3. Discussion of the results
Before discussing the results we would like to say some words concerning the validity of the approach used. An operation of an absorption profile subtraction from an observed one seems to be allowed provided that the emission and absorption lines have independent origin. It is well known that the -line intensity of ALIVARS increases while the brightness of the star become lower (Kolotilov 1977). The observations appear to favour the hypothesis that the emission is formed in a region above the dust clouds, which in its turn may lay far away above the photosphere. Unlike chromospheric lines the emission observed is formed in envelopes with radius , which in Herbig Ae/Be stars is estimated to be ranging from
(Baschek et al. 1982; Garrison 1978). More recent kinematic evaluations of a dust cloud, distant from the star, gives a value near (Herbst et al. 1994). Therefore it is obvious that the process of the formation of emission and absorption lines cannot be comparable for both geometrical and physical conditions. Hence, we may state their independence. This allows the absorption and emission intensivities to be arithmatically added. Such a procedure is widely used for the calculation of the intensity of emission component in Be stars (Dachs et al. 1990; Köppen et al. 1982). Therefore, the values of obtained were plotted against the temperature . There seems to exist a positive regression relation between and in the temperature interval 7000 K 11000 K (Fig. 1).
However two stars, namely CO Ori and EZ Ori, are lying outside this relationship, showing much more intense emission than required for the corresponding temperature. Perhaps these stars were erroneously assigned as giants, whilst they might probably be dwarfs or subgiants at least. CO Ori and EZ Ori have some years ago been included in our observing programme to make a check as marginal stars, both being HAeBes (Herbig & Bell 1988) but never ALIVARS. Moreover, CO Ori is sometimes classified as an object of the T Tau group (Herbst et al. 1994). Both CO Ori and EZ Ori are probably main (or pre-main) sequence objects, whilst all ALIVARS are giants having an average group parameter of gravity lg g 3.0 (KP1). Thus, the relationship " - " may be employed as a sensitive tool for the classification of ALIVARS and T Tau-type stars to separate them in those cases when photometric data are insufficient.
Now we would like to discuss the fact that the luminosity-temperature relation is not a quite trivial one. One would not expect at all an existence of the relatioship a priori at least for two reasons. First, neither classic T Tau stars (Cohen & Kuhi 1979; Strom et al. 1989), nor weak emission stars in the Orion Belt (Kogure et al. 1992), nor HAeBes when considered as a whole, show such a dependence. It is obvious that the HAeBes, being non-homogenious in the sense of their structure, include stars of different types with diverse physical and environmental conditions. Second, the previous result of Pugach & Kovalchuk (1993) indicates that the Balmer emission decrements of ALIVARS do not agree with the supposition that the emitting atoms are excited by radiation. The decrements appear to favour the hypothesis that the electron collisions are the preferential mechanism to excite hydrogen atoms. Since the radiative mechanism does not play a dominant part in the formation process of emission lines one hardly can suppose by anticipation that the emission line intensities would correlate with the temperature of the star.
Theoretical calculations (Boyarchuk 1966; Brocklehurst 1971) have shown that the decrement for radiative mechanism cannot exceed a certain value, i.e. . The previous conclusion of Pugach & Kovalchuk (1993) regarding the prevalence of collisional mechanism over the radiative one was based on the fact that decrements for the three stars VX Cas, V586 Ori and SV Cep were
The new data presented in Table 1 seem to confirm the former conclusion. So far as the decrements for all but one ALIVARS range in the interval
one may insist upon collisional rather than radiative mechanism of excitation.
DD Ser having the smallest value of the decrement at the same time has the most interesting profile of the -line. The point is that the sharp absorption reversal laying over the -emission profile is very strong and that the core of the sharp absorption line is lower than the continuum level (see Fig. 2).
One would can interpret such a form of a line profile as being due to the presence of an additional mass of hydrogen in the line of sight (cloud, disk, shell?). However, no traces of a shell line are visible on the and profiles. Several other stars resembling DD Ser, RR Tau and BO Cep, have a sharp absorption line component which splits the top of the emission line, but less intensive. These observations indicate that a certain portion of the -radiation emitted by the envelope is extinguished by masses of neutral hydrogen. Thus our determinations of the luminosities , listed in the Table 1, give the lower level of the possible values. In some cases the presented values of slightly differ from those published in our previous articles because new and more exact luminosities L of the stars are now employed.
The " - " relationship found is reliable, provided that no deep light fadings of the stars have taken place during the observations. The probability for ALIVARS to be at a deep light minimum does not exceed 3-5%. Our visual estimates during the spectral observations show that the stars were at their normal brightness level excluding deep light minima. Moreover, it is known that the equivalent widths of VX Cas, UX Ori and WW Vul at deep light minima grow up to values of 28.5 , 36.0 and 42.8 (Kolotilov 1977). Our observations have registered no such large values. This fact unambiguously evidences that no deep light fadings occured at the times of our spectral observations and that our visual estimations agree with this. Therefore, we are certain that our observations do not coincide with deep light minima. However, it is beyond doubt that small amplitude light variations near the point of normal brightness do occure. They may well account for the scatter of the points in the " - " diagram. Unfortunately, we have no photometric data and cannot give account of the influence of small photometric variation on the equivalent widths observed.
Table 1 contains no data on the radial velocities of the emission envelopes, because in our case such data are utterly formal. First, considerable variations of widths and forms of the emission lines do take place. Fig. 3 shows the -emission component variation of V586 Ori.
Second, the presence and the displacement of the variable sharp absorption reversal may noticeably affect the position and the form of the emission profile, and would lead to incorrect values of the radial velocity of the envelope.
It is worth to note that there is an undecisive problem concerning the luminosity classes of ALIVARS. We used class III of luminosity for all stars of the group, assuming them to be giants. However, the values of lg g for some stars (V351 Ori, V346 Ori, RR Tau) indicate that the stars should be regarded as supergiants of luminosity class II and even class I. At the same time the distance modulus (m- - ) shows that the stars are not very luminous. A similar ambiguity exists for BF Ori and UX Ori. The first has narrow hydrogen lines (Shevchenko 1989) and the latter shows some UV absorption features of supergiants (Tjin A Djie et al. 1984). But both stars in fact can't be very luminous because they are not so distant.
© European Southern Observatory (ESO) 1997
Online publication: April 28, 1998