Knapik & Bergeat proposed in Paper I, the first classification of the carbon variables in discrete photometric groups, independently of any spectral classification. The method was validated through colour-colour and colour-galactic latitude diagrams. It provides for every star sufficiently documented, a carbon variable group CVi (intrinsic SEDs from i = 1 the early one to i = 6 the late one) and the value of the interstellar extinction in the J-filter. The colour excess is for the mean extinction law of the diffuse interstellar medium (Mathis 1990) which was shown to be appropriate. A good agreement was obtained in most cases between those colour excesses and the field values from maps found in the literature. No gap was actually observed and discrete CVi-groups are adopted here only for convenience. The main features of this new pair method are
The quantity showed the correlation with true parallaxes from ESA expected for stars populating a given range in linear diameters (see Fig. 3 and Sect. 6 of Knapik et al. 1998). We must emphasize that photometric classifications do not correlate well with the spectroscopic classification of carbon stars (Shane 1928, Keenan & Morgan 1941, Keenan 1993; see Paper I). This missing correlation between photometry and spectroscopy was a strong motivation for our study. The case is completely different for oxygen-rich stars, and we were able to make use of averaged indices compiled from the literature (Johnson 1966, Johnson et al. 1968, Bessell 1979, Bessell & Brett 1988, and Koornneef 1983) in our analysis of Ba II stars (Bergeat & Knapik 1997). The obtained SEDs were named according to the corresponding spectral type and luminosity class (d for class V dwarfs, g for class III giants and sg for class Ib supergiants, e.g. G2d, G8g or M2sg). The same notation is kept hereafter whenever oxygen-type SEDs are considered.
The study in Paper I was concentrated on carbon stars with small amplitudes of variations (namely Lb or SR variables) making use of the six CVi-groups. Here we propose an extension of our work to the hotter carbon stars with bluer SEDs and to related stars. The HC stars, either variables or non-variables (the latter being a majority), are studied in Sect. 2. Six intrinsic mean SEDs we named HC0 (the earliest one) to HC5 (the latest one, close to CV1) are adopted. A few stars classified in Stephenson (1989) or usually considered as close to carbon stars (RCB variables, a few Ba II or CH or HdC stars..) do exhibit SEDs better fitted by oxygen-types SEDs (Sect. 3). Contrary to the variables of Paper I which presumably are all bright stars located on the asymptotic giant branch (AGB), many stars studied here are much fainter and possibly closer to the red giant branch (RGB). Some of them (e.g. the RCB variables) are luminous stars (bright giants or supergiants) usually considered as AGB or post-AGB objects. The cases of CS extinction and emission is discussed whenever possible, and a few conclusions are reached.
Every studied star was searched for variability in Kholopov et al. (1985, henceforth called GCVS) and the subsequent lists published in the Inf. Bull. on Var. Stars (67th to 73rd). Some of them were found in the New Catalogue of Suspected Variable Stars (Kukarkin et al. 1982, henceforth called NSV) which were not included in the GCVS as yet. They are denoted by NSV followed by the corresponding entry. Accurate HIPPARCOS photometry (ESA) was also considered when available. The remaining stars, presumably "constant" to the present accuracy, are usually named from their entries in the HD or BD catalogue.
© European Southern Observatory (ESO) 1999
Online publication: February 23, 1999