We have measured the Li abundance of 18 stars with -2 less than or similar
to [Fe/H] less than or similar to -1 and 6000 less than or similar to T-eff
less than or similar to 6400 K a parameter range that was poorly represent
ed in previous studies. We examine the Galactic chemical evolution (GCE) of
this element, combining these data with previous samples of turnoff stars
over the full range of halo metallicities. We find that A(Li) increases fro
m a level of similar to2.10 at [Fe/H] = -3.5 to similar to2.40 at [Fe/H] =
1.0, where A(Li) = log(10) (n(Li)/n(H)) + 1200. We compare the observations
with several GCE calculations, including existing one-zone models and a ne
w model developed in the framework of inhomogeneous evolution of the Galact
ic halo. We show that Li evolved at a constant rate relative to iron throug
hout the halo and old disk epochs but that during the formation of young di
sk stars, the production of Li relative to iron increased significantly. Th
ese observations can be understood in the context of models in which postpr
imordial Li evolution during the halo and old disk epochs is dominated by G
alactic cosmic-ray fusion and spallation reactions, with some contribution
from the nu -process in supernovae. The onset of more efficient Li producti
on (relative to iron) in the young disk coincides with the appearance of Li
from novae and asymptotic giant branch (AGB) stars. The major challenge fa
cing the models is to reconcile the mild evolution of Li during the halo an
d old disk phases with the more efficient production (relative to iron) at
[Fe/H] > -0.5. We speculate that cool-bottom processing (production) of Li
in low-mass stars may provide an important late-appearing source of Li, wit
hout attendant Fe production, that might explain the Li production in the y
oung disk.