The properties of circumstellar shells around LBV’s are reviewed. Some LBV’s are surrounded by very extended nebulosities, similar to classical planetary nebulae, with AG Car as the prototype. The masses of these nebulae are of the order of a solar mass and their kinematic ages are about 104 years. The origin of these nebula is still controversial, but they probably have been ejected in some major outburst in the past. It is then remarkable that the historical outburst of P Cyg has left no bright remnant. There is evidence for CNO processed matter in the ejecta. LBV’s with no detectable extended emission also commonly have nebular emission lines. Some of these nebulae are probably related to the resolved nebulae, but some certainly are very much smaller and have much higher densities.

We present the first high resolution observations of the ring nebula around the galactic LBV AG Car obtained through broad band B, V, R, I filters with the STScI coronograph on the ESO 2.2 m. telescope. Highly anisotropic continuum emission in all these bands is clearly detected, in the shape of a bipolar structure elongated along the NE-SW direction extending from a few arcsecs from the star out to ⋍ 15 arcsecs corresponding to a projected linear distance of ⋍ 0.5 parsecs, with a surface brightness in the range 17.9-19.4 mag/arcsec2. Its essentially neutral color with respect to the star Δ(V-I) = −0.07) suggests that the observed radiation is starlight scattered by dust grains of diameter > 1μ.

Recent infrared spectroscopy and imaging of LBVs lead to the following results. CO overtone emission at 2.3 μm has been found in 13 LBVs in the Galaxy and the LMC. This emission is collisionally excited in warm (3000 – 5000 K), dense (NH > 1010 cm-3) circumstellar material. Circumstellar disks offer favorable conditions for the formation and excitation of CO molecules and are very likely the location of the observed emission. It is proposed that the LBVs showing 2.3 μm CO overtone emission possess the highest density circumstellar disks.

A group of eight LBVs has been identified in the LMC with He I 2.058 μm emission stronger than H I Brγ. This group includes the CO emission star HD 37836. Helium is over-abundant in these stars with N(He)/N(H) ranging from 0.2 to >0.5. Remarkably five of the helium strong stars belong to the small class of Ofpe/WN9 stars and a further two are probably related to this class.

Slit scans of the galactic LBV AG Car have resolved the far-infrared emission from this star, clearly showing it to originate from cool dust in the circumstellar ring structure. Thermal equilibrium considerations require large grains in the ring in order to match the measured grain temperature and radial distance. Similar slit scans of the galactic B[e] star HD 87643 fail to resolve the far-infrared emission from this star at the 10″ level.

A mechanism is proposed for the formation of shells around P Cygni. These shells seem to arise at approximately two monthly intervals and each appears to undergo a steady acceleration of some 0.6 cm s-2 during the six months that it is present. In the process to be described the formation of the shells takes place in a stellar wind that moves outwards ballistically from the photosphere. Small fluctuations in wind speed are introduced there and eventually lead to the condensation of shells at a radial distance of the order of 1014 cm. The density in the shells is large enough to account for their acceleration by the Lyman continuum emitted by the star. It is also proposed here that the observed time scale of two months is set by the dynamics of the flow field in interaction with the random fluctuations imposed on it at the photosphere.

The evolution of massive close binary stars inevitably involves mass exchange between the two stellar components as well as mass loss from the system. A combination of these two processes could produce the stellar wind-modulated behavior seen in LB Vs. The possibility that LBVs are powered by accretion is examined, and does not appear to be a satisfactory general model. Instead, identification of LBVs with close binaries in high mass-loss rate or common envelope evolutionary phases shows promise.

As LBVs have luminosities close to their Eddington limits, their structure is profoundly influenced by radiation pressure. Radiation pressure effects probably cause the highly extended atmospheres and the extreme mass loss observed during the maximum states of the S Dor variables. An opacity-related instability of the radiative acceleration combined with a delayed thermal readjustement of the sub-atmospheric layers possibly explains the large-amplitude radius variations of these objects.

We demonstrate that differential rotation plays a significant role in the essential features of LBV’s, among other types of objects.

A review is given of data and theories on the motion fields in super and hypergiants with special reference to LBV’s. We show that the radiative momentum flux is incapable of driving the episodical bursts of mass loss of these stars, and that there are several indications that the LBV-phenomenon is hydrodynamically driven. The sum of turbulent and radiative accelerations in the atmospheres of the most luminous stars compensates the gravitational acceleration for stars near the Humphreys-Davidson limit. This explains their atmospheric near-instability. The motion field in the atmosphere of a typical LBV consists mainly of low-order gravity waves, while acoustic waves are rapidly damped. These gravitation waves may be stochastic rather than coherently ordered. These stochastic pulsations are assumed to be responsible for the LBV phenomenon.

The so called Luminous Blue Variables (LBVs) form a rather heterogeneous class of very active, luminous (and therefore massive) stars (c.f. R.M. Humphreys, this volume). The following two points are the aim of theoretical work concerning these objects: 1. to know their internal structure from the center up to the atmosphere, i.e. to find out the place of these objects in the course of thermonuclear evolution, or to identify the physical processes which are responsible for the different types of mass loss phenomena, and 2. to conclude which were their preceding evolutionary phases, and to predict their future evolution, i.e. to identify LBV progenitors and descendants, derive their ages, and so on. The basic tool of the theorists is a stellar evolution computer code, which allows to perform evolutionary computations (including different assumptions, which are necessary since several physical ingredients cannot yet be treated from first principles), leading stellar tracks through the part of the HR diagram where LBVs are found. But in this context we meet already a basic problem: in a certain small area of the HR diagram (which may correspond e.g. to some kind of error box) stars of several different types may be identified by the observers, each type may be even devided into several subtypes. Therefore, the correlation of computer models and observed stars is often very difficult or impossible. The various types of stars are mainly classified by spectroscopic criteria, while important “theoretical quantities” as stellar masses or the chemical composition are hard to derive from observations.

We suggest that the LBV mechanism is an essential step to “force” massive stars (M(ZAMS) ≥ 40M⊙) to finally enter the Wolf-Rayet (W-R) domain in the Hertzsprung-Russel diagram (HRD). Just as massive supergiants show increasing variability as they approach the Humphreys-Davidson (H-D)instability limit (horizontally in the HRD diagram), so the W-R stars show decreasing variability as they recede from the H-D limit (at first horizontally into the WNL domain, then, with their high mass loss rates, plunging irreversably downwards as ever hotter, smaller and fainter, strong-line W-R stars). Among the W-R stars, the luminous WNL subtypes (especially WN8) are the most variable, probably as a consequence of blob ejection in the wind. The underlying mechanism which triggers this ejection is possibly related to wind instabilities and may thus be quite different from the source of variability in luminous supergiants or LBV’s in quiescence, where photospheric effects dominate.

P Cygni was discovered in 1600 AD during an outburst when it reached 3rd magnitude. After another outburst in 1655 the star is now showing irregular photometric variations with an amplitude in V of ~ 0.2 magnitude. Because of these variations and of its peculiar spectral characteristics, P Cygni has been classified as anything between supernova and W UMa system. However, its light curve is unlike that of any other star and P Cygni could well be considered the first discovered LBV.

Over the last three years an accurate photoelectric light curve has been put together.

It is based on observations made by a number of amateur and professional observers around the world and shows the following features:

+ Variations are irregular with maxima and minima every 1 - 2 months;

+ The brightness variations seem to show more scatter when the star is faint.

The stationary features of the wind of P Cygni are considerably different from those of ‘normal’ supergiant winds with comparable luminosity. In contrast to such winds, which are generally accepted to be driven by radiation pressure, P Cygni’s mass-loss rate is higher by a factor of 5, the terminal velocity is higher by a factor of 10, and the velocity law itself is much flatter than would be expected from a first glance at glance at typical scaling relations. However, these relations depend crucially and non-linearly on the star’s distance from the Eddington limit, which for P Cyg is very small (see below). Here we investigate whether the acceleration mechanism of P Cygni’s wind can also be explained by line pressure and to what extent self-consistent wind models represent the observed quantities (especially the IR energy distribution).

Since the basic mechanisms that produce photometric variations, shell ejections, and eruptions of LBV’s are still unknown, it is worthwhile to investigate whether instabilities can occur when the improved self-consistent NLTE treatment of radiation-driven winds (see Pauldrach et al. 1986, Pauldrach 1987, Puis 1987, Pauldrach & Herrero 1988) is applied to objects lying in the LBV part of the H-R diagram. The motivation is obvious: LBV’s have lost considerable fractions of their initial masses and hence have L/M ratios close to the Eddington limit. For such objects, radiation-driven wind theory predicts not only a strong dependence of the mass-loss rate on the self-consistently calculated parameters k, α, and δ, which result from the NLTE occupation numbers of the 133 ions contributing to the line force, but also on Γ (= L/LEdd): Ṁ ~ k1/(α – δ) (1 – Γ)(α – 1)/(α – δ). Here we investigate the dependences of Ṁ on M (through Γ) and on the physical environment of the atmosphere (through k, α, δ) separately. The calculations are performed over a large model grid of stellar parameters for P Cyg, a typical LBV (see Fig. 1).

The statistical and thermal equilibrium of P Cygni’s wind as a function of radius is investigated using the model of Drew (1985) with an improved treatment of hydrogen. Provisional calculated hydrogen and helium line fluxes are compared with observation. The wind temperature is found to be cooler than observation implies, and a further heating source is tentatively identified.

Recently, cyclic variations have been discovered both photometrically in the continuum flux of P Cygni (Percy et al. 1988) and by position measurements of different components in spectral lines of this star (Kolka 1983, Markova, Kolka 1985, Lamers et al. 1985). Additionally, more stochastic polarimetric (Haves 1985) and radio flux (van den Oord et al. 1985) variations were discovered, too. It has recurrently been stressed that simultaneous multiwavelength observations are badly needed. Fortunately, we had a possibility to complement our own observational data from 1982 with some additional data of the same year from astronomical literature to compare the B, V-filter flux with the behavior of spectral lines having two flux estimates for this period, too.

The galactic LBV object AG Car is unique for assembling several interesting features: (i) a light curve typical of Hubble-Sandage variables, (ii) a largely variable P Cygni-type spectrum, (iii) a strong IR excess, and especially (iv) a small ring nebula. Caputo and Viotti (1970) found that during 1949-59 the AG Car spectrum varied between A1 and B0 equivalent spectral types. A dramatic evolution occurred in recent years, when the visual magnitude gradually faded from V=6 in 1981 to V=8 in 1985, followed by a rise of the line excitation from Aeq in 1981 (Wolf and Stahl 1982), to Beq in 1983 (Viotti et al. 1984), and to 0fpe/WN9 in January 1985 (Stahl 1986). This behaviour has represented a unique chance to follow a LBV in different spectral ranges during its major photometric and spectral changes. In the following we present the results of a long term monitoring of the star and of its nebula in the optical and UV. We propose that AG Car evolved in recent times from a red supergiant phase, and that the present LBV phase is characterized by a critical structure of the stellar atmosphere.

AG Car is one of the few Luminous Blue Variables (LBV’s) in our galaxy. The star is a photometric variable which had a recent outburst in 1981 (Whitelock et al., 1983; Wolf and Stahl, 1982) and is surrounded by a resolved ring nebula (Thackeray, 1950; Paresce, 1989). These properties indicate that AG Car is a typical LBV. However, at its canonical distance of 2-2.5 kpc, derived from its assumed membership of the Car OBI or OB2 association, AG Car has a bolometric luminosity of Mbol ⋍ −8.5 to −9.0 and is located significantly below the Humphreys-Davidson limit, and also below the upperlimit for red supergiants (RSG). This implies that AG Car might have been a RSG and that the ring might be due to the interaction of the RSG wind and the LBV wind.

VBLUW photometry of high precision of two S Dor type stars AG Car (in a quiescent stage) and HR Car was made in 1987 and 1988. Both stars show visual micro variations with amplitudes of 0m.1 – 0m.2 with time scales – 10d for AG Car and – 20d for HR Car.