2 Observations

The CO(J=1-0, 2-1, 3-2, 4-3) observations presented in this paper were carried out in the years 1993 to 1998. The long time span and the varying availability of receivers have lead to some selection effects and different completeness levels in this catalogue. Statistical consequences are discussed in Sect. 3. Here we give the basic criterias for the sample selection, describe the instruments used, and present the observational results.

2.1 The sample

A number of well studied O-rich SRa and SRb variables were selected on the basis of their IRAS-60$\mu$m flux densities (typically $S_{60} \ge 5$Jy). As expected from the small dust mass-loss rates (SR_I) only very few of the objects were already detected in CO radio lines before our survey started (L93; K94). Our first results were published in SR_IV. Subsequently, these observations have been supplemented by observations of additional sources and additional transitions for already detected objects.

We selected the O-rich IRVs also on the basis of their IRAS-60$\mu$m flux densities, but down to levels of about 2Jy in order to get a comparable number of IRVs and SRVs. Nothing was known about their circumstellar CO properties before we started this survey. Based on our experience with the SRVs (SR_IV) only objects redder then 0.36 in the zero-point corrected IRAS-colour $[12~\mu{\rm m}]-[25~\mu{\rm m}]$ were observed in order to save observing time. This biased our sample towards mass losing objects with properties similar to those of "red'' and "Mira''-SRVs. First results on these observations have been published in Kerschbaum & Olofsson (1998).

In the appendix Tables A1 and A2 list basic properties of the observed stars. Among others the source of the listed coordinates (GCVS4 = General Catalogue of Variable Stars, 4$^{\rm th}$ edition, GSC = Guide Star Catalog, HIC = Hipparcos Input Catalogue, Sim = Simbad, TYC = Tycho Catalogue) is given. Most coordinates are of sub-arcsecond accuracy. In the few cases where only GCVS4 coordinates were available the quality was checked against the IRAS-PSC coordinates. The periods listed originate from the GCVS4.

Table 1 gives the number of stars, $n_{\rm star}$, in our sample down to a given S₆₀ level and divided into the variability groups.

  

Table 1: Statistics on observations and detections

2.2 The observing runs

The observations in the CO(J=1-0) line were obtained using the Swedish-ESO Submillimetre Telescope (SEST), Chile, the 20m telescope at Onsala Space Observatory (OSO), Sweden, and the IRAM 30m telescope, Spain. The CO(J=2-1) data were obtained at the SEST, the James Clerk Maxwell Telescope (JCMT), Hawaii, and the IRAM telescope. The CO(J=3-2, 4-3) data were obtained with the JCMT. The observing runs at SEST were made in January 1993, August 1994, December 1996, and August 1998. In March 1994 and 1995, and February 1997 we had observing runs at OSO. JCMT was used in flexible scheduling mode, or in backup programme mode, during the observing periods 96B, 97A and 97B. The IRAM spectra were obtained in April 1997. Telescope and receiver data are given in Table 2. $T_{\rm rec}$ and $\eta_{\rm mb}$ stand for the representative noise temperature of the receiver (SSB) and the main beam efficiency of the telescope, respectively.

  

Table 2: Data on Telescopes and Receivers

As spectrometers we used 1MHz filterbanks at IRAM, an autocorrelation spectrometer at the JCMT (250MHz bandwidth with 160kHz channel separation), 2 filterbanks at OSO ($256\times250$ kHz, and $512\times1$ MHz), and 3 acousto-optical spectrometers at SEST (86MHz bandwidth with 43kHz channel separation, 500MHz bandwidth with a 0.7MHz channel separation, and 1GHz bandwidth with 0.7MHz channel separation). At IRAM and at the JCMT beam switching (2$^\prime$) with a chopping secondary was used. Dual beam switching (beam throws of about 10$^\prime$), in which the source was placed alternately in the two beams, was used to eliminate baseline ripples at OSO and SEST. All data were calibrated by the chopper-wheel method, and the intensity scale is given in terms of main beam brightness temperature, $T_{\rm mb}$.

2.3 Observational results

  A total of 109 stars were observed in at least one CO line: 66 were shown to have circumstellar CO line emission (7 SRa, 36 SRb, and 23 Lb variables), $\sim$ 60% of the SRVs and all but one of the IRVs were detected for the first time. Most stars were observed in at least two transitions. There is a total of 138 detected CO lines. For twelve stars any possible detections were precluded because of strong interference from interstellar CO emission.

Table 1 gives our detection statistics. All numbers are given for three selection limits in S₆₀. $n_{\rm star}$ stands for the total number of stars in the GCVS4 down to the given S₆₀. $n_{\rm obs}$ is the number of stars observed in CO by us. $n_{\rm det}$ gives the number of detected sources (not transitions!). $n_{\rm det}/n_{\rm obs}$ and $n_{\rm obs}/n_{\rm star}$ give the detection rate and the fraction of the sample observed in percent, respectively. About 75% of all Lbs, SRas and SRbs have now been observed down to an S₆₀ level of 5Jy.

Tables A3-A8 in the appendix list all our observations. The names in the GCVS4 and the IRAS-PSC are given. The first letter of the code denotes the observatory (IRAM, JCMT, OSO, or SEST), the rest the transition observed. Another code reflects the "success'' of the observation ( Detection, Non-detection, Tentative detection). An "i'' indicates contamination by interstellar CO lines.

When determening the line parameters we compared the results of fitting parabolas or 4$^{\rm th}$ order polynomials, which generally fit the lines very well, with eye estimates of where the profiles go to zero intensity. There was in general a good agreement between the results obtained with the two methods. We determined the zero intensity velocities (i.e., the velocities at the two edges of the line profile) from the best fits. The stellar velocity was then derived from the average and the gas expansion velocity of the envelope from half the difference between these velocities, respectively. Before doing this we removed the major baseline irregularities by fitting low order polynomials to the spectra. We estimate that both quantities are uncertain by about 1-2 km s^-1, but the uncertainty varies with the S/N-ratio. This means that the low expansion velocities may be uncertain by up to 30%, while for the highest expansion velocities the uncertainty decreases to about 10%. The stellar velocity is given with respect to the heliocentric ($v_{\rm hel}$) and the LSR frame [$v_{\rm LSR}$; the Local Standard of Rest is defined using standard solar motion (B1950.0): $v_{\hbox{$\odot$}}$ = 20kms^-1, $\alpha_{\hbox{$\odot$}} = 270\hbox{$.\!\!^\circ$}5$, $\delta_{\hbox{$\odot$}} = +30\hbox{$.\!\!^\circ$}0$].

The peak main beam brightness temperature, $T_{\rm mb}$, is obtained as an average of the line profile intensities in the velocity range $v_{\rm hel} \pm 0.2v_{\rm exp}$. The integrated intensity, $I = \int T_{\rm mb} {\rm d}v$, is obtained by integrating the line intensities over the velocity range $v_{\rm hel} \pm v_{\rm exp}$. Once again the uncertainty in both quantities varies with the S/N-ratio, but we estimate that it is on average 24% in $T_{\rm mb}$, 12% in I, and in the worst cases it may reach 50% in both quantities. To this should be added an estimated uncertainty in the absolute calibration of about 20%. For a non-detection an upper limit to I is estimated by measuring the peak-to-peak noise ($T_{\rm pp}$) of the spectra with a velocity resolution reduced to 15kms^-1 and calculating $I = 15T_{\rm pp}$. The Q-column gives a quality ranking: 5 (non-detection), 4 (tentative detection, or detection with very low S/N-ratio ), 3 (detection, low S/N-ratio $\approx 5$), 2 (detection, good S/N-ratio $\approx 10$), and 1 (detection, very good S/N-ratio ). The S/N-ratio given is in terms of peak main beam brightness temperature divided by the 1 sigma noise for typical cases. A spectrum must have quality 3 or better to be used in the analysis. Finally, in cases of complex velocity profiles the measured component C is indicated in the form b=broad, n=narrow, b+n=total.

All spectra are shown in Figs. A1 to A7. Their order differ from that in Tables A3-A8, where the ordering is based on the source coordinates. Since we have a sample of visually bright variables we use the GCVS4-name to order the stars in figures. Tables A3-A8 allow the reader to identify the GCVS4-name if only the IRAS-name or the coordinates are known. The velocity scale is given in the heliocentric system (Tables A3-A8 list the corresponding LSR velocities). The velocity resolution is reduced to 1kms^-1, except for some low S/N-ratio spectra where a resolution of 2kms^-1 or even 4kms^-1 is used.

As expected from our earlier work we find no significant differences in the behaviour of the SRVs and IRVs in most of the studied properties. Consequently, they are combined in a large fraction of the figures in this paper.