For extension of the present spectral classification to faint stars it is desirable to know not only their UBV values but, for comparison purposes, also have any reliable spectral estimates for some of these stars. The last requirement however involves a fair number of difficulties considering the magnitude range 17.0 - 22.0discussed. This problem can be tentatively solved in the case when interstellar light absorption in a direction towards an open cluster is negligibly small. In this case the observed colours of stars will coincide with their intrinsic colours, whence it is easy to yield spectral estimates by means of Schmidt-Kaler (1982) transition tables and then use these stars as spectral standards.
The open cluster NGC 2264 in a direction of which there is practically no dust matter (Arshutkin et al. 1990) was chosen as a test object of research. UBV values of NGC 2264 stars in the Vmagnitude range 17.0 - 22.0 were derived by Adams et al. (1983) with the U.S.A. Kitt Pick National observatory 4 m telescope, and used here for spectral classification. These data are given in the first four columns in Table 9 (the star numbers, V values and colour indexes). The fifth column contains spectra SpUBV obtained with use of the present spectral classification technique. In sixth column "standard'' spectra Sp calculated at E(B-V), assumed to be equal to 0.08 for each star and used later for comparison with results of the spectral classification, are given. The seventh column contains intrinsic colour excesses of stars E(B-V)UBV calculated on the basis of Schmidt-Kaler (1982) transition tables via spectra SpUBV that are given in Col. 5.
When computing intrinsic colours of open cluster members and
|( B-V )0 = ( B-V ) - E(B-V)||(2)|
is this value error caused by
errors of colour
X value approximation
Because the value of X for the majority of stars according to Strayzis (1977) is very close to one, the third member in the formula (4) may be safely omitted. Thus the value practically depends only on errors of colour parameters (U-B) and (B-V)measurement, which are given in Cols. 6 and 10 of Tables 13-18 presented by Adams et al. (1983). These tables contain data on approximately 200 stars which have both (U-B) and (B-V) values. About a half of these stars which satisfy the condition (3) have been involved to the spectral classification by the present technique. Other stars with , which according to Adams et al. (1983) is caused by strong ultraviolet colour excess of stars, were not considered in this study.
As an example, we shall carry out spectral classification of the first star in Table 9 of the present study. It has a visual magnitude of V=20.0 and colour parameters (U-B) =1.18, (B-V) =1.33 (Adams et al. 1983). Via the formula (1) with average E(U-B) / E(B-V) =0.85 we derive QUBV = 0.05 as the first approximation. The value identified of QUBVcorresponds to the following set of spectral estimates: Sp1 = K5 V, Sp2 = A2 V, Sp3 = A2 III, Sp4 = M4 III, Sp5 = F2 I, and Sp6 = K5 I. Spectra Sp4, Sp5, and Sp6 should be excluded from consideration in as far as they lead to such estimates of distances which move a star out the Galaxy. Other estimates Sp1 = K5 V, Sp2 = A2 V and Sp3 = A2 III correspond to distances 2.6 kpc, 8.3 kpc, and 12.6 kpc. The assumption of the star spectral type being A2 V or A2 III moves it almost to the limits of the Galaxy (distances 8.3 kpc and 12.6 kpc respectively). And if, in view of errors of distance determination towards a star , it does remain inward of the Galaxy, simple calculations show that it has undoubtedly moved far beyond the open cluster NGC 2264 and T Tau stellar grouping. This, however, contradicts observational evidence (Arshutkin et al. 1990; Adams et al. 1983) according to which the star number 302 as well as all other stars in Table 9 belong to the open cluster NGC 2264 or T Tau star grouping, located at distances about 1 kpc from the Sun.
On the other hand, Williams & Cremin (1969) specify existence of the dark dust nebula which completely cut off the light of background stars located beyond the open cluster. An opportunity for observation of stars, in particular on distances 8.3 kpc and 12.5 kpc computed for spectra Sp2 = A2 V and Sp3 = A2 III, is thus excluded. As an exception a few O-B2 high luminosity stars of extragalactic origin have been discovered in the process of the spectral classification, which is discussed later in brief.
So, spectral estimates should be excluded from the considered set of probable spectra . As a result there remains only one acceptable spectrum Sp1= K5 V. Then, using an exact value of E(U-B) / E(B-V) (Strayzis 1977) for Sp1 = K5 V, we yield a more accurate value QUBV=-0.10. With this value, via the table given by Strayzis (1977), two corresponding estimates of spectra K3 V and K7 V are obtained. Simple averaging gives K5 as a final spectrum of the star SpUBV and a new value of colour excess E(B-V)UBV =0.18 mag. Allowing for random error, the last value of E(B-V), unlike other alternative values corresponding to spectra , fits the absorption curve (Arshutkin et al. 1990; Cohen & Kuhl 1979; Walker 1956; Williams & Cremin 1969) fairly well.
Spectral classification of other faint stars was carried out in a similar way and results are given in Table 9. Colour excesses of the stars vary from 0.00 to 0.80 mag, and their spectra are all in the interval K0-M5. This agrees well with results of previous studies (Arshutkin et al. 1990; Cohen & Kuhl 1979), taking into consideration photometric and spectral classification errors.
The comparison of the present classification results SpUBV (Col. 5) with "standard'' spectra Sp (Col. 6) shows a good agreement. The differences for more than half of the stars do not exceed three subclasses. It should be noted that the accuracy of photometric observations of 17.0-22.0 mag stars is much worse than that for stars brighter than 14.0 mag, the spectral classification of which was discussed in Sect. 3. To some extent, deviations which exceed three subclasses are possibly explained by this circumstance. The deviations, however, have systematic character and may be caused by the initially accepted assumption that E(B-V) = 0.08 for each star, which may be incorrect. Actually, the real colour excesses of stars often exceed this value, and according to Cohen & Kuhl (1979) reach 0.70 mag, which agrees well with results of the present work (Col. 7 of Table 9).
Arshutkin et al. (1990), and Walker (1956) have shown that inside the open cluster NGC 2264 and at shorter distances there is practically no dust substance, and the average colour excess is here 0.08. Thus, stars with colour excesses are located a little further than the open cluster, which confirms the conclusion made by Arshutkin et al. (1990) about the existence of T Tau star grouping on 1 kpc distance from the Sun.
A study on spectral classification of NGC 2244 and NGC 2264 faint stars with the use of UBV photometric data allowed us to draw a very important conclusion. It was found that extension of spectral studies into the region of faint stars leads to the increase of percentage of unambiguous estimates of spectra. For the V=17-22 mag stars spectral classification in a direction of NGC 2264 there is practically no necessity to employ any additional astrometric criteria. Otherwise, the difficulties in obtaining reliable observational material would make up a serious problem in proper motions determination of faint stars.
The list of stars in Table 9 could be supplemented with a few of O-B2 objects whose spectra are unequivocally determined with the Q- method. Estimates of distances show that these objects are far out the Galaxy. They are observable due to their high luminosity and existence of transparency windows (Arshutkin et al. 1990). It is quite obvious that they were listed by Adams et al. (1983) as NGC 2264 members as a consequence of selection criteria used by him. These criteria probably give equivalent results when being applied both to NGC 2264 stars, T Tau star grouping, and to objects of extragalactic origin. Many of O-B2 objects have large colour excesses E(B-V) > 2.0, which exceed the threshold value of the Q-method applicability (Johnson 1958; Johnson & Morgan 1953). Because these objects do not belong to NGC 2264 or to T Tau star grouping, we did not include them in Table 9.
Copyright The European Southern Observatory (ESO)