5 Distribution of spectral types

It is instructive to compare the distribution of spectral types of the cool giants in symbiotic systems and that of giants in the solar neighbourhood. For this we restrict our sample to galactic objects, thus excluding the 13 Magellanic Cloud systems and Draco C-1. A discussion of the spectral types of the cool giants in the Magellanic Cloud systems in relation to galactic systems can be found in Mürset et al. (1996).

5.1 Normal M-giants versus miras and other types

Among the 165 galactic objects we count 129 M-type stars, including 27 d-type systems. There are 15 s-type and 5 d$^\prime$-type systems with a spectral type earlier than M. Further we have 3 systems with late type carbon stars (UV Aur, SS38, and AS210) and 2 systems with early type carbon stars (UKS-Ce1, S32). From these figures we conclude:

-

Spectral types: 80% of the objects contain a M or late C star, which indicates that symbiotic giants have a very late spectral type. A substantial fraction of the earlier types seems to belong to the halo population (e.g. Schmid & Nussbaumer 1993; Smith et al. 1996, 1997), implying that they constitute a distinct subclass of the symbiotics.

-

Variability: the relative frequency of mira type stars is much larger among symbiotic giants than among the giants of the solar neighbourhood. While only 3 out of the 122 giants in the HIPPARCOS sample (to be discussed in the next Section) are mira variables, 27 symbiotics (= 16%) are d-types which probably all contain a mira.

-

Chemical peculiarities: With 5 out of 167 the fraction of carbon rich symbiotic systems is comparable to the ratio between C and M giants in the solar neighbourhood. It is well known that the number ratio of C to M field giants is much higher in the Magellanic Clouds than in the Galaxy (e.g. Blanco et al. 1978). Correspondingly, a much larger C-star fraction among the symbiotic systems was obtained for the Magellanic Clouds by Mürset et al. (1996), who found 6 carbon-rich systems out of the 11 objects that could be classified. Thus, the frequency of carbon-rich symbiotic systems seems to reflect just the number ratio of C to M giants of the parent galaxy. Barium-star type abundance anomalies are measured or strongly suspected in the four systems S32, UKS-Ce1, AG Dra and BD-21.3873 (see Schmid 1994; Smith et al. 1996, 1997). There is no S-star known in a symbiotic system.

5.2 Distribution of M-subtypes

Figure 6 displays the distribution of M-type giants in 79 symbiotic systems based on our classifications given in the third column of Table 5. Classifications from the literature are not included in order to have a homogenous data set. Figure 6 emphasizes the differences between d- and s-types. The spectral class distribution of the d-types peaks around M6 about one subclass later than for the s-types. We found no red giant with a spectral type earlier than M4 in a dust-rich system. The differences in the spectral type distribution of s- and d-types reflects the basic property of the IR classification, which segregates the more evolved, high mass-loss giants in the d-type systems from the s-types which contain less extreme giants.

  

Figure 6: Frequency of the M-subtypes among the cool components of s-type symbiotic systems (full line) and d-types (dotted) according to Table 6The dashed line shows the distribution of M giants in the HIPPARCOS catalogue

5.2.1 M-subtypes in the solar neighbourhood

The distribution of spectral types of M-giants in the solar neighbourhood d<150 pc is derived from the HIPPARCOS catalogue (ESA 1997). A sample of M giants was selected according to the following HIPPARCOS parameters: large parallax ($\pi\gt 6.667$ mas), red colors (${V}-{I}\gt 1\hbox{$.\!\!^{\rm m}$}5$), spectral classification M, and large intrinsic brightness (${M}_{{H}p}={H}p+5\log(\pi[{\rm mas}])-10^{\rm m}<3^{\rm m}$). Hp is the mean magnitude in the broad band HIPPARCOS filter. This criterion separates the giants from the dwarfs. It also ensures a very high degree of sample completeness and helps to avoid that faint and probably more distant objects with large parallax errors enter the sample. According to these criteria we found 122 objects.

Before proceeding, we performed various tests to check for sample completeness. We found only one object, the mira variable R Cas (HIP118188), with a parallax of 9.4 mas, which failed to fulfill the above criteria. The reason is the inappropriate Hp-magnitude for this large amplitude variable with a magnitude range from $5\hbox{$.\!\!^{\rm m}$}1-9\hbox{$.\!\!^{\rm m}$}3$. The Hp-magnitude is a median magnitude of all HIPPARCOS photometric measurements. Due to unfortunate sampling of the R Cas light curve the mean value $Hp=8\hbox{$.\!\!^{\rm m}$}7$ is close to the minimum brightness and the corresponding absolute magnitude does not conform with the ${M}_{{H}p}<3^{\rm m}$ limit. In the following we include also this object in our sample. Further we excluded two objects, an M-supergiant ($\alpha$ Ori), and a faint M-giant without spectral subtype (M...) and large parallax error. Thus there remain 121 M-giants in the sample.

This sample includes 11 objects with variability flag 3 in the HIPPARCOS catalogue, which indicates large amplitude ($\Delta{H}p\gt\hbox{$.\!\!^{\rm m}$}6$) variables. 3 out of these 11 objects are in fact mira-variables, namely o Cet, R Leo and R Cas (see also van Leeuven et al. 1997). Most of the other objects in this variability group are classified as semi-regular variables.

The distribution of spectral subtypes among the selected M-giants is shown in Fig. 6. For classifications with half subclasses (e.g. M3/4 III), we have chosen alternatively the earlier or the later full subclass. Median spectral types are adopted for the two mira variables o Cet and R Leo, for which the catalogue gives a range in spectral type (e.g. ${\rm M}5-{\rm M}9 = {\rm M}7$).

5.2.2 Symbiotic giants vs. field giants

Figure 6 shows that the distribution of spectral types for the cool giants in symbiotic systems (present classifications only) differs strongly from the distribution of field giants in the solar neighbourhood. As already noticed by Allen (1980), there exists in symbiotic systems a very strong bias in favour of late spectral types ($\geq$ M5) when compared to the field giants. With our improved and more comprehensive classification we find an even stronger bias than Allen discovered. The number ratio between late ($\geq {\rm M5}$) and early ($< {\rm M5}$) type M giants is 1.7 for the giants in symbiotic systems but only 0.36 for the giants in the solar neighbourhood. Late M-stars and mira-type variables exhibit larger radii and much stronger mass loss than early M giants. Large radius and high mass loss for the cool component are possibly key ingredients for triggering symbiotic activity on a white dwarf companion. We suggest that this could explain the high frequency of late M giants and mira variables among symbiotic giants.