1 Introduction

According to the standard big bang nucleosynthesis model, lithium is one of the few elements synthesized in the first minutes of the Universe. In this scenario the primordial synthesis of lithium is very sensitive to the baryon/photon ratio ($n_{\rm b}$/$n_\gamma$), and the astronomical determination of its primordial abundance can constrain the baryonic contribution to the density of the Universe. Since the discovery of a rather uniform lithium abundance, the so called lithium plateau, in the hotter halo dwarfs (Spite & Spite 1982; Rebolo et al. 1988) at about a value $\log n({\rm Li})=2$ (on the usual scale where $\log n({\rm H})=12$), there has been a long debate on whether or not this abundance reflects the primordial one. Some theoretical models suggest that processes such as diffusion, rotational mixing etc. may have depleted the initial lithium abundance of the hotter ($T_{\rm eff}\ge$ 5600 K) halo dwarfs and reduced it to its present atmospheric value. The fingerprint of such a process has been searched and claimed to be possibly manifest in a trend of the Li abundance versus [Fe/H] and $T_{\rm eff}$ (Thorburn 1994; Norris et al. 1994; Ryan et al. 1996), or in a spread of the lithium abundance around the plateau (Deliyannis et al. 1993, 1995).

The observational evidence for a global depletion process acting on these halo dwarfs is not conclusive. Molaro et al. (1995) using a reddening-free temperature scale and Bonifacio & Molaro (1997) using direct infrared flux method temperatures provided by Alonso et al. (1996) have made a bivariate analysis of a large sample of halo dwarfs, and conclude in both cases that there is no evidence of these trends versus metallicity and effective temperature. The same conclusion has been reached by Spite et al. (1996), who re-analysed an important part of the current observations using temperatures determined from several independent methods; these authors propose that the observed spread around the plateau is due to error measurements and uncertainties in effective temperatures. On the other hand, Ryan et al. (1996) used a temperature scale similar to the scale of Carney et al. (1994) and in this case, a multiple-regression analysis shows that the correlations actually exist, especially when the most metal-poor stars are included in the sample.

This is the first paper of a series that intend to revisit the lithium problem in halo dwarfs. The paper presents new LiI $\lambda 670.8$ nm observations in 67 metal-poor stars. Here, the observations and the comparison with other works are presented. The main analyses and implications will be given in a forthcoming paper.