The 
observations were carried out with a large beam that embraces
the emission from the whole interacting system.
In most cases, the beam of the CO observations is sufficiently small
to separate the individual galaxies in an interacting system but
large enough to encompass the whole CO emitting area of each galaxy.
data
The observations were made with the 64 m Parkes antenna
in August 1994 and June 1995, and with the Australia Telescope Compact Array
(ATCA) in March 1996.
The Parkes radiotelescope has a beamwidth of 15' at 
21 cm, a sensitivity of
0.63 K Jy-1, and was equipped with a cryogenically cooled receiver
with a
system temperature of about 40 K for each linear polarization.
The backend was a 2048 channel autocorrelator, with a bandwidth of 16 MHz.
The galaxies were observed by position-switching
between source and empty sky. The reference position was taken 5' time
away from the source in right ascension.
The data were reduced using the Spectral Line Analysis Package (SLAP).
The spectra were smoothed to a final resolution of 6.6 km s-1.
The plots were made using the CLASS package (Forveille et al. 1990).
In some velocity intervals the Parkes observations were dominated by
interference and
these observations were repeated using the ATCA.
The ATCA was used in its most compact configuration
(array length of 122 m) since we were interesting in obtaining global
spectra and not maps to be as closely consistent with the Parkes observations
as possible.
The synthesized beam of the ATCA was
5',
the primary beam
33'.
The five antennas were fitted with cooled FET receivers. The backend
was a 2048 channel autocorrelator, with a bandwidth of 8 MHz.
PKS B1934-638 was used as a primary amplitude calibrator and for bandpass
calibration and was observed every day. A secondary calibrator
to correct for changes in gain and phase was observed at least once
per hour. The data reduction was done using AIPS. The uv-data have
been edited and calibrated, and global spectra were obtained using the
command POSSM.
The resolution of the ATCA spectra is 6.6 km s-1.
The final plots were made using CLASS and GRAPHIC.
The 12CO(1-0) observations have been carried out between June and
December 1995
in La Silla (Chile)
with the 15 m Swedish-ESO Submillimeter
Telescope (SEST) (Booth et al. 1989). At 115 GHz, the telescope
half-power beamwith is 43''. The main beam efficiency
of SEST is 
(SEST handbook, ESO).
We used a SIS receiver in single-sideband mode with
K.
During the observations, the typical system temperature was
300 K. A balanced on-off dual beam switching mode was used,
with a frequency of 6 Hz and two symmetric reference positions
offset by 12' in azimuth.
The pointing was regularly checked on nearby SiO masers.
The pointing uncertainties were about 5''.
The backend was a 1600 channel low-resolution
acousto-optical spectrometer with a total bandwith of 1 GHz, which
provided at 115 GHz a velocity resolution of 
.
The data were reduced using the software CLASS. The spectra
were smoothed to a final velocity resolution of 
.
Only first order baselines were subtracted from the spectra.
The uncertainties in the integrated CO and line intensities have been
computed on the following way:
they are the quadratic sum of errors due to noise in the
spectrum and in the determination of the baseline:
where 
= velocity range over which the spectrum was integrated,
= the smoothed channel width ( for the CO
spectra and 
for the spectra)
and 
= the velocity range over which the baseline was fitted.
The masses are related to the integrated intensities 
(in Jy km s-1) by:
where D is the distance to the galaxy in Mpc.
For the conversion factor from CO emissivities into H2 column densities, we use the value of Strong et al. (1988):
where 
is the main-beam line area.
For the SEST 43'' beam, this converts into
The conversion factor X has been established empirically for our own
galaxy and may differ for other galaxies (for a discussion about the
conversion factor, see e.g. Lequeux 1995).
Recent studies have reopened the question of the use of the
CO emission as a star-formation indicator: although stars form in
the dense cores of molecular clouds, it turns out that on galaxy
scale, some observations have shown that masses are better correlated
with other star-formation indicators (UV, H
) than CO luminosities
(e.g., Kennicutt 1989; Buat 1992;
Casoli et al. 1996).
For luminous objects however, a positive correlation has been found
between the molecular gas content
(as derived from CO line intensities using a standard conversion factor X)
and star-formation activity
(as traced by the H equivalent width and the far-infrared emission)
(Boselli et al. 1995).
This suggests that the value of X may vary from one galaxy to another;
for the low-luminosity ones which are less metal-rich,
H2 masses may be underestimated by using a standard
conversion factor.
The H2 masses we have computed
thus have to be taken as estimates of the molecular hydrogen
masses. The value we adopted for X has been choosen
to be consistent with previous papers
and allow direct comparison with other samples.