Astron. Astrophys. 334, 558-570 (1998)

1. Introduction

The B2e star  CMa (28 CMa, HR 2749, HD 56139, CD ) has long been known to be a light and radial-velocity (RV hereafter) variable (cf., e.g., Frost, Barrett & Struve 1926, Campbell & Moore 1928, Stoy 1959, Cousins & Warren 1963, van Hoof 1975). However, it has only become well-known after Baade's (1979a, b, 1982a, b) discovery of its remarkable RV, V/R and line-profile variation with a period of 1 365. Baade (1982a) also found that the RV of the Balmer emission lines varied in antiphase to that of the absorption lines. Baade (1984) demonstrated that the outer wings of the absorption lines showed only little or no RV changes. He used all his RV observations to improve the value of the period to 1 36673 0 00005 and noted that the RV curves from various seasons looked different. Baade (1982b) obtained uvby photometry of the star and concluded that the light variations cannot be reconciled with the spectroscopic period but rather with a shorter period of 0 435. However, Stagg (1987) obtained UBV photometry of the star which could be reconciled with the 1 365 period and noted that the period 0 435 is an alias of the spectroscopic 1.37-d period. Balona et al. (1987) obtained a long series of b observations of the star from Sutherland and La Silla Observatories. They found a monotonic decrease of the brightness over one year. Moreover, they found a low-amplitude periodic variation with a period of 1 471 from the 1986/87 data when the star was secularly rather stable, and about 0 3 fainter than at the beginning of the 1985/86 season. Clarke (1990) analyzed polarimetry and H scans and suggested that the true period of variations is twice the value found by Baade, 2 7335. Mennickent, Vogt & Sterken (1994) reported large long-term brightness variations and the presence of quasiperiodic light changes with a cycle of about 25 d at the phases of increased brightness. Balona, Stefl & Aerts (1998) analyzed a series of high-resolution electronic spectra from January 1996 and confirmed the presence of 1 37 period in the line moments of He I 6678. They were unable to model the varying shape of the line profile with either spot or NRP model and had to assume a patch with intrinsic line width differing from that outside it. Finally, Stefl et al. (1998) analyzed long series of Heros spectra of  CMa and reported the presence of two periods: a stable one of 1 37, and a transient one of 1 49 seen only in the lines affected by the circumstellar emission.

Given these contradictory results, I decided to check whether an independent analysis of available data which would also include the older RV data could resolve the question of the true period(s) of the star. Results of this analysis led me far beyond my original plan. They demonstrate an extreme complexity of the variations observed and may have some important implications for the understanding of the variability of Be stars in general.

© European Southern Observatory (ESO) 1998

Online publication: May 15, 1998

helpdesk.link@springer.de