2. Observations and data reduction
The data were obtained during the Commissioning of UVES and have been released by ESO for public use. The spectra of the two stars, whose basic data is given in Table 1, were taken on the nights 2,3,4 and 6 October 1999. The slit was , which provided a resolution of at 565 nm. We used only the red arm of UVES with the standard setting with central wavelength at 580 nm, which provides spectral coverage from 480 nm to 680 nm The detector was the mosaic of two CCDs composed of a EEV CCD-44 for the bluemost part of the echellogramme and an MIT CCID-20 for the redmost part. Both CCDs are composed of square pixels of 15 side. We used a on-chip binning, without any loss of resolution, given the relatively wide slit used. For each of the two stars three one-hour exposures were collected, under median seeing conditions, allowing to reach a signal to noise ratio of at 510 nm, confirming the excellent performance of the instrument (D'Odorico et al., 2000).
Table 1. Basic Data
The data was reduced using the ECHELLE context of MIDAS ; reduction included background subtraction, cosmic ray filtering, flat fielding, extraction, wavelength calibration and order merging. Each of the CCDs of the mosaic was reduced independently. The r.m.s. of the calibration was typically of the order of 0.2 pm for each order and in all cases less than 0.3 pm. Flat-fielding was highly successful and the single echelle orders were rectified to within % by this process, except in the vicinity of CCD defects, where the correction was not always satisfactory. The three spectra available for each star were coadded without any shift in wavelength. By cross-correlation we estimated the shift between any pair of spectra to be less than 0.5 pixel, so we decided not to perform any shift. This results in a very slight degradation of the resolution, which is not an issue for our analysis. The differences in barycentric correction were at most of 0.1 km s-1, so no appreciable shift was expected from this cause either. The normalized and merged spectra were plotted superimposed on preliminary synthetic spectra for the purpose of line identification. A single velocity shift was adequate for all the spectra, confirming that the internal accuracy of our wavelength scale is better than 0.2 km s-1. Portions of the normalized spectra of the two stars are displayed in Figs. 1 and 2.
© European Southern Observatory (ESO) 2000
Online publication: July 7, 2000