Astron. Astrophys. 345, 172-180 (1999)

2. Observations and data reduction

A multi-site campaign was held from 17 to 22 October 1989, including UV spectroscopy with the IUE satellite, optical spectroscopy at the Kitt Peak and Calar Alto observatories, B polarimetry at McDonald Observatory and UBV photometry at the Wendelstein observatory. The IUE spectra are described and analyzed by Kaper et al. (1996, 1999). Here we restrict ourselves to the data that lead to the pulsation analysis.

The experimental setup was chosen in accordance with the vsini values of the stars. With 200 km s^-1 for  Per and 214 km s^-1 for  Cep (Conti & Ebbets 1977) a resolution of 15 km s^-1 would give about 25 effective points in the line profile, sufficient to assure that modes up to 20 would be detectable. We concentrated on the weak and therefore deep-seated photospheric HeI absorption line at 4713 Å.

The Calar Alto 2.2m telescope (observers GZ and HH) was used with the 90 cm f/3 camera in Coudé focus with grating 1 (632 lines mm^-1) in second order centered at 4667 Å, giving a dispersion of 8.6 Å mm^-1. We used a RCA CCD chip () with 1024640 pixels of 15 µm size yielding a coverage of 130 Å with a 2 pixel resolution of 0.26 Å. At this setting the CCD has a quantum efficiency of 68, which gave 6.7 electrons per count with the gain set at 30. The slitwidth was 1.5 arcsec, equal to the average seeing. At Kitt Peak (observer DG) the 0.9-m Coudé feed was used with camera 5, long collimator with grating A (632 lines mm^-1) in second order with a 4-96 filter to block the first order light, yielding a dispersion of 7.1 Å mm^-1 on the Texas Instruments CCD (TI3) with 800600 pixels of 15 µm size. This yielded a coverage of 85 Å with a projected slitwidth of 21.3 µ FWHM. The CCD chip has a quantum efficiency of 70, giving 4.12 electrons per count with the used gain. The slitwidth was fixed at the average seeing of 1.5 arcsec. Typical exposure times were 2 to 3 minutes for  Per and 7 to 10 minutes for  Cep at Calar Alto, and about 1.5 times longer at Kitt Peak. The number of spectra used in the analysis was 324 for  Per (154 from Calar Alto and 170 from Kitt Peak), and 169 for  Cep (109 from Calar Alto and 60 from Kitt Peak). Fig. 1 depicts the time line of the optical spectroscopic observations for both stars.

Fig. 1. Schematic time line of the optical spectroscopic observations of the campaign used for the current analysis. In total 324 spectra were selected for  Per and 169 for  Cep

Flatfielding, bias subtraction and wavelength calibration with Th-Ar lamps were done in the usual manner. The spectra were normalized by using a linear fit through wavelength segments carefully selected to contain no traces of spectral lines. We used [4695.2, 4700.4] and [4720.1, 4723.4]Å for both stars. We also attempted higher-order polynomials, but this did not change the results of the frequency analysis. All spectra were finally smoothed and reduced on a uniform velocity grid of 15 km s^-1. Spectra taken at the same epoch at the two observatories showed excellent agreement within the errors given the signal to noise ratio (S/N) which was on average about 800 (Calar Alto) and 500 (Kitt Peak) for Per and 500 (Calar Alto) and 300 (Kitt Peak) for Cep, respectively. This agreement was reached after some small night-to-night and observatory-to-observatory wavelength corrections. These were only needed for Calar Alto spectra, because of some technical problems at the telescope that occurred at several nights, which made apparently small changes to the wavelength setting.

Throughout this paper, the reference wavelength for conversion to the stellar restframe has been corrected with 60 and -75 km s^-1 for the runaway stars  Per and  Cep, respectively (Gies 1987).

Average spectra are shown in the top panels of Figs. 2 and 3. Sample quotient dynamic spectra of  Per and  Cep are in these figures compared with the inverse Fourier transform (after removal of noise) and with folded data using the pulsation frequencies reported in this paper.

Fig. 2. Figure at the left: dynamic quotient spectra of 2 nights of Per data. Horizontal lines indicate the mid-exposure times. The upper panel shows the average spectrum used to create the quotient spectra which are shown in the panel below. Note the slight depression at the left part of the average spectrum, caused by an unidentified line. Middle figure: inverse Fourier transform for which frequencies only above 2 c d-1 were used (see Sect. 3). Figure at the right: all data folded with a 3.5 h period, rebinned to 25 points per cycle

Fig. 3. Figure at the left: dynamic quotient spectra of 3 nights of Cep data, in similar format as in Fig 2. Middle figure: inverse Fourier transform using frequencies only between 1 c d-1 and 5 c d-1. Figure at the right: sum of data folded by respectively 12.3 and 6.6 h and rebinned to timesteps of 0.0075

© European Southern Observatory (ESO) 1999

Online publication: April 12, 1999
helpdesk.link@springer.de