We construct a magnetic helicity conserving dynamo theory that incorporates
a calculated magnetic helicity current. In this model the fluid helicity p
lays a small role in large-scale magnetic field generation. Instead, the dy
namo process is dominated by a new quantity, derived from asymmetries in th
e second derivative of the velocity correlation function, closely related t
o the "twist and fold" dynamo model. The turbulent damping term is, as expe
cted, almost unchanged. Numerical simulations with a spatially constant flu
id helicity and vanishing resistivity are not expected to generate large-sc
ale fields in equipartition with the turbulent energy density. The prospect
s for driving a fast dynamo under these circumstances are uncertain, but if
it is possible, then the field must be largely force-free. On the other ha
nd, there is an efficient analog to the alpha-Omega dynamo. Systems whose t
urbulence is driven by some anisotropic local instability in a shearing flo
w, like real stars and accretion disks, and some computer simulations may s
uccessfully drive the generation of strong large-scale magnetic fields, pro
vided that partial derivative (r)Omega[partial derivative (theta)v(z)omega
(theta)] > 0. We show that this criterion is usually satisfied. Such dynamo
s will include a persistent, spatially coherent vertical magnetic helicity
current with the same sign as -partial derivative (r)Omega, that is, positi
ve for an accretion disk and negative for the Sun. We comment on the role o
f random magnetic helicity currents in storing turbulent energy in a disord
ered magnetic field, which will generate an equipartition, disordered field
in a turbulent medium, and also a declining long-wavelength tail to the po
wer spectrum. As a result, calculations of the Galactic "seed" field are la
rgely irrelevant.