The importance of maintaining the greatest possible independence of spectral classification from theoretical or other external information is emphasized anew, with reference to some historical discussions now seen with the benefit of hindsight. This ideal requirement applies equally to the development and to the application of a classification system, although in practice some well-established information may guide one’s intuition in the initial hypothetical formulation. The fundamental position of this principle in the MK approach to classification is a major reason for the value of its spectral types, and for its continuing success in uncovering new phenomena. The ability of a particular technique to produce interesting or useful results is surely the most significant criterion of its value, and from this viewpoint it appears that new techniques and methods will complement rather than replace traditional spectral classification. Finally, the unique importance at this time of applying both new and traditional methods to spectral classification in the Magellanic Clouds is stressed; they provide the only current opportunity for detailed spectroscopic examination of numerous stars in external systems. It is essential that large telescopes be utilized for this work so that the best attainable observational quality may be maintained, and the many fascinating phenomena revealed by spectral classification in the Galaxy can be comparatively investigated to the maximum extent praticable in the Magellanic Clouds

As a classification technique, photoelectric narrow-band photometry is especially effective in the case of late-type spectra, in which molecular bands furnish the most sensitive criteria. Measurements of molecular bands with bandpasses of about 50 Å can be made very efficiently, and for normal stars they can be calibrated in terms of temperature and luminosity. In the case of normal late-type giants and supergiants, two-dimensional classifications can be obtained from measurements of TiO and CN; for very cool giants and for dwarfs it is useful siso to measure VO and CaH, respectively. All these molecules have bands in the red and near-infrared spectral regions, where cool stars are relatively bright and where photometric accuracy is highest.

High dispersion (2.4 Å mm−1) spectra of chemically peculiar late B and A stars reveal a wealth of information. In this paper we attempt to organize this material by introducing a classification scheme based on the second spectra of the lanthanide rare earths. Four categories are introduced: a) Lanthanide Rich, b) Nd-Sm Anomalous, c) Iron Group-Rich, Rare Earth Moderate, and d) Lanthanide Normal. The variability that is possible in different stellar spectra of Nd II and/or Sm II relative to Ce II, Eu II, and Gd II is a prominent feature of the data. Thus far, the only means of detecting these variations is to use high dispersion spectra. The relationship between our classification and low dispersion spectroscopic and photometric criteria is a problem for the future.

For MK classification of program field stars the “normal” observer is confronted with a number of practical problems, rendering classification on the true MK system very difficult. These problems concern completeness of standard-star grid, spectrogram resolution, comparison of originals-reproductions, spectrogram density levels, spectrogram widening and emulsion. A new atlas of slit spectrograms at 74 Å mm−1 is presented, giving good coverage especially for high-luminosity stars. MK classification from spectral intensity tracings is discussed. Advantages include easy distribution and sharing of standards, relaxation of resolution requirements and wide acceptable density range. Apparent shortcomings of the method seem easy to overcome. An atlas of spectral intensity tracings is presented. Some problems regarding objective-prism classification are discussed for higher as well as lower resolution. Quantitative classification methods are commented. Work in progress is presented on systems for two-dimensional classification of low-resolution objective-prism spectrograms of stars with spectral types earlier than G0.

Future advances in spectral classification are considered in terms of the likely effects of the following four factors: 1. observation from space; 2. new instrumentation developments; 3. modern data processing techniques; 4. the impact of new observations and astrophysical theory.

A morphology analysis as well as a statistical study of the spectra provided by the S2/68 experiment is made in order to establish ultraviolet criteria of classification. These criteria, either line features or intensity ratios, permit the establishment of a two-dimensional classification to be made from the so-called B0 to G0 type star in the classic range. This classification scheme permits also “abnormal” spectra to be singled out, thanks to the definition of the “normal” spectra.

In the last ten years Molecular Astrophysics has become an extremely interesting field of interdisciplinary research as a result of unexpected discoveries of complex organic molecules made by radioastronomy in many regions and objects of the Galaxy. Since not all molecules have detectable transitions through the atmospheric windows, the development of UV and IR space astronomy will allow us to study new important molecular transitions in the interstellar medium and in the atmospheres of cool stars and comets. In this paper the present status of molecular astrophysics will be described and special emphasis will be devoted to the importance of IR space astronomy for the future knowledge of molecular species in the interstellar medium and in the atmospheres of Carbon stars where the low temperatures are particularly, suitable for infrared studies

In my earlier work (Nandy, 1977) I have described an ultraviolet photometric system which involves a two dimensional stellar classification. This system is based on constructing broad band (Δλ ≈ 100 Å) ultraviolet colors (m2700 - V), (m2190 -m2740) and (m1500 - V) from the low dispersion S2/68 spectra. This photometric system is suited ideally to extend the classification to stars fainter than V = 6m.0, because the spectral features as described by the previous speaker, A. Cucchiaro, are not easily identifiable in these spectra.

The standard classification criteria, when applied to peculiar objects, such as emission line and shell stars, cannot easily be interpreted in terms of their intrinsic physical characteristics. Observations outside the optical spectral region may provide complementary information on their nature leading to a better spectral classification.

This has been a timely conference. Astronomical spectroscopic classification is facing crossroads that make it wise for us to look carefully at where we are now and which way we should go. The on-going explosion in detection techniques for fainter and fainter objects at all wavelengths and the availability of fast analytical procedures are opening up these new roads. It is a tribute to Fr. Patrick Treanor’s vision that we have gathered here to take stock of the current status and to consider the future of spectral classification and it was fitting that we did so while remembering the pioneering work of Fr. Angelo Secchi

Ce Colloque, organisé en l’honneur du Centenaire du Père Secchi, arrive a un bon moment pour faire le point sur le passe: les classifications de Secchi, de Harvard, le systeme Morgan et Keenan et examiner les problèmes d’avenir.

D’abord, quel est le but des classifications? Certains théoriciens ont l’impression que nous sommes dans la même position qu’un zoologiste classant des animaux. Mais cette impression est fausse: d’abord les animaux existent en espèces discontinues alors que les spectres stellaires forment un continuum.

Since the publication of the MKK atlas in 1943, there has been much progress in (1) the construction of fast spectrographs of higher resolution, C2) faster fine-grain emulsions and sensitizing techniques, (3) the partial or complete explanations of certain spectroscopic effects, [4) the discovery of other spectroscopic effects, and C5) refinements in the classification system based on new spectra and data on luminosities. An interim atlas entitled An Atlas of Low-Dispersion Grating Stellar Spectra was published in 1968 by Abt, Meinel, Morgan, and Tapscott but it is out of print and was not reissued in anticipation of the forthcoming greatly-improved atlas.

Equipped with a 15° F2 objective prism the (45/77/150)cm Schmidt telescope of the Copenhagen university Observatory, Brorfelde yields spectra of 102 A/mm at Hα to a limit of approximately mpg =9.5.

Two regions in the Large Magellanic Cloud (centered atα = 5h 11m, δ= -66° 08’ (34B), α = 5h 27m, δ= -72° 15’ (47B) have been searched for OB and supergiant stars. A catalogue of 312 stars is presented in which objective prism spectral types, positions (1975), approximate B magnitues and cross identifications are given. About half the stars catalogued in each area are OB stars, the remainder are supergiants. The region at δ=-72° has two to three times as many F and G-type supergiants as the region at δ= -66°.

A spectroscopic and photometric study of 75 normal main sequence B2-B6 stars has been carried out. From the spectra (12 Å/mm, 3700-4800 Å) are determined: equivalent widths of all measurable lines, line profiles of hydrogen lines and v sin i. The photometric work includes besides uvby and β photometry, also photoelectrically determined line strengths of Hα and He I 4471 Å. Some preliminary results concerning the relation between the MK types and the atmospheric fundamental parameters (Teff, log g) of the stars are briefly discussed.

The U. K. 1.2m Schmidt objective prism with its small apex angle (23’12”) and high ultraviolet transmission is a particularly powerful tool for the study of faint objects. It has been shown that objects of 20m and fainter can be reached on a 60 minute exposure of unwidened spectra on hypersensitized IIIa-J emulsion. A description of the prism and its performance has already been published (Nandy et al. 1977).

With the two parts of the Bonn Spectral Atlas it was hoped to provide a tool for the improvement of accuracy in low-dispersion spectral classification. Earlier estimates from our spectra by different observers with classification experience indicated a rather small increase in classification errors with decreasing resolution. They amount to 0.17 spectral types per 1000 Å mm-1 increase (larger numbers) in dispersion in the range 200 - 1300 Å mm-1 dispersion added to the 0.1 spectral types which are typical errors at MK classification dispersion near 100 Å mm-1 and 0.56 luminosity classes per 1000 Å mm-1 increase (larger numbers) in dispersion in the range 200 - 1300 Å mm-1 dispersion added to the 0.6 luminosity classes which are typical errors at MK classifications near 100 Å mm-1. In order to improve on the error determinations the three first-named authors, all graduate students with no previous experience in spectral classification, classified independently between 800 and 1300 spectra with each of the three Bonn Atlas dispersions of 240, 645 and 1280 Å mm-1. The classification errors of the three students differ by very small amounts and thus are averaged in the following discussion.

The recently developed Kodak emulsions IIIa-J and IIIa-F have been used as a new tool for deep-survey photometry. Among their most useful features are the following: (i) fine to very fine granularity; (ii) very high resolving power; (iii) high sensitivity over a large wavelength range in the optical part of the spectrum.

Some numerical results obtained from scanner data of a variety of normal star types, and from the spectral sensitivities of these Kodak plates combined with various optical filters are presented. We study the effects of these filters on the resulting synthetic colors with respect to the intrinsic potential of the colors to classify faint stars. In addition we briefly discuss the effects of interstellar reddening and the calibration of the photographic colors through photoelectric UBV photometry.

This synthetic approach to multicolor photometry is discussed in the more general context of the relations between spectral classification, stellar photometry, and model atmosphere calculations. This context is the natural basis for galactic astronomy.

Spectrophotometry data of 775 early type stars in a narrow band photometric system are analyzed, looking for the best parameters for a two dimensional spectral classification in the spectral region 3575 Å - 6000 Å. A correlation is established with the MK classification. The statistical technique of Principal Componant Analysis is applied to determine the number of linearly independent parameters among the data.

Temperature classes based on photometric measurements of the 7100 Å band of TiO are compared with published spectroscopic classifications by various investigators. Dwarfs, giants, and supergiants are treated separately. The photometric types are in excellent agreement with types given by P. C. Boeshaar for M dwarfs and by P. C. Keenan and Y. Yamashita for M giants. In all other cases the scatter is larger. The usefulness of the types given by various authors for dwarfs and giants is tested by examining their correlation with color temperature determined by narrow-band photometry.