To a great extent, the spectra of the stars listed in the Catalogue have been interpreted by differential curve of growth analysis, the Sun having been adopted in the majority of the cases as reference star. In the mid-sixties, the availability of powerful computers has pushed the authors to perform detailed analysis instead of coarse analysis. The theoretical line equivalent widths and curves of growth were then computed from detailed model atmospheres, interpolated in a grid of constant flux models. Authors interested in early type stars (early F, A, Am, Ap) ordinarily use the Kurucz grid of Atlas models (Kurucz 1991a, 1991b, 1993), those interested in late F, G and K stars use mostly the Gustafsson's grid of atmosphere models (Gustafsson et al. 1975; Edvardsson et al. 1993). Both, Kurucz and Gustafsson models were computed with the assumption of Local Thermodynamical Equilibrium (LTE). In a strictly differential analysis, this is not very important, because a large part of Non-LTE effects cancel out when the programme star is sufficiently similar to the reference star.
A very important step in a spectral analysis of a star is the determination of accurate
atmospheric parameters for the selection of the appropriate model atmosphere. No abundance
can be derived unless the three physical parameters, effective temperature
, surface gravity
,
and microturbulence (
) have been obtained. The effective temperature of a star, which is the
critical parameter, is most often derived from wide or narrow-band photometry, but in
some cases, it is derived on purely spectroscopic grounds from the comparison between
observed profiles and 
computed profiles. The spectroscopic surface gravity
is determined from ionisation and excitation equilibria, as obtained from neutral and
ionized lines carefully chosen in the stellar spectrum and from wings of strong lines, broadened
by collisional damping. The microturbulence is determined by fitting a theoritical curve of growth,
computed with an appropriate model, to the observational curve of growth. The metal abundance is
obtained by comparing the observed equivalent widths of weak and medium-weak lines
(possibly between 10
and 50 mÅ), in the spectrum of a programme star to those resulting from a model atmosphere computed
with the appropriate stellar parameters (
, 
, 
) previously determined for the star
under investigation. Attention has been paid by the authors of the recent analyses to treat the
reference star in the same way as the programme star. According to our own experience, many
differential analyses with respect to the Sun show spurious effects just because the Sun
has not been analysed with the same grid of model atmospheres as the programme star
(Cayrel de Strobel
1996).