3. The CV7-group
3.1. A later intrinsic SED for carbon variables
In Paper I, we made use of the =[0.78]-[1.08] colour index derived from Baumert's (1972) data. As quoted in Table 1 of Paper I, is nearly a constant (1.06 to 1.14) along the CV1 to CV6 sequence. This remarkable property was turned into advantage by using diagrams of e.g. vs ([1.08]-K)- (see Fig. 2 of Paper I): the reddening vector is nearly perpendicular to the intrinsic narrow locus of unreddened stars. Similar diagrams were used with the H, L or L'-magnitudes instead of the K-ones. They are slightly less efficient. While preparing the present analysis of carbon Miras, the authors found well-documented variables with SEDs notably redder than the CV6 one. Some of them were interpreted as reddened CV6 stars, making use of the above-mentioned kind of diagram. The interstellar extinction could not be responsible for many other SEDs (including some of high galactic latitude stars) as shown by the inspection of published maps (FitzGerald 1968, Neckel & Klare 1980, and Burstein & Heiles 1982). Those SEDs have to be intrinsically red (mean temperatures less than 2300 K) and/or to be affected by some CS reddening. This is the case of the Miras R For (C361), V CrB (C3682) and T Dra (C3921). Finally, we were able to add a seventh group (CV7) based on this limited sample of only three unreddened (negligible interstellar extinction) Miras at maximum light (except data in K, L etc.. of T Dra since IR excesses were obvious). This new group is documented and included in Tables 6 and 7.
3.2. The example of V393 Aur = AFGL 815
We present here the detailed results for V393 Aur (C 1050) which is quite typical of the CV7-group in every aspect. It was studied as an infrared object (AFGL 815 = IRC 40140 = IRAS 05440+4311) before being catalogued as the SRa variable V393 Aur (Kazarovets et al. 1993; formerly SVS 2629). The available magnitude ranges are 14.8-16.1 and 1.3-2.3 in the V- and L-filters, respectively. The short wavelengths data of Lockwood (1974) was especially useful. Infrared observations at various epochs were taken from Strecker & Ney (1974), Merrill & Stein (1976), Gosnell et al. (1979), Ney & Merrill (1980), Noguchi et al. (1981: values shifted as described in Sect. 2.3 of Paper I), and of course the IRAS-PSC data (1988; see Sect. 2.3 of paper I for the calibration used). The vs. dereddening diagram of V393 Aur is shown in Fig. 2 for selected data at two epochs (likely maximum at JD 2441659 and minimum light close to JD 2443472). The obtained solutions are CV7 for both epochs, with , and , respectively. The mean extinction law for the diffuse interstellar medium was again used as tabulated by Mathis (1990, Table 1). According to the available maps, is adopted for the interstellar medium on the line of sight. Our results support the existence of an additional variable selective CS extinction. The diagram appears as distorted at short wavelengths if a distinct selective law is adopted. We deduce selective CS contributions of and 1.37 at maximum and minimum lights respectively. Neutral (i.e. independent of wavelength) extinction would however remain undetected from the present method. Lower limits of about and 4.9 respectively, are thus deduced for the CS extinction at the above-mentioned epochs. Due to the thermal emission from CS grains which is not considered in our pair method, IR excesses appear in Fig. 2 when a comparison is made to the corresponding extrapolated straight line. The latter is a "model" of the reddened star in terms of a given intrinsic group and SED, affected by a given selective extinction , assuming a given extinction law is verified throughout. The excesses of V393 Aur start from the H-filter (r=0.62) at maximum light and from the K-filter (r=0.38) at minimum light. Finally, we summarize the main features we found in common in many CV6 and CV7 stars:
Table 1. The cool reddened carbon variables of the CV7-group with C-entries from Stephenson (1989). A few stars not quoted in the GCVS (1985) are named with their IRAS-entry. The Miras are denoted by M. For the star is considered as unreddened. C496 is also CIT5 and C5358 is also CIT13. The E'(B-V) indices quoted are field values from published maps. Approximate phases and IR excesses (n = no excess or filter of first occurrence H, K, L, L',  or ) are found in the last column; (a) star at bright level; (b) = AFGL 1131: companion contribution subtracted; (c) = AFGL 971; (d) = AFGL 1235; (e) or 0.7 if beyond 2 kpc from sun.
Extinction variations on time scales larger than the stellar period are also observed (see Sect. 6). The reader is referred to Le Bertre (1992) and to Whitelock et al. (1997) who documented and studied a sample of carbon-rich variables on many cycles.
3.3. The results
Then we applied our pair method, as described in Sect. 2, to late carbon variables especially those whose analyses as reddened CV6 stars remained unsatisfactory. We were able to find 23 stars including 12 Miras (30 SEDs) which can be interpreted as CV7 stars reddened through the mean law for the diffuse medium as given by Mathis (1990). The results are quoted in Table 1. About 40 potential candidates were rejected because of missing data at short wavelengths. No E(B-V) excess substantially lower than the values read from maps is observed, but larger values are frequently found, especially at phases close to minimum light. The values at maximum light are smaller and eventually close to those (interstellar) from maps (see Table 1). As mentioned in Sect. 3.2, our analysis leads to a model of the observed SED (CVi-group) modified by some selective (interstellar + CS ) dust extinction. This model is illustrated graphically by the straight lines and CVi-groups obtained (see Sect. 2.2). Thermal emission from CS grains is also observed resulting in substantial IR excesses relative to our model (see Figs. 2, 6 and 7 for CV7-stars). Part of the SEDs in Table 1 received the comment "n" in the last column, that is "no IR excess detected". This statement holds only for our wavelength range (up to IRAS ). We decided (see Paper I) not to include IRAS measurements at and since the excess emission may often be due to contamination by galactic cirrus (Zuckerman 1993, Ivezic & Elitzur 1995). Intrinsic excesses contributing to detached shells are also observed in a sample of carbon stars (see Wallerstein & Knapp 1998 for a review) which belong to various CV-groups. The excesses in Table 1 presumably originate from hot dust in non-detached shells. Preliminary calibrations suggest effective temperatures in the 1850-2250 K range. Additional CS reddenings are thus more frequent in this low temperature group than they are in early CV-groups. It can easily be shown that the extinction law cannot be much different from the interstellar one we adopted. A non-selective (i.e. wavelength independent or neutral) contribution might also intervene however, which would remain undetected here. In those very cool atmospheres, CS grains are expected to condense close to the photospheric level (Salpeter 1977, Gail & Sedlmayr 1988), and the analyses of dust shells through Monte Carlo radiative transfer simulations lead to optical depths of the order of unity or larger (Lorenz-Martins & Lefèvre 1993, 1994).
Part of Miras and large amplitude semi-regular variables (SRa-type) actually show a range in CV-groups according to the phase of the data. The earlier group is derived close to maximum light or just after it, the later one being observed close to minimum light. This is just what is expected along a temperature (and opacity) sequence, if temperature variations are predominant. Opacity variations in the atmosphere and CS envelope seem predominant in other cool, large amplitude variables where the variation range of the effective temperature is probably less than 300-400 K. This latter feature was already observed in a few CV6 Miras such as C833 = R Lep and becomes the rule in the CV7-group. Miras are studied as such in Sect. 6. It should be noted that the SEDs of the CV1 to CV6-groups were derived in Paper I on the basis of non-Miras only. The CV7-SED shows a distinctive break in the CV-sequence, with e.g. The (presumably) fortuitous balancing probably responsible for the nearly constancy of the along the CV1-CV6 sequence (see Table 6) no longer works. A severe change in line of sights opacities possibly occurs between CV6 and CV7. Dust contributions close to the photosphere may play a role. The generated extinction decreases for an expanding dust shell, the grains being pushed away by radiation pressure. However, it maintains large in the extreme objects described in Sect. 6. The picture may then be confused by dust backwarming on the atmosphere. It arises the question of validity of the intrinsic SED we adopted for CV7 since it relies on few stars. We consider however that it is globally correct since:
© European Southern Observatory (ESO) 1999
Online publication: March 10, 1999