“There is, strictly speaking, no such thing as how one should conduct Bruckner. There is really only a right or a wrong way.” This quotation, from an unexpected source, Anton von Webern, suggests that at the time he made the comment, sometime in the later 1920s, the “right” way to play Bruckner should have been self-evident upon study or inquiry. After all, at that time there were still many people alive who had heard the symphonies played during Bruckner’s lifetime, with Bruckner present in the hall, and were aware of the traditions which had been inaugurated when the composer was able to give advice.

But can we, separated from these events by over a century, determine what that authentic style might be? Important volumes have been written on performance practice as it may have been in earlier times, yet the romantic period is not extensively studied, as if what we hear today were essentially the same as what one heard all of a hundred years ago. This is because the basic technique appropriate to most instruments was developed in the early years of this century–for example, for the violin by Leopold Auer and his students–at a time when romantic music dominated the concert hall, and the assumption is that these traditions are strong enough to have resisted change. Yet I can remember from my youth many expressive devices and techniques employed by violinists, pianists, and conductors which would today be considered undesirable mannerisms or affectations. The portamento or gliding of pitches by violinists, the anticipation of the bass note by pianists, and the inevitable accelerando accompanying every crescendo imposed by conductors, are heard in many historical recordings from the 1930s, but are hardly encountered at all today except in unkind caricature. As for tempos, many of the earlier recordings seem quite alien to us; they were made at a time when velocity was considered to be a necessary mark of a worthy performance, while today in the Age of the Adagio we seem to consider ourselves too refined for such athleticism.

Yet it is worthwhile to reconstruct the bygone style, the sound understood and expected by the composers of the most elaborate orchestral music ever written. The value of such an approach is shown in the spectacular recordings made by Barry Wordsworth and his New Queen’s Hall Orchestra; their account of the Tallis Fantasy of Vaughan Williams made me feel as if I had never heard this beloved piece before. Some of the restorations are simple, such as the seating of the second violins on the right, and some of them are very difficult to attain, like the creation of a gliding style of violin-playing which is both effective and tasteful.

Even if these efforts are made, though, establishing the place of Anton Bruckner’s music in the milieu of romantic musicianship is not simple. Attempts by certain historically-minded conductors to produce a more authentic Bruckner have been greeted by most listeners with skepticism, which in my opinion is to some extent deserved, as (paradoxically) these conductors seem to have made many interpretative decisions arbitrarily, without sufficient research into what evidence there may be.

The first movement (Kopfsatz) begins with a beautiful, lyrical and long-breathed melody which Bruckner told his friends he heard in a dream. The time signature is 2/2, and the tempo indication in the Gutmann score is Allegro moderato, minim (or half-note) = 58. From my measurements it can be said fairly that only a few recordings, those of Horenstein (1928), Ormandy (1935), Kabasta (1942), Rosbaud (1957), and perhaps Szell (1968), come anywhere close to following the composer’s wishes, minim = 58. The first three of these conductors use the Gutmann score; Rosbaud uses Haas, while Szell is using Nowak, and at any rate is hardly faster than the Haitink performance of two years earlier. (It is amusing that the commentary supplied with the Haitink set states that the version is Haas’s, but the brass is missing at measure 123 and there is a chord at 149 and the version is Nowak.) On the other hand, eleven conductors, using all three versions, begin the symphony at 40 or under, a tempo which cannot in any way be called “allegro moderato”. Horenstein is the only one over tempo, and upon first hearing, his speed gives the impression that the performance is to be just a rigid run-through. Yet this fast tempo enables him to take certain expressive options unavailable to the many conductors who start the symphony below 50.

The B and C themes are indicated in the first publication to be slower. The time signature changes to 4-4 upon the entrance of the B theme at measure 51, and the tempos of the two themes are to be respectively crotchet (quarter note) = 108 and 96, equivalent to minim = 54 and 48 for comparison. Although the development begins with an allusion to the A theme, it continues at the slower tempo until, again according to the first publication, the initial tempo is re-established with the A theme being played in C minor, fortissimo, in inversion, at measure 233 in the development. The indication at that point is Tempo I, and the time signature changes back to 2-2. The A theme is then treated more quietly, but presumably at the same tempo, until the recapitulation is reached at measure 281, and the theme is presented upright and inverted simultaneously, with a modest decoration in the violins and first flute. The indication of Tempo I. at measure 233 is the key to the proper performance of some 86 measures that follow it. By contrast, Nowak writes “molto animato” at this point following a manuscript indication⁴, while Haas writes nothing. What is the result? The measurements show that the average here is in the high 50s; indeed, from the calculations it is about 57. So at this point the conductors are, perhaps belatedly, playing the theme at the right speed. However the standard deviation is large. A further set of measurements established the relationship between the A theme at the beginning and the A theme at measure 233 is for each performance. According to the first publication, and also according to Haas who has none of these internal tempo changes, this ratio should be unity, while Nowak’s indication might suggest that the theme should be faster (or at least suitably fast) at 233. However, everyone has a ratio greater than one, with three of them, one using each version, occupying a place in the stratosphere. This effect is created by the slow tempos used at the beginning, not by a too-fast tempo at m. 233, as we saw from the average calculated above. The lowest ratios are Horenstein (1928), whose tempos are 63 and 66, Haitink (1966) with 52 and 56, and d’Avalos and Karajan (1989) whose data points overlap although d’Avalos plays quite a bit slower than Karajan. The high ratios are Böhm (1943) using the Gutmann edition, Hindemith (1960) using Nowak, and Sinopoli (1989) using Haas-ized Nowak parts.

Sinopoli creates quite a problem for himself with his egregious velocity at measure 233, 1.93 times as fast as his very broad opening tempo. The inversion of the theme at 233 is very stern and fast, but at measure 249, where the second violins play the melody and the firsts a decorative counterpoint, his tempo suddenly becomes slower and the listener is caught unawares. As the tempo further subsides, the indicated first-theme speed of 58 is crossed at about measure 261, at the D minor entrance. The melody is quite well shaped under Sinopoli’s direction, and the listener can observe that it would be quite practical to play it at that speed anywhere. At the recapitulation in E major, measure 281, the tempo has sagged to about 54, but the first-violin-and flute decoration is still played rather perfunctorily.

I believe that the correct thing to do would have been to start the movement at 58, return to 58 at 233, and then yield a little after 281 so as to give the decoration enough time to sound gracious. This is what one would do following the first publication, and perhaps even following Haas if one kept a steady tempo through the whole movement as, for example, Otmar Suitner (1980) does in his sturdy performance–but not using Nowak, whose “molto animato” creates a real problem. The Sinopoli performance is remarkable in that it contains many of Haas’s details, despite the statement Edition: Leopold Nowak supplied by Deutsche Grammophon. There are brasses at 123, or at any rate at 125, though the low brasses at 125 don’t seem to be playing the Haas music (they seem instead to be following the same line as is in the Karajan recording of 1970/71); the wind chord is missing at 149, the first-violin doubling is missing at 154 and 156, and for good measure, in the slow movement the strings start playing pizzicato halfway through 217, not halfway through 216 as in Nowak and Gutmann. Yet he uses the percussion earlier in the Adagio, in the famous passage beginning at 177. For his performance he seems to have started with Nowak and Haas-ized it to some extent, rather than the reverse, because of the “molto animato” which no conductor would introduce, with all of its problematic implications, into a score which did not already contain it. (Nor indeed, in spite of what Bruckner wrote to Nikisch, should one expect any conductor, no matter how inspired, to introduce the tempo changes in the A theme of the finale on his or her own.) One should not criticize the Sinopoli performance too severely, as the melodies are quite lovingly shaped, but the absence of clear tempo indications in the score he used seems to have gotten him into trouble.

Another important aspect of the large-scale first movement tempos is the relationship between the themes, each one slower, and more complex, than the preceding one. That preference of Bruckner’s, expressed in metronome markings and time signatures, is not observed by any conductor in this study. It is true that the B and C themes are usually played at tempos close enough to those specified to be considered in accordance with them, but not a single conductor slows down when entering the B theme, deciding instead to conduct a four-beat pattern where the minims are faster, rather than slower, that those in the earlier two-beat pattern. How can this be done? It must be that conductors are affected by the story about the first theme appearing to the composer in a dream. Nonetheless, if the movement is to be considered a sort of Allegro, and if the relationships among the themes are to be those specified by the composer, the speed of the opening theme must be stepped up considerably above today’s custom. And the curious fact is that the passage in the development, following the outburst at 233, shows in one recording after another that this theme can be perfectly gracious, and indeed quite dream-like, at quite a brisk tempo. This problem is somewhat similar to that encountered in the Gesangsperiode of the 1880 and 1888 versions of the Fourth, as I discussed in my 1996 paper delivered at the Manchester conference⁵. There the speeds must increase and decrease according to a complex pattern, largely unspecified in 1880, and the problem is solved by well-thought-out and internally-consistent tempo indications added in 1888. However, the difficulty in the Seventh is more severe, since it affects the whole movement. It seems simply that if the first theme is too slow, as it nearly always has been, the movement will not hang together as a whole.

It is also interesting to consider the extent to which the old practice of accelerating during a crescendo is still being done. It turns out that there is a ritenuto, indicated in all editions, occurring at the end of the B theme of the first movement; this is at measure 121 and extends for two measures. Many conductors accelerate before this ritardando over many measures, some by enough to give room for the whole ritardando. The Fried performance of 1923 has one of the stronger accelerandos, while Horenstein (1928), whose B theme is slightly slower than Fried’s although his A theme was the fastest of all, employs the effect much more discreetly. Ormandy (1935) has no accelerando, decreasing slightly through the theme and then sharply at the end, Schuricht (1938) has a substantial acceleration and not much ritardando, and Jochum (1939) is symmetrical, accelerating slowly and decelerating rapidly by the same amount. In 1966, Max Rudolf attains the highest speed seen in this theme in the study, almost 74, then slows down to 54; the effect is amazing.

Although everyone takes the ritenuto at 121, the tendency to accelerate before it is much less in the recordings since 1980, Haenchen (1991) being a felicitous exception. A similar accelerando with subsequent ritardando, specified in only the first publication of the Eighth, is cited by Korstvedt⁶ in his book on that symphony to show the persistence of the old tradition into performances using both the Haas and Nowak editions.

There is no indication of a Luftpause over the barline before m. 125, but some conductors use one as a means of separating the tempos. Among these are Ormandy (1935), Jochum (1939), Kabasta (1942), and Rosbaud (1957). Other conductors tackle the difficult task of making a sudden transition in a graceful manner without a pause. Fried (1923) solves the problem by making the final tempo of the ritardando equal to the C tempo, which incidentally is 48.1 to the minim–right on the mark. Jochum (1939) achieves parity at 52 but has a Luftpause anyway; Böhm (1943) does not achieve parity (44 at the end of B, 55 for the beginning of C) but does without a Luftpause. In recent years this device has become rare; since 1958 there have been only two substantial pauses, in Pesek (1985), and in Davis (1987), he of the reversed Adagio and Scherzo.

One of the criteria for distinguishing between Haas on the one hand, and Nowak and Gutmann on the other, is the existence of doublings in the first violin part, an octave lower, in measures 154 and 156. This effect can be difficult to hear, and sometimes one wonders why they were put in, although one does not wonder why it was done for the seconds at 339–the melody soars far more dramatically when doubled in the upper octave. At any rate, the music at 153 and after, which is structurally the codetta, is particularly light and charming, and the portamento which one hears on some of the earlier recordings heightens the effect, if one does not automatically dismiss it as vulgar. The first three recordings (Fried, Horenstein, and Ormandy 1935) have noticeable portamento, and so do Kabasta, Furtwängler in Cairo, and Rosbaud, and some of the others of that period have a very slight gliding, but after the late 1950s it is not heard at all. Once one becomes accustomed to the very special appeal of a discreet portamento, it seems a loss not to have it. Its warm and gracious effect would be a real enhancement to a performance conducted at the old flexible tempos.

At this point I confess that my original reason for studying this symphony in such detail was to investigate, as part of my work on the edition of the Second, an entirely different problem: that of the possibility that there should be an enclave of slow music in the scherzos of both symphonies, following indications in the early publications. This is not the trio; in the Seventh it is a passage within the main scherzo extending from letter E ( measure 125), through letter F (measure 157), and onward to some point before the recapitulation at letter G (measure 185). There is no indication at any of these points in either Collected Edition score of any tempo change, but the Eulenburg and Philharmonia (Wöss) scores have “Etwas ruhiger” at letter E, and the Behn piano arrangement published by Gutmann has that and an “a tempo” at letter G. Although none of the full scores has “a tempo” or “Tempo I” anywhere, there can be no doubt that the recapitulation at letter G must be at the original tempo, and that the crescendo beginning at m. 165 is intended to facilitate an accelerando to accomplish that.

Measurements show Schuricht (1938), Jochum (1939), and Kabasta (1942) taking quite slow new tempos at 125; Jochum and Kabasta, Horenstein and Ormandy as well, speed up slightly for 157, and re-establish the pre-ritardando tempo by 185. Fried makes a much greater slow-down, and that of Abendroth (1956) is truly amazing; he moves from 91, well above the specified tempo of 80, down to 60 at letter E, farther down to a lilting 54 at letter F, and up to 96 at letter G in a vertiginous acceleration which has to be heard to be believed.

By contrast, the conductors of the 1990s do not see much of an opportunity here, though Rattle slows down a little in response to the evident demands of the music. Most of those who change tempo in this region broaden slightly at letter F rather than E, perhaps seeing a hurried struggle at E, and a relaxed and lyrical reward at F. Eugen Jochum’s technique in achieving his imperceptible but wide accelerandos is amazing. Somehow he is able to gather tempo without any feeling of rushing or urgency. It is as if the music itself leads the listener to the faster tempo. And it must be that sense of graciousness and appropriateness which informed the old conducting style. Wand’s (1980) scherzo is well-organized and elegantly rendered in the same way, but without the sense of abandon which Jochum achieves. It is, incidentally, amusing to hear the Vienna Philharmonic in its recordings of this movement, for example under Karajan (1989), play such passages as mm. 93-96 with typical Viennese rubato.

Rosbaud’s (1957) scherzo is noteworthy for its slowness, which many listeners have criticized. However, I was struck by the sound of the trumpets in the normally-hurried passage stretching from 169 to 180, echoed by the clarinets in 181 through 184–that under Rosbaud’s direction, for once they did not seem raucous, but instead had a certain lyrical or melodic effect. Rosbaud only achieves 70 at the end of his accelerando, but it seems sufficient in context. Matacic (1983) is somewhat similar, though his letter F is even slower. Haenchen produces a particularly well-thought-out relaxing of the tempo, discreet but definitely present. As with Rattle, I believe that this is a reaction to the music itself. Some conductors get slow by stages in the measure or two just before letter E, but Fried (1923) establishes his new very-much-slower tempo extremely suddenly.