Tempos and proportions in Brahms: period evidence.
Still, we have reason to be grateful to those `good friends'. The Brahms MMs may not decree the ideal tempo, but they can help us evaluate hypotheses. This article uses them to assess two ideas that have gained currency: (1) that Brahms generally played his own works at faster tempos than mainstream performers of today; (2) that he wanted his works to be played according to proportional tempos.
We have a total of 44 metronome marks for eight Brahms works (see table i for the Mms and for notes on their origins). These marks might in some cases be more reliable records of performing practice than those left by Beethoven. Beethoven's are important guides to the sort of tempos he imagined, but they were tempos he imagined while metronomizing, either long after he had heard performances or without his having heard any; in real performances, such factors as acoustics and instrumental action can influence one's speed. But the Mms for the Requiem almost certainly record the tempos Brahms took at the premiere,(3) and the Mms for the Third Piano Trio record the tempos Brahms took at a rehearsal in 1890 (according to the pianist Fanny Davies, who wrote them down after the rehearsal and then double-checked them with Joseph Joachim, the violinist in the trio). As for the other metronome marks, Robert Pascall suggests that in some cases `they would have served a useful purpose where Brahms performed music on tour (as he did extensively with the Second Piano Concerto), instructing local conductors as to his own speeds before he arrived to play the work.'(4) And since Brahms typically sought to perform his works before publication, some of the metronome marks may reflect concert experience.
[TABULAR DATA NOT REPRODUCIBLE IN ASCII]
In addition to the metronome marks we have stopwatch timings of the premieres of the first two symphonies, as well as Hans von Bulow's timings of his own performances of the first three symphonies (see table 2). Unfortunately, all the timings are rounded to the nearest minute, and Bulow gives only the total time of each symphony. The timings are too imprecise, then, to help evaluate the issue of proportional tempo, but they may still be useful in examining the question of speeds.
[TABULAR DATA NOT REPRODUCIBLE IN ASCII]
1 How fast were Brahms's tempos?
In 1905 Fanny Davies, comparing current performances of Brahms with the composer's own playing, said that `the tendency is usually to play the andantes too slowly, and the quick movements, scherzos, &C., too quickly'.(5) In a 1933 book on the symphonies Walter Blume complained that tempos were becoming faster.(6) And in 1980 Max Rudolf (b 1902) wrote, `If we are to believe reports by musicians who performed Brahms's symphonies under his direction, he would not have approved of the rushed tempi we now sometimes hear. His music making was relaxed.'(7) Davies's remark supports the general belief among early-music performers that Brahms continued the tradition of 'classical fast andantes',(8) but in other respects, the verbal reports conflict with the oft-expressed idea that Brahms is nowadays played too slowly. Since verbal reports are subject to various kinds of doubt -- e.g. do the words 'too quickly' mean to us what they did to Davies? -- objective evidence might be helpful.
Timings Recorded timings of the First Symphony show a slowing trend. When Otto Dessoff led the premiere his first movement took between 13'30 " and 14'30" with repeat -- which implies tempos faster than those of any conductor on record except Weingartner and Norrington.(9) If we use as our baseline an estimate of Dessoff's timing without repeat (11'30"), we find that a sample of pre-1946 conductors average about 10 per cent longer, while a sample of post-war conductors average a full 23 per cent longer.(10) (See table 2.) Dessoff's Andante sostenuto (then marked Poco Adagio) took approximately 9'00" -- and the pre-1946 average is nearly that, but post-war recordings average about 8 per cent longer. Similarly, average pre-1946 timings in the last two movements are within the range of Dessoff s possible timings, but post-war averages are much slower -- 15 per cent slower in the third movement and 12 per cent slower in the finale.(11)
The slowing trend also applies to the first movement of the Second Symphony. A sample of postwar timings averages about 9 per cent (1'27 ") longer than Hans Richter's premiere, while the pre-1946 average is at the middle of his possible range.(12) Also, Richter's finale was faster (by as much as 2 minutes) than all but a few conductors on record. The only movement that reverses this trend is the Adagio non troppo, where Richter took longer (approx. 11') than the majority of conductors on record.
The timings, then, seem to contradict the verbal reports. While the First's Andante has been taken more and more slowly over the course of our century, it appears that early recordings took it as quickly as the premiere did, which we would not expect from Davies's comment. As for quick movements, those of the first two symphonies were -- by today's standards -- not 'relaxed' when premiered but, as Roger Norrington has written, `brisk'. The overall timings are also short, as are Bulow's timings.
Of course, this leaves open the possibility that Brahms conducted these pieces more slowly than his champions did. There is a some evidence for this in, at least, the symphonies' Allegretto (third) movements. According to his publisher, Fritz Simrock, Brahms conducted the First Symphony's Allegretto at an 'Andantino' pace.(13) Also, in the Allegretto of the Second, the critic Hanslick preferred Richter's tempo to Brahms's own -- indicating, obviously, that the man who led the premiere took a tempo different from the composer's; when informed of Hanslick's views, Brahms said that he wanted the movement to be `quite peaceful, especially at the end', suggesting that his tempo was slower than Richter's. One must not assume that this extends to the four timed Allegro movements, but for two of these movements that possibility has at least some support. In the First's finale Brahms later changed the premiere's Allegro con brio to Allegro moderato ma con brio, and, during publication, to Allegro non troppo ma con brio; and in the Second's opening movement, although Hanslick did not say that Richter's tempo differed from that of Brahms, it may be significant that Brahms's remark about the symphony's Allegretto -- that he wanted it to be quite peaceful, especially at the end' -- applied to the first movement as well.(14)
Consideration of that movement, however, brings up another limitation of the timings: they tell us nothing about tempo fluctuations. A sample of recordings indicates that initial tempos in this movement have not changed significantly over the century; but it also shows that before 1946 conductors let their tempos accelerate more often and to a much greater degree than after the war. Such acceleration turns out to be the main reason why the pre-war recordings have quicker timings.15 This may not have much relevance to the timings of Richter -- his Brahms conducting was once called `metronomic' and Brahms complained about his general lack of nuance(16) -- but the composer's own conducting style seems to have included the use of acceleration.(17) (I should note, by the way, that the post-war slowing of the First Symphony's opening movement reflects genuinely slower tempos.)
A final concern is that any conclusion drawn from the timings of the Allegros is based on only four movements. The metronome marks, of course, expand our database; they often come from Brahms himself; and they do not require us to wonder how slow 'quite peaceful' is.
Metronome marks At least a quarter of the 44 metronome marks suggest that Brahms did favour brisk tempos. To list only outer movements, the MM of the Allegro of the Second Piano Trio iS 12 tO 20 points (9-17 per cent) faster than a sampling of recorded performances," and only the quickest recordings attain the Mms of the outer movements of the First and Third Piano Trios, or the Maestoso of the First Piano Concerto.19 But at least another quarter of the metronome marks suggest a man whose tempos were indeed `relaxed'. Eleven of the Requiem's 14 Mms are slow -- sometimes moderately, sometimes extremely. For example, the Allegro (sixth-movement) fugue has an MM slower than almost any tempo on record, as does the preceding Vivace. (Subsequent acceleration, however, may explain why this Vivace's MM is slow; Brahms told the conductor Siegfried Ochs that he wanted not a tempestuous allegro but a gradual tempo increase, after having started it not too fast'.)(20) Other works, too, contain slow metronome marks -- for example, Rinaldo's Allegro con fuoco chorus `Und umgewandelt', and the Third Piano Trio's Presto non assai (about which Davies marked her score, `not the least idea of presto').
Confusing though this is, we still might locate a few tentative patterns. One may be that Brahms took choral fugues slowly -- a possibility suggested by the Mms for all three of the fugal sections in the Requiem (all of which are slower than most conductors have taken) and for the fugal `Nein, nicht langer' from Rinaldo. This may illustrate a general principle: that in determining Brahms's tempos, among the factors to be considered is topos. In this case, the topos `choral fugue' may have implied a slow tempo to Brahms. Another possible toposrelated tempo may involve what Michael Musgrave calls a `Brahmsian archetype', the 6/8 scherzo style, which Brahms often associated with C minor.(21) The first-movement Allegro in the C minor Symphony is widely held to be in this style, and the finale of the Third Piano Trio might also be -- the MM for the latter is fast, while the timing for the former suggests speedy tempos.
A firmer conclusion is that for Brahms, alla breve metres do not necessarily indicate fast speeds. This is shown by the slow Mms of the Presto non assai of the Third Trio, and of the `Herr, lehre doch mich' section and both alla breve fugues in the Requiem. (Norman Del Mar describes the third-movement fugue's MM as 'altogether too stodgy',(22) and eight of the 11 performances I surveyed exceed it handily.)(23) course, the timing of the Second Symphony finale was quite rapid, which shows that alla breve movements were not necessarily slow either. In his Handexemplare, Brahms changed three movements marked c to alla breve and one tempo the other way;(24) presumably these changes did not imply an alteration of speed, but only of the initial metre. A sense of what alla breve may have meant to Brahms is conveyed by what Davies says of the Third Trio's Presto non assai: 'the signature of the alla breve was made apparent by the rhythmical flow from the first beat of the bar to the third' -- and not, her words imply, by speed.
Another conclusion may be that while the metronome marks tend to support Davies's claim about faster Andantes, they also qualify it. When converted to crotchets, the andante MMs range from 69 to 104, slightly lower than the range marked 'andante' on a metronome (76 to 108). A sense of the range can be gained from the Requiem. Its final movement is marked 'Andante con moto' in the autograph full score, where the MM first appears, and this MM is quicker than any tempo in my survey of 11 recordings, save for the three 'early music' recordings. But the MM of the Andante moderato third-movement opening struck Del Mar as 'too heavy and dragging,(25) and eight out of the 11 recordings take a faster tempo.
These two movements suggest a clue to the speed of andante tempos: the adjective or prepositional phrase in their tempo markings relate to somewhat slow or fast metronome marks. Other clues might reside in the music's character and metre. The Andante of the Second Piano Concerto has an MM far faster than any performance on record; but the opening section of Nanie, also a 6/4 Andante, has an MM that is faster still, yet does not 'sound fast' and is often exceeded in performance. To explain this, we might note that the melody of the Piano Concerto Andante moves against its notated metre -- I hear the melody in the first three bars in 4/4 -- and it has a slow-moving song-like character; but the metre of Nanie is unambiguously 6/4, and the melody has the gentle swing of a slow dance in triple time. Perhaps that is why its MM feels more moderate than that of the Piano Concerto in spite of its being faster.
The MMs of the two 6/4 Andantes differ by a full 16 per cent, which illustrates a more general conclusion: that it is difficult to find broad patterns correlating the MMs with tempo indications and metres in Brahms, as has been done with Beethoven.(26) A possible exception are putative Brahmsian topoi that have associated tempos, but my earlier (and similar) speculations would need more data to be confirmed. Another complication is that Brahms's idea of the correct tempo changed regularly. He wrote to a friend,
I can quite easily start you on a subscription for metronome markings. You pay me a tidy sum and each week I deliver to you -- different numbers; for with normal people, they cannot remain valid for more than a week.(27)
Not only did his own tempo choices vary; Brahms also made it clear that he did not believe that there is one ideal tempo for a work. Thus he believed that the best (though by no means unambiguous) clues for performers are not his metronome marks but his verbal tempo markings, about which he was painstaking. As we have seen, however, the metronome marks may give at least a few clues about his own tempo choices, even if he did not want these marks to determine our choices.
2 Continuity versus character: did Brahms want proportional tempos?
One factor influencing our choices can be the tempos of previous movements: good musicians tend to choose tempos that not only suit the movement they are playing, but also bear a satisfying relationship to previous tempos in the performance.(28) David Epstein has made important attempts(29) to understand some of the psychological mechanisms behind such choices. These choices are often intuitive, but Epstein believes that Brahms intended something more systematic and thorough. Epstein holds that
a steady basic pulse runs continuously throughout a work of Brahms, underlying all segments, sections, and movements of the work and serving as the referential basis -- indeed, the temporal module-for explicit changes of tempo. New tempos relate to this basic pulse in such a way that the two tempos can be characterized by simple ratios of low-order whole numbers -- for the most part 1:1, 1:2, 2:3, 3:4 and their inverse.
This is in fact the principle of the tactus found in Renaissance music ... Brahms, with his interest in and knowledge of Renaissance music, may well have been impressed by the coherence and unity this system provided, for unity and coherence were cornerstones of his own musical ethos.(30)
Brahms's notation of 19 movements calls explicitly for proportional tempo relationships within the movement.(31) Epstein(32) and, before him, Bernard Jacobson(33) have pointed out that in the finale of the Second Piano Concerto, the coda's MM relates to the previous MM at a ratio Of 4:3. Other movements imply such internal relationships. Epstein discusses cases in which rhythmic motifs occur in two contiguous sections, implying, he says, how the tempos of the sections should relate.(34) An example is in the introduction to the Finale of the First Symphony. (See ex.1.) One of the composer's letters has been said to support Epstein's views on this passage, by showing that 'Brahms viewed his notation of the timpani as crucial for achieving the proper "metrical modulation" between the Adagio and the ensuing Piu Andante'. In the letter, Brahms writes, 'I will leave the triplet motion in the timpani [bar 29], for this is the main thing at the beginning [bar 30] of the next section, and this way the timpanist is more likely to grasp it.'(35)
But this sentence does not necessarily refer to a tempo relationship; it may indicate only a motivic one.36 Conductors who do maintain a proportional relationship here differ on what it should be. Epstein argues that it indicates a 4:3 ratio, but others read it (more literally) as a 2:1 ratio,(37) and some distinguished Brahms conductors maintain a 3:2 ratio here. A broader objection can be raised as well: if Brahms really did intend for large works to be united by a single tactus, it is puzzling that he never indicated this explicitly. Indeed, Raymond Knapp has argued that 'on the rare occasions when Brahms did specify precise metrical relationships ... his directions indicate an exceptional procedure rather than a general practice'.(38) The remaining Mms might be helpful in adjudicating here: they can at least indicate if the Piano Concerto coda's proportional MMs are exceptional. Let us consider this first with respect to between-movement tempo relationships, and then to within-movement relationships.
Between-movement tempo relationship We have sets of Mms for five multi-movement works -- the First and Third Piano Trios, the Second Piano Concerto, Rinaldo and Ein deutsches Requiem. If any of them could be expected to have a proportional tempo scheme, it would be Rinaldo, since most of its 13 movements flow continuously one to the next. Some of its Mms can indeed be related proportionally. The quaver pulse (MM 66) of the second movement (the aria 'Stelle her der goldnen Tage') relates to the minim pulse (MM 100) of the first movement at a 3:2 ratio. The aria moves into a 2/4 section, with a pulse of ?? = MM 88, a 3:4 ratio with the earlier section. The next marking, ?? = MM 96, does not relate to that of the 2/4 section, but may be close enough to the original 100 to stand in a 3:2 ratio with the aria's opening and a 1:1 ratio with the opening tactus; subjectively, MM 96 and MM 100 are hard to distinguish. This illustrates a problem we will encounter again -- that to posit a ratio, approximation is sometimes necessary. Epstein argues for such approximations by referring to a general principle from psychophysics called Weber's Law. Applying this law to music yields the intuitively obvious result that we notice a change in tempo only if it is sufficiently large relative to the original tempo; Epstein quotes research suggesting that this amounts to a minimum tempo change of 5 per cent -- although he makes it clear that this is not an absolute figure, but depends on musical content.(39) If we accept the principle, Rinaldo's following MM, 76, can be related to MM 96 at a 4:3 ratio; and the MM of the following movement is again at MM 96. After this point, there are some MMs that relate to the earlier MMs, but some that clearly do not, and some of those that might relate are probably too far separated by other movements to allow a proportion to register -- although the final chorus is, like the opening, an Allegro in alla breve, and it too has an MM close to the original 100. A single tactus does not run through every section of the work, then, but might be said to unify parts of it, including the opening and closing movements.
Ein deutsches Requiem, too, has some movements that relate proportionally and some that do not. The MMs of its first two movements, 80 and 60, relate to each other at an exact 3:4 ratio. Two other exact ratios involve non-contiguous sections of the work. The Requiem's seventh and final movement -- which recalls the first movement in key, metre and motif, and in the end quotes it -- has the same MM as the first. And if one converts the MM of the third-movement fugue from a minim to crotchet, it equals the MM of the second-movement fugal section, 'Die Erloseten'; Musgrave says that Furtwangler observes this relationship in his recording, which adds to a sense of unity between the movements.(40) (Gardiner, Herreweghe and a few others observe it as well.) The tempos of these two fugal sections might be related in one's memory, partly because they end contiguous movements in a similar manner, and partly because the entire third-movement pace relates to that of the immediately preceding fugal section, 'Die Erloseten'. This latter relationship is a bit approximate: if one converts the MM of 'Die Erloseten' from a crotchet to a minim, it is only 2 points faster than the third-movement opening. Even for those sceptical of Weber's Law, this degree of change is hard to discern subjectively.
One can, then, make a case for the MMs revealing proportional tempo relationships between the first three movements. The case may be strengthened by the fact that these movements form the first large unit within the overall composition.(41) (Of course, the relationships may be coincidental; e.g. the two fugal tempos may simply reflect Brahms's idea of how fast a choral fugue should go.) And the finale could be said to tie the work together not only by its material, but by a tempo relationship of 1:1.
The MMs for the fourth, fifth and sixth movements, however, do not relate proportionally to that of the first. Another possible relationship may exist, however: the fourth and six movements (which originally were contiguous) have identical opening MMs. Yet when Brahms inserted the current fifth movement between them, he gave it an MM that does not relate to theirs. This suggests either that he wanted the fifth-movement tempo to contrast with those of the surrounding movements, or that he cared too little about proportions to assign a tempo to the fifth movement that did not fit its character, or that the fourth's and sixth's identical Mms were coincidental.
The latter two conjectures gain support from an anecdote. Years after composing the Requiem, Brahms told Ochs that he wanted a much slower tempo for the fifth movement than that in the first edition: ?? = MM 72 instead of the previous 104.(42) Brahms said that this slower tempo should be set according to the chorus's quavers in bar 21. What is significant for the current discussion is that Brahms's reasoning makes no mention of proportions or of a tactus; rather, it concerns itself with an internal matter of text enunciation and expression. His later MM for the fifth movement does not relate proportionally to the surrounding movements' MMs any more than the earlier one did.
At most, then, the Requiem's proportional scheme included only part of the work. And even if we were to conclude that Brahms did intend partial proportional schemes for the Requiem and Rinaldo, we would have to note that such large-scale vocal/orchestral works are atypical of Brahms; we cannot, therefore, draw general conclusions from their metronome marks. The three other metronomized multi-movement works are more typical instrumental compositions, and with one exception they fail to support Epstein's hypothesis. In the Second Piano Concerto and Third Piano Trio, the four movements' MMs do not relate to each other proportionally. Admittedly, a 5:4 relationship can be postulated for the Mms of the Andante and Finale of the Concerto, but I question whether a ratio so high is intentional (if we include denominators higher than 4, we may be opening the possibility that any tempos can be shown to have proportional relationships). In any case, neither movement's MM relates to the those of the first two movements, which disproves the idea of a unifying tactus.
The only one of the three works whose MMs might relate all its tempos to an initial tactus is the First Piano Trio (which, perhaps not incidentally, is the only one to date from early in Brahms's career). The MM of the second movement (3/4, ?? = MM 100) can be said to relate to that of the first movement (4/4, ?? = MM 72) at a 3:4 ratio, if we are willing to be approximate. Approximation allows us to relate the third movement's 63 at about a 3:2 ratio to the Scherzo's 100; and one bar of the finale (66) corresponds approximately to one beat of the slow movement (and is at exactly a 3:2 ratio with the Scherzo). The approximations are all acceptable according to Weber's Law; but, as in Rinaldo, in each case Brahms could have created exact ratios with small adjustments of the metronome marks. While he may have considered these Mms close enough to create proportions (we will see a case in the German requiem that suggests that possibility), it is also possible that all these relationships are coincidental.
In summary, from the five works under question, three movements in the Requiem, three in the First Trio, and seven in Rinaldo might have proportional relationships to their first movement's tactus. That makes a maximum Of 13 out of the 27 non-first movements (48 per cent); if we restrict ourselves to instrumental works, it becomes 3 out of 9 (33 per cent) versus, in the choral works, 10 out of 18 (56 per cent). Three or four of the remaining non-first movements have MMs that may relate not to the first-movement tactus but to another movement that immediately precedes it. Even if we accept all the proportions mentioned as intentional rather than coincidental, however, the most obvious conclusion is that for Brahms, relating the tempos of different movements proportionally was not a typical practice, and may have applied mainly to large-scale vocal/orchestral works rather than to instrumental pieces.
Within-movement tempo relationships Seven of the metronomized movements have more than one section. In four of these movements, the metronome marks relate the tempos of two contiguous sections with fairly exact 3:4 or 4:3 proportions.(43) We have seen that this ratio connects the two Mms in the finale of the Second Piano Concerto; it also relates the two Mms in Nanie, and the Mms of the first two sections of the second movement of the Requiem. And Davies's MMs for the finale of the Third Piano Trio indicate that at the meno Allegro section beginning in bar 49 the tempo slowed to 88 -- three-fourths of the original 120 -- although this seems to contradict a statement, earlier in her article, that the section 'was started a littler slower but not very much at first; then it broadened ...'(44)
Some other cases are possibilities. In the Requiem second movement, the MM of the transitional 'Aber des Herrn Wort' (marked 'Un poco sostenuto') slows the tempo by 10 per cent; this is discernible subjectively, if only barely. Similarly, in the third movement, the fugue's MM is two points (3.8 per cent) faster than the opening section's. Both cases could be said to support the idea of a basic tactus for the movements: Brahms might have meant for each movement to be heard at a united tempo, but with (in the second) a slight slowing for the transition and (in the third) a slight sense of increased speed at the fugue.
Also, the last few bars of 'Aber des Herrn Wort' feature the head-motif of the ensuing fugal section (see ex.2), which suggests that the speed of the motif is identical in both sections. But the motif is in quavers in the first section and crotchets in the second. This implies, then, that a crotchet of the later section equals a quaver (MM 56) of the earlier one; but the later MM (108) is four points too slow to uphold this proportion. This later MM is Brahms's replacement for previous numbers (104 and 116) that were also not exact doublings, but were close. This might suggest that Brahms held that a small deviation from exactness does not undermine proportions.
The question of approximation returns with the Scherzo of the First Piano Trio. Its trio section (piu lento) is marked ?? = MM 72, and Weber's Law is necessary for seeing this as forming a 3:4 proportion with the main section, whose MM is J. = MM 100. MM 76 would have created a more exact proportion, though again 72 may have seemed close enough. Also, the trio section's dotted minim has exactly the MM of the first-movement minim (72), although again it is not certain that this is intended. Whatever one's verdict in this case, that at least four (and possibly six) of the seven movements appear to have some internal tempo proportions can be taken to support Epstein's thesis with respect to multisection movements. This might be added to the other supporting evidence -- the movements in which Brahms notated proportional relationships explicitly, and other movements with implicit proportional relationships.
Other metronome marks, however, demonstrate that Brahms did not have a strict policy of always relating internal tempos. The clearest example is in the sixth movement of the Requiem, whose opening MM does not relate to those of the two later sections. These two, admittedly, can be related to each other, with one semibreve beat (MM 50) Of the alla breve fugue standing at a 3:4 ratio with one bar (?? = MM 37.33) Of the preceding Vivace; but locating this proportion requires too much effort to give confidence. Similarly, Davies reports that at bar 66, marked 'sostenuto sempre', the Third Trio finale's tempo slowed from the (proportional) MM 88 tO MM 72, which can relate to the previous tempos only if one uses proportions with denominators of five.
Even if we do accept all these proportions, we can find unambiguous rejections of proportions in two letters Brahms sent to his publisher. The first, dated 2 July 1878, concerns the Allegretto grazioso of the Second Symphony. The score originally gave proportional tempo relationships for both of this movement's presto interludes relative to the recurrent Anegretto theme; but Brahms had his publisher remove the second presto's proportional indication from the score.(45)
In the second letter, dated 19 December 1878, Brahms asked his publisher to change the tempo marking at the end of the first movement of his Symphony no.1. He wrote,
A stupid error has struck me. At the end of the first movement it should not say Poco sostenuto. People misunderstand this and take the same tempo as in the introduction [which is marked un poco sostenuto]. Meno allegro is what should be there.(46)
This seems a clear-cut case of his rejecting a proportional tempo relationship between two sections of a movement. But Walter Frisch has suggested that Brahms originally did intend the two sections to proceed at the same tempo, and that he changed the marking only because conductors took the introduction more slowly than he had intended, so that their ending tempos were also too slow. Brahms was opposed at this point to adding metronome marks, so instead he changed the second tempo marking.(47) If Frisch is correct, it would again indicate that even when Brahms originally intended for different sections within a movement to relate to each other proportionally, he did not consider such relationships immutable and would discard them when motivated by other concerns.
Some of these concerns may involve the weighing of costs and benefits. The advantages of proportional tempo relationships are, as Epstein has explained, that they create continuity of motion and add to the unity of a work. But they also have disadvantages. Sometimes one wants to emphasize not continuities of tempo, but rather contrasts; and sometimes insisting on proportions can force one to take a tempo that does not suit the character of a particular section. The latter reason may be why Brahms removed the proportional indication for the second Presto in the Allegretto of the Second Symphony -- it may have felt too slow when played that way. Most conductors seem to think so; Epstein himself acknowledges it;(48) and the score has at least one clue suggesting a faster tempo.(49) Unifying the movement's pulse may, in the end, have seemed less important to Brahms than having a suitably fast speed for the Presto.
Epstein has argued convincingly that performers are justified in relating tempos proportionally even when composers never imagined doing so,(50) and also in employing different proportions from those that composers indicate. He might agree, then, that while the identical MMs of the outer movements of the Requiem create a 1:1 proportion, the proportion used by Gardiner and Herreweghe may be more effective. Both start the last movement faster than the first, but then slow to the first movement's tempo when its material is recalled.(51) Brahms himself seems to have substituted one proportion for another in the Andante grazioso of the Third Piano Trio. Davies's MM for the quasi animato middle section shows that in performance Brahms took a tempo that was too fast to preserve the proportion notated in the score (1:1); but instead of taking a faster tempo unrelated to the previous one, his actual tempos related at a 3:4 ratio.(52) This case, and the ratio Brahms observed in performing the work's finale, furnish striking evidence of his feeling for within-movement proportions.
Still, the case of the Second Symphony's Allegretto suggests a further degree of permissiveness: even when Brahms indicates a proportional design, performers should not be held to it if they get good results with non-proportional tempos. Otto Klemperer's tempo in the"aber des Herrn' is not only far faster than the metronome mark, but is far too fast to allow for an approximately 1:2 ratio with the following section; yet his 'Aber des Herrn' is exciting, partly because of the sheer speed.(53)
Epstein has done a service by reminding us of Brahms's interest in proportional tempos, and by illustrating some ways in which Brahms might have applied them. But it would be a disservice to demand that performers always play Brahms this way (not that Epstein insists on this).(54) And as we saw, the evidence makes it seem unlikely that Brahms had a pervasive system of tempo proportions in mind. Even when he clearly did have proportions in mind, they might not have always been meant to bind performers any more than the metronome marks; and even if they were meant to do so, Richard Taruskin and Peter Kivy have argued that there is no strong reason why composers should be the only ones allowed to implement second thoughts regarding performance instructions. Brahms's own views seem compatible with this stance: Davies says he once told a performer who had asked about the interpretation of a piece, 'Do it how you like, but make it beautiful'.(55)
(1) G. Henschel, Personal recollections of Johannes Brahms (Boston, 1907), P-78. Brahms's reply also discusses tempo flexibility.
(2) In letters Of 14 and 22 February 1894, reprinted in Johannes Brahms im Briefwechsel (hereafter Briefwechsel), xiv, ed. W. Altmann (Berlin, 1920), PP.412-13. Brahms expressed reservations about these MMs as early as 1868, in the very letter (1 October i868; ibid., p.164) by which he entered two of them, saying that one should give new MMs every year or not at all (Uberhaupt entweder jedes Jahr neu metronomisieren oder gar nicht').
(3) Brahms added the MM for the fifth movement in the letter to his publisher of 1 October 1868; in that letter he also changed the MM of the second-movement fugal section to 108 (previously it had been 104 and 116). The other 11 MMs are not in his hand, and Brahms wrote, on 14 February 1894, that they are' not from me but from Reinthaler' ('die auch nicht von mir, sondern von Reinthaler sind'); Carl Reinthaler had rehearsed the work with the composer for its premiere, conducted by Brahms on 10 April 1868. Nonetheless, Michael Musgrave has concluded that all the Requiem's MMs' clearly came from [Brahms] or had his approval'; see M. Musgrave, Brahms: A German Requiem (Cambridge, 1996), p.91, n.1. If Brahms did not approve, it is inconceivable that a composer so perfectionist about his scores would have allowed the MMs to stand for 25 years in the score of one of his most popular works, or that he would have added two MMs to the score himself. And it seems almost certain that Reinthaler would have aimed to preserve in print the tempos that Brahms had taken, rather than his own tempos. Note also that the MMs appear in the autograph full score and the copyist/part-autograph vocal score used at the rehearsals and performance at the premiere.
(4) R. Pascall, Playing Brahms: a study in 19th-century performance practice (University of Nottingham, Department of Music, 1991), p.15.
(5) F. Davies, `Some personal recollections of Brahms as pianist and interpreter' (1905), Cobbett's cyclopedic survey of chamber music (London, 1929), p.184.
(6) W. Blume, Brahms in der Meininger Tradition (Stuttgart, 1933), pp.5-6.
(7) M. Rudolf, The grammar of conducting (New York, 2/1980), p.359.
(8) Pascall, Playing Brahms, p.15.
(9) While it is possible that Dessoff took the movement's introduction much more slowly than recorded conductors and then took the Allegro much more quickly, or vice versa, either case would require one tempo so fast as to be almost inconceivable. Regarding the Andante, it should be noted that Brahms revised the piece after the premiere; but the final version is only three bars longer than the original one. See Johannes Brahms: Symphonie Nr.1 c-Moll opus 68, ed. R. Pascall (Munich, 1996), pp.194-8, 212, 248.
(10) The estimate is from W. Frisch, Brahms: the four symphonies (New York, 1996), p.180. For those who are uneasy about estimating the timing without repeat, let me note that when compared to Dessoff's timing with repeat, five modern performances with repeats average 26 per cent longer.
(11) I have not included Mackerras or Norrington in the comparisons, since they consciously buck the trends of our era. Note that the timings from Brahms's day suggest that the speeds of early recorded performances cannot be written off as the result of the limitations of the 78 rpm side or of incorrect pitching of transfers.
(12) R. Brinkmann, Late idyll, trans. P. Palmer (Cambridge, MA, 1995), pp.31, 61-2, shows that today's slower tempos in the first movement of the Second are sufficient to create a very different emotional effect than the faster tempos of earlier recordings.
(13) Simrock wrote, `die Bezeichnung ist falsch ... Der Satz wird von Brahms selbst durchaus Andantino genommen.' See Brodbeck, Brahms: Symphony no.1 (Cambridge, 1997), p.100, n.26. Still, Dessoff's letters to Brahms asking about tempo choice (Briefwechsel, xvi, esp. pp.205-6) suggest that at the premiere of the First Symphony he would have sought the composer's guidance.
(14) Dessoff wrote on 19 October 1878, `Hanslick findet, dass Richter das Tempo des Menuetts besser getroffen habe, als der Componist' (Briefwechsel, xvi, p.206). Brahms replied, on 28 October, `Satz 1 und 3 wunsche ich recht ruhig, namentlich die Schlusse' (ibid., p.207). My thanks to Styra Avins for bringing the passage to my attention and for advice on translating the tricky words `recht ruhig'. The passage also discusses the tempo of the Adagio non troppo, possibly suggesting a faster tempo than that of Richter. Regarding the First's finale see Johannes Brahms: Symphonie Nr.1, ed. Pascall, p.xvi.
(15) For example, Stokowski 1929, Fiedler 1931 and Giulini 1981 all begin the movement at MM 100, and Weingartner begins at an even slower 96; but Fiedler reaches speeds as high as MM 138, Weingartner reaches 126, and Stokowski MM 120, while Giulini stays mostly between 96 and 104, topping out briefly at 108. It is because he sticks to his slow tempo that he takes three minutes longer than Weingartner and two minutes longer than Fiedler or Stokowski to get through the same number of bars. Similarly, Furtwangler 1945, Walter 1960, Monteux 1961 and Masur 1992 all begin this movement at MM 108, but Furtwangler surges as high as MM 138, while Masur and Monteux reach a maximum of MM 120 and Walter of MM 116. Thus Furtwangler's timing is a minute faster than Masur's and Monteux's, and almost two minutes faster than Walter's. (Three of these MM sets come from table 8.2 in Frisch, Brahms: the four symphonies, p.180; others are my own estimates.)
(16) Frisch, Brahms: the four symphonies, pp.166-9.
(17) Some evidence of Brahms's use of acceleration in conducting can be found in his comments on the Vivace of the sixth movement of the Requiem (referenced in n.20), and in the autograph of the Fourth Symphony finale. (The relevant deleted pencil marks are visible in the facsimile of the autograph (Adliswil-Zurich, 1974), finale pp.24, 35.) In discussing these pencil marks, he wrote that when conducting new works that were not yet familiar to orchestras, `I cannot do enough pushing ahead or holding back to come near the desired expression, passionate or quiet' (`Ich kann mir in dem Fall oft nicht genug tun mit Treiben und Halten, damit ungefahr der leidenschaftliche oder ruhige Ausdruck herauskommt, den ich will') (?20 Jan 1886, in Briefwechsel, vi, ed. A. Moser (Berlin, 1908), pp.219-20; trans. based on Rudolf, The grammar of conducting, p.358). However, Brahms then said that once the work has been `completely absorbed' the extra markings of tempo modification become superfluous and should be discarded. Although the letter opposes tasteless extremes of tempo modification it does indicate that, when conducting, Brahms was not averse to `pushing ahead'. (We have ample evidence of his doing so in his solo playing; see, for example, W. Crutchfield, `Brahms by those who knew him', Opus, Aug 1986, pp.15, 21.)
(18) My sample include's the recordings by the Rubinstein/Szeryng/Fournier (RCA), Istomin/Stern/Rose (Sony), Beaux Arts (Philips) and Ashkenazy /Perlman/Harrell Trios (EMI) and the Trio Fontenay (Teldec). All five play the revised version of the op.8 trio, but I compare their tempos to the MMs of the first version on the grounds that the opening themes in all four movements were more or less unchanged.
(19) Another group of quick MMs is for the Violin Concerto, op.77. They come not from Brahms but from the work's dedicatee, Joseph Joachim, who left two sets that are well above the tempos of any recordings I have easy access to. One set is in the solo violin part accompanying the English translation of J. Joachim and A. Moser, Violin-schule, trans. A. Moffat (Berlin and London, 1905), pp.250, 261, 263 and 267. The MMs are: I, etwa 120; II, etwa 72; and III, etwa 96, coda, 120; in a German edition, Robert Pascall informs me, they are 126, 72, 104 and 132. My thanks to Pascall for informing me about the latter set, and for noting that the former set is found in a 1910 printing of the first edition of the solo part of the piano reduction -- but not in 19th-century printings.
(20) I quote M. Rudolf, `A recently discovered composer-annotated score of the Brahms Requiem', Bach: quarterly journal of the Riemenschneider Bach institute, vii/4 (Oct 1976), p.13, n-5.
(21) M. Musgrave, The music of Brahms (London, 1985), p.9.
(22) N. Del Mar, Conducting Brahms (Oxford, 1993), p.209.
(23) Here and later, I use the MMs (for Furtwangler, Klemperer, Walter, Karajan, Kempe, Sawallisch, Norrington and Gardiner) provided in Musgrave, Brahms. A German Requiem, pp.76, 79, 83, except for a few of the third-movement fugue MMs (which are misprinted at twice their value). To this sample I have added my own MMs for Shaw (Telarc), Levine (RCA) and Herreweghe (Harmonia Mundi).
(24) R. Pascall, `Brahms and the definitive text', Brahms: biographical, documentary and analytical studies, ed. R. Pascall (Cambridge, 1983), p.69.
(25) Del Mar, Conducting Brahms, p.205.
(26) See R. Kolisch, `Tempo and character in Beethoven's music', reprinted in Musical quarterly, lxxvii, (1993), and the summary of Hermann Beck's research on this subject in W. S. Newman, Beethoven on Beethoven (New York, 1988), pp.90-97.
(27) `... kann ich Ihnen recht wohl ein Abonnement auf Metronom-Angaben eroffnen. Sie zahlen mir was Gut's und ich liefere Ihnen jede Woche -- andere Zahlen; langer namlich wie eine Woche konnen sie nicht gelten bei normalen Menschen!' Trans. J. Eisinger and S. Avins, in S. Avins, Johannes Brahms: Life and letters (Oxford, 1997), letter 433. Avins dates the letter January 1884.
(28 )My thanks to Styra Avins for reminding me of this. My informal survey of recordings of Brahms by highly regarded Brahmsians found that almost half of their tempos relate to a previous movement's tempo proportionally. However, those tempos that did not relate proportionally were subjectively just as satisfactory as those that did.
(29) D. Epstein, Shaping time (New York, 1995). This book is surely among the most important studies of tempo, even if one questions some of its conclusions.
(30) D. Epstein, `Brahms and the mechanisms of motion: the composition of performance', Brahms studies, ed. G. Bozarth (Oxford, 1990), p.204-5. Alan Forte and Werner Schulze have also proposed proportional tempos in a few specific Brahms works.
(31) Those whose sections are related proportionally include the finales of opp.2 and 5; the Adagio of op.8, first version; the Adagio of op.36; the finale of op.67; the op.73 third movement; ops. 80 and 81; Das Madchen, from op.93a; the Andante grazioso of op.101; op.116 no.2; and op.118 no.5. Some involve changes of metre with no indication of a tempo change: Marias Wallfahrt from op.22; op.74 no.1; op.89; op.92, no.4; the finale of op.98; the Hosanna of the unpublished Missa Canonica of 1856; and Vom heiligen Martyrer Emmeraus, one of the 1869 Deutsches Volkslieder.
(32) D. Epstein, Beyond Orpheus (Cambridge, MA, 1979), pp.94-5.
(33) B. Jacobson, The music of Johannes Brahms (London, 1977), pp.51-2.
(34) Epstein, `Brahms and the mechanisms of motion', pp.210-11, Shaping time, pp.258-85, and Beyond Orpheus, pp.91-4. Evaluation of his examples would require more space than is available.
(35) `la[Beta]e ich in der Pauke die Triolenbewegung, weil diese zu Anfang des nachsten Satzes die Hauptsache ist u. Der Paukist es so sicherer fa[Beta]t'. Letter to Robert Keller, 28 September 1877. Translation and commentary from The Brahms -- Keller correspondence, ed. G. S. Bozarth (Lincoln, Nebraska, 1996), p.3.
(36) R. Pascall, personal communication, 1997.
(37) Epstein, Beyond Orpheus, p.94. Del Mar sees it as a 2:1 proportion, but says that `the "right speed" ... is not exactly double any of the previous speeds', thus requiring `an imperceptible and skillful adjustment': Conducting Brahms, p.16. Epstein (Shaping time, pp.504-5, n-14) observes that Brahms changed the last not of bar 29 to duple-time demisemiquavers in the autograph, thus creating a built-in ritardando; he argues that this supports his proposed 4:3 ratio. However, Robert Pascall (personal communication) points out that the above-quoted exchange between Brahms and Keller shows that Brahms's final choice was that the note should appear in the printed score in its original triplet motion.
(38) R. Knapp, `A review of Norrington's Brahms', American Brahms Society newsletter, xi (Spring 1993), p.6.
(39) Epstein, Shaping time, pp.166-7.
(40) Musgrave, Brahms: A German Requiem, p.77.
(41) Musgrave, Brahms: A German Requiem, p.36
(42) S. Ochs, Das deutsches Gesangverein fur gemischten Chor (Berlin, 1926), iii, pp.168-9.
(43) Another 4:3 ratio exists in both sets of Joachim MMs for the Violin Concerto. The MM of the coda of the finale relates to the earlier part's MM at a 4:3 ratio -- which, impressively, is the ratio that Epstein argued for in this movement without knowing about these MMs. See Epstein, Beyond Orpheus, p.94.
(44) Davies, Some personal recollections', p.184. The score does not clarify matters: the words `not much' appear immediately after `meno Allegro', but so does the MM. I am grateful to Peter Horton of the Royal College Library for this information.
(45) Brahms wrote: `Beim zweiten Presto (3/8) steht doch nicht mehr:. Briefwechsel, x, ed. M. Kalbeck (Berlin, 1917), p.79. Epstein arrived at precisely this proportion, apparently without knowing that the score originally indicated it: Shaping time, p.268.
(46) `Eine Dummheit ist mir aber aufgefallen. Zum Schluss vom ersten Satz nicht Poco sostenuto. Das wird mi[Beta]verstanden und dasselbe Tempo wie in der Einleitung genommen. Meno Allegro soute da stehen'. The translation is from Frisch in Brahms: the four symphonies, p.176. The original is in Briefwechsel, x, 100.
(47) Frisch, Brahms, the four symphonies, p.176. In support of Frisch's hypothesis we might note that Brahms added the qualifying `Un poco' to the original `Sostenuto' marking of the introduction.
(48) The implied tempo, he says, `may seem slow': Epstein, Shaping time, p.536.
(49) The indication `Poco-a-poco' is written over the second presto's retransition to the `Tempo primo' main theme (bars 190-93). This marking would make little sense if the second presto had been at a proportional (that is, identical) tempo to the main theme, so it appears to imply that the presto had a different tempo -- presumably a faster one, in which case the Poco-a-poco indicates a ritardando. My thanks to Walter Frisch for pointing this out to me.
Epstein is aware that the Poco-a-poco indication seems to contradict the proportion under discussion, and he suggests two resolutions on p.269 of Shaping time: `Does it imply a ritard, again of rubatolike character, or does it underline what is already built into the music, the bit-by-bit apparent ritard just described?' The second proposal seems uncharacteristic of Brahms, who was not inclined to add superfluous indications to his scores.
(50) Epstein, Shaping time, pp.202, 241-53.
(51) That Gardiner does this was pointed out in V. Hancock, `Brahms in better balance', Historical performance, v (1991), p.38.
(52) Brahms himself changed his mind about the tempo of `Aber des Herrn': in the autograph and vocal parts, he marked it `Un poco animato', which in his usage indicates a slight increase in speed; for the first edition, he changed it to `Un poco sostenuto', which indicates a slight slowing. That `animato' and `sostenuto' indicate changes of tempo in Brahms is confirmed not only by various markings in his scores, but also by the MMs: that for the `quasi animato' section of the Third Trio is faster than the preceding, and those for the Requiem's `Aber des Herrn' and the `sostenuto sempre' of the Third Trio are slower.
(53) Davies, `Some personal recollections', p.184. Davies's MM here as reprinted in Cobbett's seems to include a mistake, as it is given in crotchets instead of dotted crotchets; I assume that the latter was intended. In a further case of one proportion being substituted for another, Furtwangler, Walter, Kempe, Shaw and Levine create a 2:1 ratio between `Denn alles Fleisch' and 'So seid nun geduldig', instead of the notated 4:3 ratio. Hardly any conductors maintain the notated ratio, perhaps because the MM for `So seid nun' is extremely slow.
(54) Epstein, Shaping time, p.155.
(55) `Machen Sie es wie Sie wollen, Machen Sie es nur schon'. Quoted in Davies, `Some personal recollections', p.184.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Johannes Brahms|
|Author:||Sherman, Bernard D.|
|Date:||Aug 1, 1997|
|Previous Article:||New light on Quantz's advocacy of Telemann's music.|
|Next Article:||Mozart's Piano Concertos: Text, Context, Interpretation.|