3. The radial velocity curve
Figs. 1-5 show the heliocentric radial velocity curves of RR Lyrae during the five Blazhko phases: , , , and . The velocities are determined by fitting a Gaussian to the whole Fe II 4923.921 Å line without taking into account line doubling that occurs just before maximum luminosity (Chadid & Gillet 1996a). As discussed in Sect. 2.3, the accuracy of the measurements is around a few hundred m/s. A good idea of the accuracy is given by the dispersion of points within the phase interval 0.2-0.5. It is quite good for August 3-5, 1994 (Fig. 1) and relatively poor for July 24-26, 1996 (Fig. 3) and August 5-7, 1997 (Fig. 2).
It is clear that a velocity shift is present over the two or three consecutive nights that are required to build up a complete pulsation period. For example, over three nights at (Fig. 2), the observed velocity shift is about 4 km/s at the pulsation phase , i.e. 7% of the whole pulsation amplitude. This shift can also be smaller (, Fig. 4). A shift is also visible during the double velocity maximum, which occurs during the pulsation phase interval 0.6-0.8 (Fig. 1), but this is not always the case (Fig. 2). It is difficult to say if all of these variations are due only to the Blazhko effect. Fig. 6 shows an overplot of the five heliocentric radial velocity curves which were fitted with a Fourier series. Although the curve deformations are sometimes large at similar Blazhko phases, for instance at and curves show completely different shapes, it is difficult to say that they are not connected with the Blazhko effect because it is not a strictly periodic variation.
Table 2 shows the basic dynamical parameters which can be deduced from the integration and derivation of the velocity curves (Chadid & Gillet 1996a). Although the number of observed Blazhko phases is too small to provide any statistical relevant conclusion, there is no well marked correlation with respect to the Blazhko phase. For instance, the primary acceleration seems stronger near and than at , but the photospheric extension appears larger at than at . The variation of the -velocity is weak (smaller than 1 km/s) and does not seem clearly connected with the Blazhko effect. Because we have estimated that the standard deviation of the dispersion of radial velocities is between 117 and 299 m/s (see Sect. 2.3), it is not clear that a real variation is present. Moreover, this is not consistent with the result of Struve & Blaauw (1948) who argue for a variation of -velocity with a period of approximately 42 days. Consequently, new high quality observations over a complete Blazhko period and, if possible, over the same cycle, are required to confirm how the -velocity is variable and to determine eventually its amplitude and its period.
Table 2. Main dynamic parameters at the five Blazhko phases. The first column gives the Blazhko phase , the second the average radial velocity over one pulsation period , the third the primary acceleration, the fourth the secondary acceleration and the fifth the radius variation.
© European Southern Observatory (ESO) 2000
Online publication: July 13, 2000