It is interesting that we've all been writing about this question of
buffering (or not) by Fe2 silicates while having quite different rocks and
conditions in mind. I had not considered this to be an important issue in
prograde metamorphism of pelites, because in my experience they mostly
don't have magnetite (it can be more common early on in the garnet zone),
and Mike has explained this eloquently. So, while my own observations on
Connemara biotites many years ago supports Charlie's assertion that biotite
chemistry does slide continuously, it does not affect this assemblage, and
the effect on fO2 is finite but small.

One rock I was thinking of is illustrated on p.109 of my Picture Atlas of
Metamorphic Rocks (rock 106). This shows a regional pelite garnet breaking
down to develop seams of magnetite (along with andalusite etc.) in a later
contact aureole overprint. If we suppose that the reaction is 2 magnetite +
2 andalusite + 4 quarts = 2 almandine + O2 and assume an ideal mixing on
sites activity model for the garnet, with other phases pure, then it turns
out that if, as a result of this reaction, the garnet changes from Xalm =
0.7 to Xalm = 0.6, then this leads to a shift in log fO2 of just 0.4. Given
the very low value of fO2 itself, clearly huge amounts of fluid
infiltration are needed to achieve even this small shift if there are just
a few percent garnet in the rock. However, at the temperatures of the
contact overprint, Dugald correctly points out that there is probably no
intracrystalline diffusion in the garnet. Instead, the other garnet cations
go off and participate in other minerals, and the composition of the garnet
that reacts with the fluid remains constant (assuming it was not strongly
zoned initially). So it appears to work as an almost perfect buffer, even
to Howard's standards! In fact, we are talking non-equlibrium processes
here, so maybe not perfect... There is a nearby locality where the same
regional event simply experienced extensive greenshist overprinting, and a
couple of pages on in the Atlas (rock 109, p. 112) is an example of a
garnet being replaced by chloritoid and chlorite. The neat thing about this
texture is that the garnet has broken down without any intracrystalline
diffusion, while the Fe and Mg so released have distributed themselves in
an equilibrium manner between the chlorite and chloritoid products, i.e.
XMg chl > XMg gt > XMg ctd. Salutory.

Of course I've concentrated on garnet because it is much simpler to
quantify the effects of any change in composition on fO2. Biotite is much
more complex both because of faster diffusion rates and more importantly
because it can incorporate Fe3. In my experience there is a big colour
shift between biotite with magnetite and the normal biotite without. Even I
would (probably) not describe biotite as an fO2 buffer, but at the same
time a rock with a significant amount of Fe biotite has an enormous
capacity to retard any changes in fO2 that might otherwise be induced by
the infiltration of nonequilibrium fluid. In the absence of significant
infiltration, fO2 is simply not an independent variable in Fe2 silicate -
rich rocks, as there is just not the oxygen available to shift the
proportion of Fe2 to Fe3. It's interesting that Mike instinctively uses the
term buffering while referring to the possibilty of biotite limiting fO2 -
we can't get away from it even though its not strictly accurate!

Bruce


--------------------------------------------
Professor Bruce Yardley
School of Earth Sciences
University of Leeds
Leeds LS2 9JT
UK

Tel. 0113 233 5227 Fax 0113 233 5259
---------------------------------------------

GEOFLUIDS now exists! http://www.blackwell-science.com/gfl