MASSIVE BLACK HOLES AND LIGHT ELEMENT NUCLEOSYNTHESIS
                     IN A BARYONIC UNIVERSE

          Nickolay Y. Gnedin, Jeremiah P. Ostriker
     Princeton University Observatory, Princeton, NJ 08540

                       Martin J. Rees
       Institute of Astronomy, Cambridge CB3 OHA, England

		          ABSTRACT

We re-examine the model proposed by Gnedin and Ostriker (1992) in 
which Jeans' mass black holes ($M_{\rm BH}\approx10^6\msun$) form shortly 
after decoupling. There is no nonbaryonic dark matter in this model, but
we examine the possibility that $\Omega_b$ is considerably larger than
given by normal nucleosynthesis. Here we allow for the fact that much of 
the high baryon-to-photon ratio material will collapse, leaving the universe
of remaining material with light element abundances more in accord with the
residual baryonic density ($\sim10^{-2}$) than with $\Omega_0$ and
the initial baryonic density ($\sim10^{-1}$).

We find that no reasonable model can be made with 
random phase density fluctuations, if
the power on scales smaller than $10^6\msun$ is as large as expected.
However, phase correlated models of the type that might occur in
connection with topological singularities can be made with
$\Omega_b h^2=0.013\pm0.001$, $0.15\la\Omega_0\la0.4$ which are either
flat ($\Omega_\Lambda=1-\Omega_0$) or open ($\Omega_\Lambda=0$) and
which satisfy all the observational constraints which we apply, including the
large baryon to total mass ratio found in the X-ray clusters. The remnant
baryon density is thus close to that obtained in the
standard picture ($\Omega_b h^2=0.0125\pm0.0025$, Walker et al 1991).
The spectral
index implied for fluctuations in the baryonic isocurvature scenario,
$-1<m<0$, is in the range expected by other arguments based on large scale
structure and microwave fluctuation constraints.
The dark matter in this picture is in the form of massive black holes.
Accretion onto them at early epochs releases high energy photons which
significantly heat and reionize the universe. 
But photodissociation does not materially change light element abundances.
A typical model gives
$\bar{y}\approx1\times10^{-5}$, $n_e/n_{\rm H}(z=30)\approx0.1$ and a
diffuse 
$\gamma$-ray background at 100 KeV near the COBE limit 
of the order of 10\% of that observed which originates from high redshift 
quasars. Reionization in this model occurs at redshift 600 and reaches
$({\rm H\,{\scriptstyle II}}/{\rm H_{tot}})\approx0.1-0.2$.