We present the results of an extensive search, conducted at the Canada
-France-Hawaii 3.6-m telescope, for beryllium (Be) in the atmospheres
of lithium-deficient F and G dwarfs. We also report revised lithium (L
i) estimates for the entire sample using previously published equivale
nt widths and updated, consistently calculated stellar parameters. Abu
ndances derived from an LTE analysis of the Li and Be line-forming reg
ions confirm the suspicion that F stars which deplete Li by factors of
10-200 may also be beryllium deficient. Photospheric Be concentration
s range from near meteoritic levels in G dwarfs to factors of 10-100 b
elow this assumed initial abundance in hotter stars. Moreover, signifi
cant Be deficiencies appear in stars that populate a 600 K wide effect
ive temperature window centered on 6500 K. This Be abundance gap is re
miniscent of the Li gap observed in open clusters. Also, the discovery
of 12 probable ''110 Herculis'' stars, objects that exhibit a deplete
d, but detected, surface concentration of both Li and Be, provides a p
owerful means of differentiating between the possible physical process
es responsible for observed light element abundance patterns. Indeed,
the Be data presented here, in conjunction with the newly calculated L
i abundances, lead to the following conclusions regarding the hypothes
ized, light element depletion scenarios: Mass loss cannot account for
stars with severely depleted (but detected) Li and moderate Be deficie
ncies. The predicted timescales for surface depletion due to microscop
ic diffusion are too long for significant Li and Be deficiencies to de
velop in cool (T-eff less than or equal to 6200) stars; nevertheless,
underabundances are observed in these stars. Diffusion theory also pre
dicts Li and Be depletion rates to be comparable, but it is evident th
at Li and Be depletion proceed at different speeds. Models of mixing i
nduced by internal gravity waves cannot explain mild Be deficiencies i
n cool dwarfs. A key meridional circulation prediction regarding the e
fficiency and severity of Li and Be dilution is shown to be fallible.
However, rotationally induced mixing, a turbulent blending of material
beneath the surface convection zone due to the onset of instabilities
from superficial angular momentum loss, predicts both the observed li
ght element depletion morphology as well as the existence of 110 Her a
nalogs. These ''Yale'' mixing models provide, therefore, the most plau
sible explanation, of those presented, for the observed Li and Be abun
dances.