As far as stamina goes there's also the rather important point that we're not talking about singing on top of an orchestra but on top of a chamber group of what by today's standards were incredibly quiet instruments. Koopman's argument that you would need a Pavarotti to hold down the tenor line of the St Matthew misses that point completely...

Am I right in thinking that the B Minor Mass was never performed in Bach's day and that he wrote it for his personal satisfaction?  I understand Lutherans only used the first two sections of the common of the mass (not that there is anything for them to object to theologically in the rest.)

Well, I'll go along with that, if a little reluctantly...

... I suppose one theory which we mustn't resist is that this hypothetical treble might have been (at different times) CPE Bach, JC Bach, WF Bach etc...

I've been trying to find some kind of indication as to what Bach performed during Lent. If he didn't have much or any concerted music during Lent then of course that gave him weeks to rehearse nothing really tricky except the Passion setting... haven't found anything just yet though.

It seems in fact that the ripieno parts for the St John may have been used at the first performance but sometimes not at later ones: they don't contain the alterations he made for later performances. So I have it the wrong way around up there where I suggested he used solo voices first and added ripienists later.

Incidentally there's a rather telling titbit in the surviving ripieno parts. There's a part for a late performance which contains the music for Pilate and Peter. It specifies tacet for all the choruses contained within it. So on an occasion when Bach had another bass at his disposal he didn't use him to reinforce the chorus.

The Kyrie and Gloria of the Mass in B minor were sent to Dresden in 1733 by the way: Rifkin notes that there were only 11 singers in the Dresden Hofkapelle at the time (3S, 4A, 2T, 2B). Both movements are in 5 parts which at most they could have doubled - even that would have required an alto to sing soprano.

One particularly important point Rifkin makes in any case is that if Bach used one voice per part as the basic setup for his concertante vocal music he was only doing what everyone else in Germany at the time and for at least a century before had been doing.

This is not entirely true - there were a few that date from the Weimar period (one of which was expanded at Leipzig), written for Dominica Oculi (3rd Sunday in Lent) and Palm Sunday.

A useful list of the works concerned (together with their specifications) can be found (inter alia) in a useful liturgical listing HERE

Baz

That was more a Sudden speculation than a Rifkin/Parrott argument I think (at least I'll continue to think so until I find a bit in the book that actually says it). It's certainly Bach's Leipzig not Weimar practice that is directly relevant to the Passions. (The question of Bach's pitch level in Weimar is another thorny one...)

("Quasimodogeniti"? Did someone make that up?)

It does however seem to be at least the case that the Passion setting was typically performed by both Bach's first and second choirs - with the first choir performing all choruses and arias and the second joining them only for the choruses (and on at least one occasion not even that).

I've read Parrott's book and, as a non-specialist, find his arguments quite convincing. Indeed I have yet to read anything which attempts to rebuff them. But has there been such an attempt? If so I think I really ought to take a look at that too.

I haven't yet heard a recorded example of one of the large-scale vocal works using OVPP which has persuaded me that it has yet been done convincingly in modern times, whatever Bach did or would like to have done. Clearly the soprano and alto parts aren't going to sound anything like they "ought to", but in general I don't see how we can have an accurate idea of what any of Bach's vocalists sounded like, what for example the expressive approach to the recitatives would have been (or ought to have been), what kind of pronunciation was used (I sometimes find even a 21st century Saxon accent pretty hard going to understand, and the difference between this and what is now called "Hochdeutsch" would presumably have been even greater then) and so on. All of these are important issues too, which bear crucially on the sound of the music. In general I wonder why "period pronunciation" isn't taken more seriously in German-texted "early music".

Reiner - goodness me. 

I'm led to believe that the Protestant German countries went Gregorian in 1700, for what it's worth. No idea when they traditionally celebrated New Year - although Bach's New Year cantatas seem to have been performed on Jan 1 and his Annunciation pieces on March 25...

richard -

Pronunciation - yes. A can (as opposed to a diet) of worms. I'm also surprised that German groups haven't made more of an effort to work at historical pronunciation (goodness knows singers from outside Germany have enough trouble just getting their German to sound German...). Perhaps it's something to do with the fact that the English early music movement was a little more closely connected to universities where scholars of early pronunciation might be close at hand. But that's more rash speculation.

Vocal approach - we don't have any idea of how any vocalists sounded back then, of course. Everything I hear or read tells me that everything must have sounded much lighter, both voices and instruments (players of Baroque trumpets are still using the intonation holes in general; that doesn't just affect the intonation but affects the whole dynamic level in allowing the instrument to be played louder than it would be possible for a really historical trumpet to bend the notes). That means even with singers under modern conditions (of education and of physical development) things like vocal register work quite differently (which to me best explains why Bach's tenor parts were practicable for his students then whereas nowadays really good Bach tenors are in the hen's teeth category - I'm certain Bach's tenors used head voice quite freely). On the other hand if there are now effective performances in consort setup of Schütz and Monteverdi I don't see why it shouldn't be possible with Bach.

I do see why it is taking a while to work, though. The singers who specialise in Bach nowadays are still primarily in the world of choir + soloists, which whether the choir is 40 or 12 means a substantially different approach. And even in talking about singing one voice per part we often slip into phrasings such as 'Bach's cantatas performed by soloists', which still has a subtle bias to it. I do think it can only work if the performing setup is seen as a single ensemble of voices and instruments. That's certainly how they're written, with every line developed to a similar degree, and if anything becomes too overtly 'soloistic' or 'accompanimental' then for me the concept on which the musical fabric is based starts to get lost.

There are a lot of problems with the Rifkin Decca recordings but for me it's moments of those (the opening choruses of BWV 140 and BWV 78) that best hint at what might be possible. And the Kuijken recordings seem to be getting closer.