Lambda-Bootis stars present a considerable challenge to our understan-
ding of stellar atmospheres, structure and evolution. They are metal
poor (depletion of up to 2 dex), but yet are not Population II stars.

The original diffusion/mass loss theory advanced by Michaud and Char-
land (1986, ApJ 311, 326) produced lambda-Bootis stars only at the END
of their main sequence lifetimes. They showed that a mass loss rate of
about 10(-13) solar masses per year together with diffusion in the
stellar atmosphere can actually lead to the observed underabundances
of elements after 10(9) years.

Alternately, Venn and Lambert (1990, ApJ 363, 234) suggested that the
lambda-Bootis phenomenon is associated with the accretion of metal
depleted gas from the interstellar medium. Charbonneau (1991, ApJL 372,
33) has shown that this theory can also produce lambda-Bootis stars,
but contrary to the theory of Michaud and Charland, these stars would
be unevolved lambda-Bootis stars at the ZAMS. A flux-excess at 200
micron would be an indication for accretion and, according to our
estimates, any excess depends nearly linearly on the accretion rate.

The ISO-SWS will be well suited to find emission lines of atoms and
molecules of the accretion disk, like H I, H2-molecules, CO, etc., and
other species with yet unknown lines. These proposed observations for
the brightest lambda-Bootis stars have an exploratory character and do
have the potential of breaking new grounds.

Our proposal complements several guaranteed time programs by adding
observations with different filters and instruments, and thus improving
considerably the scientific efficiency of ISO.