4. Analysis of MK spectral classes
MK spectral classes were not discussed in the previous section and should be treated separately. We have seen from Fig. 5 of Sect. 2 (where the spectral type index was plotted against spectral classes), that this index is connected very tightly with spectral classes (except for supergiants). But the relationship for dwarfs is non-linear. This relationship reminds us of the solid line in Fig. 2 of Sect. 2 (the MK calibration "effective temperatures against spectral classes" for dwarfs). Therefore we may guess that the dependence "effective temperatures against " for dwarfs should be very simple (almost linear). Fig. 10 confirms our expectations. Appling the method of stepwise linear regression (see the previous section) to these data we found the following very simple analytical dependences ( demonstrated in Fig. 10):
for 17 stars of luminosity class V (R = 0.988) and
for 10 stars of luminosity class I (R = 0.988), the rms differences (published minus calculated values) being equal to about 100 K and 240 K, respectively.
We may guess from Fig. 10 that giants (triangles) should follow to same relationship as supergiants (circles) if , but the same relationship as dwarfs (crosses) if . Subgiants (luminosity cass IV) and bright giants (luminosity class II) are absent in this Fig. because there are no calibration lines for these classes in MK calibration data by Schmidt-Kaler (1982). Any interpolation with available calibrations seems to be very uncertain (up to 600 K) and a new calibration for the intermediate luminosity classes is very desirable (as well as improved calibrations for other luminosity classes).
Since the calibration of MK spectral types against effective temperatures and absolute magnitudes set up by Schmidt-Kaler (1982), many new data became available (or will be available in the nearest future). As to MK spectral classes, the most important new source is the catalogue of MK spectral types for 963 standard stars compiled by Garcia (1989). The more recent Perkins catalogue of revised MK types for the 1054 G, K, M standard stars by Keenan and McNeil (1989) should be added. New refined MK classifications have been published for about 200 earlier type (A and F) stars by Garrison (1989) and Gray and Garrison (1989), authors who work very closely to the Morgan and Keenan's system.
To calibrate MK types against abslolute magnitudes the Hipparcos catalogue of very accurate trigonometric parallaxes is undoubtedly an invaluable source. For distant supergiants with uncertain trigonometric parallaxes, information about their membership in clusters may be useful in estimating absolute magnitudes, and this information may be derived from another invaluable source - the Simbad database operated at Strasbourg, France, which containes references for a huge amount of stars.
Our spectral indices may be used in future MK calibration to smooth and interpolate the MK data. Application of indices from non-photographic spectral wavelength ranges may help to recognize binaries and peculiar stars. Future MK calibration will help in improving spectral classification techniques and will be used in many other fields of astronomy.
© European Southern Observatory (ESO) 1997
Online publication: April 28, 1998