Pioneering Haydn On Record
by Christopher Howell

An occasional series dedicated to commercial and live recordings of Haydn symphonies up to and including those of the first two complete cycles by Ernst Märzendorfer and Antal Dorati

Symphony no. 8 in C major Hob. I/8 “Le Soir” (1761 or earlier)

A pdf version of this article can be downloaded for offline reading and printing here.

Symphony no. 8, the last of the trilogy – preceded by “Le Matin” and “Le Midi” – has a number of common features with the others, but also some individual touches of its own. Like the others, it offers plenty of work for the orchestral soloists though not, as far as the strings are concerned, in the first movement, which emerges as a sort of overture to the concerto-like music that follows. Unlike the others, it has no slow introduction but pitches straight in with a vivacious “Allegro molto”. This has something of the gigue to it, though without the typical upbeat with which a gigue invariably starts. If you are looking for programmatic elements, I suppose the extended Andante second movement might have inspired a later composer to give it some such romantic title as “Evening Thoughts” or “The Close of the Day”. It certainly offers some warm sonorities – the rich euphony of a solo cello with a solo bassoon playing in thirds and sixths below it is surely highly original in a quiet sort of way. The minuet has mostly rather spare writing in two parts, but there is also an interesting passage for the wind choir. Like the other two symphonies, the trio is basically a solo for the violone and, here as before, Robbins Landon suggests using the double bass (if you do not have a violone), whereas early editions prescribed the cello. I certainly prefer the double bass in Symphonies 6 and 7 but I am rather concerned here that the double bass, playing an octave below written pitch, will be below the violins which should be supporting the melody line, whereas the cello will be playing above them. Some careful balancing will be necessary if we are to hear the tune and not the counter-melody. The finale is entitled “La Tempesta”. I had not thought of storms as being especially evening phenomena, but perhaps they were in that part of the world. Much has been made of the fact that the storm is punctuated by an arpeggio figure on the flute that also found its way into the storm music in “The Seasons”, but no other material is common to the two works. In the later oratorio, Haydn had the resources of as large an orchestra as existed at the time, complete with brass and drums, and creates swirling drama. In the symphony, the opening pitter-patter of the rain on the violins is extremely imaginative, but the forte outbreaks are just generalized, vigorous music not particularly suggestive of a storm. I would say that Vivaldi, with only strings, managed more convincing storm music than this, but perhaps one of the performances I am about to hear will persuade me otherwise.

It is interesting that cherry pickers have opted for one of the other two symphonies. I have not found come across a recording, commercial or live, of just “Le Soir” on its own.  The first recording, Franz Litschauer’s 1950 coupling of Symphonies 7 and 8 (HSLP 1016), was not available to me, so I begin with Max Goberman’s, set down some time between 1960 and 1962 with members of the Vienna State Opera Orchestra (HS 3M/HS 3S). He gives a lively account of the first movement and stretches the Andante as far as he dares – almost certainly much further than we are allowed to hear since we are told[1] that he played both repeats, omitted by the engineers for the LP issue. Most of it sounds absolutely lovely, though I did sometimes think of a Victorian organists improvising before Evensong, especially when the conductor encourages the soloists to delay their upbeats in the romantic manner. Yet I have my doubts as to whether an Andante in 2/4 was really intended to have four slow beats in the bar. There are passages, such as the bare staccatos for the strings, where there seems quite simply too little happening to hold the attention. Goberman plays the minuet in sturdy fashion and opts for a cello in the trio. The finale has plenty of energy, but the close recording of the flute reduces the contrast with the full orchestra when it enters, perhaps to the detriment of the music’s effectiveness as a depiction of a storm. As issued, this recording has first half repeats only in the outer movements and, as remarked above, none in the second.

Of Karl Ristenpart’s 1963 Nonesuch recording of the trilogy with the Sarre Chamber Orchestra (H-71015), I enjoyed Symphony no. 6 up to a point and no. 7 very much. His no. 8 is closer to the latter. The first movement is lively and joyful – thus far there is little to choose between him and Goberman. In the Andante, he shaves more than two minutes from Goberman’s timing. This feels more like an Andante in two and passages like the bare staccatos do not hang fire. That said, Goberman succeeds in investing the more melodic moments with a genuine depth of feeling, whereas they sound rather ordinary under Ristenpart. It would seem that neither conductor found the ideal solution here. Ristenpart’s sturdy minuet with cello in the trio is very similar to Goberman’s – except that the second repeat of the minuet is missing, no doubt a casualty of the attempt to fit the trilogy on a single LP. The finale is fast and fiery. Ristenpart and the engineers between them have placed the flute further back and this works much better – it sounds like someone scurrying for cover before the storm breaks over them. This movement is considerably more effective under Ristenpart. First repeats only in the outer movements, none in the second and, as noted, a truncated minuet.

Vilmos Tátrai, leading the Hungarian Chamber Orchestra from the first violin desk, set down “Le Soir” in July 1965 (Qualiton LPX 1241)[2]. Their first movement, with first repeat only, is slightly more relaxed than with Goberman or Ristenpart, but is full of dancing enjoyment. Their second movement completely avoids any suggestion of romantic twilight reverie. Taken definitely in two and with short staccatos, it has elegance and a sense of wistful intimacy. The dotted rhythms played by the full violin section after the initial statement on solo strings come across at this tempo as a quixotic comment. The Hungarians find a far more ambiguous mood than either Goberman or Ristenpart. They play both repeats, and the movement does not outstay its welcome. Their minuet, rather than sturdy, has a tripping lilt. With delicate staccatos and elegant slurred notes, it combines graciousness and good humour. The trio is taken by the cello and makes me wonder more than ever if it could work with a double bass. Frankly, this movement is delicious. The finale, with both repeats, is relatively steady. I was initially disappointed but came to see that Haydn’s storm, like Beethoven’s in the “Pastoral” Symphony, can benefit from having all the notes well played.  

As with Symphonies 6 and 7, I have been unable to hear the recordings by Helmut Müller-Brühl with the Cologne Chamber Orchestra (May 1966, Schwann VMS2007) and Gerhard Bosse with the Leipzig Gewandhaus Bach-Orchestra (21-24 August 1967, Eterna 820923/825923)[3].

Which brings us to Wilfried Boettcher’s 1968 recording with the Vienna Chamber Orchestra (TV 341505). Robbins Landon’s edition had come out in 1965 and several points of phrasing show that Boettcher is using it. He takes a very relaxed tempo in the first movement – if all performances were like this, no one would have suggested the movement sounds like a gigue. It nevertheless sounds delightful at this tempo, which enables Boettcher to make more than others of the harmonic excursions in the development and the uneasy pause near the end. No repeats, though. Boettcher takes the second movement at about Tátrai’s tempo but imbues the solo passages with Goberman’s warmth, while the sparer moments and the dotted rhythms have their quixotic nature. The minuet is sturdy but with a graciousness that eluded the sturdy performances of Goberman and Ristenpart. In the trio we hear the double bass at last. It just manages to dominate the texture without apparent twiddling by the engineers, while the scale passages sound fascinatingly gruff at the low pitch. I loved this, and would have loved it even more had the second repeat not been omitted. Whoever insisted that the three symphonies had to go on one LP has a lot to answer for. The finale, repeatless obviously, is taken at a steadier pace even than that by Tátrai. The pinpoint clarity of the pattering rain at the beginning is fascinating and Boettcher, too, demonstrates the advantages of having every note well played. But maybe a notch faster would have done no harm? Nonetheless, a lovely performance – what there is of it.

I stated in my article on Symphony 6 my reasons for believing the 1972 trilogy attributed to Hanns Reinartz an elaborate hoax.

After a delightful “Matin” and a disappointing “Midi”, Ferdinand Leitner’s 1972 trilogy with the Bavarian Radio Symphony Orchestra (Hänssler) seemed, for at least half the way, on the mend. The first movement is relaxed but dancing and not at all heavy, rather in the Boettcher mould, and at least it has the first repeat. The second movement offers scope for reflections on the difference between perceived tempi and actual ones. My impression was that this was about Goberman’s tempo. Leitner’s soloists have all the space in which to express the music lyrically, even romantically, though the bare staccatos do not sound static. In reality, Leitner, playing the first repeat, takes only a few seconds longer than Goberman who has neither. How is it done? On careful analysis, Goberman, though basically not much slower, takes time over phrase ends and expands with the swelling phrases, with the result that things risk grinding to a halt, if a deeply felt one. Leitner, with many years’ experience in keeping things subtly going in the likes of Bruckner and Richard Strauss, whose music can fall apart with too much self-indulgence, is able to fill the music with as much feeling as Goberman within firm overall pacing. A comparison between these two would be a valuable lesson for any aspiring conductor. There are occasionally hints from the back of the room – or from the next room – that a harpsichord is playing here. Leitner’s minuet goes at about the same speed as the sturdy ones, but achieves instead a serene graciousness that is very attractive. Though he seems to be using the Urtext edition, he evidently does not agree with Robbins Landon – as also in Symphony 7 – that the double bass is the proper substitute for the violone in the trio. Still, his cellist plays very nicely and earns a few kudos for adding modest decoration in the second repeat. But then, oh dear, the finale! This is slower even than Boettcher, or seems so, and a better behaved storm never broke. With both repeats, this is utterly dispiriting, a dismal way to end. A pity.

Ernst Märzendorfer’s first movement goes at about Boettcher’s speed, but with crisp staccatos and forward wind the music seems to be made of slightly sterner stuff. Märzendorfer maximizes the contrast between the delicate opening and the forte passages, and also between staccatos and legatos. The sudden pause towards the end of the movement is given added point by the sense of unease he creates in the passage leading to it. In short, other performances offer more sheer delight, but Märzendorfer perhaps finds more in the music. First repeat only. His second movement is definitely in two and I began by finding it unsettlingly fast. Though I adapted to it, I did rather miss the natural warmth of Boettcher and Leitner. On the other hand, Märzendorfer brings out the ambiguity of the music. Is it lyrical? Is it quixotic? Is there even a touch of menace to the marching rhythms from the full strings contrasting with the more songful music for the soloists? Here, too, first repeat only. His minuet is sturdy, even strutting with its sharp staccatos. This relatively slow tempo may have been chosen to accommodate the double bass in the trio. Like Boettcher, Märzendorfer shows that with careful balancing, this instrument can play solo convincingly against the other instruments. The finale is terrific. Taken at great speed, the raindrops at the beginning are pregnant with menace, the flute, and later the solo strings, sound like people dashing for cover, and the onslaught of the full orchestra is powerful in spite of the small forces involved. You cannot imitate thunderbolts with just strings, but Märzendorfer makes the descending figures sound like swirling gusts of wind and recreates the dynamic effect the music must have had on the Esterhazy audience. Both repeats are taken.

Dorati takes the first movement at a genuine Allegro molto, verging on Presto. It is certainly exciting. Once or twice the phrase “all sound and fury signifying nothing” came to mind but Dorati keeps it under firm control with feather light piano playing, so it is not all one dash. Both repeats are played. His second movement is fairly slow but by leaning on the first note of each phrase and then coming off with delicate staccato upbeats he expels any suggestion of romanticism in favour of simpering curlicues. Phrased in one-bar units rather than in long lines, the music becomes rather monotonous. The first repeat is played. The minuet is not one of Dorati’s flabby ones but here, as in the second movement, his leaning on the first of any slurred pair of notes and slightly delaying staccato upbeats creates a world of artificial bobs and curtseys. The double bass plays in the trio – correctly and musically but without the indefinable presence of a real soloist. The finale, with both repeats, is not much faster than Leitner’s and not much better, reawakening thoughts that Haydn’s orchestral resources were simply not up to depicting a storm – something handsomely disproved by Märzendorfer.

Looking at the commercial recording situation up to this time, if you want Urtext and all repeats, the answer is simple – nobody had done this. Unless you want to construct yourself a composite performance, you will have to forgo at least one repeat. Of the Urtext versions, Märzendorfer seems the best overall, though I would like to keep a more romantic second movement handy – if only Leitner had conducted the rest as well as he did this movement, or Boettcher had included at least some repeats.

There are just two live/broadcast versions to consider. Harry Blech, with the Orchestra Alessandro Scarlatti di Napoli della RAI, completed the trilogy which opened his concert on 24 January 1961. He pitches into the first movement with great gusto, at a speed something like Dorati’s, but he is better at making the orchestra sound as if they are enjoying it rather than being driven along. With good dynamic shading, this is exhilarating. He is particularly imaginative with the pause near the end. Haydn did not specifically indicate a dynamic here, though the preceding bars are marked forte so most conductors give a powerful accent to the note that stops the movement[4]. Blech makes a sudden piano, then has the flute play its phrases with a stuttering uncertainty, before the whole orchestra enters to brush it away. Whatever the rights and wrongs, he gives a meaning to a moment that, in some hands, stops only so it can start again. First repeat only. The second movement is similar in concept to Leitner’s – freely lyrical though not actually so very slow. No repeat but very lovely all the same. The minuet raises a curious point. The melody notes in the third bar are bowed legato. In the Robbins Landon edition they are indicated staccato, whereas the old Breitkopf score has them bowed legato as Blech plays them. This is curious because the performances by Ristenpart and Tátrai, set down before the Robbins Landon came out, play the notes staccato. Goberman has them separated but not really staccato. So how did they know the legato bowing was wrong? Were there other editions between the Breitkopf and the Robbins Landon (who also edited the Eulenberg miniature score)? I have not traced any, but since they have been rendered obsolete they would not be easy to find. Blech is the only conductor who wholeheartedly embraces the legato marking and moreover, taking his cue from this – I presume – he has many other phrases bowed legato where Haydn has indicated nothing. In the trio, for example, there is no indication in the Breitkopf score of how to play the forte interjections in octaves between the violone’s phrases – played inevitably by the cello under Blech – whereas Robbins Landon has them staccato. Blech bows them legato. He also has the passage for wind choir in the minuet played legato, though oddly, he has the notes slightly separated in the da capo. The result is to change completely the character of the music which, invested  by Blech with a certain noble graciousness, comes to sound like Handel in an old-style performance. At least Blech gives us long lines, not simpering curtseys, but it has to be admitted that, attractive though it is, it is wrong. A guilty pleasure … Blech’s finale is up there with Märzendorfer’s in its seething vitality, the only other one I have heard that truly makes this movement come off. First repeat only. Incredibly, two brief forte G major chords are added at the end. Effective but … could Blech cite any evidence – early editions or whatever – that Haydn added them at some point?

In my article on Symphony 7, I discussed a 1965 recording from a private source by Ernest Bour and the South-West German Radio SO of Baden-Baden. SWR has issued a swathe of Bour’s Haydn performances with this orchestra. These are available in various download options as well as YouTube, but do not seem to be dated. I cannot say, therefore, whether Symphony 8 was recorded on the same occasion. If Bour did Symphony 6 to complete the trilogy, it has not seen the light of day.  Various points of phrasing, as well as the use of a cello in the third movement trio, show that he is not using the Robbins Landon edition but, like everyone except Blech, he evidently knew that the legato slurs in the trio were wrong. He takes the first movement quite steadily and, while Boettcher and others showed that it can be delightful at a relaxed tempo, an element of high seriousness rather damps the spirits. The Andante flows quite swiftly. Like Ristenpart, he shows that this movement can sound rather ordinary if treated matter-of-factly. The minuet is taken unusually swiftly. This was refreshing at first, and it is the most interesting part of this performance, but in the end I found it rather dogged. The finale achieves some impact at a good tempo, without quite equalling Märzendorfer or Blech. In the flute arpeggios, either the player gets very slightly behind or Bour gets very slightly ahead – every time, the orchestra bursts in as the flute is finishing its last note. Once or twice can be excused in what was probably a studio broadcast without patching, but Lady Bracknell’s comment can be imagined when it happens every time. First repeats only in movements 1, 2 and 3. Bour, as I noted in the previous article, has a high underground reputation, but I would not cite these two Haydn performances to justify it.

The unfortunate conclusion is that the performance that gives me the most complete pleasure all through is Blech’s – and this applies to the trilogy as a whole. Unfortunate since the text used is dubious, unless it could be proved that Robbins Landon’s conclusions were more opinable than is normally supposed. I rather doubt this. For a performance of the trilogy with a safer text, Märzendorfer’s is evidently the best within my cut-off date, though he is not always as revelatory as he can be elsewhere. Perhaps the Vox people should investigate whether Boettcher actually recorded at least the first half repeats in the outer movements and, if so, reinstate them to make a very attractive record[5].   

Christopher Howell © 2026

[1] See Anthony Hodgson: The Haydn Seekers, CRC Magazine, Winter 2001.

[2] Gray’s Classical Discography https://classical-discography.org/ shows that Tátrai recorded Symphony 7 on 3-4 July 1962, coupling it with Symphony 49. Symphonies 6 and 8 followed on 26-29 July 1965. A complete recording of the trilogy was therefore not originally planned, though they have been logically brought together for CD release.

[3] Bosse’s LP contains Symphonies 6 and 8. He does not appear to have recorded Symphony 7.

[4] The old Breitkopf edition and the Robbins Landon Urtext are identical on this point.

[5] At this point in such a discussion, somebody usually pipes up to say “you can always reinstate the repeats electronically”. None of this, please. A true artist even if not one of the greatest, will never play anything exactly the same way twice. At the very least, there is likely to be a certain slowing at the end of the second half, if the music is to stop there. It would sound incongruous then to jump back and repeat the section, finishing with the same slowing down as before. If this does not convince you, what are you going to do when Haydn, as in the first movement of Symphony 7, writes a first time bar with a lead back that would not be recorded if the conductor’s intentions from the beginning were not to take the repeat? You could only try some such mad solution as taking the lead back from another performance altogether and hoping it fits.