The two issues in this movement concern the presence or absence of the first crescendo or “wave” in the coda, and the absence or presence of the extra trumpet notes very near the end of the movement. In the 1872 version, the first wave begins at measure 500 with the first violins beginning their distinctive tremolo-figure on middle C (C4), with a steady crescendo rising and leading quickly to sequences of quarter notes in a decrescendo ending on a sustained low D (D4) for the oboe and a drum roll on G. The second wave begins at measure 532, with the same violin figure but up an octave, starting on C5 and resulting in a much larger crescendo and the concluding events of the movement. For the version of 1877, Bruckner removed the first wave but retained the second wave at measure 488, reshaping the violin part so that it began on middle C as the first wave had done, but then jumping abruptly to the higher tessitura in the last two notes of the eighth measure so that the violins could continue from there as in 1872. And at that same time one measure was skipped in the continuation of the crescendo. In an attempt to present both versions simultaneously, Haas had to suppress the 1872 modification of the second wave, making both choices erroneous, and Nowak did not correct that error. And my edition is the only one to show that measure of 1872 (measure 547) deleted in 1877.
The crescendo of the second wave in the 1872 version and the similar crescendo in the single wave of the 1877 version end the same way, abruptly, followed by one of the many silences that give this composition the name “Pausensymphonie”, symphony of rests (not pauses). One of the 2¾ rest measures of 1872 in this position was deleted in 1877, a process carried out in several other places at that same time; of course Haas and Nowak present only the shorter silence. Then, marked “Langsamer” (slower) there is a plaintive recollection of the principal theme by the cellos, which had played it at the beginning of the movement, lightly accompanied by winds in 1872 and more ornately in 1877 with the clarinet in contrary motion. (Voices in contrary motion were also added in 1877 in several other places as well.) The music continues as the bassoon seems to call together the other winds, and suddenly the peroration bursts upon the listener. In 1872, the tempo for the peroration is “Tempo Imo”, tempo primo, the tempo of the waves and indeed of the whole movement; in 1877, the tempo is “Sehr schnell” (very fast). Of these, Haas chooses the first, meaning that even if all the cuts are taken to approximate the 1877 version, the peroration will be played as it would be in the 1872 version. Nowak, thinking of 1877, takes the “Sehr schnell”, which is what most conductors do anyway as we shall see.
In the sample, the 21 recordings of the 1872 score in my edition all include both waves, and the 15 recordings of my 1877 score and the six recordings of the first print of 1892 all feature one wave, which is the second wave as modified by Bruckner in 1877. Of the 52 Haas recordings, 33 use two waves, departing from the 1877 score, and 19 use one wave, in the incorrect reading supplied by Haas. And of the 54 of the Nowak score, 29 use two waves, and 25 use one wave. These divisions are probably not statistically different. It seems that conductors simply use the number of waves they want to irrespective of the edition. Here, as we shall see in many other places, one simply has no idea what will be present in this passage if one buys an unknown recording using either the Haas or the Nowak score. Two recordings conducted by Okko Kamu in 2014 are noteworthy in that in them he uses the Nowak edition with one wave, but changes the beginning of the wave and omits the extra trumpet notes to agree with my edition. Later I will explain how I know he used Nowak, which of course was obsolete at the time.
As for the speed of the peroration, one would expect that all the conductors reading “Tempo Imo” in either my 1872 edition or Haas’s score would go back from the previous “Langsamer” to the initial tempo of the movement, and that all the conductors reading “Sehr schnell” in my 1877 edition, in the 1892 first print, and in the Nowak score, would go somewhat faster, maybe quite a bit faster. Here following are the statistics. The cited percentages measure the amount by which the peroration tempo incrementally exceeds the overall tempo. For example, if the overall tempo for the movement was MM half note = 60 beats per minute, and the peroration was conducted at 72 beats per minute, the speed-up would be expressed as 20%, (pesroration ÷ overall) – 1, all times 100%.
1872 edition, 21 recordings. Only Eichhorn, Tintner, Jurowski, and Blomstedt in his first performance of the 1872 at Montréal, stay close to the initial tempo, averaging at a speed-up of 8%; the others increase the speed by an average of 26%, for example from MM half note = 60 to half note = 76. Blomstedt in later recordings goes as high as 34% (from 63 to 84), still a modest value as we will shortly see.
1877 edition, 13 recordings. All go faster, an average of 43% faster, ranging from 11% (Paavo Järvi 2016) to 111% (Janowski 2012, more than twice as fast!). It is possible that the three recent Järvi recordings, with speed-ups of 11%, 15%, and 19%, would be thought of as not speeding up, but the others definitely do make quite a difference.
1892 edition, 6 recordings. All go faster, an average of 76% faster, ranging from 32% (Lim, 2016) to an astounding 128%, MM half note = 67 to 152 (Scherchen 1964). And of course all of these speed-ups sound even more extreme because they come immediately after an area where the tempo is slower, often much slower, than the overall tempo from which the speed-ups are calculated. Now for the Haas and Nowak scores.
Haas score, 52 recordings. For this analysis, it might be valid to separate the Haas recordings into two populations, one with speed-ups of less than 20% which will seem to the listener as the maintenance of the pre-“Langsamer” tempo, and those with greater speed-ups in defiance of the Haas indication “Tempo Imo”. Twenty of the recordings fall into the slower group, a few of them with tempos slightly slower than the overall speed, and the average speed-up of those twenty recordings is a modest 8% just like the smallest performance speed-ups for the 1872 version cited above. By contrast the other 32 Haas recordings have an average speed-up of 51%, with two recordings conducted by Leonard Slatkin reaching speed-ups of 118 and 119%, again more than twice the speed of the rest of the movement. That is in spite of Haas’s direction, which follows Bruckner’s original indication to return to the original tempo. This effect is so strong that it cannot be blamed on the irresponsibility of conductors. There is something in the music of the peroration that makes people want to take it fast, and that is probably why Bruckner changed the tempo indication to “Sehr schell” in 1877.
Nowak score, 54 recordings. The score says “Sehr schnell”, and we should consider all the recordings together. The overall speed-up is 63%, as if from MM 60 to MM 98, ranging from 29% (one of the Kamu recordings) to 102% (the interesting Daniel Nazareth recording of 1980 which we will consider again shortly). One Nowak recording was excluded from this group. A 1997 recording conducted by the experienced and cosmopolitan Hiroshi Wakasugi speeds up by only 3%, tantamount to not speeding up at all. In context that seems odd; even his compatriot Takashi Asahina, using the Haas edition and known for his slow tempos, sped up by 22%.
It is possible that conductors who adopted a slow overall tempo would be more inclined to speed up the peroration by a larger amount than a conductor who was already going rather fast. A graphical analysis
might bring some sense of that effect, as revealed by the measurements. First we shall look at the overall tempos in terms of the date of the performance. As can be seen in the first graph, the tempo shows a slight downward trend in time, losing 4 or 5 beats per minute in metronome marking over the sample span of eighty years. This is a commonly-observed phenomenon in Bruckner studies, but it seems rather artificial. The second graph plots speed-up percentage against overall speed. In that graph, conductors who sped up
by only a small amount would be clustered at the bottom just above the horizontal axis, and those who sped up significantly would lie higher in the display, while conductors who used a slow overall tempo would be at the left, and the fast-tempo performances would be at the right. Thus, conductors who used a slow overall speed and a large speed-up of the peroration would appear high up on the left, and conductors who were conducting the movement more rapidly and did not speed up so much would appear low on the right. It seems that the suggestion made above is valid, considering how so many of the performances trend toward the lower right. But there are outliers, the most notable of which is the lone point at top a bit to the right of center, which belongs to the wild and historically valuable Scherchen performance, which uses the 1892 version. He makes a cut in the finale even larger than the immense excision suggested by the “vide” in the score, proceeding directly from the buildup to the recapitulation to the final peroration, omitting the A theme, the B theme, the C theme, and two thirds of the coda in one fell swoop. I am glad that he did not perform any such shenanigans in his recordings of Mahler and Glière symphonies from the early 1950s from which I learned those monumental works uncut. The outlying very fast performances at the lower right are respectively those of Carl Melles 1977, Wolfgang Trommer 1967, and Jascha Horenstein 1969. It is notable that almost 40 years earlier, Horenstein recorded the Bruckner Seventh, conducting the first movement faster than anyone ever has since then, approximating the metronome markings in the first edition which literally everyone in over a thousand recordings since then has ignored.
The bizarre history of the trumpet error at the end of the peroration began with Bruckner himself. In the spring and summer of 1872, as Bruckner was writing the symphony in Vienna, he seems to have simultaneously maintained two scores of the symphony, delivering them to two copyists in Franz Hlawaczek’s copying atelier whom we now know only by their last names as Carda and Tenschert. Carda and a few helpers copied the orchestral parts from the score they had, and Tenschert began the copy score from the score he had. This was done movement by movement during the composition, because Bruckner wrote scores in successive or sequential bifolios, four pages at a time. But when scores were copied, for ease in subsequent binding the bifolios were nested in groups or signatures of several bifolios at a time so that when the binding was stitched together many pages could be fastened together at the same time. And each orchestral part consisted of one group of bifolios stitched together in the same way. Today, we have the Bruckner holograph score that Tenschert used to make his copy score. It was Mus.Hs. 19.474, mentioned and described above, and the copy score he was working on is Mus.Hs. 6035. But we do not have the holograph score Carda was using. That Carda did not use 19.474 is apparent from many differences between it and the parts. There must have been another score, and considering the relative neatness of 19.474, it must have been the real composition score which like the other composition scores that survive, would have quite a bit of scratching-out and rewriting. The composition score of the Fifth is a case in point; it is a veritable battleground of conflicting texts, and we would expect that the real composition score of the Second would be similar. One suspects that Bruckner might have discarded it along with the sketches for all the symphonies through the Seventh and heaven knows what else, when he moved into the Belvedere Kustodenstöckl in July 1895 at the invitation of the emperor.
One of Bruckner’s compositional traits, which his enthusiasts love and his detractors criticize, is frequent repetition. And here in this peroration the first eleven of the twelve measures are identical, part by part. Bruckner’s habit in such passages was to indicate the repeat with special repeat signs, a virgule flanked by dots ̇/ˑ , for all measures being repeated. And that he did here in 19.474, and apparently by error, he put in one more repeat for the staff containing the two trumpet parts instead of entering the concluding single note. This reading was picked up by both copyists in their copy scores 6035 and 6034, and was used by Haas in his edition as shown in the upper staff of Example 2. In Haas’s score the trumpet notes produce an effect similar to a passage in the 1877 revision of the First Symphony at the end of the finale, and thus it sounds entirely plausible, though it is wrong because the notes do not appear in the parts. The fact that these extra notes also did not appear in the first printing of 1892 must not have bothered Haas a bit; it was one more difference between the first publication and his Collected Edition which he could show the court. There are 52 recordings in my sample in which one can hear what that sounds like. But Nowak, in his edition of 1965, favored the 1877 version when he could, and in 1877 Bruckner had caused to be added an extra measure after the trumpet notes, as can be seen in the lower staff of Example 2. This was done by Hlawaczek himself in 6035, in Mus.Hs. 6034 which is a copy score prepared later by Carda, in all the parts, but not in 19.474 which represents only the 1872 version. The result is that the trumpets stop without resolving on to an accented note; it is as if a cavalry charge suddenly fell off a cliff. Bruckner has surprising moments in his symphonies, but nothing else quite like this. Nevertheless Nowak put it into his score, and there are 54 recordings showing what it sounds like in this paper’s sample. But Carda’s orchestral parts did not contain the spurious extra notes, which is immediately apparent upon merely glancing at the parts since the added measure follows a measure in which the trumpets, like every other instrument in the orchestra except the kettledrum, play one note, not six as in the previous measures.
In Haas’s critical report on the Second Symphony, there are quite a number of citations of readings in the parts, and indeed when he inspected them he numbered every ten pages in pencil, and there are 1058 pages in the surviving parts. (In doing the research for my editions, I found I had to number all the other pages.) But for his edition he used the trumpets as they are in the scores, and so did Nowak. However, the reading in the two trumpet parts is valid, because Bruckner conducted from those parts twice with extensive rehearsals, and would have certainly corrected something which would have been heard so easily. It is quite satisfying to see that the eventual first print does not contain those extra trumpet notes. They must have been corrected out in the galley proofs, because there is no evidence of their cancellation in the Stichvorlage 6035. It’s amusing to think how many people could, indeed should have caught this mistake: first Bruckner himself in 1872 or 1873, then Tenschert and especially Carda, then Hlawaczek, then Haas and Nowak, and finally any of the throng of erudite conductors from 1965 to 2017 who used the Nowak score but never thought to question this strange-sounding effect. Bruckner himself finally caught it while supervising the 1892 print near the end of his life in 1892, but Haas put the error back in in 1938 and in 1965 Nowak made it worse, and there it is, still being played.
In the sample recordings, everyone plays what is on the stand: 1872, 1877, 1892 (no trumpets), and Haas and Nowak (five extra invalid but very loud trumpet notes). That is, everyone but three. In 1996, Franz Welser-Möst has the trumpets play one cadence note in the added measure which sounds better but is wrong; Daniel Nazareth in his distinctive recording from 1980 has the trumpets repeat the rhythm through both measures until the orchestral chord, which resourcefully solves the problem but sounds too long and is also incorrect. Takeo Noguchi, conducting the orchestra of the Furtwängler Institute of Tokyo, one of the more than 150 amateur orchestras in that amazing city, is the only one to remove the extra notes cleanly. His recording of the Nowak score dates from 2011 and he could have gotten the idea from my scores which were published by that time just as Kamu did, but I doubt it as he uses two waves in both the first movement and the finale. Then in two other recordings, the second trumpet could not resist adding the cadence note. These are Segerstam 1983 and Thomas Dausgaard’s 2011 performance with the RSO Stuttgart. And two measures later the accidental effect seems to be repeated by Bruckner himself, where the first trumpet alone among the instruments of the orchestra descends an octave for the final note, and gives the impression that the whole orchestra does.