Only ten years ago a symposium on the special class of stars we now call Luminous Blue Variables (LBV’s) would not have been possible. At that time we were only just beginning to recognize the similarities among the stars we call S Doradus variables, P Cygni stars, and Hubble-Sandage variables, and their importance in the evolution of the most massive stars.

About a decade ago it became clear that the observed H-R diagram has an upper luminosity limit and that η Car, P Cyg, and S Dor are all near the temperature-dependent boundary for hot stars. With hindsight the luminosity boundary can be seen in an H-R diagram for hot stars published by Hutchings (1976), but the full temperature range is necessary to show that it really exists. When more data became available in 1977—1978, including the evolved and cool supergiants, the upper luminosity limit became obvious; and we proposed that an instability causing rapid and unsteady mass loss (i.e., LBV’s) is the basic reason for this boundary (Humphreys and Davidson 1979). We emphasized the temperature-dependence of the boundary for the most luminous hot stars, the lack of cooler counterparts at similar luminosities, the nearly temperature-independent limit to the luminosities of cool hypergiants, and the critical role that objects like η Car, P Cyg, and S Dor may play in the evolution of the most massive stars.

Various evolutionary sequences leading to LBV are examined. The sequence O-Of-LBV-WR-SN is well supported by the models; some LBV with relatively lower luminosity may turn into OH/IR sources. The overall duration of the LBV phase depends mainly on the average mass loss rate; for < Ṁ > = 10−3M⊙y−1, it lasts about 104 y.

Very massive stars undergo, when they reach log Teff = 3.9, strong departure from hydrostatic equilibrium due to supra-Eddington luminosities at some depth in the outer layers. This results in heavy mass loss, as the growth rate of the instability is very fast. We suggest that the amount of mass ejected in a shell episode is mainly determined by the mass of such a layer that its thermal adjustment timescale is within an order of magnitude of the stellar dynamical timescale. Simulations of B-light curves due to shell ejections by LBV are performed and some sensitive properties are identified.

The characteristics of the Ofpe/WN9 class in the LMC, including two newly discovered members, are reviewed; a strong relationship between this class and the LBVs is established by HDE 269858 = Radcliffe 127, currently in outburst with a spectrum identical to that of S Doradus. The evidence for CNO-cycle products, i.e., enhanced N and He, on the surfaces and in the circumstellar nebulae of Ofpe/WN9 objects and LBVs is discussed, with emphasis on recent near-IR observations of He/H. Finally, the occurrence of several Ofpe/WN9 and LBV stars in compact multiple systems will enable valuable information about the evolutionary status of the peculiar objects to be derived. Strong threads unite these three topics: the same objects frequently recur in each.

If Luminous Blue Variables (LBVs) are not each unique types, three broad groups can be characterized depending on the luminosity and location of LBVs in the Hertzsprung-Russell diagram. To assist in defining the evolutionary nature of LBVs, connections are made to stars that have similar spectral character with the suggestion that some of these objects that may someday become LBVs.

The study of the context of LBV stars, that is close neighbourhood (companions) and spatial relationship to young objects such as cepheids. stellar associations, red supergiants. enriched OB supergiants. Hll regions. WR stars can provide useful clues on their true luminosity, their age thus eventually initial mass and evolutionary status.

It is, perhaps, tautological to say that the luminous blue variables (LBVs), being luminous and blue, should be ideal targets for space ultraviolet observation. In fact, many have been observed on occasion and these will be discussed in this article. Mitigating circumstances, however, conspire to make this study difficult. The LBVs are rare and tend to be apparently faint, and since they are massive and formed in the plane in our Galaxy and in active star-forming regions in other galaxies, they tend to be heavily reddened. The stars were not really studied at all in the ultraviolet until the launch of the International Ultraviolet Explorer Satellite (IUE), with the notable exceptions of photometric observations of several of the galactic stars with ANS and TD-1 and Copernicus observations of P Cygni (Hutchings et al. 1987). IUE, a small aperture (45 cm) spectrographic satellite, covering the spectral range from 1200 to 3300 A, allows for an adequate characterization of the UV properties of these stars.

This paper reviews the major new facts on microvariability of LBV’s of the last few years, and discusses some observational requirements for the future.

The basic physics of model atmospheres for Luminous Blue Stars (LBS) are discussed. The discussion comprises the model description of both subsonic photospheres and supersonic winds and the application of model calculations for the purpose of quantitative spectroscopy.

We present new determinations of the H-R diagrams for luminous, hot stars in the LMC and the SMC. Using all available photometric and spectroscopic data, we discuss the conversion of the observed measurements to temperature and luminosity. The resulting H-R diagrams for normal stars confirm the upper luminosity limit, and also show an interesting “ridge” not noted before.

The main characteristics of the LBVs with “normal” S Dor-type outbursts are reviewed. At quiescence they are luminous (MBol ≈ −9 to −10) OB supergiants (14 000 K ≲ Teff ≲ 30 000 K) occupying an inclined instability strip in the HRD. The most luminous S Dor variables are the hottest ones. During outburst dense envelopes are expelled of equivalent spectral type middle A to early F. S Dor variables occupy at maximum a vertical strip in the HRD at T≈8000 K. The observed amplitudes in the visual range increase from 1 mag (e.g. R71; Teff = 14500 K at quiescence) to 2.5 mag (e.g. R127; Teff = 33000 K at quiescence). The existence of an amplitude-luminosity relation is suggested which can be used to derive extragalactic distances.

Eta Carinae is at once our most notable LBV but also a doubtful LBV, because its Great Eruption observed 150 years ago was so unique. Since that event exceeded other major LBV outbursts by orders of magnitude in ejected mass and energy, we are not certain that it really was an example of S Dor, P Cyg, R127-type behavior. But even if it turns out to be a freak, η Car has provided either the original example or one of the earliest hints for several facets of the LBV phenomenon, e.g.:

1. -- Visual-wavelength brightening with little change in luminosity during each major mass-loss event;

2. -- The idea of a modified Eddington limit causing instability in a critical range of surface temperature;

3. -- CNO-processed surface material;

4. -- Bipolar structure in the ejecta.

Some LBV’s undergo brightness variations in the visual of about 1–2 magnitudes on time-scales of decades. These outbursts occur at approximately constant bolometric luminosity with the stellar radius and the temperature changing correspondingly. This subgroup of the LBV’s is generally called Hubble-Sandage variables or S Doradus stars and will be the focus of this talk. We present self-consistent radiation-hydrodynamic calculations to model the observations from minimum to maximum states.

B[e] supergiants show evidence for a non-spherical two-component stellar wind. The general appearance and the physical properties of the suggested disk-like configuration are discussed. The high mass-loss rates, the surprisingly large number and the location in the H-R diagram make these stars important for the understanding of the post-main-sequence evolution of massive stars.

It has been suggested by Zickgraf et al. (1986) that the outer atmosphere of some B[e] stars has a two-component structure: a fast, radiation-driven wind from the pole, and a dense, slow outflow from the equator. This is also die basic picture investigated by Poe, Friend, and Cassinelli (1989) (hereafter PFC) to explain the momentum problem associated with Wolf-Rayet stars. Here we apply the PFC multiforce wind theory to model the B[e] outflow phenomenon.

A number of yellow and red super- and hypergiants show phenomena that are similar to those shown by Luminous Blue Variables. The LBV phenomenon may not be restricted to the blue part of the Hertzsprung-Russell diagram and the conventional name ‘S Dor variables’ seems more appropriate.

One of the most luminous stars in our Galaxy (Humphreys 1988) HR 8752 = HD 217476 (G0 Ia), is similar in some respects to Luminous Blue Variables. Studies by several authors (Sargent 1965; Smoliński 1971; Smoliński et al. 1977; Stickland and Harmer 1978; Lambert and Luck 1978; Lambert et al. 1981; Smoliński et al. 1986; Sheffer and Lambert 1987) have contributed to the understanding of activities in the atmosphere of this star, but its nature is still not well understood.

Luminous Blue Variables (LBV’s) are losing mass at a rate which is higher than in normal stars of the same luminosity. This high mass loss is evident from the occurrence of P Cygni profiles in the visual spectrum, the large numbers of UV lines which are Doppler shifted or show P Cygni profiles and the large IR excess or radio free-free emission. Mass loss from LBV’s is strongly variable on a wide range of timescales from months to centuries and possibly even millenia. During these variations the mass loss may vary from values as low as 10-6 to 10-5 M⊙/yr, when the star is quiet, to outbursts of the type observed in P Cygni in AD 1600 and n Car in 1837 (for reviews see Davidson, 1987; Lamers, 1986, 1987).