1 Introduction  

Symbiotic stars are closely related to the cataclysmic variables, in the sense that a white dwarf's activity is triggered by mass transfer from a cool companion star. But the cool star in symbiotic systems is a giant rather than a dwarf as in the case of cataclysmic systems. The orbits of symbiotic systems are correspondingly wide (P>200 d). The nature of the interactions and of the activity of the white dwarf are still a matter of debate. In any case the fundamental characteristics of the red giant donor are certainly an important factor. The spectral type of red giant branch or asymptotic giant branch stars is strongly correlated with stellar radius and probably also with mass loss. Both, radius and mass loss are important parameters in interacting binaries. Thus, a significant sample of accurate spectral classifications may clarify some aspects of the interaction processes in symbiotic stars.

Spectral classifications of cool giants of symbiotic systems are widely scattered in the literature. A compilation of early classifications is given in Allen (1982). However, these spectral types are based on various classification criteria and their quality is often rather crude. Sophisticated and devoted classification work has been published by Kenyon & Fernández-Castro (1987) and Schulte-Ladbeck (1988). These papers are, however, restricted to about three dozen bright objects. Recent classifications can be found in the extended lists of Medina Tanco & Steiner (1995), Harries & Howarth (1996), and Mikoajewska et al. (1997). These studies employed, however, spectra with restricted resolution or spectral coverage which may cause some larger uncertainties in the resulting spectral types.

Determining the spectral types of cool giants in symbiotic systems is complicated by the composite nature of the spectra. Often the blue spectral region, according to which the classification of "normal'' stars is traditionally done, is strongly contaminated by light from the circumstellar nebula and/or from the hot companion. Therefore, the spectral classification of cool giants in symbiotic systems should be done in the near IR region, which offers a number of decisive advantages:

-

In the near IR the cool star is usually the dominant spectral component while the contributions from the nebula and from the hot star decrease towards longer wavelengths.

-

The near IR is full of telltale molecular bands formed in the atmosphere of the cool giant. These bands are temperature sensitive and well suited for spectral classification.

-

In the near IR the red giants are bright enough to be observed with relatively short exposure times.

After describing the spectroscopic data in Sect. 2, we derive in Sect. 3 the spectral types for the cool giants in about 100 symbiotic systems. We compare our results with previous studies and discuss the accuracy of our classification. Supplementary spectral types are taken from the literature and are included in a catalogue (Sect. 4, Table 5) which contains spectral types for the cool giants in about 170 systems (about 95% of all known symbiotic systems). In Sects. 5 and 6 we discuss properties of the distribution of the spectral types of the cool giants in symbiotic binaries and correlations between the spectral types and other system parameters. Considerations on the evolution of symbiotic binaries are presented in Sect. 7. Section 8 summarizes our results and conclusions.