NGC 3992 and its companions were observed with the Westerbork Synthesis Radio Telescope in the period May 1997 to September 1997. The observations lasted for approximately 48 hours divided up into $4 \times 12$ hour periods with different antenna spacings. Due to maintenance, on average, two of the fourteen telescopes, one movable and one fixed, were not available, which reduces the number of interferometers from 40 to approximately 27. The digital line backend was configured for 64 frequency channels, evenly spaced over a 5 MHz total bandwidth. After Hanning smoothing this resulted in a velocity resolution equal to twice the channel spacing, or 33 km s^-1, FWHM. A full listing of the observing parameters is given in Table 2.

Table 2: Observing parameters.
Telescope WSRT

Observing date May 1997 to Sept. 1997

Duration of observation 4 $\times$ 12 h

Number of interferometers $\sim$ 27

Baselines (min-max-incr.) 36-2736-36 m

Full res. beam ( FWHM, $\alpha \times \delta$ ) 14 $\arcsec$ $\times$ 18 $\arcsec$

FWHpower primary beam 37 $\arcmin$

Rms (1 $\sigma$ ) noise per channel

    full res. 1.96 K = 0.473 ${M}_{\odot}$ pc^-2

    res. = 30 $\arcsec$ $\times$ 30 $\arcsec$ 0.55 K = 0.132 ${M}_{\odot}$ pc^-2

Velocity central channel 1050 km s^-1

Bandwidth 5 MHz

Number of channels 64

Channel separation 16.6 km s^-1

Velocity resolution 33.3 km s^-1

Field centre (1950) (11 $^{\rm h}$ 55 $^{\rm m}$ 07 $^{\rm s}$ ; 53 $\degr$ 39 $\arcmin$ 18 $\arcsec$ )

K-mJy conversion,

    equivalent of 1 mJy/beam 2.62 K (full res.)

0.73 K (res. = 30 $\arcsec$ )

Adopted distance 18.6 Mpc

**Table 2:** Observing parameters.
Telescope	WSRT
Observing date	May 1997 to Sept. 1997
Duration of observation	4 $\times$ 12 h
Number of interferometers	$\sim$ 27
Baselines (min-max-incr.)	36-2736-36 m
Full res. beam ( FWHM, $\alpha \times \delta$ )	14 $\arcsec$ $\times$ 18 $\arcsec$
FWHpower primary beam	37 $\arcmin$
Rms (1 $\sigma$ ) noise per channel
full res.	1.96 K = 0.473 ${M}_{\odot}$ pc^-2
res. = 30 $\arcsec$ $\times$ 30 $\arcsec$	0.55 K = 0.132 ${M}_{\odot}$ pc^-2
Velocity central channel	1050 km s^-1
Bandwidth	5 MHz
Number of channels	64
Channel separation	16.6 km s^-1
Velocity resolution	33.3 km s^-1
Field centre (1950)	(11 $^{\rm h}$ 55 $^{\rm m}$ 07 $^{\rm s}$ ; 53 $\degr$ 39 $\arcmin$ 18 $\arcsec$ )
K-mJy conversion,
equivalent of 1 mJy/beam	2.62 K (full res.)
	0.73 K (res. = 30 $\arcsec$ )
Adopted distance	18.6 Mpc

Line emission of the main galaxy and the three companions was detected in 31 channels. Continuum emission only, was in principle available in 15 channels at the low velocity side of the emission and in 10 channels at the high velocity side. The continuum subtraction was complicated by a strong continuum source at a distance of 34 $\farcm$ 0 from NGC 3992, at position 11 $^{\rm h}56^{\rm m} 39^{\rm s}$ :54 $\degr$ 9 $\arcmin$ 48 $\farcs$ 8 (RA:dec:1950) with a flux of 1.7 Jy. First, this continuum source was cleaned away by subtracting 50 components with a gain of 0.5 found in a small region around this continuum source. This removed the worst of the grating rings. The rest of the continuum was subtracted by fitting a linear relation to the line free channels at both velocity sides and subtracting the appropriate amount from each channel in between. Next the data were cleaned. Clean components were subtracted until a level of 65% of the noise level. Data were restored by convolving the clean components with a Gaussian beam of 14 $\arcsec$ $\times$ 18 $\arcsec$ (FWHM, $\alpha \times \delta$ ) and adding the residuals. In this way a cleaned collection of line channels, or data cube, at full resolution was constructed, ready for further analysis. To give an impression of the H I intensity levels, H I extensions and of the position of the companions with respect to the main galaxy, already at this stage a total H I image at full resolution of the whole field is presented in Fig. 2. A detailed description of how this image was constructed can be found in the next section.

$\begin{figure} \par\resizebox{\hsize}{!}{\includegraphics{H3038F2.ps}}\end{figure}$

Figure 2: Greyscale image showing the full resolution total H I map of NGC 3992 and its surroundings. From top left to bottom right the three companions, UGC 6969, 6940, and 6923 are clearly visible; their H I column densities are larger than that of the main galaxy. Note the central H I hole of NGC 3992, at the region of the bar. The beam is indicated in the lower left, the greyscale is linear from $0.2\times 10^{20}$ to $34.8\times 10^{20}$ H-atoms cm^-2.

The whole field was searched in detail for H I emission other than that from the already known sources. Nothing was found; no unresolved sources above the noise level and no extended sources, nor in the full resolution field nor after smoothing to lower resolution. Considering the noise statistics of the channel maps, the threshold for detecting unresolved sources was put at five times the rms noise level corresponding to H I cloud masses of $3.5\times 10^6$ ${M}_{\odot}$ . As can be seen in Fig. 2 the three companions clearly stand out in H I and display a higher column density than the main galaxy. Thus if there is H I in the field it is associated with a stellar component, there are no free floating H I clouds around, at least not above the present detection limit. It can also be seen that the main galaxy has a faint gas extension around the stellar disc, this contrary to the companions, where the gas ends suddenly at the optical edge. A possible explanation for this is stripping of the gas from the companions when these have passed by, or interacted with NGC 3992.

2 Observations and data handling