We propose SWS spectroscopy of selected lines in a complete sample
of 56 Ultraluminous Infrared Galaxies (Lir>10E12 Lsun) at z<0.35
in order to establish the nature of their primary energy sources. The
discovery of such galaxies was one of the major results of the IRAS
survey 10 years ago but their nature remains unclear and construction
of large samples for further study has taken time.They are generally
mergers containing the largest concentrations of molecular gas known
but relatively few exhibit QSO/Seyfert1 type optical spectra although
their luminosities are in the QSO range. It has been suggested,
therefore, both that they are mostly powered by extreme starbursts and
by dust enshrouded AGN's. A statistically valid test is of the utmost
importance in view of speculation that they may represent different
phases of an evolutionary sequence from starbursts in gas rich merging
spirals via the formation of dust enshrouded AGN's to visible
QSO's following dispersal of their obscuring dust by SN or AGN driven
winds. As extinction to their nuclei could correspond to Av's of 100
-1000mag searches for dust enshrouded AGN's in the optical are of
limited value.ISO now offers the unique capability for detecting and
discriminating between starbursts and AGN via spectroscopy of the
[NeII]12.8um,[NeIII]15.6 and [NeV]14.3/24.3 lines which trace 20-60eV
photons and, based on theoretical line ratio diagrams, are optimum
for our  purpose. Their fluxes should also be comparable to Hbeta but
at wavelengths where the extinction in mags is a factor 500 lower.
([SiIX]3.95um will be intrinsically fainter and suffers higher
extinction but can be observed simultaneously with the [Ne] lines
and is of specific interest due to its high ionization potential
(303eV) and recent detection in AGN. Observations of the [Ne] lines
to the same s/n ratio in our complete flux limited sample should
establish a measure of the relative importance of starburst and AGN
activity which is statistically valid for whole UIG population.