Figs. 2-6 clearly show that we succeeded in fitting all available observational data obtained for CU Vir for more than 40 years with a combination of two different periods. This is not meant to imply that the period changed instantaneously but that it changed in a time period that is short compared to the more than 40 years of observations of CU Vir. A crude estimate of the speed of the period change gives P_ days per cycle. The abrupt change in period near JD 2446000 (1985) raises some difficult questions about the cause of such a change. Since all successful models of the variations of CP stars use rotation as their primary cause, our results imply that CU Vir abruptly slowed its rotation at this time. The problem is in how this could be accomplished, since this implies a sudden loss of angular momentum unprecedented in the study of CP stars. In fact, the only stars that show such effects are close binaries where mass exchange is taking place. The spectroscopy shows no clear evidence that CU Vir is a close binary. All radial velocity variations including those of the hydrogen line cores can be explained in the frame of the oblique rotator model with inhomogeneous abundance and temperature distributions over the stellar surface (Ryabchikova 1991). Abt & Snowden (1973) showed that any solution of the radial velocity variations originating from binary system orbits leads to an estimate for the inclination angle i . If this were the case, the equatorial rotational velocity would have to be larger than 1000 km s-1. Our new binary system solution gives a slightly larger inclination, i , but even in this case the equatorial rotational velocity, 650 km s-1 still exceeds the critical rotational velocity for main sequence stars.
Wolff (1981) considered various braking mechanisms for CP stars and found that loss of angular momentum due to accretion from a surrounding nebula takes place on time scales of about yr. Likewise, mass loss due to an intense stellar wind can result in loss of angular momentum over similar time scales. Thus, both of these mechanisms act over time scales many orders of magnitude greater than is observed for CU Vir. These mechanisms might explain a slow change of the period which possibly is presently occurring. Additionally, a stellar wind could be detected by spectral signatures, which have not been observed. However, Leone et al. (1994) found radio radiation from CU Vir at 6 cm. Later Leone et al. (1996) observed the radio spectrum at 1.3, 2, 6 and 20 cm. They interprete their observations as gyrosynchtron emission coming from the circumstellar regions which are close to the star. This could be the result of a stellar wind. In this case, the following scenario can be considered as a possible explanation of the rapid change in the rotational period. A weak stellar wind exists in CU Vir. The matter is trapped by the magnetosphere of the star so no significant mass loss occurs. When the trapped matter exceeds some critical mass, it may be thrown off, and the magnetic field works as a trigger mechanism.
Another possibility is migrating spots, since the periodic variations of the CP stars are thought to be due to the non-uniform distribution of chemical abundances in their atmospheres combined with rotation. However, none of the dozens of CP stars that have been relatively well observed over time intervals of several decades have shown any evidence of such spot motion. Again, the abruptness of the period change is difficult to explain with such models. Kuschnig, et al. (1997) discuss the atmospheric models of CU Vir in more detail.
Shore & Adelman (1976) proposed that CP stars that were not in binary systems could experience free body precession due to a distortion in the shape of the star by the magnetic field. The precession periods are predicted to be about 5 to 10 years for the shortest period MCP stars (e.g., CU Vir). Changes in both the shapes of the light curves and the times of maxima and minima are predicted. However, any precessional changes must eventually be periodic in nature and we have no evidence for that so far, although the period may still be changing (see Sect. 3.3). Thus, if a precessional period does exist it must be on a longer time scale than is predicted and there is the additional problem of the apparently constant period prior to 1984. Also, the shapes of the light curves appear to be very stable over the more than 4 decades of observations of CU Vir. It is hoped that several more years of observations will help to clarify what is happening with the period of CU Vir.
© European Southern Observatory (ESO) 1998
Online publication: October 22, 1998