**Table 2:** Global characteristics of the best observed parts of the light and colour curves. Time scales for the SD phases are given in years, those for the micro-variations in days

Star	JD-2400000	Features	Time scale	Ampl.	colour in
	(Date)			(mag)	max	min
HR Car	13000-30000	3 maxima	5.5-8y	1-1.5
	(1889-1938)
	45000-49000	3 maxima	2-2.7y	0.5	red	blue
	(1982-1994)
		micro var.
		near min	20-40d	0.15	blue	red
		micro var.
		near max	100d	0.2	red	blue
R127	44500-49500	3 maxima	1.4y	0.5	red	blue
	(1981-1994)
		1 maximum	7y	1.7	red	blue
		micro var.
		near max	100d	0.1	red	blue
R40	36000-49500	1 maximum	25y	0.2	-	-
	(1957-1994)
		rising
		branch	7y	0.6	red	blue
		micro var.
		near min.	46d	0.1	blue	red
		micro var.
		near max.	100d	0.05	red	blue
R71	36000-49500	3 maxima	10y	1	red	blue
	(1957-1994)		6y	0.25	red	blue
			3y	0.25	-	-
		micro var.
		near min.	14-24d	0.1	blue	red
		micro var.
		near max.	100-200d	0.2	red	blue
R110	36000-49500	3 maxima	5-10y	0.3-0.6	red	blue
	(1957-1994)

Together with the results of Paper I, we now have an outline of the variability of 8 LBVs over a time interval of the order of a century or more. In four cases the SD activity is reasonably well-covered by photometric observations over

: AG Car, S Dor,

(Paper I) and HR Car (present paper Sect. 3). The time scales for the VLT-SD phases lie in the order of

. For the four other cases, a few scattered observations were made in the past with long time gaps in between, while monitoring programs were carried out only within the last few decades: R40 (SMC), R71, R110 and R127 (LMC). Thus, for these stars only details of the normal SD phases and the superimposed micro-variations could be documented. The pronounced individuality of these stars, especially with respect to the SD phases, is most striking. We summarize the characteristics of the micro-variations of the LBVs as follows:

Near minimum light all LBVs investigated so far show micro-variations (amplitudes

) on a time scale of

(called

variations in Paper I). The light-curve characteristics change from cycle to cycle as for normal

variables and quasi-periods may differ significantly at different epochs. This could mean that multi-mode pulsations are present. Detailed analyses of the light variations of

and

offer strong additional evidence for this statement (Sterken et al. 1996a, 1997). These micro-variations are generally blue in the maxima and red in the minima.

Near maximum light LBVs show micro-variations (amplitudes also

) on a longer time scale, viz.

. These are generally red in the maxima and blue in the minima. It is a pity that we could not establish whether AG Car's maxima showed the same type of micro-variation because of the lack of sufficient photoelectric monitoring.

The change of the short-time scale micro-variations to the long-time scale ones and vice versa happens rather abruptly, within at most a few hundred days, somewhere along the rising and declining branches at some temperature between 10000K and 15000 K. The corresponding change in the time scale is relatively small (e.g. a factor two for R40), while one would expect this factor to be much larger if the brightness variation is due only to the expansion of the star (i.e. a decreasing density) and if both types of micro-variations are due to radial pulsations. Apart from that, the

micro-variations do not vary much in time scale as the steller brightness continues to increase (e.g., by 05 for HR Car and by

for R127). Obviously, short- and long-time scale micro-variations are caused by different instability mechanisms. The long-time scale variations could be caused either by a thick shell responding to possible stellar pulsations, or by the expanded stellar photosphere being subject to a new type of photospheric instability.

In this context it is very relevant to mention the spectroscopic studies by Wolf (1972) and Hillendahl (1970) of HD 33579 (non-LBV), by Leitherer et al. (1985) and by Wolf & Stahl (1990) of S Dor in maximum and by Wolf (1992) of S Dor and R127 in maximum. They analyzed emission-line variations pointing to pulsation-like movements and depth-dependent velocity fields on a time scale of months. At the dates denoted by arrows in Fig. 5 (click here)a (R127), Wolf (1992) has obtained spectra between 4450 and 4560Å. Sometimes these stars exhibit inverse P Cyg-type line profiles pointing to a very unstable mass outflow, with large clumps of matter falling back to the star. In this case kinetic energy can be released. Attempts have been made to link these spectroscopic changes to the observed photometric micro-variations, thus far with little success. Only in a recent study of the LBV P Cyg, some evidence was found for changes in the brightness of the star at the epochs of shell ejections (Israelian et al. 1996).

It is not unlikely that all these phenomena, partly or largely of a stochastic nature, contribute to the long-time scale micro-variations and their sometimes intricate colour behaviour. These problems can only be solved by simultaneous spectroscopic and photometric monitoring campaigns for which medium-sized and small telescopes, respectively, are most useful.

It should be recalled that the above-mentioned A-type hypergiant HD33579 shows long-time scale micro-variations like the LBVs in maximum and with a quasi-period of

, generally, however, with a reversed colour behaviour, i.e. blue in the maxima and red in the minima (van Genderen 1979). According to Nieuwenhuijzen et al. (1997) this star is a redward evolving star showing an atmosphere permeated by shock waves covering a large fraction of the visible surface and following each other in time intervals of some 100d. This supports the multi-colour photometric evidence that the 100d-type variations of HD33579 cannot be explained by pulsations (van Genderen 1979). However, the preference for the reversed colour behaviour compared to that of the LBVs is very puzzling. Evidently, normal A/F-type

variables and LBVs in maximum, although having the same temperatures, behave not quite similar.

It is no exaggeration to state that LBVs can be considered as the most complicated stars among the pulsating variables. They form an excellent and well-equipped laboratory for the study of the different kinds of photospheric and atmospheric pulsations and instabilities of a massive star within a large range of temperature, radius and mean density.

CS acknowledges a research grant from the Belgian Fund for Scientific Research (NFWO). Research at the Armagh Observatory is grant-aided by the Department of Education for Northern Ireland, and by the UK PPARC through the provision of the Starlink network. The authors are indebted to the Association Française des Observateurs d'Etoiles Variables (AFOEV) for making public their magnitude estimates of HR Car. Part of the data were obtained during observing time allottments 55D-0317, 56D-0249 and 57D-0133. We are much indebted to the referee Dr. M.W. Feast for his critical evaluation.

8. Discussion and conclusions