3 Detectability

 

3.1 Variability type

As shown in Table 1 the $n_{\rm det}/n_{\rm obs}$ value differs somewhat between the three variability groups (59% for the Lbs, 47% for SRas, and 64% for SRbs). This does not necessarily have a physical reason, but may be due to a selection effect. The relatively low value for the SRa variables (the smallest group) is probably due to the fact that we tried to observe all objects down to the S₆₀-limit, whereas in the cases of the SRbs and the Lbs we avoided observations of "blue''-objects (see below), and objects with very short periods. Consequently, any statistics derived from these observations has to be very preliminary until we arrive at more complete samples, but at this stage we find no differences between the "reddened'' IRVs and SRVs in terms of CO detectability. The total detection rates are high, 71% for the 10Jy-limit, 63% for the 5Jy-limit and decreasing only marginally to 61% for all observations.

3.2 Period

In Fig. 1 we present the detection rate as a function of the period for the SRVs. The statistics is poor in some period ranges, but it appears that the detection rate decreases for the shorter periods (75$^{\rm d}$), that it is hard to detect stars in the intermediate period range ($200^{\rm d}-300^{\rm d}$; although the fraction of stars with interstellar CO confusion is high here), while for the longer periods ($\gt 300^{\rm d}$) the detection rate approaches 100%. For the Miras observed by Young (1995, Y95) the detection rate drops drastically when going to periods shorter than 325$^{\rm d}$ (from about 70% to 30%), i.e., a very similar behaviour. Since we do not expect large luminosity differences among our stars (i.e., the sampled volume does not depend on the period), our detection statistics should be mainly caused by differences in mass-loss rate.

  

Figure 1: Period distribution of the observed SRVs. The different shadings denote the success of detection

3.3 IRAS-colour

In Fig. 2 we have plotted all our observed stars in the classical IRAS two-colour diagram (van der Veen & Habing 1988; including the regions introduced by them). There is no apparent trend in the detection rate with far-infrared colour, except that the six "bluest'' SRVs, located in region I, were not detected. For Lbs no objects bluer than 0.36 in $[12~\mu{\rm m}]-[25~\mu{\rm m}]$ were observed. Our sample objects populate, as expected, mainly regions II and IIIa. We note that the fraction of stars where the spectra are confused by interstellar CO emission increases with redder $[12~\mu{\rm m}]-[25~\mu{\rm m}]$ and $[25~\mu{\rm m}]-[60~\mu{\rm m}]$ colours, suggesting that the IRAS fluxes are not entirely free from an interstellar contribution.

  

Figure 2: IRAS two-colour diagram for the observed stars. The different symbols denote the success of detection

3.4 IRAS LRS-class and M-subclass

The IRAS LRS-class is an independent quantity measuring dust mass-loss in a qualitative sense. Figure 3 gives the detection statistics for the IRAS LRS-classes. No clear trend is visible. We detect objects with a silicate emission feature (2n class) as well as without a clear feature (the latter sources showing only a "flat'' continuum, LRS-classes 14-17). However, from recent ISO results (e.g., Kerschbaum et al. 1997) we have indication that many 1n-class objects show another dust feature at 13 $\mu$m, normally attributed to corundum (crystalline Al₂O₃, e.g. Begemann et al. 1997). Generally, one finds a stronger 13 $\mu$m feature for the warmer objects, whereas the colder ones are dominated by broad silicate emission at 10 and 18 $\mu$m.

Figure 4 gives the detection statistics for the M-subclasses. There is a trend of increasing detection rate with increasing sub-class (hence decreasing temperature), although we have very few objects with early and late subclasses. This again is comparable to the results on Miras by Y95.

Consequently, the detectability appears not to be very affected by the composition of the circumstellar dust or the stellar atmospheric temperature. However, we are still dealing with a small sample of objects, where we have concentrated on those expected to have circumstellar envelopes.

  

Figure 3: IRAS LRS-class distribution of the observed stars. The different shadings denote the success of detection

  

Figure 4: Spectral M-subclass distribution of the observed stars. The different shadings denote the success of detection