Lorimer et al. have recently reported that the spin-down age (similar
to 7 x 10(9) yr) of the millisecond pulsar PSR J1012 + 5307 is much lo
nger than the cooling age (3 x 10(8) yr) of its white dwarf companion.
A scenario where the neutron star magnetic held decays spontaneously,
and conservative accretion of all the mass lost by the secondary then
'recycles' the neutron star as a millisecond pulsar, cannot explain t
his discrepancy: the pulsar spins too rapidly at the end of the accret
ion phase to reach the current value within the white dwarf cooling ti
me. However, if the decay of the neutron star magnetic field is induce
d by the accretion process itself, spin-up and held decay occur on sim
ilar time-scales, and the neutron star spin period cannot in general r
each the usual short equilibrium value. The pulsar is thus able to acc
rete all the mass lost by the companion, reaching its current period a
nd period derivative within the white dwarf cooling time.