For each star we have collected the following additional information. Table 1 (click here) gives the distribution of the stars according to these additional data.
The General Catalog of Variable Stars (Kholopov et al. 1985, here after GCVS) divides long period red variables into two groups, those with amplitudes greater or less than 2.5 mag. The small-amplitude stars are often called "Semi-Regular'' and the larger amplitude ones "Miras''. It appears that this amplitude rule is not followed very strictly and that the division between these two groups of LPVs is somewhat artificial (see also Kerschbaum & Hron 1992). However, strong correlation between amplitude and emission features at some specific phase in the visible spectra has been found earlier. Using emission line information, Feast (1972) obtained a similar but not exactly identical division.
Both Mira and SR variables include M-, S-, and C-type stars.
The C/O abundance ratio was the first criterion discriminating
the M type (also called oxygen-rich stars,C/O 
1) and
the S stars (C/O 
), from the C type (C/O 
1,
also called carbon stars). This criterion corresponds
to the chemical composition of the surface layers. The second
criterion was based on their spectral characteristics (for
a review, see Jaschek & Jaschek 1987). Let us remark
that the classification of the C stars is less certain
than that of M stars, while S stars are the most difficult
to classify.
Intermediate objects also exist, as those between S and M stars, called MS and SM stars, or those between C and S stars, called SC and CS. Less common are those between C and M stars, called CM and MC stars (for AGB chemical composition, see Lambert 1988).
Masers from late type stars are essentially emitted by three
molecules (Elitzur 1980): OH (hydroxyl radical), SiO (silicon
monoxide), and H
O (water).
Detection of such maser emission depends both on its
intensity and on the distance of the star (Dickinson et al.
1978), thus non-detection of emission can be due either
to non-emission by the star or to greater distance of an
emitter. Also, the possibility that some stars have
asymmetric shells (perhaps because of an asymmetric mass
loss) can make detection of maser emission dependent on
the star's orientation. Furthermore, in a binary system
the companion could supress the maser emissions (Lewis
et al. 1987). Despite all this, maser emissions
are statistically meaningful.
The pumping mechanism responsible for OH masers is associated with the IR radiation. The properties of type I OH LPVs (characterized by main line emissions) and type II (showing strong satellite emission lines) have been studied earlier and some correlations have been found (Bower & Kerr 1977; Ukita 1982; Sivagnanam et al. 1989), including a strong correlation with mass loss. So, there is some evidence for the following sequence of increasing mass loss rates: (non OH)-(type I OH)- (type II OH)-(OH/IR) stars.
SiO maser emissions seem to be an atmospheric phenomenon (coming probably from the inner regions of the envelope) and thus provide direct information on the motion in it (Elitzur 1981; McIntosh et al. 1989). Hall et al. (1990) detected changes in the intensity of the SiO masers. Some suggest that the maser luminosity may depend strongly on the mass loss rate (cf. Bowers 1985).
The H
O maser emissions are still the less-studied
ones; the difficulties come from the strong variability
of their profiles (e.g. Bowers et al. 1993). Even so,
a relationship has been suggested between these variations
and the mass loss rate (Engels & Lewis 1990; Bowers
& Johnston 1994).
Table 1: Summary statistics
