The following topics will be discussed: a) A few historical comments; b) MK classification - the most important stage of classification work; c) Recently revealed peculiarity features and the problem of further differentiation of the classification scale; d) Classification work in the USSR; e) The role of classification results with respect to galactic structure studies; f) Low dispersion spectra and faint M-type stars and the missing mass problem; g) Extraterrestrial spectral observations: new promising means for research.

The classification of Of stars presents a number of problems, in part due to the small interval of the spectrum explored up to now. With J.M. Vreux an extended study has been made in the infrared region, up to 1.1μ. Be stars have been studied with L. Houziaux, The hydrogen Paschen lines and lines of helium I and II can be studied in the region and used to determine temperature class and luminosity.

Four methods for finding emission-line galaxies have been compared. Method (a) uses the ultraviolet excess, as found either by filter photography (Haro 1956) or by objective prism spectra (Markarian 1967). glanco (1974) introduced a thin prism with the CTIO Schmidt (1740 Å mm-1 at Hβ) which with IIIa-J plates [Method (b)] gave enough resolution for Smith (1975) and MacAlpine et al. (1977a, 1977b) to detect and classify galaxies by strong emission lines. Following a suggestion by McCarthy that even higher dispersion might be useful, I have used the CTIO Schmidt with [Method (c)] the 4° prism, a GGl+55 filter and IIIa-J emulsion and with [Method (d)] the 10° prism, an RG630 filter and IIIa-F emulsion. These latter give about 400 Å mm-1 at Hβ and Hα respectively which improves the visibility of emission lines against the galaxy continuum so that [0111] 5007 and 4959 and Hβ can be seen on the green plates and Hα and [SII] 6725 can be seen on the red plates.

The Michigan Curtis Schmidt telescope at Cerro Tololo Inter-American Observatory has been used with the “thin” prism (Blanco 1974) to survey the Large and Small Magellanic Clouds. Four plates, covering approximately 90 square degrees, and three plates, covering approximately 70 square degrees, were taken of the Large and Small Magellanic Clouds respectively. One hour exposures on nitrogen baked, IIIa-J plates were obtained. The limiting magnitude of the plates is V = 16 mag. At the dispersion of 1360 Å/mm OB stars can be recognized by their long uv extension, in which no sign of a Balmer discontinuity can be seen. The supergiants can be recognized by the appearance of a Balmer discontinuity and classified into temperature types by the strength of the hydrogen lines.

This paper reviews first of all the various problems connected with the establishment of a relationship between MK type and various photometric parameters: reddening, binarity, rotation. Mention is then briefly made of the various relationships proposed in the literature for the UBV, uvbyβ, DDO, Vilnius and Geneva systems. The MK classification is a bidimensional classification which provides also indication of peculiarity (Ap, Am, Ba stars, for example). The problems arising from stars having a peculiar chemical composition are examined. The establishment of a relationship between a color index and the MK type or the location of the various MK types in a photometric diagram are both classical methods to obtain a correlation between a photometric system and the MK schema. However a new solution is proposed by the method of stellar boxes, which allows us to find all the stars having the same colors. If many photometric systems are limited in their possibilities of classifying all stars from O to M, the GENVIL (or VILGEN) system, formed with four filters of the Geneva system and three of the Vilnius system and at present under consideration, will certainly provide a very good means to this end. Finally, the new spectroscopic and photometric catalogues available at the Stellar Data Center in machine-readable form offer us the possibility of improving the present correlations.

We have started at the OHP observatory an observing program for classifying stars in Golay’s boxes of photometrically identical stars. So far 380 stars have been observed at 80 Å/mm dispersion, covering the range from BO to G8. The main result is that stars belonging to the same box are spectroscopically identical at this dispersion if one considers classifications taken from different well experienced observers. This remains true within the current errors of spectral classification. The problem presented by the Ap and Am stars is discussed in detail.

Intermediate band multicolor photometry now is one of the principal methods to obtain information on stellar temperatures, luminosities, metallicities, peculiarities and interstellar reddenings. The method has some advantages over the method of stellar classification by visual inspection of low dispersion spectral images:

Sophisticated algorithms of modern statistical analysis are not yet widely applied to astronomical data. For the moment, these methods are mainly used by scientists in fields where the physical basis is less well determined than in astronomy. The algorithms could however be quite useful in new fields of our science where the physics is still being investigated. In more classical applications we have found them quite useful in the study of the relationship between photometric and spectroscopic data.

Hauck and Mermilliod (1978) have compiled a new catalogue of approximately 20,000 observations in the Strömgren four-color system. Using the facilities of the Stellar Data Center in Strasbourg, Egret and I have revised the computer program that dereddens the indices to take into account recent revisions made by Crawford (1978) in the routines for handling B stars. Then, following the method of Golay (1978), we have constructed stellar boxes in the four-color system.

Modern digital spectrophotometric devices allow for a much more rapid gathering of spectroscopic data on stars and galaxies. Colors and line strengths can lead to new classification schemes, and/or physical interpretations of much fainter objects than previously possible. However, the same old conceptual problems of interpretation (e.g. in abundances of G and K stars) remain a serious difficulty. Illustrations of these accomplishments and problems are given.

A spectrophotometer employing linear arrays of Reticon photodiodes has been used at Asiago observatory since August 1977. The first version, with a 128 - element array, has already been described in detail (Barbieri et al. 1978). The final configuration has 512 photodiodes, each 50.8 μm high, so that 2-5 cm of spectrum can be recorded in a single exposure. The spectrophotometer is mounted at the focal plane of the Boiler & Chivens spectrograph of the 182 cm telescope; with the 600 groves/mm grating the diode width corresponds to 5.9 Å. Two other gratings are available, respectively with 400 and 7200 grooves/mm. A fairly sophisticated computer system permits a remarkable flexibility of operation, with real time display on the CRT of the acquired data and recording on mag. tape. To subtract the thermal current and the sky, the star is taken out of the slit by tilting a transparent plate in front of the spectrograph, always under computer control; the astronomer can choose different star/sky exposure times so as to optimize the observing procedure. Guiding can be performed from the control cabinet with an integrating SEC-TV system.

The need for establishing classification criteria at long wavelengths is stressed. The usefulness of doing this is illustrated with a discussion of the composite spectra of FU Orionis stars. Spectra of these pre-main-sequence stars from 1.5-2.5μ were obtained with a Fourier Transform Spectrometer. Luminosity criteria in the l-2μ range are also discussed with application to M stars.

Two separate Balmer jumps are observed in some Be stars. The first one, in absorption and unchanging, is attributed to the central star and gives its spectral type and luminosity class in the λ,D system. The second one, occuring at shorter wavelengths, either in absorption or in emission, originates in layers at a low pressure. It is shown that these layers modify the colors of the central star on both sides of the Balmer jump and that they are not optically thin even in the continuum.

A thin prism (with a dispersion of 2480 Å/mm at Hγ and Hβ) is now available for the UK 1.2m Schmidt telescope. Å 60-min. unwidened exposure on a Ilia-J (hypersensitized) emulsion reaches fainter than 20 mag. The spectra are measured and digitised with the fast measuring machine COSMOS. A method to determine the red shift of faint galaxies and the spectral classes of faint stars and galaxies from the digitised spectra will be described.

A slitless field spectrograph for Cassegrain telescopes has been designed. It is based on the focal reducer optical arrangement and grating prisms. Together with the “multiple diaphragm method” not only a gain in limiting magnitude of 3 to 4 magnitudes has been achieved, but also other drawbacks of the classical objective prism camera are avoided.

A survey is given of the present state of digitization and automation of spectral classification from low and intermediate dispersion objective prism plates and an outlook to automatic classification work by a photoelectric multi-element detector scanning system.

A computer program for the automatic classification of stellar spectra is described. The aim of this program is to derive accurate absolute luminosities and spectral classes of individual stars for improved spectral parallaxes. Preliminary results are presented and future applications are discussed.

A scheme of automatic MK spectral classification is presented. It is based upon standard spectrograms obtained for visual spectral classification. Computer results are given for all the MK standard stars earlier than Bl. They are well correlated with the MK system. However, the results might be improved by using a rapid-scanning Michelson Fourier Transform Spectrophotometer to obtain the stellar spectra. The analysis of the interferograms should also be made by a computer.

We study the conditions under which a two-dimensional spectral classification according to Teff and ne is possible and when 3- or It-dimensional spectral classification is necessary because the classification criteria are unavoidably influenced by the metal and helium abundances. For late type stars, where molecular absorption becomes important for the continuum a seven-or more- dimensional classification may be necessary. At this point spectral classification may not be useful any more. We find generally that a derivation of the gravitational acceleration from the spectroscopically determined electron density is only possible if the helium abundance is known.

It is stressed that the ultimate aim of spectral classification is stellar classification. MKK classification is briefly described and possible additional parameters or dimensions discussed. Considering various new developments, a new approach to classification which is highly computer dependent, is suggested.