The primeval baryon isocurvature (PBI) model for the origin of cosmolo
gical structure is explored with the aid of detailed numerical simulat
ions. in this model we assume there is no exotic dark matter and that
we live in an open universe with baryonic content initially in the ran
ge OMEGA0 = 0.1-0.2. The amplitude of the primeval entropy fluctuation
s is normalized to the COBE measurements, and the spectrum of the entr
opy fluctuations is taken to be a power law with index in the range m
= -0.5 to 0.0. We use H-0 = 80 km s-1 Mpc-1. Shortly after decoupling
in this model, a large fraction of the mass must condense to a dark co
llisionless component such as low-mass stars or massive black holes. W
e follow the two-component (gas+collisionless) mixture with a hydro+pa
rticle-mesh (PM) code in a (64 h-1 Mpc)3 box containing 128(3) = 10(6.
3) cells and particles, and a full nonequilibrium treatment of radiati
on, ionization, heating, and cooling. Additional large PM simulations
are made in a box with size 400 h-1 Mpc containing 250(3) = 10(7.2) pa
rticles. The PBI model has more primeval density fluctuation power at
both long and short wavelengths (and less at inter-mediate wavelengths
) than the standard cold dark matter (CDM) model, and it has a peak at
approximately 600 h-2 Mpc, the Jeans mass prior to decoupling. The po
wer spectrum, when convolved with the temperature history of the gas,
allows gravitational growth of structure at three characteristic epoch
s: at 1200 > z > 250 the collapse of mass concentrations at DELTAM alm
ost-equal-to 10(5)-10(8) M. produces the assumed dark collisionless ba
ryonic component; at 40 > z > 6 galaxy spheroids form at masses in the
range DELTAM almost-equal-to 10(10.5)-10(12) M.; and at z < 1 large-s
cale structures form at AM greater than or similar to 10(12.5) M.. For
our parameters, 10%-15% of the residual baryons collapse to form gala
xies at 10 greater-than-or-equal-to z greater-than-or-equal-to 5, givi
ng an acceptable mean luminosity density for M(stellar)/L(B) = 3-4 (so
lar units), and an acceptable galaxy mass function for M(tot)/L(B) alm
ost-equal-to 200. The position correlation function of galaxies in the
model is acceptable with a modest bias of galaxy candidates over dark
matter (b = 1.1). The abundance of and correlation function among ric
h clusters also are acceptable, with modest amounts (10%) of merging c
ontinuing at low redshift (z < 0.3). Principal differences between thi
s model and the standard OMEGA = 1 cold dark matter model are (1) the
small-scale relative velocity field is much lower (300-350 km s-1 vers
us 800-1000 km s-1 at 5 h-1 Mpc), reflecting a systematically smaller
value of the true dynamical density parameter; (2) the large-scale coh
erence length of the peculiar velocity field is considerably greater (
bulk flow in 100 h-1 Mpc sphere of 500 km s-1 versus approximately 200
km s-1 in CDM); (3) early ionization makes it much more likely that t
he thermal cosmic background radiation has been scattered well after d
ecoupling, considerably reducing cosmic background radiation fluctuati
ons on scales smaller than about 2-degrees; and (4) galaxies and clust
ers of galaxies are assembled at much higher redshifts. The central pr
oblem for the PBI model is the dynamical evidence from large-scale pec
uliar motions that the density parameter may be close to unity. If thi
s is so, the PBI model is uninteresting. In all other cases we can see
observational advantages for PBI over CDM and other OMEGA = 1 models.