For the Sr and Nd decay systems, the depleted mantle and continental crust appear to be complimentary and the isotope data are consistent with a simple model in which the crust has been created by extraction of partial melts from the mantle. Pb isotope ratios are not easily accommodated in such models, however. Pb isotope ratios record an increase of U/Pb in the depleted mantle, whereas extraction of partial melts should decrease in this ratio. White [1993] concluded from a study of U, Th, and Pb in MORB that the U/Pb ratio in the depleted mantle is lower than the bulk Earth value, consistent with the incompatible depletion of this reservoir. Furthermore, the low value of U/Pb he inferred is consistent with continental mass balance. The problem then apparently lies with Pb isotope ratios. White [1993] concluded that the depleted mantle is an open system, arguing depletion of the upper mantle is maintained at pproximately steady-state by a combination of extraction of partial melts and input of new material from mantle plumes.
Further evidence of early mantle depletion has been published
within the last 4 years. Collerson et al. [1991] reported a
Sm-Nd isochron for a metamorphosed ultramafic suite from Labrador that
gave an age of 3815 Ma and an initial 
of +3.0. Collerson
et al. interpreted these rocks as tectonically emplaced slices of
Archean mantle. They are slightly light rare earth-enriched, in contrast
to the positive initial 
that indicates a light
rare earth-depleted history. A second suite from the same region, which
Collerson et al. [1991] interpret as meta-komatiites, is light
rare earth-depleted yet has negative initial
(
). There has also been a preliminary report
of +4 initial 
in the 3960 Ma Acasta gneisses in the
Slave Province of Canada [ Bowring et al., 1994].
as high as +4 at 3800 Ma requires extensive light
rare earth depletion of the mantle. This in turn requires the existence
of a sizable light rare earth-enriched reservoir, perhaps
volumetrically comparable to the present continental crust.
However, pre-3800 Ma continental crust is extremely sparse. Instead
of continental crust, some have focused on basaltic oceanic crust as
the incompatible element-enriched compliment of the early Archean
upper mantle. Galer and Goldstein [1991] pointed out,
however, that the large degrees of melting associated with modern
oceanic crust production do not sufficiently fractionate Sm and Nd
to produce the depletion observed in the early Archean mantle.
They suggested thick, alkalic oceanic crust was created by smaller
degrees of melting in a regime where no seafloor spreading occurred.
A third possible explanation for the high 
in
early Archean rocks is that they are simply wrong, and there is
increasing skepticism about their validity, e.g., Carlson
[1994]. While the actual analytical data are not questioned, present
day 
Nd/
Nd and 
Sm/
Nd, not
initial 
, are the measured quantities. Calculation
of initial 
from these data assumes the samples in
question have been closed for their entire history and that the age
is accurately known. The suspicion is that subsequent metamorphism
has changed the Sm/Nd ratios. Nearly all early Archean rocks with
high initial 
have undergone subsequent metamorphism and
the high initial 
represent only one 
of
a spectrum of initial 
values. A closely
related controversy is the question of 
Nd anomalies in early
Archean rocks. 
Nd is produced by 
-decay of 
Sm,
an extinct radionuclide with a half-life of 103 Ma. A
Sm/
Sm ratio of 
in the early Solar
System has been established from meteoritic studies. If mantle
depletion began before about 4.3 Ga, then excesses in 
Nd due
to 
Sm decay should be observed in those samples with high
positive initial 
. Harper and Jacobsen
[1992] reported an average excess 
Nd of 
ppm for
six replicates of an Isua, West Greenland sample that has an
initial 
of +4 at 3.81 Ga. This would appear to confirm
that depletion of the upper mantle got underway very early.
However, excess 
Nd in other early Archean samples have not
been found, nor has the Isua analysis been replicated in
another laboratory. As a result, the existence of 
Nd
excesses remains controversial.