Measurements of the mass-to-light ratios in
clusters suggest that 
, the ratio
of the total density of the universe to
the critical density, exceeds 0.2 [9]. This determination
of 
is consistent with measurements
based upon the large-scale
velocity fields and the dynamics of
the large-scale structure [67].
Values of 
less than 0.2
are very difficult to reconcile with the
500 km/s random velocities seen in large
scale structure surveys and even harder to reconcile with
large-scale streaming motions.
The observed (presumed cosmological) abundances of deuterium, helium and
lithium
are only consistent with standard big bang nucleosynthesis
if the baryon density is much less than 
.
The best fit value for
, which is nearly
an order of magnitude below the dynamical
values [72].
For example, if 
km/s/Mpc, 
implies
that Y, the Helium/Hydrogen abundance
ratio, is 
and 
, the Deuterium/Hydrogen
abundance ratio, is 
[72]
while if
km/s/Mpc, 
implies
and 
.
There are many extragalactic HII regions with 
and best estimates imply 
. These
observations appear to require either a significant
modification of our ideas about big bang
nucleosynthesis or the existence of copious
amounts of non-baryonic dark matter. (See, however,
Goldwirth & Sasselov [30] for a dissenting view).
All of the proposed
modifications of BBN appear to violate known
observational constraints.
For example, Gnedin & Ostriker [32] proposed
that an early gamma-ray background
photodissociated some
of the primordial Helium. This model predicts a spectral
distortion of 
and a fully
ionized universe. y describes the deviation of the
observed spectrum from the thermal spectrum and
is a measure of the energy injection in the early universe.
COBE [42] found that the observed spectrum was consistent
(within the experimental errors) with a thermal
spectrum and constrained 
.
Inhomogeneous nucleosynthesis models have been studied
extensively in the past few years. However, Thomas
et al. [68] found that
even models with large
inhomogeneities imply 
for 
.
Thus, they are also not consistent with 
.
While the theory of big bang nucleosynthesis is well developed, there is still uncertainty in converting the observed line ratios to abundances. Most of the abundances for external systems assume a spherical clouds with constant rates of ionization. It would be interesting to study a nearby system such as the Orion nebula and estimate the error associated with this approximation in the analysis. Goldwirth & Sasselov [30] have made an important first step in studying the sensitivity of these element abundances to model uncertainties. There is a need for more work.
While none of these three arguments is incontrovertible, they all do suggest that most of the universe is in non-baryonic matter. The rest of this paper will review the most popular proposed candidates for non-baryonic dark matter and consider various schemes for detecting its presence.