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. 83 in G minor “La Poule” Hob. I/83 (1785)
Part 2: From 1950 to Märzendorfer and Dorati
A pdf version of this article can be downloaded for offline reading and printing here.
In the first part of this article, I gave a general introduction to the work and discussed the first recording, made by Sir John Barbirolli and the Hallé Orchestra at the end of 1949. I followed this with an analysis of six live recordings by the same conductor from 1965 to 1970, the last year of his life. The second part of the article will take up the story from 1950 to the arrival of the first two complete Haydn symphony cycles.
Barbirolli did not have to wait long for a successor. The first LP recording was set down for the Haydn Society by Anton Heiller and the Collegium Musicum Vienna in 1950 (HSLP 1015). This orchestra had been founded in 1945 by Kurt Rapf. The Haydn Society generally concentrated on unrecorded symphonies, so perhaps they were unaware that Barbirolli was in the can. Or perhaps they thought the symphony should be recorded in a more authentic edition. The Andante gets its missing bar and the phrasing at the beginning of the finale is as in the Robbins Landon edition, but there are slurs in the first movement similar to those in the Barbirolli performance but not found in the Urtext.
Perhaps, too, the Haydn Society thought they would be getting a more upfront performance compared with the romantically inclined Barbirolli but, if so, they must have their heads in dismay as the orchestra started playing. Some of Heiller’s Haydn recordings remain undimmed by time – his no. 26, for example – but this one found him on an off-day. I was prepared to give him the benefit of the doubt in the first movement, which is world-weary rather than dramatic, but not unattractively so, and builds up well in the development section. There is no particular delicacy or charm to the clucking second subject, but, thus far, Haydn’s own genius carries the day. At the end of each part, we do just about hear the strings’ dotted rhythms where with Barbirolli we hear only three sharp chords. The first repeat is played. In theory, the Andante moves at a tempo midway between Barbirolli’s fastest and slowest performances, but the problem is that for much of the time it does not really move at all. Even in his most extended versions, Barbirolli was a master of orchestral phrasing and ensured that we hear three beats in the bar. Under Heiller, the repeated quavers become a plodding six-in-a-bar. The first theme can just about take this, but later on the music becomes too static to hold the interest. The minuet is formal, even doleful, with surprisingly little lilt for a Viennese band. The finale has a tempo similar to Barbirolli’s disappointing Baden-Baden performance. Like that traversal, this enables Heiller to invest the forte passages with some weight, but the main theme is impossibly pedestrian. G. C. Burke, in his 1952 survey of available Haydn recordings[1], remarked that “The pleasure one must undoubtedly feel in hearing this recording must be gravely compromised if one has opportunity later to hear a really proficient performance. Conductor and orchestra are not at one here, nor is the latter alluring in tone. Projection is flabby and the dryness of the bass contaminates the sound.”
One hopes Burke found consolation when the Barbirolli was transferred to LP, for the 1950s produced only two further recordings, neither of which had wide circulation. I have been unable to hear the version set down by Mathieu Lange and the Berlin Chamber Orchestra on 5 May 1957 (Eterna 720042).
Also in 1957, Serge Baudo recorded the symphony with the Collegium Musicum de Paris (Contrepoint MC 20148). This was a chamber orchestra founded in 1952 by Roland Douatte, whose own recordings with the ensemble were mostly of baroque music. Baudo is well remembered for his interpretations of impressionist and contemporary music. This early foray into the classical repertoire is an improvement on Heiller but less so on Barbirolli, especially since he seems to be using the same edition, with altered phrasing and the missing bar in the second movement. He plays the string triplets near the end of each section of the first movement with brilliant separate bows as in the Bärenreiter edition instead of slurred, one bow to each triplet group, as in the Urtext score[2]. I must say, though, that we hear the dotted rhythms at the end of each section clearly at last.
Baudo takes a steady tempo for the first movement, obtaining a certain majestic drive. It is interesting that, though the Collegium Musicum Vienna and the Collegium Musicum de Paris were chamber orchestras specializing in baroque and early classical repertoire, there is more litheness and transparency to Barbirolli’s Hallé – evidently this depends more on the conductor than on the number of players. The clucking second theme is not particularly charming, but it has delicacy. No repeats. The Andante is kept flowing. Barbirolli has shown that a great master of orchestral phrasing can make a slower tempo work, at least sometimes, but even he moved towards a less indulgent tempo in his last performances, and this at a time when his slow tempi in romantic music were exasperating even his most fervid admirers. If Heiller proved that a very slow tempo can fall dead in the hands of a lesser conductor, Baudo shows that a proper Andante, even without especial distinction of phrasing, can produce more than acceptable results. Where Baudo does score is in the minuet, which he takes fairly swiftly with a merry, gay atmosphere. Unfortunately the trio, which he takes marginally slower and with a ritardando at the end, has not the character of the best Barbirolli performances. The finale, with neither repeat, is barely faster than Heiller, but manages to be more light hearted, even jolly. This orchestra has a superb flautist, who converts me entirely to the tempo every time he gets the tune on its own – not that it happens often.
The 1960s saw an upturn in the recorded history of “La Poule” and began splendidly with the recording by Szymon Goldberg and the Netherlands Chamber Orchestra. Goldberg (1909-1993) was a Polish-born Jew, later naturalized American. He was the concertmaster of the Berlin Philharmonic under Furtwängler from 1930 to 1934 but was forced to leave Germany on account of his Jewish roots. As a violinist, his collaborations with major pianists stretch back to pre-war duos with Lili Kraus and ahead to 1970s collaborations with Radu Lupu, both preserved on record, as are his 1960s performances with Ingrid Haebler. He formed the Netherlands Chamber Orchestra in 1955 and conducted it until 1979. He also taught violin in several major American institutions.
This recording raises further questions over editions. Several things, such as inauthentic pizzicatos and the phrasing of the opening theme of the finale, are as in the Robbins Landon Urtext, and there is no added upbeat before bar 85 of the finale. Yet some of the phrasing in the first movement is still at variance with the Urtext and the triplets towards the end of each section are played with separated bows, clearly articulated. In the already-discussed four bars of repeated notes in the second movement, there is no alteration to the harmony in the third bar but the fourth bar, played by Heiller, is cut.
That said, the performance is splendid. The first movement is unhurried, but avoids stateliness and has the litheness of the best Barbirolli performances. The clucking theme is delightful. The first repeat is played. In the Andante, Goldberg risks a slowish tempi, but with a heartfelt, withdrawn reverence that never lets it stagnate, or that causes the repeated notes to chug. Maybe it needs a string player on the rostrum to get these repeated notes to sing like a melody, but we will see how this theory holds as the article develops.
Two further textual points are raised by the minuet and trio. The Robbins Landon edition marks the opening forte in square brackets – indicating that this is an editorial edition. The autograph, in fact has no dynamic marking. Since the orchestration is fairly full, forte seems a reasonable assumption and the old Bärenreiter edition unashamedly added forte without comment. Goldberg plays it piano and it sounds absolutely delightful. He achieves a gentle lilt in this movement that has eluded the performances discussed so far. Delightful, too, is his trio, not least for the way the two upbeat quavers introducing it are played a perky staccato. They are indeed marked staccato in the Robbins Landon edition, but legato in the old Bärenreiter and so they were played by Barbirolli, Heiller and Baudo. The autograph has no indication at all, though it has quite detailed staccatos and slurs later in the theme. Make of that what you will. The finale, with the first repeat, goes at a joyous pace, beautifully controlled. On the whole, Haydn lovers in 1960 who looked back nostalgically to Barbirolli but wanted a more modern-sounding recording, could happily replace their scratchy old discs.
Karl Münchinger’s recording, set down on 10-12 April 1961 (Decca LXT 5647 / SXL 2284 in UK, London CM 9297 / CS 6230 in USA), has the added allure of the Vienna Philharmonic. In the main, he seems to be using the Bärenreiter edition “as is”, so we have separate bows for the triplets at the end of each section of the first movement, all four bars of the repeated chords in the second movement, a forte start to the minuet, a slurred upbeat to the trio and the upbeat to bar 85 in the finale.
I found Münchinger’s Symphony 104 an unexpected pleasure, and in three of the four movements I can say the same of his 83. The first movement, though clearly beaten in four, is quite brisk – refreshingly energized in the forte passages and with the clucking theme dancing along neatly. At the beginning, the conductor makes sure we hear that the sforzatos are in the lower strings as well as the violins, something I had not noticed in previous performances. Decca’s excellent recording obviously helps here. Münchinger’s strengths become his weaknesses in the second movement. It is true that Haydn indicated Andante in three, not in six, so Münchinger could say that his swift tempo is exactly that. But three or six is a matter of feeling as much as of tempo. The movement is dominated, in this performance, by the obsessively equal repeated quavers, to which the more melodic passages remain chained, unable to express themselves freely. The minuet is sturdy and formal but with a pleasing lilt and the finale is a real success, vigorous without sounding over-driven.
The first complete cycle of the “Paris” Symphonies came from Ernest Ansermet and the Suisse Romande Orchestra. The entry for this recording (Decca LXT / SXL 6021) in Gray’s Classical Discography gives a generic date of 6-11 and 19-25 April 1962 which presumably refers to all six symphonies, so it is not clear whether Ansermet or Bernstein, whose version was made on 9 April 1962, came first in this particular work. Bernstein took longer to complete the entire set.
Ansermet’s Haydn has not found universal praise. So far in these articles, I have admired his unvarnished truthfulness, combined with an instinct for the natural movement of the music. I must say he does some odd things here. On the plus side, he seems to be using the Robbins Landon edition, albeit with a few personal variants, such as a crescendo into the recapitulation of the first movement and some fussy chopping up of the phrasing in the Andante. So we have all four bars of repeated chords in this latter movement, a staccato upbeat to the trio and no unwritten upbeat to bar 85 in the finale. For the first time in this survey, I have heard the crotchets in bars 27-29 of the first movement separated, not slurred. And he begins finely, with a vigorous tempo similar to Münchinger’s and close attention to sforzati, in upper and lower strings alike. The less mellow sound of the Suisse Romande Orchestra might be thought a benefit in this music. But then Ansermet, of all people, takes the clucking second subject at a notably slower tempo. Taken on its own terms, it is charming and graceful, and Ansermet moves in and out of his two tempi very dextrously. All the same, other conductors have shown that a single tempi works perfectly well in this movement.
Anyone who thought Münchinger too fast is in for a shock when they hear Ansermet. He is not only faster still, but dabs at the single repeated notes with delicate, well-separated staccatos, creating a kind of stuttering effect. Does he see this movement, too, as illustrating the pecking hen? Haydn put staccato dots over the notes, it is true, but he also put a slur over each bar. The two string-player conductors, Barbirolli and Goldberg, agree that this combination of staccato and slurring means that you stop the bow so that we hear the repeated notes, but without leaving gaps between the notes and allowing the notes to sing as a musical phrase. However, Ansermet brings his idea off very well, treating the movement with balletic lightness. The more melodic phrases do not sound squeezed, as they do under Münchinger, because Ansermet does not aim to make them expressive. And there is no danger of the second subject hanging fire at this pace. While I do not really like this, I find it more convincing than Münchinger because, while the latter sounds like the usual interpretation speeded up, Ansermet gives the music a different character entirely.
Ansermet’s minuet is slow and formal. The conductor’s usual sense of balletic rhythm is not enough to keep this afloat. The finale, on the other hand, is excellent. It does not quite spin off the page as did Münchinger’s, but it is thoroughly vivacious in its own right. A curious performance overall.
As noted, Leonard Bernstein made his recording with the New York PO (Columbia Masterworks MS 6609 in USA, CBS SBRG 72240 in UK) at about the same time as Ansermet, but took his time over the full cycle of “Paris” Symphonies and this one was issued only in 1964. There is the usual question over editions. Bernstein seems to be using Robbins Landon, but reinstates the slurred upbeat to the trio and adds a few personal variants, such as a crescendo into the recapitulation.
His first movement is not especially swift but it is taut and fiery, with every phrase leaping into life through his detailed phrasing. The clucking theme is not exactly elegant but it is bright and cheeky, especially when the oboe enters with its dotted rhythms. First repeat only.
The second movement has a fairly flowing tempo, definitely three-in-a-bar, but retains the expressive character of slower versions. Bernstein has the violins sing their repeated notes as a musical phrase, not chugging as with Münchinger or pecking as with Ansermet. I find no reference to Bernstein having studied any instrument but the piano, but evidently he made it his business to understand the instruments he was conducting, thereby showing that my theory about conductors who do or do not play an instrument from the violin family can be circumvented by a great musician. This is a very beautiful performance, notable for its tender restraint, apart from the forte outbursts, into which Bernstein pitches dramatically. One oddity – bar 8 is played at a notably slower tempo than the rest. Maybe it was spliced in from a different, slower, take?
Bernstein was always inclined to take minuets slowly and heavily, but this is better than most, with a certain forward-moving vigour. The trio is delightfully pointed. The finale, with the first repeat, has tremendous drive, even a tad too much – Münchinger seems to me to have got this absolutely right, yet the difference is a matter of seconds. Perhaps it is Bernstein’s characteristic intensity that makes it seem a little relentless. All the same, a very fine performance.
Leslie Jones and the Little Orchestra of London recorded “La Poule” twice at a distance of two years, once for Pye in 1964 (GGC4014/GSGC14014 in UK, Nonesuch H71083 in USA) and again in the context of a complete cycle of the “Paris” Symphonies for Nonesuch in July 1966 (HC-73011).
The 1964 raises the usual doubts over editions. The Andante is as per Robbins Landon (and Breitkopf), with all four bars of repeated chords, while in the first movement we have the inauthentic slurs in bars 27-29, the authentic slurs removed from the triplet quavers at the end of each section and a crescendo into the recapitulation. The upbeat quavers to the trio are slurred and an unwritten upbeat is added to bar 85 of the finale. So it looks like he is using Breitkopf. A very odd thing happens at bars 116-117 of the first movement. Haydn has a written out pause here, lasting for the remaining three beats of bar 116 and the whole of bar 117. Long enough for anyone, you would have thought, but Jones adds an entire extra bar’s pause[3].
That said, there is quite a lot in his favour. The first movement is swift and fiery, its terseness accentuated by the omission of both repeats. This tempo gathers up even the clucking theme in its urgent drive – it sounds more impertinent than elegant. Effective if a little one-sided. One casualty of a faster tempo is the risk, not everywhere avoided, that the dotted rhythms, so trenchantly articulated by Barbirolli at his more majestic tempo, become blunter triplet rhythms, as though the movement were in 12/8. Bernstein showed that a master conductor can avoid this risk at a faster tempo and I suppose the risk is there at any tempo if the conductor does not keep a watchful ear on the proceedings.
The second movement is swift, at approximately Ansermet’s tempo, though without his pecking staccato quavers. It is neatly done, though I continue to find that the music lacks depth at this pace. The minuet is delightfully rustic, with a perky trio that is marginally – but very marginally – faster than the minuet. The finale, too is a complete success, at a rip-roaring tempo that manages not to sound overdriven.
Two years later, Jones seems by and large to be using the Robbins Landon Urtext. The inauthentic slurs in bars 27-29 have gone and the separate crotchets are punched out gleefully. The slurs over the triplets quavers near the end of each section are in and the extra bar’s rest is out. The two-quaver upbeat to the trio is staccato. But there is still an unwritten upbeat to bar 85 of the finale and there are some other divergences common to performances using earlier editions. The last three quavers of bar 37 in the first movement are legato-bowed by the violins. A small but curious point. One of the very few differences between the 1950 Urtext edition which Robbins Landon edited for the Boston Haydn Society and his 1963 Urtext edition is the slurring of these three quavers. They are not slurred in Haydn’s autograph but they are slurred there in the analogous passage in the recapitulation (bar 150). This might seem to justify retroactively slurring them in the exposition. Whatever doubts Robbins Landon had in 1950, he took this view in 1963. However, in bars 109-112 of the development Haydn makes much of this little phrase, not slurring it, and here Jones quite rightly has the three notes played strongly with separate bows. So there might be a case for courting inconsistency and playing what Haydn wrote. A little later, the last two crotchets in the lower string parts of bars 59-60 of the same movement are slurred. They are likewise slurred at the corresponding point in the recapitulation, bars 171-172. Here, the Robbins Landon edition, which had no slurs the first time round, has the last three – not two – crotchets slurred in the lower strings, and also for the oboes and bassoons. The justification seems to be that, at bar 171, but not bar 172, Haydn’s autograph has the last three crotchets slurred in the bassoon part, though not in the others. Robbins Landon takes this as licence to apply this same phrasing uniformly to all parts in both bars, though without applying it retroactively to bars 59-60. It is indeed difficult to see why the bassoon should play these notes differently from the other instruments. Whatever, the solution adopted by Jones of playing the first crotchets notes of all these bars well separated and the last two slurred is musically effective but is not what Haydn wrote, either in the exposition or in the recapitulation.
But what of the performance? The first movement is more traditionally majestic this time, with the first repeat included. This tempo allows for clear articulation of the dotted rhythms. The clucking subject is more purposeful than charming, even at the slower pace. I could not help feeling that some of the previous verve and terseness had been lost, though it remains a more than decent performance. The Andante takes about a minute longer, though it is still a good deal faster than some performances discussed earlier on. With the repeated crotchets sustained and cantabile, this sounds ideal to me, and remarkably close to Bernstein. So far so good. Unfortunately, the minuet takes about a minute longer too, and is a limp, drooping affair. Whatever happened? The finale has much of the life of the previous one, but seems slightly less well controlled. I can only advise admirers of Jones to get both and make for themselves a composite version with the first movement (probably) and the second movement (definitely) from 1966 and the others from 1964.
In between Jones’s two versions, in 1965, came the cycle of “Paris” Symphonies by Denis Vaughan and the “Orchestra of Naples” – actually the Orchestra “Alessandro Scarlatti” di Napoli della RAI out of contract (RCA LM/LSC 6805). This seems another case of the Robbins Landon edition not quite “as is”. The points raised à propos the second Jones recording are as in the Urtext and Vaughan has no upbeat before bar 85 in the finale. He makes a few adjustments to the bowing and phrasing in bars 33-40 of the first movement and adds some quite effective crescendos and diminuendos, while showing that a crescendo into the recapitulation of this movement is unnecessary – he actually tapers the sound away so as to break into the recapitulation dramatically. It seems, though, that nobody can conduct this symphony without tinkering somewhere. In bars 50-51 of the second movement, the dotted crotchet plus quaver rhythm in the first violins’ melodic line becomes a double dotted crotchet plus semiquaver, and I wondered why the quavers in bar 72, leading into the recapitulation, are separated rather the bowed legato as indicated. These things sound quite musical, but what is written sounds quite musical too.
In my previous encounters with this series (Symphonies 85 and 86), I was not unduly impressed and wondered if the orchestra was really inspired to give the visiting Maestro their best. These sessions seem to have got them all on a better day and the playing is attentive and lively. A typical Italian vibrancy to the string playing, not always evident when Vaughan conducted them, is further proof that they were in collaborative mood. The first movement is fairly majestic and actually a tad slower than the second Jones, but sounds faster because it is punchier in the forte passages and finds more contrast in the elegantly turned clucking theme. Vaughan’s Andante is among the slower ones and he is also the first so far in this survey to give at least the first repeat. The considerable length is vindicated by beautifully expressive cantabile playing and there is a finely graded diminuendo over the four bars of repeated chords – some conductors have already reached minimum volume about halfway through. The minuet is about as slow as Jones II, with a trio slightly slower still, but with sufficiently alert playing not to sound limp. The finale is also unhurried, but with crisp playing from the strings and pastoral sounding flute and oboes, it makes the best case for not driving this movement too fast. One of the best performances so far.
Gray’s Classical Discography[4] lists a recording of this symphony from the 1960s by Walter Schulten and the Westdeutches Mozartorchester (Susato SD-2001). No other evidence of this recording has come to light so perhaps this is a reissue of the Bertelsmann recording of Symphony 73 by this conductor, which I have discussed in my article on that work.
The 1960s conclude, therefore, with the cycle of “Paris” Symphonies conducted by Günter Wich with the Sudwestdeutches Kammerorchester (February 1969, Mace MAV 3606), of which no. 86, in particular, has so far impressed me very favourably. Unfortunately, the gallant YouTuber who has made this available has a CD copy that is going bad, with the music getting stuck or skipping several bars in each movement. Did he not listen to what he was posting? Under the circumstances I can only make some general comments based on the parts that are in the clear.
The edition is the usual mish-mash of things that correspond to Robbins Landon and things that do not. Separate bows in the triplets near the end of each section of the first movement, for example, but staccato quavers in the upbeat to the trio. One of the four bars of repeated quaver chords in the second movement is omitted.
So far as I can tell, the performance is nice but not to the extent that it becomes imperative to hunt down a copy that plays properly. The first movement is swift and textures are lighter, with the wind more forward, than in many other performances using a chamber orchestra. A particularly interesting point is that in the clucking theme, Wich, more than anyone else so far, makes us hear that the semiquaver pairs in bar 50 and elsewhere, are not acciaccaturas, which is how they usually sound. His second movement is swift and light – Bernstein and Vaughan have shown that a relatively fast tempo here need not preclude depth. The minuet is slow and formal with a trio slower still, but with rustic wind well forward it comes off better than I feared. The finale, like Vaughan’s shows that this movement can benefit from an unhurried tempo if the playing is alert and spry.
Kurt Sanderling’s recording with the Berlin Symphony Orchestra (Eterna 826345) was set down on 17 and 19 March 1971 as part of his cycle of the “Paris” Symphonies. At this date, the Märzendorfer and Dorati cycles were still ongoing, though Dorati at least had covered this symphony in sessions from 11 to 28 November 1970. The dates of Märzendorfer’s individual symphonies are not known. Aside from a few very minor tweaks, this seems a faithful rendering of the 1963 Robbins Landon Score. As might be expected with Sanderling entering his late phase, tempi are broad but, in the first movement, combined with rhythmic resilience and pointed attention to detail. The clucking theme has a good deal of grace. It is for the second movement, however, that I would return to this performance. The tempo is broad but songful and with an almost Sibelian sense of unhurried awe. About the slow minuet I am less certain, though Sanderling manages to give it a certain formal grace and phrases the trio deliciously. The finale is steady and far from heavy handed, yet I sense a lack of joy compared with several other versions.
November-December 1973 is just outside my terms of reference, but I would have listened to Menuhin’s recording, part of a cycle of “Paris” Symphonies with the Menuhin Festival Orchestra, if I had been able. Unlike his LP of Symphonies 26, 44 and 48, it achieved a CD issue in the 1990s, but has long disappeared and none of the usual blogs offered it. I should have been interested to hear how a great string player dealt with the repeated notes theme in the second movement.
Märzendorfer seems to be playing the 1963 Robbins Landon score pretty literally – he adds no dynamics of his own, leaving Haydn to speak for himself. The first movement is a firmly driven two-in-a-bar, fierce in the minor-key passages and delicate but still terse in the clucking theme. I have to say I prefer this movement a little more Maestoso – Haydn’s marking is four-in-a-bar, after all, but I find this preferable to the first Jones if a swift performance is wanted – perhaps because the inclusion of the first repeat allows it more space. The second movement has the repeat notes pecked at Ansermet-style. I do not really like it this way – detailed biographical information about Märzendorfer is hard to come by, but I find no evidence that he was a string player. His tempo, however, is a tad slower than Ansermet’s, so if you like the idea of the hen pecking away gently during this movement, this seems preferable to Ansermet. I would point out that, done this way, there is no differentiation between the staccato dots with a slur over them and the straightforward staccatos in the second theme. The minuet is quite slow, but my attention was caught by the virile, stamping rhythm – no formal court dance this, but a country festival. If it is to be done slowly, this is how to do it. The trio is pertly pointed, drawing attention to the fact that there are repeated pecking staccato notes also here – and in the finale too. I am not sure if it is productive to tease out hen noises in all four movements when the title was not Haydn’s anyway. Märzendorfer’s finale is swift but not overdriven. He is the first to give us both repeats, but this only emphasizes that the conductor is exerted too tight a control over the orchestra to achieve a real explosion of joy. A fine performance after its lights, but not one of my favourite Märzendorfer discs.
Dorati, not unexpectedly, gives us the 1963 Robbins Landon edition as is, but adds quite a lot of dynamic tweaks of his own. In the first phrases of the first movement, for instance, which he takes briskly but more steadily the Märzendorfer, he imposes a diminuendo on the third bar of each phrase. An original idea, but the more raw drama of every other conductor, in their various ways, seems more effective. He fusses no end with the dynamics from bar 25 to bar 40. He does ensure, as only Wich has done so far, that the slurred semiquavers in bars 50 and 57 do not seem acciaccaturas, but on the other hand, he is not always attentive over the dotted rhythms. They do not quite lapse into 6/8 time, but they are not always incisive. In the Andante, he differentiates between the slurred staccatos of the theme and the straightforward staccatos of the second subject – bar 60 is punched out almost aggressively – but he does not make the music sing, preferring to concentrate on short phrases. Despite a good midway tempo, he somehow manages to sound fussy and perfunctory at the same time. The minuet is slow, full of preening rococo graces. I found it tiresome. In the trio, the lower strings are often fractionally behind the upper ones, suggesting that the players were rather cheesed off with it. The finale is swift and over-phrased, with too many little crescendo-diminuendos in almost every bar to sound either joyful or eager.
Anyone who wants to hear this symphony with all its repeats will have to look among the later versions. Of the recordings made up to this point I should be inclined to go for Vaughan, with Goldberg or Bernstein well in the running. Editions apart, I would still have a lot of time form Barbirolli in Naples, or else his last Prom performance. As has been seen, many of the others have at least one especially fine movement, such as Sanderling’s Andante, the 1964 Jones minuet or Münchinger’s finale.
I discussed six live Barbirolli performances in the first part of this article. Four further live/broadcast versions remain for consideration.
Hans Rosbaud’s performance with the Grosses Orchester des Südwestfunks, shortly to be renamed the SWR Sinfonieorchester Baden-Baden und Freiburg and usually referred to in English as the Southwest German Radio Symphony Orchestra, is a studio recording dating from 1950. The text seems to be an almost pure rendition of the 1950 Robbins Landon Urtext, with just a few tweaks. Whereas Robbins Landon, in 1963, retrospectively applied the slur over the last three violin notes of bar 150 to the similar phrase in bar 37, Rosbaud goes the other way and, presumably courting consistency, plays no slur in bar 37 and removes that from bar 150. He makes an unwritten crescendo into the recapitulation of this movement and adds an upbeat to bar 85 in the finale. The presumed use of this edition raises an interesting point in bar 19 of the second movement, where Rosbaud has an A flat at the apex of the first violins’ phrase. This is how it was in Bärenreiter and that is how it still was in the 1950 Robbins Landon, perhaps by oversight since the natural is clearly marked in Haydn’s autograph and appears in the 1963 Robbins Landon. Most conductors, whatever edition they are using, play an A natural, perhaps by analogy with the following phrase in the second violins. However, this is an unusually clean text for the date.
Rosbaud is terse, driving and crisply articulated in the first movement, pushing briskly though not insensitively through the clucking theme. His second movement is quite broad and differentiates between the slurred staccatos of the first subject and the straight staccatos of the second. It is well phrased but somewhat gaunt. Perhaps this is deliberate, since the music momentarily becomes almost lush in bars 64 to 69. The minuet is swift and purposeful, without the slightest pause before the trio. Only Jones in 1964 was faster than this, though he achieved a greater sense of enjoyment. The finale is again brisk and driving. It does not seem overdriven, as some of the faster ones do, but pure exuberance seems not to be on the agenda. It is for this last reason that I would not quite place this very fine performance at the top of my list. There is something a little forbidding to its gaunt single-mindedness.
Nicolai Malko’s performance with the BBC SO, given in the Maida Vale Studio on 31 January 1960, was included in the Cameo Classics/Richard Itter Collection of historical Haydn performances. It is not one of the best in the series. Malko uses something very close to the edition played by Barbirolli, with the missing bar and changed harmony in the second movement and numerous smaller alterations. This might matter less if the performance was equal to the best of Barbirolli’s. The first movement is heavy and listless with a particularly charmless clucking theme. The Andante is slow and uncommitted. Momentarily, Malko wakes up for the minuet, which goes at a lively speed, a sort of stamping peasant dance. The conductor drops off again for the finale, which begins very slowly. The tempo picks up during the movement – or did the players take things into their own hands? By the time of the recapitulation it is going a quite a decent pace. Malko did much good work, but he must have been out of sorts that day.
Back to another German radio orchestra, that of Cologne, for a performance by Christoph von Dohnanyi. This opened a concert on 26 November 1965 and was followed by Theodore Berger’s Violin Concerto (soloist: Ricardo Odnosopoff) and Brahms’s Second Symphony. The Haydn and Brahms circulate privately with radio disturbance in the second movement that would make this tape unsuitable for an official issue. It would be interesting to know where the concert took place, since there is reverberation so long that an entire bar’s rest is scarcely enough to clear the sound.
There is a video of Dohnanyi discussing Haydn’s 88th Symphony and singing the composer’s praise, but no LP or CD recordings seem to have been made. He took the trouble here to use the quite recent 1963 Robbins Landon Urtext and to play it without frills, but he emerges as a slightly old-fashioned Haydn interpreter. The first movement is majestic in the Barbirolli manner, but with rhythmic resilience, and he relaxes the tempo a little for the clucking theme – not as much as Ansermet did and more convincingly. The second movement is fairly swift. The slurred staccatos are not pecked at but neither do they really sing and there is no differentiation between these and the straight staccatos of the second subject. The music is nicely phrased but others have found more depth. The minuet is slow and formal, with a slower tempo for the trio, but there is an agreeable lilt that not all slower performance achieve. Given the reverberant acoustic, it is as well that the finale is taken at a fairly relaxed tempo. It is interesting that, while his timing is identical to Dorati’s the music dances along much more amiably.
Although the Boston Haydn Society had published their pioneering Urtext edition of the “Paris” Symphonies in 1950, the news was slow in crossing the city to Symphony Hall. The Boston SO did not perform La Poule until 16 August 1969, when David Zinman conducted a performance. It was repeated the next season, on 25 and 26 July 1970, under Kenneth Schermerhorn. It cropped up from time to time over the following years but has been heard only once since 1990 (under Nelsons in 2017). I have not heard the Zinman but the second of the Schermerhorn performances can be heard in streaming on request to the BSO Archives[5]. From the phrasing it is evident that the old Bärenreiter edition was used – but at least not the edition that cut a bar and altered the harmony in the Andante.
Kenneth Schermerhorn (1929-2005) is mostly remembered as the long-term conductor (1983-2005) of the Nashville Symphony Orchestra, with which he recorded a well-regarded legacy, mainly of American music for Naxos. He served as trumpeter in the Boston SO in the 1950s and conducted several concerts with the Tanglewood Music Center Orchestra in 1955-56. He returned to the BSO as guest conductor just twice, in 1970 and 1974. The 1970 programme would seem to have been performed with a reduced orchestra, since the Haydn was followed by Mozart’s Piano Concerto K.459 (with Christoph Eschenbach), Stravinsky’s Concerto in D and Tchaikovsky’s Mozartiana.
Schermerhorn is swift and spirited in the first movement, maintaining his drive through the clucking theme. The second movement has an agreeable flow and staccato quavers with a slur over them are differentiated from those without. No great depth emerges. The minuet has a lively lilt with a nicely pointed trio. After three admirably sane and sound movements, a spirit of devilry seems to have got into the conductor and orchestra, leading them to experiment with how fast the finale could go. As a demonstration that a virtuoso orchestra can play it at this breakneck speed without coming of the rails – for just one tiny moment I thought they were going to – it has its uses, but it sounds too pressurized to my ears. Commenting this concert, Michael Steinberg of the Boston Globe felt that “together with all the thoughtfulness and clarity he brought to his work, there was … a tightness that kept the level of vitality generally a bit low”[6].
Of these four live/broadcast traversals, it is the Rosbaud that I would add to my performances that count, but it sounds as though a few Haydn recordings under Dohnanyi would have been worth making.
The BBC Programme Index shows that this symphony got a fair amount of attention during these years, beginning with Ian Whyte and the BBC Scottish Orchestra in 1943. This was followed by Harry Blech and the London Mozart Players in 1951 and the Bournemouth Municipal Orchestra under a conductor not named in the Radio Times (maybe Rudolf Schwarz) in 1953. The BBC cycle of all the Haydn symphonies using their regional orchestras under an array of conductors allotted no. 83 to Maurice Handford and the BBC Northern Orchestra in 1965. Christopher Seaman seems to have developed an affection for this work in his BBC Scottish SO days, with performances in 1969, 1970 and 1973. These do not seem to be re-broadcasts of the same performance, since the programmes are otherwise different. There were also relays under Frederik Prausnitz (BBC Northern SO, 1969), Szymon Goldberg (Northern Sinfonia, 1970) and, in 1971, Uri Segal (BBC Northern SO), Andrew Davis (BBC Scottish SO) and Charles Groves (BBC Northern SO). I suppose some of the later ones might survive somewhere.
| I | II | III | IV | |
| Heiller | 07:54 first repeat only | 08:18 neither repeat | 04:12 | 04:29 first repeat only |
| Baudo | 05:45 neither repeat | 06:42 neither repeat | 03:44 | 03:08 neither repeat |
| Goldberg | 07:42 first repeat only | 07:53 neither repeat | 04:00 | 03:44 first repeat only |
| Münchinger | 07:18 first repeat only | 05:47 neither repeat | 03:59 | 03:46 first repeat only |
| Ansermet | 05:47 neither repeat | 05:07 neither repeat | 04:21 | 03:57 first repeat only |
| Bernstein | 06:52 first repeat only | 06:32 neither repeat | 03:59 | 03:41 first repeat only |
| Jones L 1964 | 04:55 neither repeat | 05:12 neither repeat | 03:31 | 03:53 first repeat only |
| Jones L 1966 | 07:08 first repeat only | 06:14 neither repeat | 04:22 | 03:49 first repeat only |
| Vaughan | 07:35 first repeat only | 11:17 first repeat only | 04:23 | 04:19 first repeat only |
| Wich | (damaged copy) first repeat played | (damaged copy) neither repeat | (damaged copy) | (damaged copy) first repeat only |
| Sanderling | 07:58 first repeat only | 07:27 neither repeat | 04:31 | 04:18 first repeat only |
| Märzendorfer | 06:48 first repeat only | 05:54 | 04:11 | 06:15 both repeats |
| Dorati | 07:46 first repeat only | 05:46 neither repeat | 04:27 | 03:54 first repeat only |
| Rosbaud | 06:33 first repeat only | 06:47 neither repeat | 03:43 | 03:42 first repeat only |
| Malko | 06:04 neither repeat | 07:47 neither repeat | 03:43 | 02:56 neither repeat |
| Dohnanyi | 07:46 first repeat only | 05:15 neither repeat | 04:12 | 03:54 first repeat only |
| Schermerhorn | 07:07 first repeat only | 05:55 neither repeat | 04:14 | 03:22 first repeat only |
Christopher Howell © 2026
[1] G. C. Burke: Joseph Haydn on Microgroove Records, High Fidelity November-December 1952,
[2] To be pernickety over this, in the autograph score, viewable at IMSLP, Haydn himself wrote slurs over some of the triplet groups but not all of them. Robbins Landon presumably took the view that, since MS scores in those days tended to be shorthand rather than finished products, Haydn slurred a few as an indication that they were all to be played that way. Probably this is correct, but Haydn was remarkably consistent, when rewriting this same passage in the recapitulation, over which groups he slurred and which he did not. So there might be a case for actually phrasing the music as Haydn indicated. Heiller plays them all slurred. Barbirolli seems to have been aware there was a problem here. In his 1949 Hallé recording, he seems to be doing something very similar to what Haydn wrote, though since he has those played with separate bows executed smoothly, not brilliantly articulated as by Baudo, it is not entirely clear what he is doing. In his Naples performance, he has them all slurred, while in Baden-Baden he has them separated and clearly articulated.
[3] This might be an editing error, but did Jones not listen to the edited master before it went for pressing?
[4] https://classical-discography.org/
[5] https://archives.bso.org/. Interested listeners may make a personal request to hear digitalized material for private research only – the recordings are not to be copied or shared.
[6] Boston Evening Globe, 28 July 1970.
