Tempo Studies on Bruckner’s Sixth and Eighth Symphonies

William Carragan
Contributing Editor, Anton Bruckner Collected Edition, Vienna

A paper delivered at an international conference on
The Wagnerian Symphony
Hudson Valley Community College, Troy, New York
November 22, 1996
Revised 2006, 2010, 2016–2019

An earlier, much shorter version of this paper has already appeared in American Record Guide, November/December 1996.

On January 13, 1995, Leon Botstein conducted the American Symphony Orchestra in a most remarkable all-Bruckner concert. The program included the dramatic anthem Germanenzug (The Germans' Campaign), the première of Paul Hawkshaw's new edition of the early Psalm 146, the pastorale Abendzauber (Evening Magic) for tenor soloist, humming chorus, horns, and yodelers, and the Fifth Symphony in the famous (or notorious) Schalk version of 1894. For an old Bruckner hand like me, indeed for the whole audience, the program was full of interest and provocation. But just as interesting was Botstein's proposal that a style of Bruckner performance existed in former times, with quicker and more flexible tempos, and climaxes handled with grace rather than bombast. He went on to speculate that this early style was a casualty of Bruckner's endorsement by the Third Reich, with an impressive, monumental, triumphalist interpretation being then urged by Nazi ideologues as more in keeping with the uses to which Bruckner's music was then being put. Botstein then proceeded to conduct the Fifth very fluently, but at a vertiginous speed, especially in the first movement. After I got over my shock, I began to see the virtues of his approach, and by the time the wonderful chorale arrived at the end, with Schalk's (or the composer's?) amazingly successful supplemental reorchestration, I was quite ready to entertain thoughts of an earlier, leaner, more Schubertian Bruckner. However, I was not sure when and how such a change from early-lean to modern-heavy might have taken place, and I resolved to undertake a tempo study of some of the symphonies to try to find out.

Previous work on the Fourth

I had first to choose a symphony with an extensive recording history, going back into the 1930s, which had copious and explicit tempo indications in the first published score to serve as a basis for evaluating changes. I decided to work with the Fourth, which was recorded by Böhm in 1936, Eugen Jochum in 1939, Walter in 1940, Furtwängler in 1941, and Kabasta and Knappertsbusch in 1944. Nearly 100 recordings of this symphony, in five major versions, have been made since!1By 2010, well over 300.

The first published version of 1888 derives from a manuscript prepared by Franz Schalk (I, II, III) and Ferdinand Löwe (finale) and then given to Bruckner for his comments and revisions. Bruckner entered many modifications in his own hand, including extensions of ideas proposed by Schalk and Löwe. When he was finished, the score was taken to the publisher, Gutmann, with perhaps some final adjustments by Schalk and Löwe.2My source for the history of the Fourth is the recent dissertation of Benjamin Korstvedt (U. Penn., 1994), who is now on the music faculty of the University of Iowa. (2006: Clark University, Worcester, Massachusetts) In the Finale, in which lay my principal interest, the Löwe score contains metronome markings at the beginning (half note = 72) and also at the great unison theme at the first climax (half note = 66). Then explicit directions call for exact half-tempo at the second theme (thus quarter note = 66), with stepwise changes to somewhat faster tempos during the second theme group. At the third theme, the tempo of the unison theme (half note = 66) is re-established, and the tempo of the very beginning (72) is to be resumed only at the mysterious onset of the development. The manuscript specifications of this tempo scheme, at first in Löwe's writing, are continued through the movement in Bruckner's own hand. Do any of the recorded versions carry out these instructions? No!

Still, the earliest performances show some influence of the 1888 tempos, even when the conductor was using the newly-published Haas edition of 1936. Tempos tended to be brisk, though nobody made the great distinction between the tempos of the first and second theme called for in 1888. Then, in about 1953 or 1955, tempos started to become slower, particularly at the great unison theme near the beginning, and at the coda. In the last 40 years, the slowing has become extensive and inexorable, even in the work of a single conductor. At the same time the unison theme near the beginning has in modern performances become so distended and pretentious as to lose all shape and interest.3I presented these results in a paper at the Perspectives on Anton Bruckner II conference at the University of Manchester in England in April 1996. The paper included the following sentence: "This downward trend, already seen in whole-movement timings and conclusively shown in individual tempos, cannot proceed much further without fatally damaging the symphony."

The great virtue of the 1888 tempo scheme is its clear delineation of the structure of the exposition: first theme fast, second theme slow, first theme fast again. These stipulations cannot be ignored without lessening the capacity of the form to carry the listener along. I do not mean to suggest here that the listener must be trained so as to appreciate explicitly the details of formal analysis. Rather, the composer and the interpreter must collaborate in order that the listener may hear the composition most effectively. Thus the composer must create a simple, self-consistent, and complete system of tempo indications, and the obligation is particularly incumbent when the forms are large. In this respect Bruckner's manuscripts, which have been used as the basis of the Collected Edition, are essentially private, unedited documents, with rather few tempo indications. Only the first publications, largely ignored by the editors of the Collected Edition, contain sufficient tempo indications to tie up all the loose ends and bring out the form successfully.

To illustrate these points, my friend Aaron Snyder and I prepared a composite tape, taking excerpts from several performances to recreate the Löwe tempos. The result has a combination of stern dignity and warm, flexible lyricism that added up to much more than the individual excerpts might lead one to expect. We used recordings by Dohnányi and Janowski for the first and third themes, and Walter (1960) for the second theme. Dohnányi's steady grandeur, Janowski's meteoric intensity, and Walter's gentle urbanity together made a remarkable listening experience.4Maybe someone will soon put it all together in one performance! And in listening to the composite tape, one finally understands the secret of the 1888 scheme: the basic pulse is maintained throughout the movement, because of the strict half-tempo, and the movement sounds like a continuous and integrated structure rather than a concatenation of different and disjunct formal elements. Thus, conductors who do not follow these tempo indications, for whatever cause, are producing fragmentary, episodic readings as a result, and are contributing to Bruckner's undeserved bad reputation as a poor formalist.

An incidental result of the study was the demonstration that National Socialism had nothing to do with the change of tempos from fast to slow. Conductors with Nazi sympathies played both fast and slow; so did stern anti-Nazis! If the Nazis wanted more monumentality in their Bruckner, at least in this symphony there was not sufficient time during the thousand-year Reich to develop such a tradition. But, as we shall see, other symphonies have different histories.

The Sixth

The Sixth is a particular favorite of mine, perhaps dating from my conversion to Bruckner's music in 1951 while reading Donald Francis Tovey's two splendid essays on the Fourth and on the Sixth. As with the Fourth, the first publication of the Sixth contains detailed tempo indications not present in the manuscript. This edition was prepared by Cyrill Hynais before Bruckner's death, though it was not published until 1899 after Joseph Schalk had made many more contributions. The tempo indications are rather more modest in number than in the Fourth. The most important ones are those for the first, second, and third themes: Maestoso5Majestoso in the manuscript, and in the Haas and Nowak editions., Bedeutend langsamer, and Gemässigtes Hauptzeitmaß, etwas breit. The first word means "grandly" but a moderately quick (allegro) tempo is implied by the nature of the music. The second theme is to be played "significantly more slowly", the third at a tempo slightly slower than the beginning and rather broadly". Although the first two indications are present in the Collected Edition publications as edited by both Haas and Nowak, the designation indicating that the third theme is to return nearly to the opening tempo is missing. Yet following that instruction is essential in bringing to life the form of the movement.

We do not have many early recordings of the Sixth, but in my own library I have two scores that offer some early information. One is a copy of the Universal-Edition printing of the Hynais score of 1899, which at one time belonged to Kapellmeister Heinz Norfolk in Vienna, and at another time to Heinrich Huber, probably also in Vienna, and in it someone entered time indications every 15 seconds from the beginning of the first-movement recapitulation to the movement's end. These time indications must have been taken down during a performance. My friend Mark Kluge checked the published records of the Vienna Philharmonic, and found that there was a broadcast of the Sixth conducted by Franz Schalk in 1927. Now a broadcast, heard in a brightly illuminated home, would provide a much better opportunity to make such notations than a dimly-lit concert hall. It would be fascinating if these notations were in fact relics of that Schalk performance! Certainly they do come from a quite early event, and they correspond to no recording of the Sixth of which there is any record. They conclusively indicate a sharply faster third theme.

Reconstruction of the movement from the Norfolk tempos, making reasonable assumptions for the development and calculating the exposition to be just like the recapitulation (allowing for different length of theme groups) gives a time of 14'23" for the whole movement. In discussions with me, Benjamin Korstvedt has insisted that the underlying logic connecting the tempos of the A and B themes is that the quarter note be maintained steadily, that is, the ordinary quarter note of the A theme is equal in duration to the triplet quarter note of the B theme. This concept is borne out adequately well by the pencil indications present in the Norfolk Hynais score, where A = 72 and B = 50. Later, we will use it as the cornerstone in our attempts to reconstruct movement durations simply from a postulated initial tempo.

The other score is a copy of the Haas score signed on the first page of music by none other than Paul Hindemith in 1952 at Linz, when he conducted the Sixth there. On the facing page someone kept a record of the movement timings for both the performance and the rehearsal, and these timings reflect an extremely rapid reading. Indeed, in the score Hindemith encircled the word Majestoso on the score, and wrote next to it “12 Minuten einsparen”, in other words, “Can you spare 12 minutes?” (really, “save 12 minutes”).

Now, he did not quite attain a speed which could result in a duration of only 12 minutes. Instead, the timing of the first movement was 13'0" in the rehearsal, 12'45" in the concert. Still, we may conjecturally reconstruct the tempos he used, in the following way: Assume a tempo at the start, 58, or 66, or even 72 (!). Then the B theme is at 2/3 of the initial tempo, the C theme is at a moderated 11/12 of the initial tempo, and the codetta is at 10/12 of the initial tempo. Then the development begins at the codetta speed, and continues at the C-theme speed (at the A minor place), attaining the A-theme speed at m. 195. Actually the C-theme tempo is reestablished at m. 175, the C major place. Incidentally, the Norfolk conductor didn't seem to accelerate here much, though. Anyway, whole-movement durations depend on the initial tempo in this way:
Initial tempo 72, movement length 12'28"
Initial tempo 70.4, movement length 12'45"
Initial tempo 69.1, movement length 13'0"
These leave room for plenty of rubato, and a really big ritard at the end. As for the individual tempos Hindemith used, putting in 50 for the B theme, and also adjusting the ratio at measure 147 to be 25/36, means the following:

Hynais Hindemith perf. Hindemith reh.
A 72 69.4 68.0
B 50 48.2 47.2
C 66 63.6 62.3
movt. length 12'17" 12'45" 13'0"

The Hindemith score provides a number of interesting insights. First, some industrious character was interested enough in this symphony to time his performances, and get his autograph. Secondly, this same person compiled a list of errata, some of which apply to the score and were corrected by Nowak. This was either before or after the appearance of the Nowak score in 1953. Perhaps it was before, because (a) the Hindemith performance was in 1952, and (b) it would only have mattered to historians what errors lay in the Haas materials after the appearance of the Nowak edition, not to practical musicians.

But most important to me is the expectation that Hindemith seems to have had, to conduct the movement in 12 minutes, i.e. in a time compatible with the tempos of 72, 50, [66] suggested in the manuscript Hynais score, but never entered into the published score. Now, Hindemith was a very rigorous person, and he certainly had an industrious and analytic mind of his own besides having been in his youth quite a rebel. But he was also deeply dedicated to the opportunities of performance. However, I submit that he never could have proceeded from the Hynais tempos to a movement length of 12 minutes by calculation, as I have done. It would simply require too much work with the tools then available (slide rule at best). Maybe one could say, 48 plus 50 measures at 72, 52 plus 8 plus 40 measures at 50, and the rest, 171, at 66; this would be 98/36 plus 200/50 plus 171/33 equals 2.72 plus 4.00 plus 5.18 equals 11.90 minutes, actually  11.904 by the calculator. Perhaps it could have been done. But would he have trusted it to be proper to assign all of both third theme groups, most of the development, and all the coda to 66? If one breaks it down much farther, the ability to do the calculations belongs to a scientist, not a musician of even Hindemith=s capabilities, I would think. (The reason the time is not longer is that most of the material assigned to 66 would be played no faster, and likely slower than that.)

Hindemith is much more likely to have gotten 12 minutes as an ideal from experience, not calculation. And this experience is quite probably connected with a tradition of playing the piece fast enough so that Hynais=s tempos, and a movement length of 12 minutes, are not ridiculously unattainable. In this way we obtain a look into the past, just as we had with the Fourth where there were many more early recordings (for the complete Sixth we have only Georg Ludwig Jochum and Volkmar Andreae before 1950, a sad fact). So Botstein was wrong when he suggested that the Nazis always slowed Bruckner down, but he most definitely was right when he said that much of Bruckner used to be played quite a bit faster than it is today.

Recordings that have third themes at a tempo comparable to that of the first theme include those of G.L. Jochum (1944), Swoboda (1950), Swarowsky (1960), Kubelík (1971 and 1981), Sawallisch (1981), Dohnányi (1991), and Barenboim (1994). The Barenboim has been extravagantly praised, but I think the Dohnányi is even better, with a distinctively faster third theme in a very bright and elegant performance. The Sawallisch has usually gotten bad reviews, based on the confused ambience and a certain rough-hewn quality to the playing; but I enjoy it for its good overall organization. Answering an internet request I made for specific comparisons, David Lampson wrote to me that “Sawallisch’s reading of the first movement is energeticCfull of intensity and passion. Even with the brisker tempo, the introduction of the theme brings a dark, almost mystical feel to the music. Klemperer's approach is entirely different. With a more relaxed tempo, the deliberate rhythmic pulse that opens the movement evokes majesty, while the treatment of the thematic material brings to mind the romantic notion of the grandeur of nature. At times pastoral, at others dynamic, Klemperer's vision is wide open and un-mysterious.” The Klemperer recording to which he refers is the EMI from 1964, where the performance starts very slowly, continues with one of the faster second themes, and barely speeds up at all for the lumbering, heavy third theme. As for the ending of the movement, Lampson says: “Sawallisch’s is a nearly violent climax, a resolution to all the energy that has been building in the movement. Klemperer's is noble and magnificent.”

A totally different reaction is offered by David Griegel: “I listened to the Dohnányi again the other night in order to refresh my memory, and the most impressive thing is how well it’s played. The Cleveland Orchestra seems unmatched for precision, and the balance between winds and strings sounds pretty much ideal. (The brass isn’t allowed to get punchy in the first movement, for example.) But Dohnányi’s interpretation of the Sixth is pretty ordinary, quite unlike his incredible, exciting interpretation of the Ninth. I don't particularly like Dohnányi’s tempos; they seem too fast to me. I like a slower approach to the Sixth, one that brings it into the same sound world as the Seventh, Eighth, and Ninth. For that reason, my two favorite recordings are Celibidache/Munich Philharmonic Orchestra/Sony 48348 (VHS) and Eichhorn/Bruckner Orchester Linz/Camerata 345 (CD). I think I like the Celi best. The Adagio is very slow, but this makes it sound more like the Adagios of the later symphonies. The center of gravity of the entire symphony is thus shifted to the Adagio.” Finally, Benjamin Korstvedt had this to say: “I sat down and listened to the Cleveland/Dohnányi recording of the Sixth. It is truly remarkable. At first it struck me as a bit dull, but soon the marvellous clarity of the texture became apparent. Dohnányi really balances the orchestra very well, they play extremely well (markedly better and much more cleanly than the BPO for Barenboim), and the attention to detail and inner voices is great. Chamber music Bruckner. ...For really taking the score apart and putting back together they do well indeed. I can't imagine that Sawallisch and the Bavarians do quite the same thing.”

I know all these people and respect their opinions, but my opinion is closest to the last. For me the structural issue alone makes the Dohnányi greatly superior, although the precision and elegance of the playing of the Cleveland, on which all agree, do no harm. From the structural standpoint the Dohnányi is equalled only by the glorious Kubelík performances, with the Bavarian State Orchestra on Originals, the Chicago Symphony on their own label, and the Cleveland Orchestra (a radio performance of which some people have tapes). The issue here is not one of creeping slowness, but instead whether the third theme is to be fast. And certain conductors make it so, with no other indication of familiarity with the oft-derogated but immensely valuable Hynais score. Perhaps the criterion being applied is that of a consistent beat, in which the quarter note is uniform throughout the movement. This would require the tempo to change from a beginning tempo of 60 (say) to a tempo of 40 for the second theme group, returning to 60 (or perhaps 56) for the third theme. It is curious that the printer's score of the Hynais edition includes a pencil-scrawled 72 for the first theme, and 50 for the second theme!6Benjamin Korstvedt, personal communication. Some other performances that do not get much faster at the third theme still make enough of a difference there for a new world to be created. One of these is Georg Tintner's recording with a Moravian orchestraCan interesting performance that bodes well for the huge forthcoming Naxos series.7Tintner plans to record all symphonies in all versions! He has started with another recording of the Sixth, and recordings of the early versions of the Second and Eighth. (Note in 2010: before his death he did get through all the symphonies, but not in all versions.)

The Eighth

The manuscript of the finale of the Eighth, found in Wn Mus.Hs. 19.480, has three metronome markings. They are reproduced in the Nowak editions of both the 1887 and 1890 versions, the composite Haas edition, and the first published edition of 1892. One occurs at the beginning (half note = 69), one is at the second theme (half note = 60), and one is at the recapitulation of the second theme (60 again). In the 1887 score there are eleven other stipulations, in the 1890 yet six more, while in the 1892 score there is a grand total of 65 tempo indications including the three metronome markings and their associated verbal expressions. The 1892 indications are not in conflict with the earlier scheme, unless one insists that absence of an indicated tempo change requires that no change be permitted. Just as in the previously-studied cases of the finale of the Fourth and the first movement of the Sixth, the third theme group is indicated to be played at the opening tempo (Erstes Zeitmaß); this is at m. 129 of the 1892 score (1890 m. 135).8If the indication had been kept in the Collected Edition scores, it might have served as a curb on today's conductors, who universally take the opening of the movement at tempos far exceeding 69. This matter is further discussed below. The added indications go on to specify what should happen concerning the lovely chorale in E major which appears 24 measures further on (1892 m. 153, 1890 m. 159). The chorale is marked Feierlich, innig in all three scores (1887, 1890, and 1892), this indication surely meaning that the tempo should be slowed substantially. However, only the 1892 score states where the normal tempo of the third theme is to be resumed, and that is not until an additional 24 measures have elapsed after the beginning of the chorale. By contrast, modern conductors who do not take account of the first published version habitually wait only eight measures to resume the previous tempo. Thus the great outburst in B flat, serving ultimately as the dominant of the exposition's closing tonality of E flat, and containing Bruckner's only use of the contrabassoon, loses the significance it would otherwise have as a place where both the tempo and the dynamic level must be suddenly stepped up.9The dynamic in the 1892 score at this point (m. 129) is mf, with a gradual crescendo, not attaining ff until 20 measures later. This is in direct conflict with the ff marcato of 1887 and 1890 at the beginning of the B-flat section, and can perhaps be derogated as a result. This kind of “orchestral crescendo” was also suggested for places in the Second and Sixth in the first printed editions, always in conflict with Bruckner's earlier specifications. These crescendos might be related in style to the huge swellings asked for by Löwe in the Adagio of the Ninth. I am extremely skeptical about them. Unlike the tempo indications, they have neither the necessity from the standpoint of form, nor the consistency with Bruckner's own methods, which should be required for them to gain general acceptance.

Another area in which the 1892 tempo indications clarify matters is in the alternation of thematic material in the first part of the development. This alternation is a trick which goes back to Schubert and before; Bruckner was to use it again in the first movement of the Ninth, and almost certainly also in that symphony's tragically uncompleted Finale. Here it consists of free development of the C1 theme, starting with its inversion at m. 285 (1890; the 1892 numbers are 6 less) and resuming with its tail-motive at m. 309, alternating with statements of the A2 theme, first at m. 301 and later at mm. 323 and 333. This structure is brought out by the tempo stipulations as shown in the following table.

Tempo Indications at the Beginning of the Development

291 C1 E flat minor bewegt, doch nicht eilend (1892)
297 nachlassend (1892)
301 A2, C1 E flat minor Erstes Zeitmaß
309 C1 →V(F minor) etwas breiter
317 nach und nach wieder belebter (1892)
323 A2, C1 F minor a tempo (ziemlich lebhaft added in 1892)
332 Column 2 Value rit. (1892)
332/333 Column 2 Value [Luftpause!] (1892)
333 A2, C1 G major = V(C minor) a tempo (1892)
345 new section Column 3 Value
Column 1 Value Column 2 Value Column 3 Value

The 1890 stipulation etwas breiter at m. 309 is the only indication in that score that the C1 sections are to be taken at a slower and more flexible tempo than the sections in which the discursive C1 accompanies the grand, stern A2. However, the markings in the 1892 score make the desired contrast in feeling between the two types of music very clear and explicit. And the Luftpause of 1892 is high drama indeed when properly prepared.

One more distinctive feature of the 1892 tempos must be detailed. This is a most remarkable alternation of beat lengths, reserved for the central climax of the developed first theme. This passage, which has perhaps the highest level of perceived conflict in the entire symphony, reaches through sequential extensions of A1, accompanied by a string figure derived from A2, the point at 1890 m. 468 where A2 itself must soon be heard from in the brass. In that measure the time signature is changed to 4/4 (sehr breit), and in the next back to 2/2 (a tempo). From this point on the measures alternate in time signature until in a three-measure group beginning in 4/4 an accelerando (belebend bis...) is made to an a tempo and 2/2. At the same time the thematic material alternates between A1 (broad, stately) and A2 (rapid, brilliant). Few conductors attempt to achieve this amazing and phenomenal effect, and fewer still are able to make it work. One of the most notable is Furtwängler who brings it off with incredible panache. Of course there is nothing like it in the 1890 score, but listeners with open minds will be thrilled by it when they hear it.10This passage requires for its execution on these terms a degree of faith in Bruckner's revision and a courage and selfconfidence which many conductors do not have. The case is similar to that of the mooted ritardandos restored by Nowak in the A theme of the finale of the Seventh, prompted by a letter of Bruckner and using the first printed edition as a source. Recently a young but well-known English conductor told me that these ritardandos were too difficult. I countered by saying: "Maestro, that may or may not be true, but they are authentic." Suggested in the 1892 score (by use of vi- -de marks) is a substantial excision of 58 measures (1892 measures 519-576, corresponding to 1890 measures 523-580). The effect of this cut is to remove the concluding, questioning climax of the first theme group, as well as the entire second theme group; the movement cannot be considered to be in sonata form if this cut is made. The cut must have been taken more often than it should have been. I possess an Eulenburg score of the 1892 version of the Eighth, sold to Fritz Eigl around 1920, in which in pencil someone suggests that the cut be taken six measures earlier, at 1892 measure 513 (corresponding to 1890 measure 51911There are 6 measures of introductory material in 1892, but only 4 in 1890. In this respect the 1892 score is closer to the 1887 score.). This additional cut substitutes the 6/4 beginning of the third group for the 6/3 continuation of the material of the first group, creating an ugly solecism in voice leading which if heard in performance could not enhance Bruckner's reputation.12The shortening of the "fantasy on the second theme" in the development of the Finale of the Second Symphony, executed ad hoc for the performance of 1876, also created a voice-leading problem, in this case parallel fifths. In 1877, the cut was revised slightly and ratified, with a removal of the parallel fifths in word if not in spirit. Apparently Haas thought this adjustment was late and corrupt, because he did not follow it and thus forced Bruckner to stay with the parallel fifths he loathed; this was between measures 327 and 328 of the Haas edition, and was retained by Nowak. The new Collected Edition score of the late version will contain the linkage as modified under Bruckner's instruction in 1877 and printed in 1892. Note, however, that this is not the only case where a first-published edition suggests a huge finale cut that would eliminate the Gesangsthema! There is one suggested for the finale of the Second that eliminates the entire recapitulated first and second theme groups and skips to a point well within the third, and of course the Schalk version of the Fifth also eliminates the reprise of the Gesangsperiode, creating a much longer stretch of continuous fugal writing than Bruckner ever envisioned by himself.13I believe strongly that these vi- -de cuts were intended for emergency use only, as Bruckner himself wrote when proposing a similar cut in the Second in 1877: “Nur im höchsten Nothfalle!” Otherwise, there would be no way to explain why the editors of the first published versions provided tempo indications which so elegantly support the listener's comprehension of the form, while at the same time providing conductors with the means to mutilate that form beyond recognition. The 1894 score of the finale of the Fifth is of course sui generis and must be loved or hated on its own terms.

How are the tempo indications carried out in performances? To start with, there are the three metronome markings present in every edition. Every conductor of this symphony has seen these markings. But do any of them follow them? No, of course not! Instead, a tradition has developed in which the A theme is played quite rapidly and dramatically at the outset, and after about 12 minutes the Gesangsthema (B theme) enters in the guise of a renewed slow movement. Unfortunately the slight resemblance of the B themes of the Finale and the preceding Adagio contributes to the impression that the Adagio has been resumed. It is curious that although only in the rare 1892 score is there an indication for the C theme to return to the initial tempo, most conductors do get faster there. But they rarely come anywhere near the tempos they use for the opening theme, which can be as fast as 94! To my mind such hurried tempos falsify the movement completely, no matter how exciting they may seem at the moment. At present I know of only two recordings in which the opening is performed slower than 70. These are the Knappertsbusch studio performance of 1963, an exact half note = 69, and Klemperer's 1970 performance on EMI, an astounding half note = 63. The Klemperer is outrageous not only for his molasses-like tempos, but also for his cuts; he excises the codetta and three quarters of the development, as well as the entire third theme group in the reprise. (His Cologne recording has none of these flaws.) Koussevitzky made similar cuts, but they were necessitated by the 50-minute length of the radio presentation for which the recording was made. It is too bad that we do not have the complete recording.

The 1963 Knappertsbusch recording was described by the late Jack Diether in American Record Guide as sounding “exactly like a slow-motion film.” He went on to say, “It's a magnificent bore, but magnificent!” Perhaps his reservations came from the fact that after directing the first theme at exactly the specified tempo of 69, Knappertsbusch goes on to conduct the second theme at 42 instead of 60. In order to apply a systematic test to the recordings of the study, I applied two criteria to the measured tempos. The first was the “Bruckner criterion,” or criterion of 1887 and 1890, which measures the deviation of tempo of the opening from 69 and that of the two Gesangsperioden from 60. The other was the “Oberleithner criterion,” or criterion of 1892, which also included the deviation of the C theme from a resumed 69. The “winners” (low scorers) when these two criteria are applied are:

Bruckner Criterion

Column 1 Column 2 Year Column 4 Br.Cr. Ob.Cr.
Furtwängler Vienna 1954 Hunt 355 6.7 9.5
Furtwängler Berlin (Titania) 1949 M&A 624 10.2 9.8
Karajan Berlin 1975 DG 419196 11.5 11.5
López-Cobos Cincinnati 1993 Telarc 80343 11.9 12.3
Asahina New Japan Phil. 1993 Fontec 9054 12.1 8.2
Furtwängler Berlin (Dahlem) 1949 Ent. 4135 12.3 15.0
Karajan Vienna 1988 DG 427611 12.5 11.7
Karajan Vienna 1965 Arkadia 705 12.6 15.3

Oberleithner Criterion

Column 1 Column 2 Year Column 4 Br.Cr. Ob.Cr.
Asahina New Japan Phil. 1993 Fontec 9054 12.1 8.2
N. Järvi London 1986 Chandos 8843/4 13.6 9.4
Furtwängler Vienna 1954 Hunt 355 6.7 9.5
Furtwängler Berlin (Titania) 1949 M&A 624 10.2 9.8
Kubelík Bavarian 1963 Orfeo 203891 14.6 11.4
Karajan Berlin 1975 DG 419196 11.5 11.5
Karajan Vienna 1988 DG 427611 12.5 11.7
López-Cobos Cincinnati 1993 Telarc 80343 11.9 12.3
Schuricht Vienna 1963 EMI 767279 13.4 12.5
Horenstein Vienna 1955 Vox 5504 16.8 12.8

As the Oberleithner table shows, the recording that comes closest to the 69-60-69 contour specified in 1892 is Takashi Asahina's with the New Japan Philharmonic on Fontec 9054. As with the Fourth, Furtwängler's readings are close to the old stipulations, though the clear winner is the 1954 recording in which he actually used the 1892 version. The high scorers, those conductors most out of compliance with the metronome markings, include Knappertsbusch, Koussevitzky, Walter, Albrecht, and the egregious EMI Klemperer of 1970, with Bruckner and Oberleithner scores in the high 20s or even over 30. Some will be puzzled to see Karajan on the list of low scorers; he comes nowhere near the short lists for the other symphonies. Karajan does tend to play everything legato, which is particularly irritating in the scherzo.14Many listeners would rather have the tempos wrong and the articulation and the mood correct! The 1944 Karajan (Prussian State Orchestra) is extremely slow, bearing out Botstein's ideas, but over the years Karajan's Eighth grew steadily faster, tending to come closer to satisfying the demands of the metronome markings in the score. It is particularly interesting that there is no historical trend of tempos, or criterion scores, over time. Evidently the non-authorial tradition described above, that of a very fast A, a very slow B, and a moderate C, was already established before recordings started to be made. If the slow and ponderous Karajan performance of 1944 is a product of National Socialist ideology, it did not catch on at the time, and its perpetrator gradually abandoned its principles over the many years of his continuing career.

The performances led by Knappertsbusch are endlessly fascinating and enriching, even though they satisfy hardly any tempo criterion one may be able to establish. With little of Furtwängler's intellectual approach, they still have a personal quality of earnestness and lyrical grace that overcomes any objection one may have to the occasional sloppy entrance. Other performances of interest must include the Mravinsky, now available in the second large set of CDs (the Ninth was in the first set), the SuitnerCa splendid reading newly availableCand the Eichhorn with the Bruckner Orchestra of Linz in top form. Asahina's Eighths are his best work, and the more recent recordings seem to contain somewhat better playing than the earlier ones. The Fontec set of Symphonies 4, 5, 7, and 8 is lavishly presented and quite attractive. The performance of the first theme at 69 and the second theme at 60, returning imperceptibly to 69 as the third theme approaches, would produce a far more authentic allegro feeling than the frequent present practice, which as we have seen is to start with a minute-and-a-half of hysterical activity at 88, and then begin another slow movement at a tempo of 46. One must remember that Bruckner, a symphonist among symphonists, revered and imitated Beethoven, and Beethoven concluded his symphonies with robust allegros. Robert Simpson's praise of Celibidache for transforming the finale of the Fourth into another adagio is misplaced. It is possible, of course, to see beauty in a performance at any speed, but I should think that the beauty that proceeds from the composer's own specifications might have a chance to be greater and more profound. I must conclude this paper with an earnest plea to conductors: become acquainted with the complete and self-consistent stipulations in the first publications, and follow the tempo indications in the score!


I am grateful to David Aldeborgh, the director of the Bruckner Archive15Note from 2010. With Mr. Aldeborgh=s gracious permission, the Bruckner Archive name has recently been taken over by John Berky of Windsor, Connecticut, for his huge and developing collection of Bruckner recordings, scores, books, and memorabilia. at Poughkeepsie, New York, and to Mark Kluge, both strong supporters of the first published editions, for supplying many recordings not in my collection, and to them and Aaron Snyder, Benjamin Korstvedt, Paul Nudelman, and Julia Carragan for helpful suggestions. I am also thankful for the help given by Hirochika Mochizuki in obtaining for me an almost complete set of the Bruckner recordings of Takashi Asahina.