Best Recording 2010
Dunedin Consort & Players
John Butt - Director
Susan Hamilton - soprano 1
Cecilia Osmond - soprano 2
Margot Oitzinger - alto
Thomas Hobbs - tenor
Matthew Brook - bass
Mass in B minor BWV 232 - Johann Sebastian Bach
Bach's Mass in B Minor is undoubtedly his most spectacular choral work. Its combination of sizzling choruses and solo numbers covering the gamut of late-Baroque vocal expression render it one of the most joyous musical experiences in the western tradition. Nevertheless, its identity is teased by countless contradictions: it appears to cover the entire Ordinary of the Catholic Liturgy, but in Bach's Lutheran environment the complete Latin text was seldom sung as a whole; it seems to have the characteristics of a unified work, yet its origins are perhaps the most diverse for any of Bach's large scale compositions; it was written in an age when composers generally prepared music for specific occasions, yet we have no firm evidence that the whole work was designed with a performance in mind. Somehow, a mystique grew around the Mass soon after Bach's death, and C.P.E. Bach performed the Credo section during the 1780s; but it was nearly a century before it was available in print. The first performances in the early decades of the nineteenth century were presented by institutions of which Bach could hardly have conceived - amateur choral societies with a vast number of performers. And, over the last century it has often been at the centre of major disputes in the field of Bach scholarship: the question of its original function, its chronology, the legitimacy of the various manuscripts and, of course, its performance practice.
Even the title ‘Mass in B Minor' was not applied until the nineteenth century. Bach's autograph contains four discrete sections: the Kyrie and Gloria are together entitled Missa, these movements being the regular part of the sung Lutheran mass of Bach's time; the second section is called Symbolum Nicenum - the Nicene Creed. Then follows the Sanctus - again an independent manuscript (a slightly modified version of a pre-existing setting); the fourth section contains the remaining texts of the Mass, ‘Benedictus' to ‘Dona nobis pacem'. The fact that Bach gave each of these sections separate folders and title pages suggests that if the work were ever performed it would most likely not have been in a single sitting. On the other hand, there are obvious musical coherences suggesting that, in some sense at least, Bach viewed the work as a musical whole. Perhaps he conceived it along the lines of keyboard collections such as the Well-Tempered Clavier, which do not necessarily have to be performed as a whole yet show an obvious overall plan (equally analogous is the Christmas Oratorio, sung on six separate occasions during the Christmas season).
The Sanctus was first performed as an independent work on Christmas Day 1724 and Bach completed the Missa (i.e. Kyrie and Gloria) in 1733, while he was seeking an honorary title from the Elector of Saxony in Dresden; this would have elevated his status back in Leipzig. He took the opportunity occasioned by his son Wilhelm Friedemann's appointment as organist at the Sophienkirche in Dresden to travel with several family members and present his petition to the Elector in person. He included a beautifully presented set of performing parts as an example of his music, namely the Missa. When measured against some of the music sung in the Catholic liturgy at the Dresden court, Bach's music is not immoderately proportioned; indeed there are several factors - virtuoso horn writing, florid vocal lines, musical similarities with some of the works sung in Dresden - to suggest that Bach tailored the work to the capabilities and demands of the Dresden musicians.
Bach reused some of the Gloria in Cantata 191, c.1745 - it may well have been this performance (possibly for the Peace of Dresden on Christmas Day), also including a repeat of the Sanctus, that gave Bach the idea of setting the remaining texts of the Latin Ordinary - The Creed, Sanctus/Benedictus and Agnus Dei. The handwriting of the latter movements shows that the composer was severely hampered by physical problems during the last year or so of his life. We still know of no reason for Bach's final compilation; possibly he intended it - like the Missa - for the court at Dresden, since similar forces are required. Possibly there were events in Leipzig that demanded this sumptuous music (certainly Bach had performed Latin settings of the Creed during the 1740s); some have suggested that it was commissioned by a distant patron. Other reasons - such as Bach's desire to write a sort of personal memorial, demonstrating his lifelong achievement in modern and historical styles - we can only guess. Quite possibly there was a combination of motives, some practical, some speculative, that led Bach to complete this project.
But despite all these questions, and the warts and wrinkles in the surviving manuscripts, the Mass in B Minor has somehow transcended the murky conditions of its origins. Bach seems purposely to have compiled some of his choice choral pieces to fit into the larger context of the full Mass. Some might balk at the fact that so much of the piece was taken from earlier works: the ‘Gratias' from the lost model for a chorus also used in Cantata 29; the ‘Qui tollis' from Cantata 47; parts of the Creed from Cantatas 12, 120 and 170; the ‘Osanna' from Cantata 215 (a secular cantata), the ‘Agnus Dei' from the lost model for an aria that is also used in the Ascension Oratorio. The evidence of the autograph score suggests that many of the other movements are parodies too - although in these cases the originals are entirely lost. Indeed only certain sections (and only the ‘Confiteor' as a whole) show signs of fresh composition, and were probably the last things that Bach wrote. In his time there was no shame in reusing earlier music. It was the actual use that was important - whether the music was suitable for the new context, whether it was skillfully reworked. Indeed, perhaps part of the enduring quality of the Mass lies in the fact that so much of its music was essentially ‘composed twice'.
Bach achieved tremendous variety and, almost paradoxically, a sense of unity in the complete Mass - complementary qualities that became central to the aesthetic judgment of art over the coming century. The historical styles range from Renaissance-style textures (some with plainsong cantus firmus) to those current in the eighteenth century, such as the Italianate concerto style. But many idioms are unusual in traditional sacred genres, particularly those with dance-like allusions: e.g. ‘Qui sedes', a sort of Gigue; ‘Quoniam', a Polonaise; ‘Et resurrexit', a Réjouissance. Even the expressive ‘Crucifixus' alludes to the Passacaglia. It seems that Bach often sought to unite the sacred with the best that the secular world could offer - a sort of sublimation of religion within art that was soon to resonate with Romantic aesthetics.
Bach worked assiduously to integrate the existing music within the new setting, often lopping off sections (the music for the ‘Osanna' and ‘Et expecto' originally began with an instrumental ritornello) or adding new lines. He also often paired movements from disparate sources and adapted them to match each other in length - the ‘Quoniam' is carefully pruned of its final part (presumably a da capo in the original) so that its length works in direct proportion to the succeeding ‘Cum sancto spiritu'. Then there are musical coherences: the return of the music for the ‘Gratias' for the ‘Dona nobis'; the ‘Osanna', which contains motives relating it to the ‘Sanctus'; the ‘Agnus Dei', which recalls the opening ‘Kyrie' in affect and melodic gestures. None of these pairings would have been envisioned when the music concerned was first composed. There are also several symmetries in the key structure of the whole piece, which suggest that Bach sought a form of musical coherence working beyond the textual divisions.
All in all, then, it seems that Bach fortuitously anticipated the values of later ages - creating something of a symphonic sense of cohesion that was hardly required in his own time. We may sense that Bach was aiming to sum up everything that music could offer, of pushing the language he knew to its limits. The Mass - of all the music he left - survives as a dense but miraculously clear musical nexus, one which has shown surprising resilience in a variety of cultural and historical environments.
Bach's final version, 1748-50, and issues of performance practice
The Mass was, in the early 1980s, the key work that Joshua Rifkin used to demonstrate his revolutionary theory that the choruses in Bach's vocal works were generally performed with only the principal singer (‘concertist') on each part. Rifkin's recording of the complete Mass produced a cumulative effect in vocal scoring: the four- or five-part vocal scoring of the Kyrie, Gloria and Creed leads to the six-part scoring of the Sanctus, to the double-choir scoring of the ‘Osanna', and finally to the doubled four-part scoring of the closing ‘Dona nobis pacem'.
Some twenty-five years later it was Rifkin again who provided the first fully scholarly edition of the Mass as a complete setting of the Roman ordinary, the work as Bach left it on his death in 1750. It is this edition, published by Breitkopf & Härtel (2006), which is recorded here for the first time. Two issues in particular distinguish this from any previous edition: first, Rifkin has removed certain ‘improvements' that crept into the score after 1750 (most by C.P.E. Bach, particularly in preparation for his own performance of the Creed). Secondly, Rifkin took account of the fact that Bach had not seen the Dresden Missa parts since 1733, so that the various refinements and alterations he made in them never made it into his own score. Moreover, Bach made other revisions to this score and arranged parts of the Gloria for the independent Cantata BWV 191. In all, Rifkin argues that the work as finished just before Bach's death is essentially a different entity from the 1733 Missa, and that a combination of the ‘best' readings from both does not really correspond to Bach's final (and virtually completed) conception of the work. Many of the numerous differences between this final version and that presented in all previous editions are not likely to be heard in casual listening; but noticeable surprises occur in the soprano line of the ‘Crucifixus' and the bass line of the ‘Et in spiritum sanctum', for instance. Two flutes (rather than the single one for the Dresden Missa) are used in the ‘Domine Deus', and the bassoons, which have a striking obbligato with horn in the ‘Quoniam', are employed nowhere else, thus making this movement stand out all the more for its unique sonority.
If we accept that the complete Mass in B Minor is a specific text with its own integrity, we still have to acknowledge that there are many uncertainties concerning how, and whether, this might have been performed in a Leipzig context. Obvious differences with the Dresden version might include the addition of dou-bling parts for both violin lines in Leipzig practice and also the use of a second string bass instrument (usually violone). What, then, of the vocal scoring?
As Rifkin and, later, Andrew Parrott have both exhaustively demonstrated, the number of sources showing where Bach may have employed ripienists in his church music accounts for barely 10% of the total. Extra singers often seem connected with larger works on major feasts, such as the John Passion, in which they are employed throughout, and the Matthew Passion, in which the second choir both performs a ripieno function and sings as an independent choir. The complete Mass might come into this sort of category, particularly since it requires eight singers for the ‘Osanna' (something not anticipated when Bach wrote the Missa in 1733). Bach indicates that these sing together in the final ‘Dona nobis', thus resulting in two voices to a part. This immediately raises the question of whether the same music, heard earlier in the work as the ‘Gratias', should also be sung in this manner. As it stood in 1733, Bach indicated that the two sopranos together sing the top line, thus suggesting that this line was sung with doubled voices, the others without. But in 1731 this music (part of the town-council cantata BWV 29) had been furnished with ripienists in all parts. So Bach countenanced the same piece of music being sung with 8 voices in 1731, 5 in 1733 and 8 again in 1750 (in the ‘Dona nobis' at least). In the Mass it is striking that Bach made a distinction between the use of both sopranos together and one alone in the movements with only four voices. Therefore, in the ‘Kyrie' II and ‘Gratias' (in both 1733 and 1750 versions) Bach indicated that both sopranos sing, while in the ‘Qui tollis' only Soprano 2 is indicated. Likewise, in the Creed, the ‘Crucifixus' is assigned only to Soprano 2 while both sing in the four-part ‘Patrem omnipotentem'. So far, then, there is evidence that Bach countenanced a ‘doubled' sound (if only for the sopranos) in the movements with trumpets (or in stile antico ones with instrumental doubling) and kept to single voices for the gentler, more expressive numbers.
As Janice Stockigt has shown, some ripienists were used as a matter of course in Dresden, and if Bach's offering were ever to have been performed there, a new set of parts would have been prepared for the main singers (with castrati on the upper parts), doubled at strategic points by ripienists. Bach in Leipzig also tended to employ ripienists strategically (the exception being the John Passion, where they double the main parts throughout). A possible model for the concerto-style movements of the Mass is provided by the score of the final movement of BWV 191, ‘Sicut erat in principio', which is an arrangement of ‘Cum sancto spiritu' from the Mass. Here a wavy line at the bottom of the score (a device Bach used for BWV 71, in 1708, to indicate ripieno participation) corresponds - at least in the main - with the sort of ripieno scoring that Bach employs elsewhere, most significantly in Cantata 195, in the version which Bach performed around 1748/9. This includes ‘call and response' scoring, in which the ripienists are added at the points where the full orchestra (including trumpets) responds to the opening entries. The indications are absent when there are fewer vocal parts, particularly in fugal expositions; they return with the later entries.
The combined evidence of this, the history of the ‘Gratias' music, the selective doubling of the soprano parts in the Mass as a whole, and the tendency to add more singers in larger pieces (and for the major feasts of the church year), suggests that it is not inappropriate to propose a hypothetical ‘late Leipzig' (or indeed ‘Dresden') ripieno scoring for the complete Mass in B Minor. Ripienists might be added in motet-like textures that are doubled by instruments (‘Kyrie' II, ‘Gratias', ‘Dona nobis pacem') and, following the recent models provided by BWV 191 and 195, in the concerted movements with trumpets. The doubled expositions in the opening ‘Kyrie', in which instrumental doubling is avoided for the initial entries, might mean that ripienists could be added for the later entries. In keeping with Bach's explicit directions for using only one soprano part in ‘Qui tollis' and ‘Crucifixus', the gentler, quieter movements would be sung with single voices. What remain are the stile antico movements, ‘Credo in unum Deo' and ‘Confiteor'. While the motet-like texture might imply the doubled voices of ‘Kyrie' I and ‘Gratias', the relatively unusual lack of instrumental doubling seems more consistent with single voices (the only counter-example is the John Passion, where the first and last choruses contain passages of doubled vocal lines without instrumental support).
Scoring the entire Mass along these lines (a solution similar to that adopted by Andrew Parrott in his own 1985 recording, although his recent writing is more skeptical of ripieno involvement) gives the work a variety of scorings from 1 to 10 voices (with a maximum of 8 separate parts), which corresponds closely to that of the Matthew Passion (with its 8 main voices and the ripieno soprano(s), added to two choruses). If our contemporary experiments are anything to go by, the difference between single vocal lines and doubled ones is not as great as many might expect; the difference is more in sonority than volume (and, given the history of the ‘Gratias' music, Bach was perhaps more casual over the issue than many scholars - on either side of the debate - might assume). What single-voice performance does enable, though, is a level of clarity and soloistic expression that is not traditionally a feature of modern choral performance. With this as a model, the sporadic doubled textures could be heard positively as an enhancement of solo performance rather than negatively, as the impoverishment of an assumed monumental sonority.
One further area of discussion relating to the performance of the Mass as a whole relates to whether Bach designed it with some degree of connection between movements, and whether any are related in terms of tempo. Certainly, the score contains the indications ‘segue' or ‘sequitur' at several points (e.g. between the movements of the Kyrie section, between the later movements of the Gloria, ‘Qui tollis' to the ‘Quoniam' and between the central movements of the Creed). Don Franklin has suggested in a study of the Missa of 1733 that Bach's system of tempo relationship might share something with the proportional system of the Renaissance era, although differing from this in significant ways. According to Bach's pupil, Johann Philipp Kirnberger, tempo should be based primarily on the choice of time signature and the notational values used. Each signature relates to a ‘normal tempo' (tempo-giusto) as held by its principal beat, and this is modified by the predominance of shorter or longer divisions (with more shorter divisions it would thus be slower, with longer notes it would be faster). This rule of thumb is then further inflected by Italian words, as necessary, which modify what might have been expected from the time signature and predominant note values. Franklin develops this theory by observing that Bach's use of the fermata might serve to cancel a prevailing tempo applying to several movements in succession. Thus the lack of a fermata (together with the ‘segue' signs) could imply a significant relationship, such as a doubling of the time between the ‘Christe' and ‘Kyrie' II (following the traditional 1:2 relationship suggested by the time signatures) or a consistency of beat between the ‘Domine Deus' and the ‘Qui tollis'. Such relationships might sometimes be substantiated by the number of beats they generate in the corresponding sections: thus the ‘Gloria in excelsis Deo' and the ‘Et in terra' would relate to one another 1:2, in terms of length, if the crotchet of the first (i.e. the hemiola of the 3/8 metre) becomes the crotchet of the second; the same relationship could apply in reverse for the ‘Sanctus' to ‘Pleni sunt coeli', generating two halves of roughly the same length.
It is possible to envisage that long sequences of movements could be related in terms of tempo, through a common or relational beat (e.g. from the ‘Domine Deus' through to the end of the Gloria section, or throughout much of the Creed). Following the pattern in the ‘Gloria in excelsis Deo' and the ‘Sanctus'-‘Pleni sunt coeli', the hemiola of compound times could relate roughly to the beat of simple times (i.e. compound times would have a beat that is a quaver longer than that of simple times). But it has to be acknowledged that any such system is only loosely connected to the very patchy and contradictory historical evidence. There is little proof that Bach ever had a fully rationalized system of tempo relations, even if he may have experimented in various ways. Nevertheless, the idea that some such experimentation might be applied to the Mass can provide the starting point for interpretation if it contributes to a sense of coherence and continuity, something that the work as a whole might seem to demand. It is hardly likely to be very productive as an end in itself.
My warmest thanks are due to Joshua Rifkin, not only for his edition, but also for his ever lively discussion of many of the issues raised above.
© John Butt, 2010