The radial velocity and spectral line bisector variability of Polaris

We present precise stellar radial velocity measurements for the Cepheid-typ e star Polaris taken in 1992-1993. The peak-to-peak amplitude of the pulsat ions was 1.555 +/- 0.056 km s(-1) in epoch 1992.4 and 1.517 +/- 0.047 km s( -1) in epoch 1993.2. These amplitudes are comparable to the one measured by Dinshaw et al. in 1987.75 and recent measurements by Kamper & Fernie for 1 995-1997. Consequently, these do not support the continued drop in pulsatio nal amplitude first extrapolated by Dinshaw et al. A periodogram analysis f or our data yields a pulsational period of 3.(d)9726769 +/- 0.(d)00011, sli ghtly higher than the value found by Dinshaw et al. and consistent with the value derived from the Kamper & Fernie data. The pulsational period of Pol aris may thus be increasing. Residual radial velocity measurements after re moval of the component due to radial pulsations show the presence of a 40 d ay period with a 2K amplitude of about 400 m s(-1), first reported by Dinsh aw et al. Spectral line bisectors of the Sc II lambda 5525 and Mg I lambda 5528 lines also show variations with the 40 day period, strongly indicating that this period is indeed real. The presence of line bisector variations excludes a low-mass companion object as a cause of the residual radial velo city variations. Two models are considered for the residual radial velocity and bisector span variations: starspots and nonradial pulsations. Although both can reproduce the amplitude and shape of the residual radial velocity variations, the spot models considered were unable to reproduce the overal l shape and amplitude of the bisector span variations. The best-fit spot mo del was provided by a single spot with a temperature 500 K cooler than the photosphere with a macroturbulent velocity of zero in the spot. If surface features do exist on Polaris and are responsible for the residual radial va riations for this star, then possibly they have a temperature structure and distribution more complicated than the simple models considered here. The best fit to all the observed variations was provided by a nonradial in = 4 mode, although these were not entirely satisfactory having a phase shift of about 0.25 between the observed and predicted bisector variations. Possibl e explanations for this shift include temperature effects, a source functio n originating in a dynamic stellar atmosphere, or a different pulsation mod e than the one that was considered. Although the exact pulsational mode is yet to be identified, we believe that the residual radial velocity variatio ns are due to a long-period pulsation mode. Supporting evidence comes from the fact that three different residual variations with periods near 40 days have now been seen in Polaris and it is unlikely that these variations are due to rotational modulation by surface features on a star that is differe ntially rotating.