"The facies concept is robust also because it contains no implication of P,T or other physical conditions. It is simply an association of rocks or mineral assemblages, from which we infer physical conditions."
If I understand this correctly: Facies has no implications of P or T. Yet, from facies (= an association of ...) we infer P and T (i.e., physical conditions). I have a problem with logic here. It may have been that at the time the facies concept had been devised, P and T could not be quantified so easily, but this has changed. Just about every modern textbook in Metamorphic Petrology has a facies diagram which is a quantitative P-T diagram (not that they would look the same, but still...).
I cite from Eric's mail to put the nail in:
"How often amphibolites form in the eclogite facies would depend on how low the P-T field is drawn for the eclogite facies boundary. If drawn down to 10-12 kbar at 600-700°C, amphibolites should be common. I think that the boundary should be raised by 4-5 kbar."
I hope I don't tread on Eric's toes if I use his statement to say that: facies (in a modern sense) can be delineated by P-T boundaries, broad and poorly defined as they may be here and there. It is just a matter to decide what reactions in what type of (common) rocks are best suited to define any particular boundary/transition. Eskola and subsequent workers must be credited for what they have achieved. There is no question about it However, if I want a sensible definition of blueschist facies versus adjoining facies in a P-T diagram, I would pull out Bernard Evans paper on blueschists, for example.
Some of the ongoing discussion is a bit too abstract for my taste. To me, the facies concept is a convenient subdivision of the P-T plane, which helps us to communicate (perhaps not always, it seems) about metamorphic conditions in a crude way. It is no substitute for thermobarometry and doesn't tell us much about the dynamic aspects of metamorphism, such as P-T trajectories. The facies concept had been developed when quantification of P-T-X relations was not the standard procedure it is today. Whatever the historical background is, we need facies for mostly practical purposes and not much else.
School of Geological & Computer Sciences
University of Natal