2. Infrared colors of S stars

2.1. The sample

 

The sample of S stars considered in this paper was selected from the list of Chen et al. (1995), which provides associations between IRAS sources from the PSC and S stars from the General Catalogue of Galactic S Stars (GCGSS, Stephenson 1984) and from an additional later list (Stephenson 1990). Among these, only stars having flux densities of good quality (i.e. with a quality flag of 3 in the PSC) in the 12, 25 and 60 m bands have been retained.

Several stars that have probably been misclassified as S stars were removed from the final sample. For example, stars from the original Westerlund & Olander (1978) sample were later recognized by Lloyd Evans & Catchpole (1989) as actually being heavily-reddened M giants or supergiants. A few other cases of M giants or supergiants misclassified as S stars were identified by Keenan & McNeil (1989; HR 3296 = GCGSS 500) and Winfrey et al. (1994; GCGSS 1314 and star 41 in Stephenson 1990). A detailed heavy-element abundance analysis (Lambert et al. 1995) has shown that the stars DE Leo (HR 4088) and HR 7442, although often considered as S stars, have normal abundances. According to Meadows et al. (1987), GCGSS 886 (IRAS 15194-5115) is now classified as a carbon star and thus has not been retained in the final sample of S stars. Our final sample may still be somewhat contaminated by M supergiants misclassified as S stars and by a few carbon stars. The star T Cet for example was classified as M5-6Se in the original paper by Keenan (1954) defining the S class, but it was reclassified as M5/M6Ib/II in the Michigan Spectral Survey (Houk & Cowley 1975). Since at the low plate dispersions used in classification work, these two types of spectra look similar (e.g. Lloyd Evans & Catchpole 1989), T Cet has been kept in our final list until higher resolution spectra resolve these conflicting classifications. The same holds true for TV Dra and OP Her (see Table 2 of GCGSS). Note, however, that SC stars like RZ Peg, FU Mon and GP Ori are retained in the sample, since they may provide important clues to the evolutionary status of S stars as a whole. Similarly interesting are the two CS stars TT Cen and BH Cru that are known to exhibit ZrO bands at some times and C₂ bands at others (Stephenson 1973; Lloyd Evans 1985).

2.2. The IRAS flux densities

  The final sample consists of 124 S stars having flux densities at 12, 25 and 60 m flagged as being of good quality in the PSC. These stars are listed in Table 1. However, PSC flux densities suffer from several shortcomings that make them inadequate for the present study. First, the 60 m flux density may in some cases be seriously contaminated by Galactic cirrus emission, as shown by Ivezi & Elitzur (1995). Second, the PSC flux densities are not appropriate in the case of slightly extended or variable sources, as several S stars appear to be. Moreover, hysteresis of the detectors hampers the search for extended sources and should be properly identified. Finally, the detectors may saturate on very bright sources, those with flux densities in excess of 1000 Jy (like  Cyg, see Appendix A).

In order to correctly handle these effects, the raw IRAS data for all S stars in our sample were reprocessed through the ADDSCAN procedure provided by the Infrared Processing and Analysis Center (IPAC). This procedure has been used to co-add all scans passing within 17 of the target position. Before being co-added, the raw data are first interpolated using cubic splines and resampled at 10 data points per arcminute in all bands. A baseline is then defined for each individual scan by fitting a parabola to the data in a window extending from to from the target ( and in the 12, 25, 60 and 100 m bands, respectively). The rms residual of the data around the baseline fit is an indication of the background noise in a given scan i (note that possible nearby sources with a peak flux density exceeding are automatically removed and do not enter the final residual calculation). The different scans are then co-added with weighting factors equal to . The noise in the co-added scan, computed in a similar way as for the individual scans, is smaller than that in any individual scan, and the reduction may be substantial when scans with very different orientations, sampling different regions of the sky around the target, are available, since structure in the Galactic cirrus emission contributes to the noise. Finally, a point-source template with adjustable width is fitted to the co-added scan in the target window . Three different estimates of the flux densities in each band are then computed: the peak flux density , the template flux density and the "zero-crossing'' flux density . The template flux density is derived from the template fit to the data, and generally agrees with , unless the source is extended or hysteresis effects are important. For extended sources, the "zero-crossing'' flux density must be preferred. It corresponds to the integrated flux density between the "zero crossings'', which are defined as the first locations, moving outwards from the peak location, where the source profile intersects the baseline. In Sect. 2.4 (click here), a criterion based on the comparison of and has been designed to identify sources with possibly resolved shells. If that criterion is met, the flux density (identified by a "+'' in Table 1 (click here)) has been adopted instead of . In the case of bright sources, detector hysteresis results in a trail extending along the scan direction, thus disturbing the template fit. In that case, has been adopted. The template fit to sources embedded in strong Galactic cirrus emission generally resulted in an abnormally narrow profile, because the adopted baseline is then too high with respect to the base of the source signal. These contaminated sources were thus readily identified by their narrow template fit, and were rejected. Finally, individual scans clearly contaminated by nearby sources have been eliminated.

The adopted IRAS flux densities are listed in Table 1 (click here) in Cols. 5-8. The 2.2 m flux density is listed in Col. 3. The calibration of Beckwith et al. (1976; 620 Jy corresponds to K = 0) has been used to derive the 2.2 m flux density from the K magnitude provided by different authors, as listed in Col. 4: 1: Catchpole et al. (1979); 2: Neugebauer & Leighton (1969; Two-Micron Sky Survey); 3: Mendoza & Johnson (1965); 4: Price (1968); 5: Chen et al. (1988); 6: Noguchi et al. (1991); 7: Guglielmo et al. (1993); 8: Epchtein et al. (1990); 9: Epchtein et al. (1987); 10: Lloyd Evans & Catchpole (1989).

The Tc content (from Paper I, and Lambert et al. 1995) is listed in Col. 9. The Cols. labelled "LRS'' and "VC'' provide the classification of the IRAS low-resolution spectrum, according to the IRAS Low Resolution Spectrometer Catalogue (Olnon et al. 1986) or to the Volk & Cohen (1989) schemes, respectively. The optical spectral type is from the GCGSS. The variability type, the period and the range of variation of the magnitude (lowest minimum - highest maximum) are from the General Catalogue of Variable Stars (Kholopov et al. 1985; GCVS). The variable name and/or the HD/HR designations, when available, are given in the last column. If maser emission has been detected (see Sect. 3 (click here) and Table 3 (click here)), the maser type (SiO or OH) is given between parentheses after the stellar designation.

Several sources observed by IRAS a few months apart turn out to be strongly variable in the IRAS bands. For these, the ADDSCAN procedure has been run separately on the two groups of data, and the flux densities are listed on two separate lines in Table 1 (click here). The corresponding approximate Julian dates have been derived from the "Satellite Operation Plan'' number attached to the scans and from the mission chronology provided in Table III.C.1 of the IRAS Explanatory Supplement (1988).

The flux densities derived from the ADDSCAN procedure generally agree with the PSC flux densities within 20%, as shown in Fig. 1 (click here). In all bands, the scatter is larger at low flux densities. Several stars show much larger deviations because of intrinsic variability. For these, it turns out that the PSC flux densities correspond to one epoch, whereas the flux densities at the other epoch may differ by as much as a factor of 2. An example of a strongly variable source,  Cyg, is discussed in Appendix A. In the 60 m band, large discrepancies are also found for extended sources, as expected (see Sect. 2.4 (click here)).

 
Figure 1:   Comparison between the PSC flux densities and those derived from the ADDSCAN procedure (see text) in the 12, 25 and 60 m bands. Flux densities of variable sources at two different epochs are connected by a vertical line. At 60 m, open circles correspond to extended sources (see Sect. 2.4 (click here))

2.3. The infrared color-color diagrams

 

The use of the ([12] - [25], [25] - [60]) diagram to probe the circumstellar shells surrounding late-type stars was first demonstrated by Hacking et al. (1985) and by van der Veen & Habing (1988, VH). VH defined various regions (labelled I to VII) corresponding to circumstellar shells with relatively homogeneous properties. Figure 2 (click here) presents the ([12] - [25], [25] - [60]) diagram for the sample of S stars defined in Sect. 2.1 (click here). In this paper, the color index [i]-[j] is defined as (where F(i) refers to the non-color corrected flux density in the IRAS band i, and F₀(i) is a normalization flux density as given in the IRAS Explanatory Supplement 1988). With this normalization, black bodies in the Rayleigh-Jeans limit have a color index of 0. The colors for stars whose IRAS flux densities show large variations are plotted as two symbols joined by a line segment.

 
Figure 2:   The ([12] - [25], [25] - [60]) diagram for S stars. The size of the square is proportional to the ZrO/TiO or C/O spectral index of the S star (whichever is available), in the spectral classification scheme of Keenan (1954) or Keenan & Boeshaar (1980), respectively. MS and SC stars were assigned classes 1 and 9, respectively. Crosses correspond to stars with no spectral type available in the GCGSS. Tc-poor S stars are represented by filled squares, and colors of variable stars measured at different epochs are connected by a line segment

There is a strong correlation between the position of an S star in the color-color diagram and the intensity of the spectral features distinctive of the S star class (as measured by the ZrO/TiO or C/O spectral indices in the spectral classification schemes devised by Keenan 1954 and Keenan & Boeshaar 1980, respectively; both will be called "ZrO index'' for simplicity in the following). In particular, S stars with ZrO indices larger than 3 are mainly found in Regions VII and VIa with a few more in Region III, while S stars with ZrO indices smaller than or equal to 2 are found mainly in Regions I and II. This segregation of S stars in the ([12] - [25], [25] - [60]) diagram is a clear indication of the inhomogeneous nature of this family of peculiar red giants. In the center of Region I lie the S stars with photospheric colors. These are in fact the binary and Tc-poor S stars ("extrinsic'' S stars) that owe their chemical peculiarities to mass transfer in a binary system (Sect. 1). Another group of weak S stars is found at the boundary between Regions I and II, a zone generally devoid of stars in the ([12] - [25], [25] - [60]) diagram (see e.g. VH, and Lewis 1989). This group includes the prototypical Tc-rich Mira S variable Cyg. The few S stars located in Region II, defined by VH as comprising O-rich stars with "young'' circumstellar shells, indeed have small ZrO indices. Most S stars populate Region VII, where C-rich circumstellar shells are found according to VH, and it is therefore not surprising that those S stars have large ZrO indices. Finally, several stars (among which are many of type SC) are located in Regions VIa and b, many of them having conspicuously resolved circumstellar shells (see Sect. 2.4 (click here)).

 
Figure 3:   Same as Fig. 2 (click here) for the (K - [12], [25] - [60]) diagram. Note that the K - [12] index may be inaccurate for variable stars, since their K and [12] magnitudes were not measured simultaneously

Several authors (see e.g. Habing 1996) have argued that the K - [12] index is superior to the [12] - [25] index for tracing the mass loss rate, the focus of the present study, because of its greater wavelength range and the fact that the photosphere emits more strongly at K while the shell emits more strongly at 12 m. Therefore, we felt that it was more meaningful to use the (K - [12], [25] - [60]) diagram to define groups of S stars with homogeneous IR properties, as follows (Fig. 3 (click here)):
Region A: S stars with photospheric IR colors ("extrinsic S stars'');
Region B: S stars with no excesses at 60 m, and small ZrO indices;
Region C: S stars with excesses in all three 12 m, 25 m and 60 m bands, and large ZrO indices;
Region D: S stars with large ZrO indices and 60 m excesses, but small 12 m\ excesses;
Region E: mainly SC stars with large 60 m excesses and often resolved shells.

 
Figure 4:   Same as Fig. 2 (click here), where the stars have been drawn with a symbol corresponding to the region they belong to in the (K - [12], [25] - [60]) diagram (Fig. 3 (click here)), as follows: open squares: Region A; filled circles: Region B; filled triangles: Region C; crosses: Region D; open circles: Region E

 
Figure 5:   Same as Fig. 3 (click here), with symbols referring to the spectral class of the IRAS low-resolution spectrum as defined by Volk & Cohen (1989): S: stellar continuum; F: featureless spectrum; E: silicate emission; C: SiC emission; ?: not available. Spectral type assignments for S stars are from Chen et al. (1995)

The stars classified this way are plotted in the ([12] - [25], [25] - [60]) color-color diagram in Fig. 4 (click here), and it is seen that the above regions almost exactly correspond to those of VH, with the exception of Region B which encompasses both Regions I and II, and Region VII which is a blend of Regions C and D (though stars from Region D are concentrated in the upper left of Region VII). The motivation for creating Region D is apparent in Fig. 5 (click here), which presents the distribution of the various IR spectral types (as defined by Volk & Cohen 1989) across the (K - [12], [25] - [60]) diagram. Region D differs from Region C in having many stars exhibiting a stellar continuum in the IR (class S) and none with silicate emission (class E). As it will be shown in Sect. 4, these features suggest that the circumstellar shells in Region D contain C-rich grains, whereas silicate grains are found in the circumstellar shells of Region C. Note that Regions C and D are also quite distinct in the (K - [12], [12] - [100]) diagram (Fig. 6 (click here)).

 
Figure 6:   The (K - [12], [12] - [100]) diagram. Symbols are as in Fig. 4 (click here)

Finally, let us remark that the different regions defined above have been denoted by uppercase letters to avoid confusion with the Regions a to d defined in Paper I from the lower-accuracy PSC flux densities. The new classification into Regions A-E is actually not very different from that in Paper I (matching uppercase with lowercase letters, and with Region d splitting into D and E), but it is clearly superior since several stars that appeared exceptional in Paper I fit well into the present classification. For instance, Cyg, NQ Pup and o¹ Ori were exceptional as being Tc-rich stars in Region a. With the more accurate ADDSCAN flux densities, Cyg moves to Region B (see also Appendix A) while NQ Pup and o¹ Ori move to Region E. The only remaining outliers in this respect are HR Peg, the only Tc-rich star in Region A, and DY Gem, the only Tc-poor star not in Region A (but in C). DY Gem is however exceptional in many other respects: it is a very cool star (S8,5 corresponding to = 3000 K; Smith & Lambert 1990) and a SRa variable with a very long period (1145 d). Moreover, it has the largest [12]-[100] index among stars in Region C (see Fig. 6 (click here)).

2.4. S stars with envelopes resolved at 60 m

 

The possibility that some S stars may have circumstellar shells resolved by IRAS is now examined. As argued by Young et al. (1993a; YPK), the 60 m band is best suited for that purpose, because it is not contaminated by Galactic cirrus emission as severely as is the 100 m band. The 60 m co-added scans were examined, and characterized as follows. First, the width of the template profile fitted to the source is compared to that expected for point sources, namely 205 and 144 full widths at the 25% and 50% levels, respectively (Levine et al. 1993). However, this criterion is not sensitive to resolved shells showing up as a weak plateau at the base of the profile. A simple criterion has therefore been devised, based on the comparison of the "peak flux density'' with the "zero-crossing flux density'' (see Sect. 2.2 (click here)). These quantities are standard outputs from the ADDSCAN procedure. In the case of a point source, and are identical within a few times the noise, measured as the rms of the residuals along the baseline outside the signal range (i.e. between 25 and 10' from the target position in either directions) after baseline subtraction (see Sect. 2.2 (click here)). In the case of an extended source, the fraction of flux density in excess of that of a point source is expressed by - )/. For estimating the significance of this excess, one has to be aware of the following effects. First, very bright (> 500 Jy) sources may have a characteristic six-pointed star shape due to reflection from the telescope secondary mirror struts. Since approximately 5% of the peak flux density may be contained in the star pattern, / values of the order of 1.05 may be of instrumental origin (Levine et al. 1993). Bright sources are also affected by hysteresis in the detector which causes a trail in the signal along the scan direction in the outgoing part of the scan. This trail is easily recognized on individual scans as it causes an asymmetry in the template fit. However, if the co-added scan results from individual scans made in opposite directions, as is often the case, trails will be present in both directions and will mimick an extended plateau at the base of the profile. For bright sources (> 100 Jy), this effect may thus also cause spurious excesses of the order of a few percent. Finally, for fainter sources, an inhomogeneous background may also cause spurious detections. In this case, a way to estimate the significance of the excess is to compare it with the inverse signal-to-noise ratio /, which is nothing more than the relative flux density excess expected from the background noise. The significance, or "quality factor'' QF, of the observed flux density excess can then be expressed as , so that QF > 5 for a detection at the level. Figure 7 (click here) presents the ratio / vs. SNR for S stars not confused by nearby sources, and several stars with QF > 5 are found. One has to be aware, however, that the co-addition process will be mostly effective in lowering the baseline noise far away from the target, where scans with different orientations sample different regions of the sky. But since they all intersect approximately on the target, the noise-averaging process will not be as effective on the target. A criterion based on the absolute noise level present on-target has therefore been considered as well, by requiring that, to be considered significant, an excess - should not be smaller than some given threshold value of the order of the background fluctuations in the 60 m band. It was found that meaningful results are obtained by adopting 0.3 Jy as the typical background fluctuation on-target, combined with a quality factor of 5. In Fig. 7 (click here), sources satisfying the - > 0.3 Jy criterion have been represented by open squares. Since very few square symbols are located below the dashed line in Fig. 7 (click here), both criteria are fulfilled simultaneously for most of the stars, and thus provide consistent conclusions regarding the resolved nature of these sources.

 
Figure 7:   The ratio / between the "zero-crossing flux density'' and the "peak flux density'' in the 60 m band, vs the S/N ratio along the baseline (see text for details). Stars lying above the dashed line have a flux density excess in the 60 m band with a quality factor larger than 5. Stars with a flux density excess above the 0.3 Jy threshold are represented by squares. Squares lying above the dashed line thus satisfy both criteria defined in the text and are probably truly resolved sources. Filled squares and crosses correspond to sources flagged by YPK as extended or non-extended, respectively, at 60 m. Observations of the same star at two different epochs are connected by a line segment. Sources confused with a nearby source are not plotted

Sources flagged as extended at 60 m by the above criteria are listed in Table 2 (click here). Columns 1, 2 and 3 give the IRAS name, the variable star name and variability type when available, respectively. The zero-crossing flux density is in Col. 4. Column 5 lists the flux density ratio /, and Cols. 6 and 7 the full widths W25 and W50 at the 25% and 50% levels, respectively, to be compared with 205 and 144 for a point source (Levine et al. 1993). The quality factor QF is listed in Col. 8. Mass loss rates and wind terminal velocities are given in Cols. 9 and 10 (see Sect. 5 (click here)).

A similar search for late-type giants with resolved shells was performed by Young et al. (1993a,b). These authors used a more sophisticated method based on the possibility of successfully fitting the signal by a point source surrounded by a circumstellar shell having "reasonable'' properties. Our simpler approach has the advantage of being applicable to fainter stars, and several have indeed been detected, as seen in Fig. 8 (click here). However, for bright sources, our method is more vulnerable to spurious detections due to hysteresis (YPK used only the scan data taken prior to passing over the source).

As can be seen in Table 2 (click here) (in Col. 11, YPK+ and YPK- denote sources flagged by YPK as extended or not at 60 m, respectively), the two methods give conflicting results for four stars, R And, S Cas, W Aql and T Cet among the 11 bright objects common to the two samples. In the first three cases, our detection is probably an artefact due to detector hysteresis, since individual (as opposed to co-added) scans show an extended tail only on the side posterior to the passage over the source (see Fig. 10 (click here) below). The situation is less clear for T Cet, as its 100 m profile is wider than the point-source template, suggesting that this source may be truly extended.

Figure 8 (click here) presents the flux density excess / vs. F(2.2), the flux density at 2.2 m, and reveals that the properties of the resolved shells in Regions B and C are very different from those of Regions D and E. In Regions B and C, the flux density excess is of the order of a few percent, with a maximum of 15% for Y Lyn. Because the flux density excess is so small, the envelopes around stars in Regions B and C can be resolved only for the stars closest to the sun [i.e. with large F(2.2)], as shown by Fig. 8 (click here). By contrast, stars in Regions D and E have much larger flux density excesses at 60 m, which make them detectable at much lower total flux density levels. Stars in Regions B and C also differ markedly from those in Regions D and E as far as the [60] - [100] index is concerned (Fig. 9 (click here); see also Fig. 6 (click here)): the resolved envelopes in Regions D and E go along with large 100 m excesses, suggestive of cool dust in detached envelopes, contrary to the situation prevailing in Regions B and C (the only exception being T Sgr, which appears as a border case between C and D). In the case of RZ Sgr, which is extended in both the 60 m (YPK) and 100 m bands (according to the IRAS Small Scale Structure Catalogue 1985), an optical nebula has even been reported by Whitelock (1994).

 
Figure 8:   Flux ratio / vs. F(2.2). Symbols are as in Fig. 7 (click here), except that large squares now denote stars fulfilling simultaneously our two criteria for extended envelopes. Stars have been separated according to Regions B, C (upper panel) and D, E (lower panel)

In Region E, o¹ Ori and S929 do not follow the general trend. In those cases, there may be a confusing background source responsible for the strongly asymmetric 60 m profile (see Fig. 10 (click here)), as the 100 m profile is offset by about 1' in the direction of the 60 m asymmetry. The extension observed for S929 may be real; its [60] - [100] color index changed by only 2% between the two IRAS observations, while the 60 m flux density changed by 20%. This color stability is observed in other variable IRAS sources like Cyg (see Appendix A), but would probably not be preserved if the excess flux density were due to a background source.

Finally, the flux density excess in Regions D and E is associated with a widening of the whole 60 m (and sometimes 100 m) source profile, whereas the excess for stars in Regions B and C is caused by a weak plateau at the base of the profile, as apparent in Fig. 10 (click here). For the resolved shells in Regions D and E, the full widths at the 25% and 50% flux density levels (Table 2 (click here)) are indeed much larger than those expected for a point source. HD 191630 and RZ Sgr are listed in the IRAS Small Scale Structure Catalogue as extended at 60 m and 100 m, respectively. In fact, this distinctive feature of the resolved shells in Regions D and E has been used to include in Table 2 (click here) two stars (BI And and AA Cam) with wide profiles, despite quality factors QF < 5 that would normally not qualify them. However, these are distant stars with small 60 m flux densities, so that the / ratio cannot be determined accurately (and is therefore not listed in Table 2 (click here)).

 
Figure 9:   Same as Fig. 8 (click here) for / vs. [60] - [100]

 
Figure 10:   The 60 m co-added, spline-interpolated scans (dashed lines) for all stars with a possibly resolved shell in Regions D - E, and for selected cases in Regions B - C (see Table 2). The dotted line is the template 60 m profile as provided by IPAC. Note how larger the deviations from the template are in Regions D - E as compared to Regions B - C. In the latter case, the deviation of the observed profile from the template profile is due to a weak extended tail. Such detections are therefore vulnerable to detector hysteresis (see text)

 
Figure 11:   Position of the sources with resolved shells (filled circles) in the the (K - [12], [25] - [60]) diagram. The filled circles represent the color computed from the "template'' flux density at 60 m, i.e. they roughly correspond to the flux density of the star alone. The upper end of the vertical segment is located at the color computed from the "zero-crossing'' flux density at 60 m, i.e. the shell + star color

Stars with resolved shells are represented as black dots in the (K - [12], [25] - [60]) diagram (Fig. 11 (click here)). This figure illustrates the relative contribution of the extended shell to the [25] - [60] index: a line segment joins the [25] - [60] indices computed from the "zero-crossing'' flux density at 60 m (measuring the combined contributions of the star and its resolved shell) and from the "template'' flux density , which is assumed to be a rough measure of the stellar contribution alone (represented by a black dot in Fig. 11 (click here)). Although that assumption is certainly a very rough one, it is not totally unreasonable as stars from Region E now move down, and most reach Regions B, C and D when adopting instead of to represent the 60 m photospheric flux density. Some stars, however, do not quite leave Region E, presumably because the point source fitting yielding does not entirely remove the contribution of the resolved shell in those cases.

Especially interesting is the fact that Y Lyn and OP Her move from Regions C and D, respectively, to Region B, which is well in line with their small ZrO index (see Fig. 3 (click here) where they appear as outliers). With this adjustment, all three SRc variables in our sample (RS Cnc, Y Lyn and T Cet) now belong to Region B, and moreover have resolved shells (tentative in the case of T Cet). As noted by Young et al. (1993a,b) and Habing (1996), resolved shells are a common property of semi-regular variable stars.

Finally, it should be mentioned that several of the stars with a shell resolved by IRAS turn out to have an extended CO shell as well, as derived in Sect. 5.3 (click here) from the modelling of the CO data. The inferred radius of the CO shell (Table 6 (click here)) is larger than 10'' for  Cyg (18''), W Aql (23''),  Gru (15'') and FU Mon (60'').

 

Region A: Stellar photospheres

GCGSS IRAS F2.2 Ref. F12 F25 F60 F100 Tc LRS VC Sp Var P Name

(Jy) (Jy) (Jy) (Jy) (Jy) (d) (mag)

26 01113+2815 139.4 2 11.95 3.02 0.63 no F S3/2 HR 363

79 03377+6303 502.0 2 42.23 10.86 1.82 no 18 S4/2 Lb 0.1 BD Cam, HR 1105

133 05199-0842 92.1 2 7.94 2.06 no 16 S S4,1 HD 35155

382 07392+1419 331.4 2 25.91 6.51 1.08 0.37 18 M3S SR 0.2 NZ Gem, HD 61913

729 11098-3209 112.8 2 9.21 2.45 0.39 S Swk 0.1 NSV 5129, HR 4346

796 12272-4127 207.2 1 17.77 4.47 0.69 18 S M3-IIIa NSV 5655, HR 4755

804 13079-8931 167.6 1 15.23 3.88 0.67 17 S S5,1 Lb: BQ Oct, HD 110994

826 13372-7136 251.4 1 24.13 6.43 1.08 18 S S6,2 HD 118685

938 16425-1902 150.1 2 12.24 3.06 ^a no 31 S Swk HD 151011

1315 22521+1640 283.4 2 27.58 7.33 1.23 0.67 yes S S4/1 SRb 50 0.3 HR Peg, HR 8714

1322 23070+0824 938.4 2 84.67 20.31 3.20 0.96 no 18 M4S SRa 93 0.3 GZ Peg, 57 Peg

a: Strong cirrus contamination.

Region B: Small ZrO index and no [25] - [60] excess

GCGSS IRAS F2.2 Ref. F12 F25 F60 F100 Tc LRS VC Sp Var P Name

(Jy) (Jy) (Jy) (Jy) (Jy) (d) (mag)

8 00192-2020 1356.0 2 199.1 81.66 14.57+ 6.08+ yes 16 M5-6Se SRc 159 1.9 T Cet

134 05208-0436 80.2 8 20.56 10.85 1.65 43 U M4Swk V535 Ori

168 05495+1547 26.1 1 11.35 3.52 0.62 F S7.5,1e M 494 4.9 Z Tau

221 06266-1148 27.6 7 F S-*2e

JD 2445420: 8.97 4.10 0.94

JD 2445620: 14.57 7.64 1.14

265 06466-2022 87.2 2 8.34 2.41 0.47 16 M4S HD 49683

323 07117-1430 40.6 2 4.17 1.52 0.30

408 07461-3705 14.11 6.25 0.87 14 F S6*1

436 07545-4400 51.6 1 22 E S4,2 M 340 4.7 SU Pup

JD 2445470: 15.44 6.88 1.29

JD 2445645: 25.03 11.95 1.96

446 07573-6509 27.6 1 5.45 1.72 0.27 S7,2 M 280 2.7 X Vol

474 08098-2809 55.5 2 7.07 2.23 0.42 S S4,2 -27:5131

533 08348-3617 55.5 1 11.25 3.44 0.53 18 S S5,2 -36:4827

589 09076+3110 2808.0 2 489.88 210.29 34.57+ 11.71+ yes 22 M6S SRc 120 1.5 RS Cnc(OH?)

626 09411-1820 10.0 4.10 0.71 F M0S M 300 5.0 FM Hya

704 10436-3459 133.2 1 31.32 13.98 1.93 1.09 42 E S5,4 SR 104 2.0 Z Ant

903 15492+4837 1295.0 2 205.41 98.75 17.13+ 7.14+ yes 41 M6.5S SRb 148 1.5 ST Her

914 16097-6158 12.24 4.51 0.70 F S4,1 M 323 5.0 Y TrA

948 16552-5335 39.1 1 8.80 2.90 S4,4

1099 19008+1210 105.8 2 11.54 3.62 yes S S5/2 Lb 0.6 V915 Aql

1131 19226-2012 20.5 7 7.32 2.94 0.50 F M8Swk M 332 7.0 TT Sgr

1165 19486+3247 5813. 2 yes E S7/1.5e M 408 6.

JD 2445450: 1781.65^a 552.84 95.47+ 21.32

JD 2445640: 1215.31^a 414.40 71.64+ 14.50

1211 20213+0047 152.9 2 22.54 7.27 1.29 17 S S7,2 M 364 4.5 V865 Aql

1346 23595-1457 120.3 2 13.11 3.88 0.73 yes 16 F S5-7/1.5-3e M 351 7.7 W Cet

a: detector probably saturated
+: the "zero-crossing'' flux density has been adopted instead of the template flux density , indicating a possibly resolved
shell (see Sect. 2.2).

 

Region C: [12] - [25] and [25] - [60] excesses, silicate emission common

GCGSS IRAS F2.2 Ref. F12 F25 F60 F100 Tc LRS VC Sp Var P Name

(Jy) (Jy) (Jy) (Jy) (Jy) (d) (mag)

6 00135+4644 37.4 2 15.96 6.24 1.57 0.85 16 F S4/7e M 346 6.9 X And

9 00213+3817 453.3 2 335.74 175.66 26.41+ 9.45 yes E S5-7/4-5e M 409 9.1 R And (SiO)

14 00445+3224 96.5 2 37.75 18.58 3.97 2.10 22 F S6/2e M 430 7.8 RW And

28 01159+7220 147.4 2 343.76 192.52 30.14+ 10.88 22 E S4,6e M 612 8.2 S Cas (SiO)

36 01266+5035 34.1 6 6.59 1.93 0.45 S6/8e M 355 5.3 RZ Per

49 02143+4404 275.7 2 166.42 70.94 14.01 5.43 yes 22 E S7/1e M 396 7.9 W And (SiO)

149 05374+3153 280.8 2 45.33 23.32 5.00 yes 43 M2S Lc 0.2 NO Aur

231 06331+1415 136.9 2 21.95 10.35 2.61^a 3.90 no 42 F S8,5 SRa 1145 1.4 DY Gem

283 06571+5524 118.1 2 16 F S5/5e M 378 7.1 R Lyn (SiO)

JD 2445420: 17.55 5.61 1.25

JD 2445620: 28.46 9.46 2.01

307 07043+2246 88.0 2 yes 16 F S3,9e M 370 8.0 R Gem

JD 2445435: 22.23 7.53 2.21 1.40

JD 2445620: 42.33 16.81 3.73 1.48

316 07092+0735 12.23 5.63 1.06 0.36 E Se M 420 4.8 WX CMi

326 07149+0111 70.5 2 16 M7se: M 395 7.4 RR Mon

JD 2445440: 27.76 11.75 2.31 0.85

JD 2445625: 15.77 7.21 1.69 1.03

341 07197-1451 20.24 10.47 1.86 27 E M 314 4.0 TT CMa

347 07245+4605 964.7 2 133.50 65.92 13.20+^d 5.58+ yes 23 M6S SRc 110 2.5 Y Lyn (SiO)

614 09338-5349 68.0 1 01 E S7,8e M 408 3.0 UU Vel

JD 2445510: 11.53 5.98 1.77

JD 2445580: 17.79 9.37

649 09564-5837 156.15 71.37 14.31 15 E S6.5/1- SRb 109 1.3 RR Car

656 10017-7012 43.7 1 6.02 1.72 0.37 S5,6 M 3.5 KN Car

816 13136-4426 356.8 1 56.55 20.77 4.04 1.78 43 C S6/8=CS SR 114 2.0 UY Cen

821 13240-5742 13.40 6.27 ^b 14 F S6*3 SR 198 1.9 EE Cen

861 14372-6106 43.3 9 21.38 9.13 15 E CSV2170

872 15030-4116 111.8 1 21.81 9.23 2.18 F S7,8e M 326 5.6 GI Lup

931 16334-3107 377.0 2 52.49 21.88 4.22^{c b} 16 E S8/4var SRa 194 3.5 ST Sco

954 17001-3651 453.3 1 22 E S7,2 M 449 8.2 RT Sco

JD 2445410: 170.65 69.64 15.37

JD 2445600: 269.51 118.53 25.01

- 17081+6422 377.0 2 61.96 25.52 4.87 2.27 Lb 0.4 TV Dra

1093 18575-0139 37.7 1 17.18 5.92 1.83 SC M: 3.8 VX Aql

1096 18586-1249 242.3 2 51.78 19.19 4.03 21 E S6/3e M 395 8.8 ST Sgr

1112 19111+2555 18.7 6 41 SC M 438 5.8 S Lyr

JD 2445440: 43.81 21.43 4.39

JD 2445630: 84.17 38.23 7.53 2.73

1115 19126-0708 286.0 2 22 E S6/6e M 490 7.0 W Aql (SiO)

JD 2445440: 1397.52^e 719.87 132.67+ 36.87

JD 2445630: 1057.16^e 481.93 100.58+ 24.90

1117 19133-1703 166.1 2 42.27 14.62 4.76+ 4.76+ yes F S5/6e M 394 5.8 T Sgr

1150 19354+5005 107.7 2 107.44 52.86 12.22+ 5.52 yes 22 E S6/6e M 426 8.3 R Cyg (SiO)

1159 19451+0827 15.3 7 7.69 3.53 0.84 I Se M 607 2.0 QU Aql

1175 19545-1122 7.59 3.80 29 E M6S V1407 Aql

1200 20114+7702 I S5/6e M 326 6.9 SZ Cep

JD 2445394: 7.27 2.71 0.68

JD 2445580: 9.67 3.45 0.72

1268 21172-4819 14.9 1 7.15 2.34 0.63 I S2,5 -48:13866

a: close weak source
b: profile badly distorted by close source
c: close weak source apparent on some scans
d: moves to Region B after removing the extended shell contribution
e: detector possibly saturated
+: the "zero-crossing'' flux density has been adopted instead of the template flux density , indicating a possibly
resolved shell (see Sect. 2.2).

 

Region C (continued)

GCGSS IRAS F2.2 Ref. F12 F25 F60 F100 Tc LRS VC Sp Var P Name

(Jy) (Jy) (Jy) (Jy) (Jy) (d) (mag)

C3107^a 22036+3315 44.1 2 15.92 7.28 1.60 C SC M 439 6 RZ Peg

1294 22196-4612 4211. 1 yes 42 E S5,7 SRb 150 1.3

JD 2445470: 969.23 471.63 88.71+ 23.88

JD 2445645: 1421.10^b 419.22 92.25+ 25.80

1345 23554+5612 66.7 2 52.07 28.10 8.80 5.28 42 E S6/6e M 476 6.9 WY Cas

1347 00001+4826 18.2 6 50.54 24.54 4.15 21 E S5/6e M 396 3.3 IW Cas

a: star number from the General Catalogue of Carbon Stars (Stephenson 1989)
b: detector possibly saturated
+: the "zero-crossing'' flux density has been adopted instead of the template flux density , indicating a
possibly resolved shell (see Sect. 2.2).

Region D: No silicate emission, weak K - [12] excess, some [25] - [60] excess

GCGSS IRAS F2.2 Ref. F12 F25 F60 F100 Tc LRS VC Sp Var P Name

(Jy) (Jy) (Jy) (Jy) (Jy) (d) (mag)

12 00435+4758 64.9 2 8.47 2.49 ^{a a} yes 01 S S5/3e M 277 7.7 U Cas

17 00486+3406 36.3 5 4.56 1.44 0.35 S6/3e M 328 7.2 RR And

20 00578+5620 169.2 2 21.87 7.81 2.43 16 F S6/3 SRb 136 1.8 V365 Cas

29 01186+6634 46.6 2 5.10 1.90 0.42

57 02228+3753 95.6 2 9.89 2.80 0.66+ 0.70 01 S S8,8 SR 159 1.1 BI And

103 04352+6602 398.5 2 42.56 11.64 3.70 2.21 yes 17 S S4,7e M 373 7.1 T Cam

116 04543+4829 175.5 2 12.81 4.27 1.38 yes F S5,8 SRb 182 0.7 TV Aur

160 05440+1753 4.72 1.55 0.50 SRa 364 2.5 EI Tau

237 06347+0057 51.1 2 6.18 2.21 0.85 Lb 0.9 CX Mon

312 07095+6853 138.2 2 14.57 5.98 1.94+ 1.87 yes M5S Lb 0.6 AA Cam

387 07399-1045 177.2 2 19.48 7.25 2.08 18 S S3,6 SRb 2.8 SU Mon

524 08308-1748 65.5 2 5.85 2.00 ^a M 250 1.5 SZ Pyx

556 08461-7051 53.0 1 5.70 1.49 0.46 0.33 SC Lb: 0.4 UX Vol

617 09358-6536 39.5 1 4.00 1.22 0.27 S7,8

696 10389-5149 109.7 1 10.61 3.21 1.11 18 S S5,2 Lb 0.4 HP Vel

787 12135-5600 19.78 6.55 1.65 18 SC6/8e M 4.3 BH Cru

803 12417+6121 39.5 3 4.39 1.38 0.35 0.39 yes S3/6e M 226 5.6 S UMa

830 13440-5306 129.5 1 15.60 5.66 2.08 18 SC Lb 1.2 AM Cen

834 13477-6009 387.6 1 49.16 14.57 3.42 17 S S8,5 SR 307 3.3 VX Cen

904 15548-7420 58.1 1 6.12 1.93 0.65 0.59 SC SRa 152 3.5 VY Aps

923 16209-2808 71.2 2 7.71 4.10 1.27

940 16472-6056 49.2 1 5.71 2.06 0.72 SC M: 2.3 LQ Ara

978 17206-2826 319.4 2 39.93 11.48 2.72 17 S S5,4 SRb 320 1.4 V521 Oph

17553+4521 545.0 2 55.41 17.50 4.14+^d 3.20+ yes 17 MII-III SRb 120 0.9 OP Her

1011 17562-1133 8.45 3.30 ^c 01

1056 18310-3541 53.0 7 5.99 2.05 0.62 S5,4 L 0.5 V3574 Sgr

1141 19311+2332 258.2 2 32.48 13.87 3.06 16 F S6/5 Lb 3.7 EP Vul (SiO)

1188 20026+3640 356.8 2 40.22 15.45 5.37 yes 31 S S7,5 SRb 212 3.0 AA Cyg

1310 22479+5923 91.3 2 10.03 3.25 ^a F S6/2 SRb 60 1.4 CV Cep

1308 22476+4047 766.3 2 100.51 32.66 10.25+ 10.07+ 16 S S7.5/1 SRb 174 1.7 RX Lac

1339 23380+7009 60.9 2 7.95 2.89 0.86 F M6S

a: profile badly distorted by close source
c: strong cirrus contamination
d: moves to Region B after removing the extended shell contribution
+: the "zero-crossing'' flux density has been adopted instead of the template flux density , indicating a
possibly resolved shell (see Sect. 2.2).

 

Region E: Strong [25] - [60] excess, many resolved shells

GCGSS IRAS F2.2 Ref. F12 F25 F60 F100 Tc LRS VC Sp Var P Name

(Jy) (Jy) (Jy) (Jy) (Jy) (d) (mag)

82 03452+5301 66.7 2 7.66 3.36 2.40^b 01 F S5,8 Lb 0.8 WX Cam

89 04123+2357 258.5 2 25.13 7.63 3.30+ 2.56 yes S +23:654

114 04497+1410 1010. 2 87.25 22.19 7.52+ 3.57 yes 18 S3,1 SRb 30 0.2 o¹ Ori

117 04599+1514 88.8 2 11.11 3.45 1.44 17 SC SRb 370 1.1 GP Ori

212 06197+0327 144.7 2 17.16 4.63 2.66+ 5.88+ 18 S SC SR 310 1.1 FU Mon

422 07507-1129 79.5 2 7.21 1.82 0.48 yes I S5/2 Lb 0.1 NQ Pup

540 08388-5116 3.9 1 5.39 2.02 0.97 S6,8 S325

652 10015-4634 58.7 1 6.31 1.84 1.23 S5,8

817 13163-6031 94.7 1 21.52 12.91 7.01 SC5:/8 TT Cen

929 16316-5026 71.2 8 42 Se

JD 2445380: 65.34 31.24 16.70+ 30.18+

JD 2445600: 98.46 45.58 20.72+ 38.74+

1189 20044+2417 129.5 2 19.16 7.55 4.95+ 7.30+ 16 F S4,2 SRa 370 2.0 DK Vul

1195 20100-6225 133.2 1 11.03 2.82 1.77+ 3.27+ yes 18 S S4,4 HD 191630

1196 20120-4433 187.2 1 39.13 25.58 13.35+ 25.88+ 16 F S4,4 SRb 223 2.6 RZ Sgr

b: close weak source apparent on some scans
+: the "zero-crossing'' flux density has been adopted instead of the template flux density , indicating a
possibly resolved shell (see Sect. 2.2).

 

IRAS Name Var F60 W25 W50 QF Rem.

(Jy) (') (') Region B

00192-2020 T Cet SRc 14.6 1.05 1.96 1.44 14.5 4.6(-8)^a 6.9 YPK-,b

19486+3247 M 2.6(-7) 9.5 YPK+,f

JD 2445450: 95.5 1.01 1.96 1.45 9.5

JD 2445640: 71.6 1.002 1.93 1.42 0.4

09076+3110 RS Cnc SRc 34.6 1.04 1.97 1.44 29.2 1.9(-7) 7.4 YPK+,b,f

RS Cnc 5.2(-8) 2.6 e

15492+4837 ST Her SRb 17.1 1.02 1.95 1.43 6.7 2.0(-7) 9.5 YPK+,b Region C

00213+3817 R And M 26.4 1.05 2.01 1.47 30.6 8.6(-7) 9.3 YPK-

01159+7220 S Cas M 30.1 1.01 1.94 1.42 10.4 4.0(-6) 20.0 YPK-

07245+4605 Y Lyn SRc 13.2 1.15 2.00 1.45 30.9 2.3(-7) 8.6 YPK+,d

19126-0708 W Aql M 1.3(-5) 18.7 YPK-

JD 2445440: 132.7 1.006 1.93 1.44 5.8

JD 2445630: 100.6 1.02 1.95 1.44 7.5

19133-1703 T Sgr M 4.8 1.11 2.05 1.48 8.0 4.0(-7) 14.1 b

19354+5005 R Cyg M 12.2 1.03 1.96 1.44 10.0 1.7(-6) 10.5 YPK+,f

22196-4612 SRb 4.6(-7) 11.0 YPK+,f

JD 2445470: 88.7 1.04 1.98 1.44 37.9

JD 2445645: 92.2 1.07 2.02 1.49 38.3

02228+3753 BI And SR 0.7 - 2.17 1.58 2.6

07095+6853 AA Cam Lb 1.9 - 2.08 1.57 4.1 2.9(-7) 17.9

17553+4521 OP Her SRb 4.1 1.15 2.16 1.53 42.4 b

22476+4047 RX Lac SRb 10.2 1.16 2.11 1.52 23.6 4.0(-8) 3.4 b

04123+2357 - 3.3 1.20 2.18 1.53 7.2

04497+1410 o¹ Ori SRb 7.5 1.48 2.89 1.71 34.8 <2.0(-7) c

06197+0327 FU Mon SR 2.7 1.40 2.42 1.62 12.6 1.0(-7) 2.8 b

16316-5026 S929 -

JD 2445380: 16.7 1.50 3.08 1.81 24.8 b

JD 2445600: 20.7 1.89 - 1.87 24.1 b

20044+2417 DK Vul SRa 4.9 1.22 2.32 1.60 25.3 2.9(-7) 5.0 b

20100-6225 HD 191630 - 1.8 1.35 2.57 1.67 24.0 <1.0(-6) b,c

20120-4433 RZ Sgr SRb 13.3 1.26 2.28 1.58 62.1 2.1(-6) 8.8 YPK+,b,d

Notes:
a: numbers between parentheses refer to power of ten
b: 100 m profile wider than point-source template
c: listed as extended at 60 m in the IRAS Small Scale Structure Catalogue
d: listed as extended at 100 m in the IRAS Small Scale Structure Catalogue
e: two-components wind
f: circumstellar shell resolved in CO (Sahai 1992; Stanek et al. 1995).

 

Star IRAS SiO H₂O OH Refs. Region A

BD Cam 03377+6303 - N - B96

NZ Gem 07392+1419 - N - B96

HR Peg 22521+1640 N N - B96,  Z89

GZ Peg 23070+0824 N N - D73,  H89 Region B

T Cet 00192-2020 N N N A89a,  A90,  D76, H94,  L78,  P71

V535 Ori 05208-0436 - N N D89,  H91

Z Tau 05495+1549 N - N B78,  J91

RS Cnc 09076+3110 N N Y? D73,  D76,  A92, D82,  K77,  L78, L90,  P92,  R76

Z Ant 10436-3459 - - N H91

ST Her 15492+4837 - N N D73,  D76,  D82, P79

S Her 16496+1501 N N - B77a,  B84,  B87, B90,  E88,  S81

19486+3249 Y N N A89b,  A92, B87,  B94,  C83a, C96,  D76,  J84, J85,  J87,  J91, N85,  O85,  P92

W Cet 23595-1457 N N - B77a,  B84,  L76, L78 Region C

R And 00213+3817 Y N N B77a,  B78,  B87, B94,  B96,  C96, D78,  F75,  J91, T94

RW And 00445+3224 N N N B77a,  B78,  C83b, C96,  F75,  F78, K77,  W72

S Cas 01159+7220 Y N N B77a,  B94,  F73, T94

RZ Per 01266+5035 N - - B77a

W And 02143+4404 Y N N A89b,  B77a,  B87, B94,  C96,  D96, D78,  K77,  O80, T94,  W72

NO Aur 05374+3153 - N N B75,  B81,  B96

R Lyn 06571+5524 Y N - B77a,  B94,  K84, S81

R Gem 07043+2246 N N N B77a,  B78,  B96, J91,  L78,  W72

RR Mon 07149+0111 N - N K77,  L78

TT CMa 07197-1451 N N N B90,  B96,  H91

Y Lyn 07245+4605 Y N N A90,  B75,  B81, D73,  D82

RR Car 09564-5837 N N - H94,  L77

ST Sco 16334-3107 N N N B77a,  B94,  J88, L77,  W72

RT Sco 17001-3651 N N N B77b,  C71,  L76, L77,  L78,  S88

TV Dra 17081+6422 N - N B94,  W72

ST Sgr 18586-1249 N N - B77a

W Aql 19126-0708 Y N N B77a,  B84,  B87, B94,  D89,  L78, W72,  Z79

 

Star IRAS SiO H₂O OH Refs. Region  C (continued)

T Sgr 19133-1703 N N - B77a,  B90

R Cyg 19354+5005 Y N N B94,  B96,  C96, P92,  W72,  Z79

RZ Peg 22036+3315 - N - B96

22196-4612 N N - D89,  H90

WY Cas 23554+5612 - N - B77a,  C83b Region  D

U Cas 00435+4758 N N - B77a,  C83b,  J91, S81

RR And 00486+3406 N N N B77a,  F78

S29 01186+6634 - N - C83b

BI And 02228+3753 - N - C83b

T Cam 04352+6602 N N - B77a,  B94,  D76, J91

TV Aur 04543+4829 N - - B77a,  B94

EI Tau 05440+1753 N - - B77a

AA Cam 07095+6853 - N - B96

SU Mon 07399-1045 N - - B94

SZ Pyx 08308-1748 - N - C83b

S UMa 12417+6121 N N - B77a,  B96,  D78

VX Cen 13477-6009 N - - P92

OP Her 17553+4521 - N - B96

EP Vul 19311+2332 Y - N B94,  R76

AA Cyg 20026+3640 N N - B77a,  B94,  B96, D82

RX Lac 22476+4047 - N N D73,  D82,  W72 Region  E

o¹ Ori 04497+1410 - N N B96,  D73,  P71

GP Ori 04599+1514 N - - L78

FU Mon 06197+0327 N - - B77a

RZ Sgr 20120-4433 - - N C71

References to Table 3. The parentheses at the end of each reference indicate which maser is probed by the paper

A89a: Allen D.A., Hall P.J., Norris R.P., Troup E.R., Wark R.M., Wright A.E., 1989, MNRAS 236, 363 (SiO)

A89b: Alcolea J., Bujarrabal V., Gallego J., 1989, A&A 211, 187 (SiO)

A90: Alcolea J., Bujarrabal V., Gomez-Gonzalez J., 1990, A&A 231, 431 (SiO)

A92: Alcolea J., Bujarrabal V., 1992, A&A 253, 475 (SiO)

B75: Bowers P.F., 1975, AJ 80, 512 (OH)

B77a: Blair G.N., Dickinson D.F., 1977, ApJ 215, 552 (SiO, )

B77b: Bowers P.F., Kerr F.J., 1977, A&A 57, 115 (OH)

B78: Bowers P.F., Sinha R.P., 1978, AJ 83, 955 (OH)

B81: Bowers P.F., 1981, AJ 86, 1930 (OH)

B84: Bowers P.F., Hagen W., 1984, ApJ 285, 637 ()

B87: Bujarrabal V., Planeses P., del Romero A., 1987, A&A 175, 164 (SiO)

B90: Benson P.J., Little-Marenin I.R., Woods T.C., et al., 1990, ApJS 74, 911 (SiO, , OH)

B94: Bieging J.H., Latter W.B., 1994, A&A 422, 765 (SiO)

B96: Benson P.J., Little-Marenin I.R., 1996, ApJS 106, 579 ()

C71: Caswell J.L., Robinson B.J., Dickel H.R., 1971, Ap. Lett. 9, 61 (OH)

C83a: Clemens D.P., Lane A.P., 1983, ApJ 266, L117 (SiO)

C83b: Crocker D.A., Hagen W., 1983, A&AS 54, 405 ()

C96: Cho S.H., Kaifu N., Ukita N., 1996, A&AS 115, 117 (SiO)

D73: Dickinson D.F., Bechis K.P., Barrett, A.H., 1973, ApJ 180, 831 ()

D76: Dickinson D.F., 1976, ApJS 30, 259 ()

D78: Dickinson D.F., Snyder L.E., Brown L.W., Buhl D., 1978, AJ 83, 36 (SiO)

D82: Dickinson D.F., Dinger A.S., 1982, ApJ 254, 136 ()

D89: Deguchi S., Nakada Y., Forster J.R., 1989, MNRAS 239, 825 ()

E88: Engels D., Schmid-Burgk J., Walmsley C.M., 1988, A&A 191, 283

F73: Fillit R., Foy R., Gheudin M., 1973, Ap. Letters 14, 135 (OH)

F75: Foy R., Heck A., Mennessier M.O., 1975, A&A 43, 175 (OH)

F78: Fix J.D., Weisberg J.M., 1978, ApJ 220, 836 (OH)

H89: Heske A., 1989, A&A 208, 77 (SiO)

H90: Haikala L.K., 1990, A&AS 85, 875 (SiO)

H91: te Lintel Hekkert P., Caswell J.S., Habing H.J., Haynes R.F., Norris R.P., 1991, A&AS 90, 327 (OH)

H94: Haikala L.K., Nyman L.A., Forsstrom V., 1994, A&AS 103, 107 (SiO)

J84: Jewell P.R., Batrla W., Walmsley C.M., Wilson T.L., 1984, A&A 130, L1 (SiO)

J85: Jewell P.R., Walmsley C.M., Wilson T.L., Snyder L.E., 1985, ApJ 298, L55 (SiO)

J87: Jewell P.R., Dickinson D.F., Snyder L.E., Clemens D.P., 1987, ApJ 323, 749 (SiO)

J91: Jewell P.R., Snyder L.E., Walmsley C.M., Wilson T.L., Gensheimer P.D., 1991, A&A 242, 211 (SiO)

K77: Kolena J., Pataki L., 1977, AJ 82, 150 (OH)

References to Table 3 (continued).

K84: Kuiper T.B.H., Swanson P.N., Dickinson D.F., et al., 1984, ApJ 286, 310 ()

L76: Lépine J.R.D., Paes de Barros M.H., Gammon R.H., 1976, A&A 48, 269 ()

L77: Lépine J.R.D., Paes de Barros M.H., 1977, A&A 56, 219 ()

L78: Lépine J.R.D., Le Squeren A.M., Scalise E., 1978, ApJ 225, 869 (SiO)

L90: Lewis B.M., Eder J., Terzian Y., 1990, ApJ 362, 634 (OH)

N85: Nyman L.A., Olofsson H., 1985, A&A 147, 309 (SiO)

O80: Olnon F.M., Winnberg A., Matthews H. E., Schultz G.V., 1980, A&AS 42, 119 (OH, )

O95: Olofsson H., Rydbeck O.E.H., Nyman L.A., 1985, A&A 150, 169 (SiO)

P71: Pashchenko M.I., Slysh V., Strukov I.,  et al., 1971, A&A 11, 482 (OH)

P79: Pashchenko M.I., Rudnitskij G.M., 1979, Astron. Tsirk. 1040, 4

P92: Patel N.A., Joseph A., Ganesan R., 1992, J. Astron. Astrophys. 13, 241 (SiO)

R76: Rudnitskij G.M., 1976, Soviet AJ 20, 693 (OH)

S81: Spencer J.H., Winnberg A., Olnon F.M., et al., 1981, AJ 86, 392 (SiO)

S88: Sivagnanam P., Le Squeren A.M., Foy R., 1988, A&A 206, 285 (OH)

T94: Takaba H., Ukita N., Miyaji T., Miyoshi M., 1994, PASJ 46, 629 ()

W72: Wilson W.J., Barrett A.H., 1972, A&A 17, 385 (OH)

Z79: Zuckerman B., 1979, ApJ 230, 442 (SiO)