Based on UVES observations made with the European Southern Observatory VLT/Kueyen telescope, Paranal, Chile, collected during the programme ID No. 70.B-0258(A).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

Tables 2-8, Figs. 1-14, and Appendix A are only available in electronic form at http://www.edpsciences.org

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... Database''

http://kingpin.ucsd.edu/~hiresdla/

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... uniformity

This uniformity is even stronger when the abundance ratios [X/Si] and [X/Fe] are computed by considering all of the gas in the Si II and Fe II profiles (values in parenthesis in Table 9).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... errors

The statistical errors are defined as the average of the 1 $\sigma$ errors on measurements.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... dispersion

The number of $\sigma$ is computed as the square root of the reduced $\chi ^2$ . This is a first order approximation of the probability to have a statistically significant dispersion assuming a Gaussian distribution.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.