will first introduce some global emission properties of both galaxies, before turning to details. The total H I line spectra of both NGC1792 and NGC1808 are displayed in Fig. 2. The resolution of the current data, of 6.6 km s^-1, is about 3 times as high as that of earlier observations (D92; K93). From these new spectra we derive the total widths of the integral H I lines at the 20%- and 50%-level of the maximum, W₂₀and W₅₀, respectively, the heliocentric systemic velocities, $v_{\rm sys}$ , and the total H I line fluxes, $f_{\rm HI}$ . $f_{\rm HI}$ is calculated as the sum over all bins with line emission from each galaxy, multiplied by the bin width of 6.6 km s^-1. These values are listed in Table 2.

All measurements of $f_{\rm HI}$ , W₂₀ and $v_{\rm sys}$ confirm, within the error margins, earlier results obtained from VLA observations (D92; K93). A few global properties derived from these global H I data and other data are collated in Table 3.

Kinematic parameters, such as the kinematic position angle, PA, and the inclination angle, i, and their behaviour as a function of radial distance from the galaxy centre, can be best constrained from the earlier, higher-resolution VLA data.

3.1.1 NGC1792

$\begin{figure} \par\psfig{figure=h2765f3a.ps,width=6cm}\hspace*{0.5mm} \psfig{figure=h2765f3b.ps,width=6cm} \end{figure}$

Figure 3: Contours of the total H I line emission of NGC1792, overlaid on a DSS optical image (left panel) and contours with underlying grey scale (right panel). The contour levels are 0.027, 0.11, 0.32, 0.54, 0.76, 0.97, 1.19, 1.41, 1.62, 1.84, $2.06\times 10^{20}$ cm^-2; the angular resolution is indicated in the lower left.

The H I distribution in NGC1792 (Fig. 3) is compact. Not only is the extent of the H I disk small compared to the stellar disk (D92), but with values up to about $3\times 10^{20}$ cm^-2 the H I column densities are quite high (see Fig. 1). Note that we use here, contrary to Fig. 1, the total intensity map created by routine MOMNT, using a smoothed version of the original datacube as a mask to avoid summing up noise in areas where no line emission is present. This leads to an improved signal-to-background ratio compared to the map in Fig. 1 (but leaves no sensible background "noise''). Even with its lower angular resolution our current map corroborates the steep outer gradient of the H I emission distribution seen in the earlier VLA observations (D92), which is reminiscent of a sharp radial cutoff.

Our data also confirm the regularity of the H I velocity field of NGC1792 (not displayed).

3.1.2 NGC1808

$\begin{figure} \par\psfig{figure=h2765f4a.ps,width=6cm}\hspace*{0.5mm} \psfig{figure=h2765f4b.ps,width=6cm} \end{figure}$

Figure 4: Contours of the total H I line emission of NGC1808, overlaid on a DSS optical image (left panel) and contours with underlying grey scale (right panel). The contour levels are 0.027, 0.054, 0.11, 0.22, 0.32, 0.43, 0.54, 0.76, $0.97\times 10^{20}$ cm^-2, the angular resolution is indicated in the lower left. The cross again marks the position of SN1993af.

Our H I map of NGC1808 (Fig. 4) clearly resolves the inner disk and the outer spiral arms. Note that we use almost the same contour levels in Figs. 3 and 4, with only a few more densely spaced low-level contours to display the outer regions of NGC1808. The H I distribution is consistent with earlier results (e.g., Saikia et al. 1990; K93). The surface brightness and thus column density of the H I emission from the inner disk is relatively low (about $0.2{-}1.0\times 10^{20}$ cm^-2, typically). That of the outer spiral arms is yet lower, with values below $0.2\times 10^{20}$ cm^-2, typically. Part of the central H I gas is not visible in emission because of intrinsic self-absorption along the line of sight towards the nuclear region (see K93 and Koribalski et al. 1996 for details). Therefore, the total H I line flux measured is a lower limit (Table 2). The cross marks the position of SN1993af, which will be discussed below.

The velocity field obtained from our data (not displayed) exhibits the same general features as observed in earlier VLA observations (K93), including the skewed appearance due to non-circular gas motions.

3.1.3 A comparison of NGC1792 and NGC1808

It is directly visible from Fig. 1 that the distribution of H I in NGC1792 is much more compact than that of NGC1808. The extent of the H I distribution in NGC1792 is smaller than that of its stellar disk and at the same time the surface brightnesses in the disk is also higher. After correction for the different inclination angles, the H I column densities in NGC1792 are still higher than in NGC1808, indicating that the mean gas density is higher.

These features of the H I distributions in both galaxies can account for the observed distribution of SF activity: while NGC1808 has a circumnuclear starburst ring, NGC1792 exhibits widespread SF over large parts of its disk. The low H I column densities in the $\theta$ -shaped outer spiral arms of NGC1808 suggest that there are no signs of SF there because of low gas densities. A comparison of the SF properties of both galaxies will follow below, in the context of our radio continuum images (Sect. 3.3.3).

The rotational behaviour of both galaxies, NGC1792 and NGC1808, and the relative alignment of their spin vectors was decribed by us in D92 (Fig. 5).

3.2 No intergalactic HI between NGC1792 and NGC1808

Thus, the current observations corroborate earlier results from dedicated studies of H I in both NGC1792 and NGC1808. But, more importantly, they represent the up to now most sensitive search for intergalactic H I gas in the vicinity of this pair of spirals.

The sensitivity of our data is illustrated by the detection of H I in the outer spiral arms of NGC1808, despite the fact that the galaxy was observed in only part of the pointings. These spiral arms have relatively low column densities on the order of 10¹⁹ cm^-2. Near the centre of our FOV, the 1- $\sigma$ rms noise is 5 mJy beam^-1 per channel so that structures with column densities of $\mbox{$\stackrel{\scriptstyle >}{\scriptstyle \sim}~$ }3\times 10^{18}$ cm^-2 should have been detected at the 5- $\sigma$ confidence level. Based on a fiducial extent of roughly $5\times5$ kpc (one beam size) and a requirement to have a 5- $\sigma$ signal in three consecutive channels, this translates into a lowest detectable line flux of $f_{\rm H\,I}({\rm min}) \simeq 0.5$ Jy km s^-1. Using the relation by Roberts (1975),

We used not only the data with the highest possible angular and velocity resolution, but also various smoothed versions of the data cube. No significant H I emission was detected that is not displayed in Figs. 1, 3 and 4.

It is evident from Fig. 1 that, despite the use of a very compact interferometer configuration, no H I gas could be detected in the space between NGC1792 and NGC1808 and in their immediate surroundings, at the sensitivity of the data. Only small amounts of diffuse H I gas might have eluded detection, as quantified above. This is not entirely unexpected because, with the gas-to-total mass ratios listed in Table 3, neither of the two galaxies is prominently H I-deficient when compared with the large sample used by Roberts & Haynes (1994) to determine the "typical'' properties of galaxies of different Hubble types.

3.2.1 Absence of tidal features

At the boundaries of the H I emission distributions of both galaxies, NGC1792 and NGC1808, no plumes, tails or other previously undetected tidal features were found at a significant level in addition to the H I emission features already known from dedicated observations centred on the individual galaxies, with a detection limit of order $3\times 10^7\ M_\odot$ .

These results imply that the interaction of the pair NGC1792 and NGC1808 that might have caused the observed disturbances (D92; K93) must have left the galaxies with most of their gas. Therefore, a far-field encounter is still the most likely explanation of all observations of these systems.

Our results leave no doubt that the NGC1808 group does not lend itself to dynamical studies by means of numerical modeling, because there are no distinct features that could sensibly constrain the boundary conditions for n-body simulations to model a putative interaction.

3.2.2 No dwarf galaxies detected

It is also interesting to note that our observations have not revealed any signs of H I emission from previously unknown dwarf galaxies near NGC1792 or NGC1808, as found in other groups like for example the NGC4666 group of galaxies (Walter et al., in preparation), see also e.g. Brinks (1990).

This again implies that there are no significant amounts of debris in the form of tidal dwarf galaxies in the NGC1808 group similar to those found for the first time at the end of the tidal arms of the "Antennae'' galaxies (NGC4038/39, Mirabel et al. 1992) and later e.g. in the Leo Triplet of galaxies (Chromey et al. 1998).

3.2.3 No primordial intergalactic HI clouds

Our results also bear some significance in the context of the hypothesis originally proposed by Oort (1966) and recently refined by Blitz et al. (1999) that Galactic H I high-velocity clouds (HVCs) might be primordial H I gas clouds raining down on the Galaxy. Assuming that similar H I gas clouds exist in external groups of galaxies, these should be detectable with the sensitivity of our observations.

Although our data do not span the entire NGC1808 group of galaxies, the non-detection of intergalactic H I at the sensitivity stated above implies a constraint on the possible existence of intergalactic primordial H I clouds. No clouds are detected down to a mass of about $10^7\ M_\odot$ . This implies that such clouds, if existent, cannot form a significant fraction of the total H I gas mass of that part of the NGC1808 group of galaxies around the two spirals NGC1792 and NGC1808 sampled by our observations. This, in turn, constrains the role that such primordial gas can play in forming HVCs raining down on the two galaxies NGC1792 and NGC1808, following the reasoning by Zwaan (2001). In the NGC1792/NGC1808 system intragroup H I gas clouds cannot play a significant role as the potential progenitors of HVCs. This implies that our observations cannot lend any support to the theory by Oort (1966) and Blitz et al. (1999). Detailed calculations and simulations based on H I observations of five groups of galaxies are given by Zwaan (2001) and in the references therein.

3.2.4 No intergalactic medium at all?

This non-detection of H I emission on different spatial scales does not exclude the possibility that an intergalactic medium might exist. However, our data put very tight constraints on the properties of such a medium. If located between NGC1792 and NGC1808 and if in the form of H I, this gas must have very low column densities ( $${\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil $\displaystyle ...$ cm^-2). There might also be H I gas outside the field of view of our observations, farther away from the two galaxies, towards the other group members (Garcia 1993). Although these are at yet larger projected distances from NGC1792 and NGC1808 and/or have lower total masses, it is conceivable that other members of the group, such as e.g. the late-type system ESO305-G009 at a recession velocity of 1025 km s^-1, might have taken part in a past interaction.

One might hypothesise whether any intergalactic gas might exist in the form of other phases. For example, at least part of the gas could possibly be too warm to be visible in H I emission. Donahue et al. (1995) detected a diffuse H $\alpha$ -emitting medium around NGC4631, which might also be related to the softest observed X-ray emission at $kT \simeq 0.05$ keV (Dahlem et al. 1998). Searches which could detect warm, H $\alpha$ -emitting gas in the NGC1808 group of galaxies have not yet been conducted. But it is unlikely that such gas exists, because warm ionised gas should normally be accompanied by a neutral component, as in the case of the NGC4631 group, which we did not detect around NGC1792/1808. Both the sensitivity and the field of view of the ROSAT observations of NGC1792 and NGC1808 (Dahlem et al. 1994; Junkes et al. 1995) should have led to the detection of a hot ( ${\sim} 10^6$ K) intergalactic medium, if such should exist. However, no hot ionised gas was found.

At the other end of the energy scale, the gas might be too cold to be seen in emission. Our H I emission observations trace only warm H I gas. Cold neutral hydrogen gas would not show up in our data, being only observable in absorption against background sources. The gas might be yet colder and exist in the form of molecular gas. However, such gas, outside self-shielding gas clouds and in the absence of significant amounts of shielding dust in galaxy disks, is exposed to the intergalactic UV radiation field and thus expected to be heated to temperatures that would make it visible in emission (Shull et al. 1999).

A search for signatures of stellar plumes or tails on deep optical images (see Chromey et al. 1998 for an example) in member galaxies of the NGC1808 group of galaxies, including ESO305-G009, also led to a negative result (D. Malin; private communication).

Although being far from conclusive, all these arguments make it very unlikely that there are significant amounts of intergalactic gas in the NGC1808 group of galaxies, at least in the form of gas phases that are presently known to us (and observable).

3.3 1.34 GHz continuum imagery

$\begin{figure} \par\includegraphics[width=7cm,clip]{h2765f5.ps} \par\end{figure}$

Figure 5: Image of the 1.34 GHz radio continuum emission of NGC1808 (upper left) and NGC1792 (lower right), the two dominant members of the NGC1808 group of galaxies, and their surroundings. The cross in the northeastern spiral arm of NGC1808 marks the position of SN1993af. The angular resolution of the data is indicated in the lower left; the linear grey scale ranges from 0 to the maximum observed surface brightness of 484 mJy beam^-1. The rms noise is about 1 mJy beam^-1 near the centre and 2 mJy beam^-1in the vicinity of SN1993af.

The setup of the ATCA allows for parallel H I observations at one intermediate frequency (IF) and continuum observations at the second IF. Therefore, 1.34 GHz radio continuum images with a bandwidth of 84 MHz were obtained simultaneously, which we can compare with the existing data in the literature (cf. Table 2). The map of the same field of view as the H I total intensity image (Fig. 1) is displayed in Fig. 5. NGC1792 and NGC1808 are the most prominent radio continuum sources in the field of view. In the lower left an arc of an imperfectly cleaned beam fringe of NGC1808 is visible. The rms noise of the map is about 1 mJy beam^-1 in the central region and about 2 mJy beam^-1 further out, near NGC1792 and NGC 1808.

3.3.1 NGC1792

The 1.34 GHz continuum emission distribution of NGC1792 derived from the present data is consistent with our earlier map (D92). The total flux measured from the present data is $325\pm10$ mJy. This is compatible with our earlier measurement of $324\pm20$ mJy at 1.41 GHz (D92), which implies that the flux calibration of our observations is consistent with the earlier VLA observations.

The radio continuum surface brightness is high, which is consistent with the high H I surface brightnesses (D92 and above) and the "extremely high-surface-brightness arms'' visible in optical images (Sandage & Brucato 1979).

3.3.2 NGC1808

The 1.34 GHz continuum emission of NGC1808 is almost unresolved by our observations. The resulting image does not add new information to previously published results (Dahlem et al. 1990; Saikia et al. 1990). The total flux derived from the present data is $616\pm15$ mJy, which is about 18% higher than our earlier result of $523\pm10$ mJy at 1.49 GHz (Dahlem et al. 1990). Partly, this might be explained by the different observing frequencies, but part of the discrepancy might be due to an intrinsic variability, possibly of the low-luminosity AGN (Véron-Cetty & Veron 1985; Awaki & Koyama 1993; Awaki et al. 1996).

We checked our data for continuum emission from the type Ia supernova SN1993af at $\alpha,\delta(2000) = 05^{\rm h}08^{\rm m}00\hbox{$.\!\!^{\rm s}$ }79, -37^\circ29'12\hbox{$.\!\!^{\prime\prime}$ }8$ (Hamuy & Maza 1993; Biegelmayer et al. 1993), in the north-eastern outer spiral arm of NGC1808. We have not detected any significant radio continuum emission at the position of the supernova, with a 1- $\sigma$ noise level of our map in this region of about 2.5 mJy beam^-1.

3.3.3 Comparing the radio continuum properties

Considering that NGC1808 is a classical starburst galaxy, while its companion NGC1792 is a "normal'' spiral, it is noteworthy that NGC1792's total radio continuum flux density at 1.34 GHz is more than half that of NGC1808 (Table 3; see also Junkes et al. 1995). The major difference between the two galaxies does not lie in the total energy output, but in the size of the region over which the energy is produced in stellar winds and supernovae (Leitherer & Heckman 1995) and released into the ISM.

In NGC1808, about 90% of the radio continuum emission arises from the central starburst, i.e. inside a region of $16''\times10''$ (Saikia et al. 1990). Assuming an inclination angle of about $60^\circ$ and thus circular symmetry of the starburst, with a major axis extent of 16'', this corresponds to an area of about 0.56 kpc². Based on a total radio continuum flux of 616 mJy at 1.34 GHz, this implies that about 555 mJy arise from the central region.

At the distance of NGC1808 of 10.9 Mpc, this corresponds to a radio luminosity of $L_{1.34}\ = 8.75\times10^{20}$ W Hz^-1. Using Eqs. (18), (20) and (21) by Condon (1992), and assuming that practically all radio continuum emission at 1.34 GHz is synchrotron radiation, this radio luminosity can directly be translated into a supernova rate, $\nu_{\rm SN}$ , and a star formation rate for stars with $M\ \geq\ 5\ M_\odot$ , SFR (see Table 3). For NGC1808 a mean radio spectral index of $\bar\alpha = -0.88$ ( $S\ \propto\ \nu^\alpha$ ) was adopted (Dahlem et al. 1990). Strictly speaking, this value is valid only for the central starburst, but since the radio emission of the starburst dominates the total flux, it should be a good approximation of the overall spectral index. From this, it follows that the normalised SFR within the circumnuclear starburst in NGC1808 (based on 90% of the total nonthermal radio continuum flux density) is approximately 34.1 $M_\odot$ yr^-1 kpc^-2.

In NGC1792 the SF activity is much more wide-spread, out to a radius of $2\hbox{$.\mkern-4mu^\prime$ }2$ (7.0 kpc; D92), leading to an area over which SF is observed, again under the assumption of circular symmetry, of 153.1 kpc². Based on a radio spectral index of $\bar\alpha = -0.82$ (D92), the total SFR following from the observed 1.34 GHz flux density of 325 mJy is 11.0 $M_\odot$ yr^-1, which then translates into a normalised SFR per unit area in NGC1792 of about 0.072 $M_\odot$ yr^-1 kpc^-2.

Thus, the primary reason for the different excitation conditions of the ISM in NGC1792 and NGC1808 and the prominent outflow from the central starburst region of NGC1808, which has no equivalent in NGC1792, is the difference in area over which the energy produced by stellar winds and supernovae is released, as reflected by the observed difference in radio continuum surface brightnesses in Fig. 5. This dilution effect alone leads to a difference in the average energy density by a factor of roughly 270, while the total difference in the normalised SFR between the disk of NGC1792 and the central starburst in NGC1808 is a factor of 474.

For more detailed information on the SF activity in both galaxies see e.g. Saikia et al. (1990), Forbes et al. (1992), Phillips (1993), Collison et al. (1994), Dahlem et al. (1994), Krabbe et al. (1994), Mazzolini & Webster (1996) and Kotilainen et al. (1996). The high-resolution rotation pattern of H I gas seen in absorption against the starburst in NGC1808 is described by Koribalski et al. (1996). The distribution of stars in the central $2'\times2'$ of both galaxies can be seen in the NIR H band images by Jungwiert et al. (1997).

3 Results and discussion

3.1 HI in NGC1792 and NGC1808