The observations began in September 1993 during a test run on the newly-built spectrometer "ELODIE ", and were carried out during 12 different observing runs at the Observatoire de Haute-Provence (OHP, France), Pic du Midi (France) and La Silla (Chile) between September 1993 and January 1995.
2.1. Radial velocities
Radial velocities have been collected at OHP, with the spectrometer ELODIE (Baranne et al. 1996) installed on the 1.93-m telescope, for the faintest half of the sample, and with the CORAVEL spectrometer (Baranne et al. 1979) on the 1-m Swiss telescope at OHP for the brightest half. A few measurements were also made with the southern CORAVEL on the Danish 1.5-m telescope at ESO (Chile).
ELODIE is a high-resolution echelle spectrometer. Radial velocities are calculated from the spectra with an on-line reduction procedure using numerical correlation with a binary mask imitating a F0 dwarf. With this method reliable velocities can be obtained from spectra with a signal-to-noise ratio as low as 2. Typical exposure times were from 20 to 45 minutes. ELODIE can reach an accuracy of 15 m s-1 or better on bright objects, with frequent calibration, but since such accuracy is not necessary for our programme, we have decided to use a mean calibration of the spectrograph zero-point to speed up the observations. This increases the average instrumental uncertainty by about 100-200 m s-1. By adding the photon noise we then have a typical radial velocity uncertainty of 300-400 m s-1.
The mean uncertainty of the CORAVEL set of measurements is 0.55 km s-1.
157 radial velocity measurements were made with ELODIE, and 174 with CORAVEL . The average number of measurements per object is between 7 and 8, and a particular care has been taken to sample regularly the whole pulsation cycle. For galactic kinematics, only the centre-of-mass velocity ( velocity) is needed, and in most cases 5 well-spaced measurements are sufficient to determine the velocity with a sufficient accuracy ( km s-1, see Pont et al. 1994).
The measurements are available in electronic form (ftp 184.108.40.206 or http:://cdsweb.u-strasbg.fr/Abstract.html). Fig. 2 displays the resulting radial velocity curves.
The cepheid CI Per, originally included in the sample, was removed because the cross-correlation function showed some unexpected characteristics that may show the star to be a spectroscopic binary. Spectra with a higher signal-to-noise ratio would be necessary to settle this question.
Although our main effort concerned radial velocities, we also obtained photometry for most objects in the sample, since the existing data were not sufficient to compute reliable reddenings and distances. The observations were made in the southern hemisphere using the CCD on the Swiss 0.7-m telescope at ESO (Chile) in BVRI filters, with Geneva BV (Rufener & Nicolet 1988), Gunn R (Thuan & Gunn 1976) and Cousins I (Cousins 1974). The CCD and reduction procedure are described in Blecha et al. (1990). The raw flux is obtained by integration with an aperture of 40 pixels (16 arc-seconds), and the sky background determined individually for each measurement as a clipped mode of the intensity distribution on the peripheral annulus of the aperture. Extinction coefficients were calculated with a quadratic time dependence for each night using standard stars observed over a large airmass range. The method is similar to that described by Manfroid (1985). Two sets of standards were used: all-sky standards from Rufener (1988) for B and V, and equatorial standards from Menzies et al. (1991) for R and I. Linear colour equations are determined for each run (Bratschi & Blecha 1996). One CCD exposure was taken through each filter. The exposure times are adapted for each filter according to the colour of the observed star, as entered by the observer, with a target error of 1%. As the exposure time is limited by the telescope tracking system to a few minutes, the uncertainties on B for the faintest stars in the sample are higher ( 3%).
Altogether, it amounts to 226 measurements for 21 stars, an average of about 11 measurements per star.
In the northern hemisphere, inclement weather and technical problems prevented us from getting any data during a one week run on the 2-m telescope at Pic du Midi (France). Most of the data come from Arne Henden at USRA/USNO Flagstaff, who very kindly agreed to add some of our objects to his measurement campaign and made available to us more than 200 BVRI (Johnson and Kron-Cousins system) measurements prior to publication.
The visual magnitude curves are shown on Fig. 3, using the data of this article and of Henden. The raw data are available in electronic form (ftp 220.127.116.11 or http:://cdsweb.u-strasbg.fr/Abstract.html).
2.3. Mean values
Mean photometric values and velocities are presented in Table 1 together with general data for the programme stars. Galactic coordinates and periods are taken from the GCVS and the SIMBAD database except when indicated in the notes. The periods of FT Mon, VZ Mon and V510 Mon were slightly modified. V510 Mon in particular seems to have undergone a significant period change since its photographic detection by Wachmann (1966). Mean values for magnitudes and radial velocities were calculated either by fitting a 2nd to 5th order Fourier curve on the data when the number of measurements was sufficient, or by fitting the curve of another well-measured cepheid of similar period (a method described in Pont et al. 1994) for low numbers of measurements. The number of measurements used in each fit is indicated. The Geneva B were not used since the conversion from the CCD to the photoelectric Geneva system was found to be non-linear for the faintest objects.
Table 1. General, photometric and radial velocity data for outer disc cepheids
© European Southern Observatory (ESO) 1997
Online publication: July 8, 1998