The authors of this Edition decided to give to the Catalogue a more spectroscopic aspect than in the previous versions, keeping only as photometric parameters, the visual magnitude V and the colour index B-V. As a great work has been done by Pr. Hauck and Drs. Mermilliod to bring the GCPD up to date, it would have been superfluous to repeat the same information in this spectroscopic Catalogue. Then, why has a Catalogue of atmospheric parameters of stars been called a ``Catalogue of [Fe/H] determinations"? Because the understanding of the chemical composition of stars, interstellar matter, and galaxies has become one of the central issues of modern astronomy. Therefore we wanted to emphasize the importance of the abundance parameter [Fe/H].
As was already mentionned in Sect. 2, the atmospheric parameters listed in the Catalogue are deduced from models generally computed with the assumption of LTE. But not all the time. The reader is urged to consult the individual references for more details on the techniques used in each analysis.
The colour index B-V is a convenient photometric parameter
because it has been determined
for a very large number of stars (in File 1, 98% of the stars have a colour index B-V).
This index is commonly used as a temperature indicator in galactic studies in which
a large number of very faint stars are considered. The plot in Fig. 1 (click here) shows the distribution of the
field stars of the Catalogue (File 1) in the
plane (B-V, 
) for stars ranging between 
and 
.
There is a clear trend of
with respect to B-V, but the dispersion about the
correlation is very high.
The relation between
and B-V is particularly
uncertain around B-V=0.5
where the corresponding 
spans more than 
.
The dispersion in this relationship is undoubtedly heavily influenced by the
sensitivity of B-V to the metallicity of the star, showing the limitation
of B-V as a temperature indicator.
Figure 1: 
plotted against B-V
colour for field stars (File 1) of the
Catalogue
In most cases, the Sun is used as the reference star in the analyses. However, for
early type stars, peculiar Ap and Am stars, yellow giants etc... other well
analysed stars like 
Vir, 
Boo or o Peg, are also
used as reference.
In some cases, the
authors have not observed their reference stars with the same
observational and optical equipment as their program stars. Also
they have not always analysed their program and reference stars
using the same model
atmospheres and the same physical and atomic parameters. As a result,
even for most recent analyses, the abundance determinations from different
authors may show significant disagreement in spite of the excellent quality of the
observations. Instead of taking the mean of the abundances listed for a star with
many determinations, the reader is encouraged to carefully examine the details
of individual analyses to understand the differences. An illustration of this problem
can be given with the well-known star 
Vir (HD 113226). This star is frequently
used as a comparison star in detailed analyses. The Catalogue gives 15 values of [Fe/H] with
respect to the Sun for
Vir, ranging between -0.10 dex and +0.21 dex. What value should be taken as the
true metallicity of 
Vir? Due to such disagreements, we did not
attempt to standardize the [Fe/H] values of the Catalogue to the solar one.
We have been requested several times, between the successive editions, to include an average value of the different determinations of [Fe/H] for each star. But also for this Edition we have upheld our initial decision to publish the original values from the literature, the mean value being subjected to change at each edition, and a straight mean between values of unequal quality being physically unjustified. A weighted mean would have involved a number of subjective judgements and the development of some scheme for compensating for systematic differences and normalizations between individual investigations. For those who want to use a mean anyway, without making an "educated" weighting, we recommend keeping only recent determinations obtained with high S/N ratios, from spectroscopy with Reticon or CCD detectors. However it must be noted that [Fe/H] determinations are also affected by the adopted effective temperatures, gravities and microturbulent velocities, and that a stellar metal abundance can be in error, even if the observations are of excellent quality.