The Organizers of this Symposium asked that I make some introductory remarks on the general problems of extended atmospheres, the degree of common behaviour among stars having such a feature, and how the Wolf-Rayet stars may contribute to the progress of our inference in this area. In a Symposium entitled Wolf-Rayet and High Temperature Stars, not ‘Stars With Extended Atmospheres’, the implication of your charge to me is clear. You are implicitly assuming that the Wolf-Rayet phenomenon is primarily one of extended atmospheres. I can indeed give you a succinct summary of my outlook on these three points under such an implicit assumption, elaborating in more detail only to the extent as is necessary in order to be specific. And, as I am sure you had planned, because my outlook is considerably contrary to this implicit assumption, at least my remarks will spark controversy from the outset of this Symposium. I would like to emphasize that it is the summary of outlook to which I attach most weight and which I think is correct. The more specific details may change as we develop the mosaic of the summary, which has two main points. First, when you ask the physical picture of the general structure of a stellar atmosphere, you will eventually reach the conclusion that part of this general structure is an extended atmosphere for all stars. A focus upon some class, or classes, of them as exhibiting ‘extended atmospheres’ simply reflects the degree to which particular observations focus upon a particular part of the atmosphere as a function of the particular characteristics of the considered star. Second, in this context, I think that the implication that the Wolf-Rayet phenomenon is primarily one of extended atmospheres is incorrect. You are confusing a single system with a more deep-seated disease. My emphasis will lie on the disease as a whole, and the Wolf-Rayet stars as a guide to its diagnostics. I want to emphasize that this outlook on the general model of an atmosphere has developed jointly with K. B. Gebbie, J-C. Pecker, and F. Praderie; so most of what I say simply reflects this joint work.

In the three years since the Boulder Wolf-Rayet Conference, a great deal of observational material has been collected and processed. That conference was, of course, the stimulus for a great deal of the work, since it made clear where some of the large gaps in our observational knowledge lay. As a result the greatest progress has been in directions which were previously underexplored, and some subjects which were actively pursued before the last conference have been left almost untouched since. Thus, much of what was said in 1968 need not be reviewed again; I can confine my attention to topics which have been significantly changed by new observations. The new observations comprise, in particular, the systematic observation of spectra of many stars at moderately high dispersion, observation of far ultraviolet spectra, and discovery of some WR stars in M33 - the first outside of our own Galaxy and the Magellanic Clouds.

Very few observations are available. Of early rocket work on γ Vel (Stecher and Milligan, 1962; Smith, 1967) the most detailed spectra are by Stecher (1968) in the range λ 1200–3100 Å and by Carruthers (1968) in the range λ 1050–1216 Å. West's (1971) spectrum of γ Vel with the OAO-2 spectrometer (Code et al., 1970) in the range λ 1050–2000 Å appears to be somewhat different from Stecher's. Comparison must await full details.

We have learned from Lindsey Smith's talk how useful it would be for our understanding of Wolf-Rayet stars to have reliable effective temperatures and bolometric corrections. The ring nebulae surrounding a few WN stars provide one method for estimating the stellar temperatures. I have assumed each nebula is a normal H ii region excited primarily by the Lyman continuum radiation of the central star. Then, if the nebula is optically thick shortward of 912 Å, each stellar photon with more energy than 13.6 eV is absorbed by the nebula and re-radiated in a predictable way as hydrogen lines and free-free radio emission. Consequently, nebular radio measures such as those recently obtained by Hugh Johnson (1971) provide a direct indication of the number of Lymancontinuum photons emitted by the central star. The exact equations have been described in the Astrophysical Journal (Morton, 1969; Morton, 1970).

The Wolf-Rayet stars have perhaps the most spectacular spectra among the various celestial species that have been examined with the aid of the spectrograph. The wide emission lines that provide a striking display continue to be the enigma they have been for decades. Progress, however, in evaluating the contributions to the spectrum by different ions has been fairly complete, thanks to the splendid efforts in the laboratory by Edlen and his collaborators. Many lists of wavelengths of individual features exist from the studies of Plaskett, Beals, Cecilia Payne and Swings. Efforts at identification have been such as to provide a list of likely contributors, that by wavelength position and plausible intensity could be present in the emission band at a specified wavelength. The large width of the lines form the principal limitation. For, one can have a wide limit of coincidence in wavelength with resultant emission features that are complexes covering well over a hundred Ångstrőms. The wide nature of these complexes have perhaps encouraged in the past the use of low spectral resolution only, for the many studies that have been carried out. Seldom has one used the resolutions and dispersions that have been usefully employed for the study of the more common relatively narrow absorption lined objects. This situation pertaining to the Wolf-Rayet stars is happily undergoing a change in recent years when coudé spectra with the larger telescopes are becoming increasingly available. The southern hemisphere has been particularly rich in these objects both in variety of behaviour and in having many bright ones, and one can, therefore, justifiably hope that our information on these objects will progress henceforth with remarkable rapidity.

My intention here is to discuss the ‘high temperature’ portion of this symposium and call attention to those stars that are called Of. There are some similarities in spectral appearance to WR stars, e.g. emission lines. I should first like to define what I think are the essential differences among four groups of hot stars;

O stars: Stars that have only absorption lines in the visible spectrum. Type O is distinguished from type B by the presence of He ii 4541 at MK dispersion. It may be that some (supergiants) O stars will have emission lines in the rocket UV region but this description will be primarily concerned with ground based observations.

Of stars: These are O type stars that also have λλ 4634,40 N iii in emission above the continuum. In addition to normal O star absorption lines and N iii emission, they may also have other lines in emission. I will discuss this further below.

Oe stars: These are O type stars that have emission in the hydrogen lines (or at least at Hα), but with no emission in N iii or in other lines. I personally think that this small class of objects is related to the Be stars in their evolutionary status and in their emission mechanism.

WR stars: These stars are primarily characterized by emission lines. The only absorption lines seen are violet shifted (P Cyg type). Although in some cases emission lines appear which are similar to those found in some Of stars, the latter types always have some unshifted absorption lines present. Several Of stars have P Cyg profiles in some lines.

In taking up the subject of the relation between different types of early type stars one inevitably comes on ground set foot upon before by C. S. Beals. Already in 1940 he published an extensive paper entitled ‘On the Physical Characteristics of the Wolf-Rayet Stars and Their Relation to Other Objects of Early Type’ (Beals, 1940). The purpose of the present contribution is mainly to make the discussion more complete by including observations of later years.

In 1969 Aller and I collected information about the spectra of emission-line nuclei of planetary nebulae. (See also Aller, 1967, for classification of both emission and absorption spectra.) We classified them and compared them to the spectra of Population I stars. Our idea was to follow this general classification by detailed observation and comparison of stars in each of the classes. Most of the detailed studies are not complete, but already the picture has changed from the way Aller and I saw it.

Problems related to Wolf-Rayet stars were recently reviewed by Underhill (1968) and were discussed at length during the symposium held at the Joint Institute for Laboratory Astrophysics in Boulder (Gebbie and Thomas, 1968). Therefore, I shall not review here the historical development of ideas, and I shall not give detailed references in those cases when all the information is available in the sources mentioned above. I shall discuss first those observational data that are most important for the understanding of the evolutionary status of these stars. Later I shall discuss the problem of the interior structure and the origin of WR stars. And finally theories of the driving force for the observed mass loss will be presented.

From the time of their discovery in 1867, Wolf-Rayet stars have been objects of great interest for theoretical astrophysicists. Their spectra, dominated by extremely wide emission lines, set them apart from the run-of-the-mill stars. Moreover, the classic assumptions that have enabled accurate model atmospheres to be constructed for hot main sequence stars: hydrostatic equilibrium, radiative equilibrium, local thermodynamic equilibrium, and negligible curvature effects, are probably all violated in Wolf-Rayet stars. As early as 1894, Scheiner proposed that

Considerable effort has been expanded on both of these topics ever since the binary nature of most bright WR stars was revealed by radial velocity and eclipse observations and the basic correlation of line-width with ionization potential was discovered. I would like to describe briefly the methods of detection of binaries, to review the percentage of known binaries and to summarize the mass determinations that result from the study of WR binaries.

I wish to draw your attention to a (quite unexpected) property of the WC binaries. I believe that it gives us a clue to one of the parameters that controls the subclasses of the WR stars.

The Wolf-Rayet stars were the subject of a symposium held at the Joint Institute for Laboratory Astrophysics in Boulder, Colorado in June 1968. Here there was considerable discussion about whether one should talk about a class of objects called ‘Wolf-Rayet stars’ or whether it was more appropriate to talk about something called the ‘Wolf-Rayet phenomenon’ (Gebbie and Thomas, 1968). The point of view taken in this review is that it is more advantageous for obtaining an understanding of Wolf-Rayet objects to collate the available material under the broad category Wolf-Rayet phenomenon than to attempt to demonstrate that there is a homogeneous set of stars which can be called Wolf-Rayet stars and which occupy a significant stage in the evolution of stars, this stage being traversed routinely by all stars of a certain range of mass.