3 UVES observations

Due to its high efficiency UVES (Ultraviolet-Visual Échelle Spectrograph) on ESO-VLT (Very Large Telescope) UT2 (Kueyen) is the ideal instrument for studying MC MS B stars ( $m_{V}>15^{\rm m}$). We initially considered using one of the standard templates for full optical coverage (DIC1 390+564 nm), but rejected this option as it would have resulted in the loss of the strategic Si III triplet near 4560 Å (dichroic gap). We finally chose the UVES standard setting Blue437 nm ( $3760\,{\rm\AA} <\lambda<498$0Å) covering all the strategic lines for a classical abundance fine analysis of B stars. The atmospheric dispersion corrector (ADC) was used and the slit oriented in such a way as to minimize contributions from neighbouring stars. The slit width was set to 1.2 $^{\prime\prime}$ (corresponding to a resolving power of $R=40\,000$) not to lose light under moderate seeing conditions.

One to two spectra of each of the six preselected objects were exposed in Service Mode between November and December 2000. We make use of the pipeline-reduced spectra: they turn out to be of sufficient quality in terms of rectification, even for the use of H$\beta$ as a gravity indicator. We note, though, that the blue part of the exposure (3800-4600 Å) is much less well-defined in terms of the continuum. After rebinning to a 2-pixel resolution of $R=20\,000$ the spectra were normalized interactively and - in the case of two exposures - co-added. Table 1 gives a brief observation log including measured signal-to-noise ratios (S/N) and projected rotational velocities ($v \sin i$). Both sets of quantities fall in the range expected from the UVES exposure time calculator on the one hand and the preselection on the other. Figure 1 shows the pipeline reduction result for the two spectra of C9 in the region surrounding H$\beta$.

Except for B18, the heliocentric radial velocities $v_{\rm rad}^{\rm hel}$ are fully compatible with that deduced for the cluster by Freeman et al. (1983). B18 (just like B15 from Paper I) has a radial velocity some 40 km s^-1 higher than the cluster average. These two stars probably do not belong to the cluster. Unfortunately, the cluster membership of the other programme stars is not certain either, since the difference in radial velocity between the cluster and the surrounding field is not significant within the errors.

Figure 1: The red part of the Blue437 exposures of NGC 2004/C9. The blaze residuals are very constant from order to order allowing for a thorough rectification using a low-order polynomial.