Before we are able to judge whether the and Hipparcos absolute magnitudes are consistent, a sample of apparent "normal" type stars has to be selected. There are two recent spectral surveys in the relevant domain (B - F): 1) the papers of Gray & Garrison (1987, 1989a, 1989b) and Garrison & Gray (1994) dealing with B6 to F4 type stars; 2) the extensive classification of 1700 A-type stars by Abt & Morrell (1995). From these sources we have extracted all "normal" type stars (exluding all kinds of peculiarities) with measured colors (taken from Mermilliod et al. 1997) and Hipparcos parallaxes. In total, 1147 stars were selected. The absolute magnitudes were derived as described in Sects. 3.1 and 3.2.
As a next step we have taken into account a possible influence of the relative Hipparcos parallax errors on the absolute magnitudes. For this purpose a statistical analysis of was performed. The results are shown in Figs. 2 and 3. We have divided each sample in two parts according to a 50% limit in the total number distribution of . The exact values are and for the CP2 and normal type stars, respectively. Both samples are very similar (beside an error tail of the latter) allowing one-to-one comparison.
Fig. 4 shows the overall result for both samples divided into three subgroups (early, intermediate and late) and two accuracies of the relative Hipparcos parallax errors.
It is immediately evident that the agreement for the "normal" type stars is acceptable although the scatter for the hottest stars is very large. This scatter might be introduced by the intrinsic band width of the main sequence itself (Paunzen 1999). No influence of the relative parallax errors on the derived absolute magnitudes can be detected.
The result for the peculiar stars is rather surprising. For the early type group we find that 85 % of all stars fall above the linear relation, on the average at a level of 0.5 magnitudes. This means that either the (Hip) magnitudes are too faint (= the absolute parallaxes are too small) or the () magnitudes are too bright. Adopting the photometric calibration we would therefore obtain distances which are 25 % too large. This situation is not influenced by the error of the parallaxes. It would be desirable to calibrate the absolute magnitude difference versus a suitable astrophysical parameter in order to derive correct distances for these stars.
This trend reverses as we go to the intermediate and late type groups. The most extreme deviant cases for the late type group even reach three magnitudes. While it is interesting to obtain a physical explanation for these remarkable differences, they do not bear on any substantial influence for investigations concerning the galactic distribution.
© European Southern Observatory (ESO) 2000
Online publication: March 21, 2000