In this work we studied the long-term spectrophotometric variations of 49 classic Be stars. The parameters used in this study are: the U and V magnitudes of the UBV photometric system, total Balmer discontinuity D, and the gradient of the energy distribution defined for to m. To obtain these quantities for as long a time base as possible, we used genuine BCD spectrophotometric data and photometric data obtained in most cases since 1950 in five different photometric systems: UBV, UBVRI, Geneva system, Strömgren's uvby and 13-colour system, which we reduced to a common spectrophotometric scale given by the BCD system.
Variations of the () parameters were qualitatively compared with the spectroscopic variation of a small number of well-studied Be stars, which were mostly observed either in Be phases or in shell phases. From this comparison it generally follows that line emission is stronger when stars are brighter and redder, and that for strong reddening a veiling in the lines is also observed. Total Balmer discontinuity is smallest at phases of strong emission. In Be-shell phases the strength of the shell is always stronger when total Balmer discontinuity is greatest. RV are also highest in these cases.
For SPh-Be phases total Balmer discontinuity is smaller than the stellar component and for each star there are well-defined (U,D), (V,D), () and () linear-like correlations which can be single or double-valued. Slopes of these correlations are and in most cases, but sometimes their signes are changed. In SPh-shell phases the total BD is higher than the stellar BD, and generally no variation of V and accompanies the increase of D. In some rare cases a slight variation of V and and even a bluening is observed near . However, when a spectroscopic shell spectrum develops, reddenings or bluenings may sometimes be spurious effects due to shell lines entering the photometric filter pass-bands which are not considered in the calibration of photometric indices in spectrophotometric quantities.
The fact that the spectrophotometric patterns are single or double-valued does not seem to correlate with .
For a number of Be stars a unique pattern was observed for both ranges of D, , even after having passed through several shell phases and apparent loss of emission characteristics. Slopes of these patterns are different whether it is D < D* or D > D*.
SPh-shell behaviours such that D > D* have been observed for several stars which have moderate parameters. If these values of are due to small aspect angles i, this will mean that in such stars the CE in the neighbouring stellar regions cannot be strongly flattened.
Sometimes, the spectrophotometric correlations are somewhat dispersed. This phenomenon may possibly be due to high photometric variability, produced by activities in the underlying stellar surface, which does not obey the "smoothed'' long-term variation due to the CE.
We showed that the spectrophotometric variations in SPh-Be phases may well be produced by CE in low opacity regimes (, ). Those observed in SPh-shell phases are due to CE in high opacity regimes (, ), where the temperature decreases as their opacity increases. The low opacity CE seem to be more extended than those of high opacity. The high positive values of RV observed in the stronger shell phases should probably favor the appearance of compact formations near the star.
The formation of CE in Be stars, as well as their structure, are still open questions. They may not only depend on the mass loss mechanisms, but also on the way the mass is lost and its dynamics. Polarimetric data cannot be interpreted if CE are not somewhat flattened. Nevertheless, recent interferometric measurements of some Be stars (Stee et al. 1995, 1997; Quirrenbach et al. 1997) reveal that the effective ellipticities of CE for the H line formation region are , which clearly rule out the spherical geometry of CE regions exceeding . However, for the visible continuum, for which (Stee et al. 1997), such an approximation can still be reliable. The present study allowed us to appreciate some systematic spectrophotometric behaviours of Be stars. In particular, it showed that not only SPh-Be and SPh-shell phases can be present in a same star, but that they can both take place in stars with low values of . If these low are due to , CE geometries preventing the shielding of stellar polar regions: high flattened disks, lemniscate shaped CE at , should probably be not appropriate to model the spectrophotometric behaviour of these objects. In order to better constrain models of CE for Be stars (their geometries in particular) and/or understand the phenomena underlying the observed spectrophotometric behaviours, a number of complementary observational and/or quantitative studies have still to be done. Correlations of spectrophotometric vs. spectroscopic behaviours, time lags between both, as well as a study of the properties of spectrophotometric relations presented in this paper against the stellar fundamental parameters and the aspect angle i will be presented in a next contribution. Using different model envelopes, a quantitative discussion in terms of opacity, size and temperature of CE will complete the present work.
Many thanks are due to Dr. Ch. Sterken for having helped us to manage the data archives of the Long-Term Photometry of Variables (LTPV) program at ESO. We are indebted to Dr. J. Percy, who kindly provided us with some unpublished photometric data and to all those observers whose work is included in the IAU Archive of Unpublished Photoelectric Observations (accessible by CDS). We are particularly grateful to Dr. P. Harmanec and Dr. K. Pavlovski for having provided us before publication with some UBV data collected at the Hvar Observatory. Discussions with Dr. S. Štefl on 59 Cyg are warmly acknowledged. Useful comments and painstaking reading of the manuscript by the referee, Pr. J. Dachs, which helped to improve the presentation of the paper are greatly acknowledged. This research has made use of several databases and abstract services: Centre de Données Stellaires de Strasbourg (CDS), The General Catalogue of Photometric Data (GCPD, University of Lausanne), Astronomical Data Analysis Center (ADAC, NAO Japan) and NASA's Astrophysics Data System Abstract Service (ADS).