From the point of view of observed chemical abundances and of the theory of radiative diffusion, the scenario which considers Am stars as progenitors of metallic F0-6 giants is completely justified. It was worth the effort to try to settle this idea on firmer grounds, or alternatively to question it, by considering two other fundamental characteristics of Am stars, namely slow rotation and high rate of binaries. The main results of our work may be summarised as follows:
. Therefore, if they are descendants
of Am stars, they have not retained their chemical peculiarity. Most
of them have probably never been Am in the past.
, while practically all Am stars are slow rotators
with 
. Moreover, the shape of the 
distribution is completely different in each case: for metallic F0-6III
stars it is flat with a cut-off at about 160 kms-1, while that of Am
stars shows a maximum between 30 and 40 kms-1 with a steady decrease
towards 100 kms-1. This still holds valid when the giant F sample is restricted to stars not known as 
Scuti type.
distribution as the
metallic ones.
variations observed might be due to
Scuti-type variations rather than orbital motion.
Scuti stars), the
best data in the literature together with our measurements indicate a rate
of binaries with P<1000 days of less than 30%. This is smaller, though
not significantly so, than for normal giants. We may have missed binaries
with a fast rotating primary, since the best data in the
literature concern sharp-lines stars (hence slow rotators). However, one
sees no reason why most binaries should be in this case, especially as tidal
friction would tend to slow down axial rotation. For Am stars, this rate is
75% (Abt & Levy 1985).
It seems therefore difficult to admit that Am stars can be progenitors of metallic F giants. If such was the case, one should indeed expect:
distribution of metallic giants strongly peaked at very
small values, with a tail extending to less than 
.
Indeed, the giants are all about to leave the main sequence and have, on
average, larger radii than Am stars. Therefore they can only rotate more
slowly, by conservation of angular momentum.
distributions for the normal, than for the
metallic giants, as is the case for the normal A dwarfs compared with the
Am stars.
None of these three expectations is fulfilled. As a whole, the metallic giants cannot be considered as evolved Am stars, although the previous considerations do not exclude the possibility that some of them (especially the slower rotators) may have been Am stars in the past.
One might object that even fast rotating metallic giants may have main sequence metallic progenitors, but the latter have gone unnoticed by the classifiers because of fast rotation. But such progenitors would have been detected by Geneva or Strömgren photometry (through the m2 or m1 parameters), while significant photometric metallicity is observed only in stars classified spectroscopically as Am, except for the metallic F giants. The assumption of overabundances remaining steady (apart from that of Ca) up to the very end of the main sequence life therefore seems wrong, and diffusion theory indeed predicts that they may change drastically on shorter timescales, at least near the ZAMS (Alecian 1996).
But if metallic F giants are not evolved Am stars, what is their origin then? The fact that they are not especially slow rotators is intriguing, because it suggests that no special initial conditions are required to produce them. The same can be said about the rate of binaries, which does not seem special either. In view of this, the very simplest alternative to the idea of Am progenitors is to speculate that every late A and early F star goes through a short phase of enhanced atmospheric metallicity around the end of its life on the main sequence.
We have seen that for some FIII stars the metallicity can coexist with high
projected rotational velocity (
). The explanation
of this fact is a real challenge addressed to theoreticians of diffusion,
because for main sequence stars metallicity can appear only in slow
rotators (
).
Acknowledgements
This research has made use of the Simbad database, operated at CDS, Strasbourg, France. We thank Drs. Michel Mayor and Stéphane Udry (Geneva Observatory) for providing us Coravel data for standard stars. We thank Mrs B. Wilhelm for the correction of the English text. This paper received the support of the Swiss National Science Foundation.