Thirty Years of HIP

by Bernard D. Sherman

reprinted from Goldberg, November, 1999 (updated August 2004)

Bernard D. Sherman is the author of Inside Early Music and of many articles.









Since 1968, when Nikolaus Harnoncourt led the first period-instruments recording of the Mass in B Minor, no fewer than 21 historically inclined maestros have followed suit. If the early-music movement has a central work, it might well be this one.

You could explain away this preoccupation by referring to the status of the B Minor Mass in the standard repertory. But few music lovers would deny that its status reflects its merits. In 1818 a would-be publisher advertised the B Minor Mass as "the greatest musical artwork of all times and people"; and while such hyperbole and ethnocentrism have gone out of style, the Mass has not. It has easily withstood a century and a half of changing attitudes and probing scholarship. Nineteenth-century devotees might have been disillusioned had they known that Bach assembled most of the Mass by re-crafting earlier works (partly in 1733, and finally in 1748-49), but modern enthusiasts have learned as much and do not mind at all.

The Mass has also withstood widely varied performance styles. Indeed, the 21 "historically informed performers" (HIP) who have recorded their B Minors join an even larger number of "mainstream" conductors in the record catalog.

In the process, the HIP conductors have changed the way people tend to hear the work. A new listener wandering into a record store is more likely to be steered to Gardiner than to Karajan. If anything, it's the period-instruments performances that have become the mainstream in Bach.

It would be a mistake, however, to assume that the 20-plus early-music recordings demonstrate a monolithic new orthodoxy about how Bach should sound. True, the recordings differ from their predecessors in some common ways, notably the smaller size of the ensembles, faster tempos in certain movements, and the un-Wagnerian style of the singers. But the recordings show more variety than critics sometimes assume. The variety results at least in part from disagreements over how Bach would have performed the Mass. On top of that, even when the historical evidence is clear, performers differ over how, and to what degree, an artist should use it.

This article will not give a history or analysis of the work - excellent ones are readily available in books and on the World Wide Web (see insert, "For further reading"). Instead, it will look at some of the specific problems a performer of the Mass must grapple with.

Tortoise and Hare

Half a century ago the conductor Sir Thomas Beecham, at the beginning of a ballet rehearsal, looked over at the dancers and asked, "What shall it be today? Too fast or too slow?"

Some critics might apply the question to tempo fashions in Bach's choral music. The pendulum in this repertory has swung from the tortoise-like Otto Klemperer to the hare-like Frans Brueggen, and while these maestros represent the extremes, they personify a trend. The fastest of the mainstream B Minor Mass recordings, that of Sir Georg Solti, is on average slower than the slowest of the HIP recordings, that of Gustav Leonhardt.

Are the faster HIP tempos closer to Bach's own, or do they reflect nothing more than a change in taste? Or are they a reaction against the monumentality of some of the previous interpretations?

Some people have argued that using period instruments and smaller choruses naturally gives rise to faster speeds. A survey of recordings suggests otherwise. Andrew Parrott, using only two singers per part, takes the Kyrie 2 and Gratias (both marked "Alla breve" and referring to older polyphonic styles) as slowly as Klemperer. The same is true of Leonhardt in another "stile antico" chorus, the Credo; René Jacobs takes it and the Kyrie 2 even more slowly than the old man. Thomas Hengelbrock unfolds the first Kyrie at quarter=44, slightly slower than Karajan. Conversely, huge modern choruses and orchestras don't necessarily put a brake on fast tempos. In Vivace choruses like the "Cum sancto spiritu" and "Et expecto," Solti and Karajan run neck and neck with the period-instrument group. Instruments and ensemble size, then, tell us little about Bach's tempos.

How about period treatises on music making? Alas, tempo is hard to specify in words - just how fast is "fast"? Such difficulties add to another limitation of treatises: that they are written for beginners, and are not meant to capture the quirks of a master like Bach. Still, we can learn a great deal from a thorough survey of the literature.

One principle that emerges is that the time signature a German Baroque composer chose - 3/2 versus 3/8, and so on - conveyed central information about tempo. Studying German texts from Bach's era, and examining his scores, can yield at least some insight into the tempos implied by his time signatures. (Click here for a long article discussing this and other issues of tempo notation in Bach.)

To apply this principle to the B Minor Mass, consider the Sanctus. A recent study by the Bach authority Robert Marshall suggests that when (as here) Bach notated a piece in common time with triplets he was implying a tempo on the fast side. (Common time was notated with the signature C, or what we would call 4/4.) Bach could instead have notated the Sanctus in 12/8, but Marshall argues that the 12/8 signature would have implied a slower speed. Moreover, an earlier version of the Sanctus is notated mostly in , indicating an even faster tempo. If these arguments are right, conductors who have taken this chorus at a quick pace - Harnoncourt, Rifkin, Brueggen, Thomas, Junghaenel, and Christophers - have more historical support than those who take it slowly, such as Koopman, Gardiner, and the majority of their colleagues.

Why do only a few performers read the Sanctus as suggested above? One reason may be one of those disagreements I mentioned at the outset. Many authorities, such as the late Robert Donington, have denied that Baroque time signatures implied much about tempo. These authorities made the same denial with respect to Italian tempo words, like "Adagio" or "Allegro" (and Bach left more tempo words for the B Minor Mass than is typical in his vocal works). But more recently, a number of authorities have come to the opposite conclusion about both issues.

Of course, they don't all agree on what each signature or word meant. An example of a contested tempo word is "Andante," which Bach affixed to the "Et in unum" duet. Some prominent musicologists have claimed that this word denoted something about execution but nothing at all about tempo. Yet consider the relevant German sources - F.E. Niedt (who appears to have studied with Bach's older cousin and whose treatise may have been used by Bach students), Martin Fuhrmann, Johann Mattheson, Bach's cousin and associate J. G. Walther, and Bach's student J. P. Kirnberger. They are unanimous in suggesting that "Andante" indicates a slower-than-ordinary tempo. When Bach bothered to write "Andante" over the "Et in unum" duet, then, he was probably cautioning musicians not to play it as quickly as an ordinary movement in common time. Among HIP performers, however, only Harnoncourt, Leonhardt, Jacobs, Hickox, Robert King, and Herreweghe (in his 1989 recording) read the tempo as suggested here. The others take it more quickly, sometimes at a downright jaunty pace. I doubt that these quick tempos, whatever their artistic merits, have a strong historical basis. But I can see why some performers might disagree.

Other factors also modify the implicit tempo, of course: genre, harmonic tension, resolution, and rhythm, text and affect, and prevailing note values (for example, the beat tends to be slower when the lines generally move in unusually small notes, like thirty-second notes in 4/4). On the whole, taking all these factors into account, early music tempos do seem generally more historically plausible than those favored by a Klemperer. But that is not always the case, as the zippy "Et in unums" suggest.

Of course, the value of historical plausibility is itself controversial. Most musicians prefer a historically dubious tempo in the hands of an inspired artist, whether a Klemperer or a Brueggen, to a plausible tempo realized drably. Besides, composers change their own views on the right tempo often enough; why, some critics ask, shouldn't performers have the same freedom? And tempo is always relative to such variable factors as acoustics, the peculiarities of instruments, and phrasing and articulation. Subjective preference, many conclude, is the final arbiter of speed (and for that matter, of all the rest of performance).

All the same, historical investigation can reveal expressive possibilities that subjective preference by itself might miss. The Sanctus sounds so grand and impressive at the moderate tempo favored by mainstream tradition that only a historical kick in the pants might tempt musicians to try speeding it up. The conductors who have done so demonstrate that when taken quickly, the Sanctus has a joyousness that can be irresistible.

Theories of Relativity: Should the Tempos Relate Mathematically?

Performers must also consider how to relate the tempo of one movement to those of other movements. We don't have conclusive evidence that Bach intended "proportional" tempo relationships from one movement to the next - that is, with the different movements' tempos related mathematically relative to an underlying beat. All the same, the idea seems at least plausible for certain paired movements, namely the Gloria (in 3/8)-to-"Et in terra pax" (in "C" - common time, aka 4/4) [that's it on the right] and the Sanctus (in C)-to-"Pleni" (in 3/8). In these pairs, Bach's manuscripts lack even a bar line separating the first movements from the second movements (in this notation he was typical of his time). It's possible that the continuous notation suggests continuity of tempo.

But if so, how should the tempo continue? Should it keep the surface moving at a steady pace, with the 8th note of the C movement going at the same speed as the 8th note of the 3/8 movement? (More accurate technical parenthesis: the Gloria "ritornello," played before the voices first enter and also later in the movement, ends with what musicians call a "hemiola." That means the three 8th notes per bar become doubled in length - that is, half as fast, or three quarter-note beats spread over units of two bars. That may sound complex, but you can hear it clearly in the orchestra just before the first entry of the chorus. By what I've called the "8th note" theory, what I really mean is that the quarter-note beat in the hemiola at the very end of the Gloria section should continue unchanged in the quarter-note beats of the "Et in terra pax." Got it?) You can hear this approach in the Gloria/"Et in terra pax" of Andrew Parrott, and in the Sanctus/"Pleni" of Brueggen, Christophers, Harnoncourt, and Rifkin.

An alternative theory holds that Bach meant a full bar of 3/8 to equal a beat of C. You can hear this approach in the Gloria/"Et in terra pax" of Brueggen, Hengelbrock, Herreweghe, King, Koopman, and Leonhardt, and in the Sanctus/ "Pleni" of Hengelbrock, Herreweghe 1998, Hickox, Jacobs, Koopman, and Perlman. With this bar-equals-beat approach, the movements in C sound much slower than the related movements in 3/8.

I, personally, favor the first approach. (For one thing, as the outstanding choral conductor Timothy Stalter has pointed out to me, Bach and his contemporaries loved to use "hemiolas" at final cadences -they are a sort of built-in retardation - so keeping the doubled beat constant into the next movement would fit Bach's style.) But in spite of such interesting musicological arguments, nobody has yet proven beyond reasonable debate which relationship Bach intended - or even if the notation does indicate continuity. Perhaps, as George Stauffer argues, the tempo was to change in an unsystematic way. HIP performers have tried every solution mentioned, and at least one other, rather unfortunate, one. Jacobs slows the beat of the "Et in terra pax" to little more than half the pace of the Gloria eighth notes; the languorous result seems to portray not peace on earth so much as siesta time.

Modern performers of all stripes may sometimes be more eager to keep the tempos continuous from one movement to the next than Bach himself was. In the Credo/"Patrem" and the "Et Incarnatus"/"Crucifixus" most performers ignore notational evidence that Bach wanted the tempos of the paired movements to contrast. In both cases, the earlier movement is separated from the second one by a fermata and double bar, and the second movement has a new time signature that (I believe) suggests a different tempo. The majority of HIP performers, however, favor the assumption that Bach wanted the tempo to remain unchanged. Again, performers may have their own aesthetic reasons for preferring continuity to contrast, and such reasons need not answer to historical data. But if anyone makes historical claims, don't take them on faith.

My own conclusion (which one should also not take on faith) is that Bach meant the "Patrem" to be faster than the Credo - a solution used very effectively in Brueggen's recording. This conclusion reflects my reading of the time signatures in the autograph: the Credo signature is but the "Patrem" signature is cut-2 (the numeral "2" with a line through it. The 19th-century Bach Gesamtausgabe edition "regularized" the "Patrem" to ).

I also suspect that when, as an afterthought, Bach inserted the "Et incarnatus" chorus after the "Et in unum" duet, he expected the new chorus to be faster than the following "Crucifixus." One reason is that the latter has a time signature, 3/2, that implies something slower than the 3/4 of the "Et incarnatus." Also, the original source of the "Crucifixus," in Cantata 12, was marked "Lente"; and the movement's genre, the lamento, was associated with slow tempos.

As Rifkin, Jacobs, and Hengelbrock demonstrate, slowing down for the "Crucifixus" can have a potent effect. Once again, historical reasoning should not be used as a straitjacket - other approaches might be just as moving - but in this case, it can lead to a deeply expressive result.

How Small is Beautiful?

HIP musicians all perform the Mass with small ensembles, compared to the huge choral societies that have propagated the work since the early 19th century. The small ensembles help one to hear all the lines of Bach's complex polyphony.

But how small? Goldberg readers will have heard (more than some would like to) of Joshua Rifkin's argument that Bach expected most of his choral music to be sung by only one singer to a part. Rifkin and his defenders have skirmished repeatedly with scholars and performers who insist that Bach preferred something larger. As of 2000, there appears to be no consensus on the subject within the community of Bach scholars. The disagreements represent progress for Rifkin, because ten years ago the consensus was strongly against him. (Disclosure: I once dismissed Rifkin's idea but now consider it the more convincing reading of the evidence. Click here for an article I published on it.)

Whichever side one takes on this issue, it adds a twist to the question of historical fidelity. Suppose that Rifkin's position became generally accepted. As with the question of tempo, some musicians would ask, "Why should I be constrained by what Bach did?" But a more common argument for the use of a chorus adds something: it suggests a gap between what Bach seems to have done in fact and what he would have done if possible. Even when Bach was forced to use one singer per part, some musicians argue, he would still have preferred three or four per part. Proponents usually claim that this ideal is laid out in Bach's famous memo to the city council of Leipzig complaining about his working conditions.

Yet Rifkin, John Butt, and others have shown that when read carefully, the memo might even be said to support Rifkin's view, and in any case tells us nothing clear about Bach's preferred choral size. Nor does any other evidence. As Rifkin puts it, if Bach had heard Gardiner's or Herreweghe's choirs he might have said, This is fantastic - it's ideal for all my sacred works. Or he might have said, Hmm, interesting - perhaps one of my lads will find a way to use it someday. Or his response might have fallen somewhere in between. We can only guess.

Supporters of the one-per-part argument characterize choral performances as transcriptions, like a Haydn string quartet played by an orchestra. In this view, Bach tailored all the writing to solo voices. Opponents might respond that listeners sometimes prefer transcriptions to originals. More likely, they will disagree with the supporters about how Bach tailored his writing.

The controversy, no matter how it is resolved, has led to a great deal of choice for listeners. Several conductors, for example, use one-per-part for specific sections only. This idea was proposed and tried out before Rifkin, but his writing has inadvertently given it new life. Brueggen, Gardiner, Hengelbrock, Jacobs, King, and Perlman use solo voices in a few selected choral passages or movements, to contrast with a prevailing choral sound. Parrott and Junghanel use solo singers as their base, adding a second echelon of solo singers for effect in certain passages (a practice that we know Bach used in certain works).

Speaking and Singing

If Baroque specialists have a favorite metaphor, it is that phrasing and articulation in Baroque music shouldn't "sing" so much as "speak." The metaphor may be simple, but its application is not. Speak, for example, the word "Kyrie," then listen to some HIP recordings of the B Minor Mass's first movement. You will, on the one hand, notice that the performers avoid the singing style you might learn from a conservatory voice teacher: they will not extend each vowel until the very beginning of the next consonant. In rejecting that kind of operatic super-legato, they have strong historical support. But a few of them go to the opposite extreme, cutting the syllables "Ky-" and "-e" short to a degree that sounds mannered - and not at all like a person speaking the word.

It would be nice to say that the autograph score and parts show us exactly where Bach fell on the legato/choppiness spectrum. They don't, of course, but they provide at least some evidence, such as numerous slurs and dots. Alas, while Bach's articulation markings sometimes have unambiguous meanings, they sometimes do not.

One example, out of many, illustrates the ambiguities. In a recent (and excellent) book on the B Minor Mass, George Stauffer argues that the oboe slurs in early copies of the "Et in Spiritum Sanctum" show that the third note of each triplet should be separated from the previous two notes and played short. This "2+1" slurring would sound distinctly choppier than the smoother "3+3" articulation used on even the HIP-est recordings. But a look at the autograph score [reproduced on the right] raises doubts about Stauffer's claim. As Butt points out, the only way to indicate "2+1" slurring unambiguously was with a dot over the third note (you can find an example in the opening of the St. Matthew Passion), and Bach was more explicit in his markings later in his life. Bach does not use a dot in the "Et in Spiritum" oboe slurs. Indeed, these slurs are placed haphazardly enough to suggest that Bach may be calling only for a smoother, more general slurred effect (the opening of the St. John Passion is an analogous case). It's not clear, in the end, what Bach had in mind; Stauffer may be right, but may not be.

Even when a marking is clear, we may not agree on just how "articulated" it should be. How strongly do you bring out a two-note slur - say, those in the violin lines of the "Et incarnatus" or in the Kyrie 1 fugue subject? Historically inclined performers have varied widely in how they solve this, and the historical evidence does not settle the matter. As in other repertory, performers differ in how much to underline and emphasize musical events. Such tendencies may be as much a matter of temperament or of the "school" one belongs to as of one's reading of history or of its importance.

 Words and Meaning

The survey above hardly exhausts the controversial issues. Debates rage about such questions as what instruments played in Bach's continuo group, whether boy sopranos are preferable in his music, whether Bach wanted performers to double the voices with instruments in the Credo and "Confiteor" choruses, whether his alto lines were sung by boys or men, and so forth.

But all such debates may seem remote from a performer's ultimate concern: what is the music communicating? What is it about? I spoke earlier of how tempos that are derived historically can reveal previously unsuspected emotional veins in the music. But more often it's our sense of the music's emotion and meaning that affects our tempos, as well as our phrasing and articulation. And rightly so.

What the B Minor Mass means is, of course, a topic with many layers. Certain parts of the text leave little doubt as to the prevailing emotion, such as the joyous shouts of "Hosanna" and "Glory to God," or the pleas for mercy in the Kyrie and Agnus Dei. But some thinkers consider even this seemingly clear water to be muddy. They question whether the emotional implications such texts held for an orthodox Lutheran like Bach were the same as the implications of these texts for modern post-Romantics like us.

Other passages have had no obvious musical implication in any age. In the "Et in Spiritum Sanctum," for example, the pre-existing music Bach chose doesn't express the text in any apparent way.

A line like "I acknowledge one baptism for the remission of sins" might seem particularly dry and has always challenged the imaginations of composers, no matter how devout. Yet on that line Bach creates (from scratch, not from an earlier piece) one of his most profound choruses, the "Confiteor." He bases it on a Gregorian chant, and sets it in a Palestrina-like style - a musical metaphor for the antiquity of the institution of baptism.

While that musical metaphor is clear, certain other aspects of Bach's symbolism may be more arcane. Donald Tovey once suggested that the combination of the two themes in the "Confiteor" symbolizes a concern of the text - the connection of the one baptism with the ensuing remission from sins. While the Mass contains other such musical metaphors - e.g., the two voices representing the unity of Father and Son in the "Domine Deus" - Bach scholars have in general become skeptical of claims about Bach using numerology and esoteric symbols.

To return to our questions, then: does consideration of the work's meaning resolve - or at least give significant insight into - any of the debates about performance practice that the evidence leaves, to varying degrees, unsettled?

Let me imagine how such considerations might be used. A few HIP recordings take the Kyrie 1 downright quickly, and I feel on somewhat solid ground when I assert that a tempo of quarter=63 is faster than the evidence suggests for the tempo marking ("Largo"). I might, however, also add my interpretive belief that too fast a speed undermines the pathos of the pleas for mercy. Similarly, I could argue that the gravity of the "Et in unum" text is better served by an unhurried "Andante" tempo than by a jaunty Allegro.

Yet such arguments could easily be stood on their heads; one can find emotional justification for almost any tempo, phrasing, or articulation. A faster Kyrie might, for example, strike some listeners as compellingly intense. To mention another movement, a slowed-down "Et in terra pax" (beat equals previous bar) communicates peacefulness, but a fast one (quarter-note equals previous quarter note) may convey joyousness as well. Both are convincing emotionally.

Besides, imaginative responses to the text might lead one beyond historical evidence. Jeffrey Thomas, for example, treats the string chords in the "Crucifixus" as a depiction of the nails being driven in. It seems unlikely that Bach had such pictorialism in mind (since the chords also appear in the original version of this chorus, with a different text, in Cantata 12), but some listeners will appreciate Thomas's attempt to communicate meaning.

In the end, the B Minor Mass, like any masterpiece, has too many contradictory expressive qualities to allow a single performance to do justice to every one of them. Thus, no amount of historical digging - even if the controversies could all be resolved - will ever reveal a definitive way of performing it. Nonetheless, exploring what evidence we have has enriched our understanding of the B Minor Mass. It has shed new light on the joy of the Gloria, the delicacy of the "Domine Deus," the mystery of the "Confiteor," and even the pathos of the "Crucifixus." And it has allowed us to hear the counterpoint to a degree that the large-scale traditional performances rarely approach. We will never know for certain how the Mass would have sounded if Bach had conducted it; but our attempts to imagine that performance have yielded rewards that could be gained in no other way.


Further Reading

Two excellent books on the B Minor Mass, both in English, give a great deal of insight into its genesis, history, structure, nature, and reception: John Butt, Bach: Mass in B Minor (Cambridge Handbook, 1991); and George Stauffer, Bach: The Mass in B Minor (Schirmer, 1997). In addition, a fine brief essay on the Mass by Butt is available online at, and in the excellent new Oxford Composer Companion: Bach.

The complete autograph score and parts for the B Minor Mass can now be found on the Web, at The best modern edition yet published is the one edited by Christoph Wolff for Peters (No. 8735). Many listeners will be able to afford only the piano reduction (No. 8736). The 19th-century Bach-Gesellschaft edition - reprinted in an inexpensive Dover edition - is unreliable.


Selected Recordings (in alphabetical order by conductor's name)

When I wrote this article I was asked to add short reviews of four favorite recordings of the work. I complied, with the following proviso: "Any choice of four 'best' recordings would be, ultimately, foolhardy and unjust. I might, on another day, rank other recordings just as highly. Besides, were you to ask ten Bach aficionados for their favorite recordings, you would get ten different lists.Those listed below, then, are by no means my idea of the four best on record. Instead, they are four that, when I wrote this article, attracted me. That is because each possessed, for me, some special emotional quality."

To reduce the "injustice," I hope to add some more thumbnail reviews in the future; there are now a total of seven. And my changing thoughts since publication prove my point about changing rankings. At the moment, my top recommendations (were I pushed to a wall) would be two recordings not even on the original list-- Richard Hickox on the Chandos label if you want a chorus and Konrad Junghaenel on Harmonia mundi if you prefer one-per-part, sometimes augmented. (Rifkin, Thomas, and Bruggen would compliment these two on my short list.)

Netherlands Chamber Choir; Orchestra of the Eighteenth Century / Frans Brueggen. Philips 426 238-2PH2 (1990). Recorded at a 1989 concert, this performance is slightly less polished chorally than some of its competitors. I could hardly care less, because when everything comes together the inspiration deserves that overused metaphor, "white hot." The Symbolum Nicenum becomes a gripping drama. I can't recall a more joyous Sanctus and "Osanna"; the choristers sound as if they really mean what they are singing. The Agnus Dei, sung by Michael Chance, is among the most moving on record.

In some of his other choral recordings, Brueggen's nuanced approach to Bach hasn't worked - his St. Matthew and St. John Passions show how easy it is for the style to create a fussy effect. But in the B Minor Mass this approach works magically almost throughout.

The Gloria/Et in terra sequence provide a good barometer for listeners deciding if this recording is for them. In the Gloria section, the dance rhythms have a particularly natural lift. In the "Et in terra pax," like many conductors, Brueggen slows considerably for the opening transitional passage, and remains slow for the fugal exposition. He entrusts the exposition to soloists alone. The effect is unusually breathtaking. What's also unusual is that he then increases the tempo (and brings in the full choir) after the exposition. This is not what Bach notated, but the effect is powerful. It is typical: more than most HIP conductors in Bach, Brueggen uses tempo fluctuation to express feeling and underline structure.

Critics have complained that the instruments receive undue prominence, but I find the emphasis revealing, and appreciate Brueggen's attempt to make the voices and instruments perfect equals. In the continuo and instrumental parts, Brueggen shapes cadences subtly, with a feeling for harmonic tension and resolution. You will hear very clearly what I mean when I talk about the syncopations and dissonances that the bass line contributes to the Kyrie II.

In spite of this being the fastest recording yet made of the Mass, the tempos are convincing throughout. Not everything sets the speed records: the Credo chorus is a bit slower than in many HIP recordings, and similar in speed to Harnoncourt and Rifkin. It is much faster than Klemperer, whose Credo moves in eight to the bar, and than Leonhardt and Jacobs, who move at four to the bar. But while at Brueggen's tempo the Credo is fast enough that it moves at an appropriate two beats per bar, it is not so fast as to sound hectic, a danger to which several other recordings (e.g., Martin Pearlman's) fall prey.

The Sixteen Choir and Orchestra / Harry Christophers. Collins Classics (CD) 7032-2 (1994) What disqualifies some recordings from my list is articulations exaggerated to the point of mannerism; the relative absence of mannerisms here helps make the result palatable. But what puts this recording on my short list is the presence of more savory attributes. I think of the mysterious "Confiteor," for example, and of the powerfully shaped incarnation-crucifixion-resurrection sequence, and especially of the Sanctus. In that chorus, Christophers finds a tempo and expression that allows him to convey both majesty and vitality; most performers aim for only one or the other.

There are weak points, of course: The Kyrie 1, though it has plenty of power, feels a bit hurried, the "Christe" is undistinguished, the Gloria in excelsis unsteady. But things improve markedly, and the fine choral work and vocal and instrumental soloists make this performance especially appealing. The use of a theorbo (large lute) in the continuo stretches matters historically (yes, Bach knew lutenists in Leipzig, and used them in a few works, such as BWV 198, but....) Still, I like the effect of the instrument in the Crucifixus and a few other spots.

Balthasar-Neumann-Chor/Freiburger Barockorchester / Thomas Hengelbrock DHM 77380 Thomas Hengelbrock takes the Kyrie 1 as slowly as Karajan, Jochum, and Richter - which is to say, far more slowly than any of his period-instrument colleagues. He takes the Cum sancto Spiritu quite a bit faster than any of his colleagues or predecessors. And he nuances the articulation more than I usually like, as in the word "Crucifixus," where he has his singers set off the "fixus" as an accented "sigh." If I had read the above sentences without having heard the recording, I would have concluded that this one is not for me. But the results won me over. Hengelbrock and his group maximize expressiveness, with an view not primarily to how Bach might have done it but to modern experience.

Collegium Musicum 90 / Richard Hickox Chandos 0533/4 The British professional mixed chorus is as polished as you would expect, and so is the orchestra of British early-music stalwarts. What makes the recording a good "first choice" or basic recording for one building a collection, however, is the glowing interpretation. Hickox finds an unusually convincing balance between interpretive possibilities. He shapes movements effectively. He often gives just the right degree of emphasis to dissonant relationship between voices (neither too much nor, as in many recordings, too little). His soloists are a strong group. And inspired movements like the "Cum sancto spiritu" are worth the price of the CD. Hickox is among the handful of conductors to use a harpsichord as well as an organ in the continuo. The effect is sometimes obtrusive, but generally reinforces one of the performance's strengths - that it brings out the continuo line and its relationship to the other voices.

Cantus Colln / Konrad Junghanel Harmonia mundi HMC 901813.14 A key argument in support of Joshua Rifkin's assertion that Bach used only one singer per line is that Bach's style of writing in the choruses is so similar to his writing for solo voice in arias. The musical implication of one-per-part, then, would be to have choruses sung like Baroque madrigals, with the individual singers responding to one another and interacting with the instrumentalists. No performance on record has come as close to achieving this ideal as this one, performed by a group noted for its work with early-Baroque madrigals. Some of the singing in arias may not impress lovers of modern operatic singing, but the choral singing lives in a unique way. The choruses are sung either one or two per part; the former are particularly transparent. The overall feeling of the performance is intimate, but with plenty of emotional responsiveness: the D major trumpet-and-drum choruses are as exuberant as you could want. One peculiarity: Junghanel is the only conductor I know of who is consistent in taking fast tempos in the alla-breve time signature in "old style" choruses. I like the results in some cases (the Kyrie 2), but not in others (the Gratias and Dona nobis pacem, where the "small" notes are too fast for my comfort). But the latter cases detract little from this outstanding release. The recorded sound is both transparent and warm.

Bach Ensemble / Joshua Rifkin. Nonesuch 9-79036-2 (1981-82) Some early critics complained about fast tempos in this 1981-82 recording, but the tempos now seem moderate. I mentioned above the effect of Rifkin's taking the "Crucifixus" more slowly than the preceding "Et incarnatus," as implied by the time signatures of the two movements. I should add that his shaping of the extraordinary climactic modulation and cadences in the "Crucifixus" is perfectly judged and deeply expressive. His natural approach to articulation, too, expresses the mournful affect of this movement better than do the exaggerated articulations found on some other recordings. In general, this recording conveys an inwardness and intimacy to which no other recording seems to aspire. The singers nonetheless bring a subtle expressiveness to their lines. Some movements, like the "Domine Deus" and the "Confiteor," have never really been surpassed. Other movements have, not surprisingly, held up less well: the "Quoniam" seems a bit poky and the "Cum sancto" doesn't take wing (the band seems strained by it at times - this was, after all, the early days of period-instrument playing). And some of it is a matter of taste: as a friend of mine noted, the Kyrie expresses awe rather than pain.

A privately circulated concert tape from the 1997 Regensburg Festival captures Rifkin and company in the Mass. The concert taping has poor sound quality, but the performance is clearly superior to the older recording. The soloists in 1997 sing more freely and the phrasing is more articulate. The Gloria in excelsis, "Quoniam," and "Osanna," are, quite effectively, faster than in the earlier versions, which now seem too slow. Still, the studio recording has retained a certain appeal over the years. The 1997 taping, by the way, has contrapuntal balances that equal the excellent ones in studio version. It thus supports the claim that the earlier balances are the work of the performers, not the studio engineers, since the 1997 taping was done by an amateur sitting in the audience, using a single microphone and a home machine.

American Bach Soloists/Jeffrey Thomas, Koch 3-7194-2, (1992). This recording was preceded by several concert performances; Jeffrey Thomas sounds almost proud when he tells you that a well-known musician walked out of one of them, to protest the slow tempo of the Kyrie 1. That tempo (quarter=51) is about 10% slower than the median HIP tempo (57), but is just as plausible historically. Thomas, however, does not consider historical plausibility important.

He does seem to value expressiveness and spontaneity, and responsiveness to text, meaning, and feeling. Some listeners may find the opening four-bar invocation mannered, with its diminuendo on the second chord (a feature found in Hengelbrock and Jacobs as well), but others find it expressive. And if the Agnus Dei seems too slow, and the "Et in unum" too fast, well, I would make the same complaint about most HIP recordings, but in this recording it misses the point. Thomas has assembled a team of excellent soloists and instrumentalists, and he knows enough not to keep them on too tight a rein, even while he leads them in some interesting directions. (For example he seeks longer phrases and lines than many of them might choose on their own.) The blend of insights, of HIP with drama and feeling, of musicians with differing ideas somehow brought into synergistic unity, makes this a plausible first choice as well.


John Eliot Gardiner, 1985, Archiv 415514-2

Nikolaus Harnoncourt, 1968, Teldec Das Alte Werk (CD) 4509-95517-2

Nikolaus Harnoncourt, 1986, Teldec 2- 2292-42676-2

Philippe Herreweghe, 1988, Virgin CDCB 59517-2

Philippe Herreweghe, 1998, Harmonia Mundi 901614.15

René Jacobs,1992, Berlin Classics 1063-2

Robert King, 1996, Hyperion CDA67201/2

Ton Koopman, 1995, Erato 45099847982

Gustav Leonhardt, 1985, Deutsche Harmonia Mundi 77040-2-RG

Andrew Parrott, 1985, EMI ZCDB 47292-2

Martin Pearlman, 2000, Telarc 80517


return to Bernard D. Sherman's web site