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. 86 in D major Hob. I/43 (1786)
A pdf version of this article can be downloaded for offline reading and printing here.
Symphony 86, the fifth of the “Paris” Symphonies, is conceived, by the standards of the day, on the largest scale[1]. The Adagio introduction begins innocently enough but its brief, pithy phrases rise to a dramatic climax, after which the Allegro spiritoso slips in piano and off-key, though after four bars it bursts into an exuberant tutti firmly in the tonic major. This and Symphony 84 are among the first in which Haydn does not require both parts of the movement to be repeated – which must be a comfort to those conductors who would not have repeated the second part anyway. The finale, too, in binary form, has only the first part repeated. Evidently, as his symphonies grew in complexity and length, Haydn realized that his first movements, especially, would become unwieldy with both halves repeated. The fact that he evidently gave much thought to this, and indicated second half repeats in the other four symphonies of the “Paris” set, implies that he expected the repeats to be taken when he wrote them.
The second movement is marked Largo, but with the additional title of “Capriccio”. A later composer might have called it a “Scena”, since there is something of operatic scene painting in the stealthily rising arpeggio in the strings that opens it. Operatic, too, is the freely developing melody that follows, the violins duetting lyrically with a slightly sardonic bassoon. There is no repeat in this movement, though when Haydn repeats the stealthy string figure after a dramatic cadence he is presumably teasing listeners with the expectation that this is to be the customary repeat. The stealthy figure comes a third time, announcing a recapitulation, though the material is immediately varied. It returns yet a fourth time, now in the minor key. The nearest formal equivalent to this highly original movement would be an aria with four stanzas. The minuet is unusually developed – practically a piece in sonata form. The trio has a bucolic melody shared between strings and wind, accompanied by pizzicato strings and staccato horn chords, practically establishing the pattern for Viennese dance music of the next half century. The finale is relatively straightforward in its exuberance.
The first recording, set down on 13 September 1938, had Bruno Walter conducting the London Symphony Orchestra. Haydn did not figure largely in Walter’s repertoire, but he was not the man to miss the point of anything he conducted. The string-dominated warmth of the opening raises doubts as to whether he will be too “gentle” for this music, but come the Allegro spiritoso and the playing is as lithe as can be, while the first forte passage has the orchestra firing on all guns. As with many old-world conductors, Walter adjusts the tempo slightly here and there, but I only find this disturbing if the conductor fails to carry me with him, which Walter always does, or if the changes seem imposed from without, which they do not. In any case, the adjustments are small. Felicities are too many to mention – the flute phrase just after the double bar is one. No repeat. I compared the Capriccio to an opera scena and here is Walter, master interpreter of Mozart’s operas, bringing fresh character to every episode. The minuet trips along nicely, albeit without the second repeat – for lack of space on the 78 side I presume. The trio gets both repeats, and just as well, for I could have listened to this lilting Viennese Ländler many times more. If I ever get to see what Heaven is like, I should hope to find Walter conducting this trio somewhere not too far in. The lightness and vivacity with which the violins open the finale, with its repeated notes, shows that Walter, in his relatively early days, was also a superb technician. It cannot be easy to get this opening so perfect, yet I would make a bet that Walter simply raised his baton and got it. In second subject territory there is some knowing pacing that suggests one of the better Strauss conductors playing a polka on New Year’s Day. Anachronistic, maybe, but Haydn himself seemed to think he was writing a Viennese polka ante litteram. No repeat here, either. I have got myself into trouble before in this series by saying of the first performance I heard that I cannot imagine a better one, only to find that some other conductor not only imagined it but did it. So let me say nobody could possibly produce another performance like this.
Not even Walter himself? There are two live versions around, that of 10 February 1940 with the NBC SO and a performance with the New York PO, dated 12 December 1948. The latter has been issued by AS and there is a Memories set containing both. Neither seems available. I have listened to a version of the 1948 performance downloaded from a long-defunct site called “Musique Ouverte”. If there is a better source, I might be able to reassess this performance. The version I have is crumbly with severe distortion and a few glitches. That said, certain disconcerting aspects of the performance cannot be put down to the sound. Walter begins very broadly indeed and even the Allegro spiritoso starts cautiously. Then, with the forte passage, Walter whips up the tempo to a degree that sounds hysterical. What were slight tempo adjustments in 1938 are now practically an alternation between two different tempi. No repeat. The Walter of old emerges to some extent in the Capriccio but the conductor seems to want to dot every I and cross every T, so the sublime naturalness of the earlier performance is lost. The minuet is very heavy-handed, almost like a National Anthem. There is the second repeat this time and I cannot say I welcomed it. Even the trio lacks the easy grace of the 1938 recording. The finale begins lightly, but from the first forte passage it is just a race to the finish. To be fair, the recording here is so bad I would scarcely have noticed if they had been playing it backwards. The “polka” passages are more heavy-handed and there are some inserted portamenti as though Walter wants to find a parallel with Perpetuum mobile. The repeat is played but no, let us remember Walter by the 1938 recording.
The Walter LSO recording soon had a competitor, set down on 29 May 1940, in which the Leipzig Gewandhaus Chamber Orchestra is conducted by Paul Schmitz. A few recordings by this conductor and orchestra have been issued by dedicated bloggers, but not this one. Likewise lost to view is a 1950 Remington with Paul Walter conducting the Salzburg Mozarteum Orchestra.
This brings us to Harry Blech’s recording of 22 December 1952 with the London Mozart Players (HMV CLP 1009). This can be heard in a nice transfer by Forgotten Records (FR2012). It should be evident from other articles in this series that I have derived a good deal of pleasure from Blech’s Haydn as well as retaining warm memories of some of his concerts that I attended[2]. However, even his nearest and dearest would not have rated him alongside Bruno Walter. Perhaps it unfortunate that I heard Walter’s 1938 recording shortly before Blech. In reality, it would be difficult to fault this efficient and brilliant performance, with all repeats into the bargain. The trouble is, all the countless felicities to be found in Walter’s performance are breezed by. It is a painful example of the difference between a great conductor and a merely good one. But in truth, that warmth and generosity, together with a twinkle in the eye, for which I have praised Blech and which might have enabled me to set aside memories of Walter, do not seem to be present here. Curiosity is aroused by Blech’s phrasing of the minuet. What edition is he using?[3] All the same, in the words of The Record Guide[4],“the performance is in excellent style” (and earned their maximum two stars) and I am sure I will have heard far worse before this article is at an end.
It was a pleasant surprise to find that Zoltán Fekete’s Mercury recording (MG 10071), set down on 1 January 1954 with the Salzburg Mozarteum orchestra, was not one of them. Fekete (1909-c.1976) was born in Budapest and studied with Kodály and Bartók. By 1938 he had moved to the USA where he formed the Midtown Symphony Orchestra of New York and conducted it for five years. He also appeared with the NBC SO[5]. He aimed to be an all-rounder – composer, researcher and arranger as well as conductor – and had some success in persuading American orchestras to take up his suites based on then-little-known works by Handel and others. In 1950 he married Ann Freshman Ehrman in Salzburg, though the couple planned to return to New York shortly. This was Ms. Ehrman’s fourth marriage and evidently she decided to try at least one more, since by the 1970s Fekete was living in Munich and married to one Alma Hoehn. Fekete’s records, many of them pioneering versions of rare but worthy works[6], were issued on various labels, often bedevilled by poor sound. As heard on YouTube, his recording of Haydn 86 has thin sound and dynamic levels that go up and down disconcertingly. The performance is very attractive. The Adagio introduction is carefully graded towards its climax and the Allegro spiritoso, taken a tad slower than by Walter or Blech, finds time to shape the second subject nicely and to insert some dynamic variations in the forte passages. Logically, the Salzburg Mozarteum would have had similar numbers to the London Mozart Players, but it sounds more like a chamber orchestra. No repeat here or in the finale. The conductor shapes the Capriccio with some care and manages a formal but sprightly minuet. The trio has not Walter’s magic but is nice nonetheless. The principal divergence from previous interpretations comes in the finale. Haydn marked it Allegro con spirito in four, not Presto in two, and Fekete’s steadier tempo gives the music a more humorous character. The second subject no longer sounds like a Viennese polka and we can hear the droll bassoon accompaniment properly. Maybe CRQ or Forgotten Records could let us hear Fekete’s Haydn record – which also had Symphony 88 – in better sound.
On 29 September 1955, this symphony was set down by the Orchestra dell’Associazione Alessandro Scarlatti di Napoli under its founder-conductor Franco Caracciolo (Columbia CX 1378). The orchestra’s rather unwieldy name was abbreviated for the record-buying public to “Orchestra Alessandro Scarlatti”. The ensemble was formed in 1949, the same year as the London Mozart Players, and shared its objective of presenting the 18th century repertoire with a scaled-down band, though not necessarily accompanied by rigorous research into authentic editions and period playing styles. In November 1956, the orchestra was absorbed by RAI and extended its repertoire to more modern and even romantic works within reach of its chamber orchestra dimensions. Caracciolo remained its conductor, apart from a brief period (1964-1971) when RAI shifted him to its Milan orchestra and Massimo Pradella took over. RAI disbanded the orchestra at the end of 1992 in the context of a general retreat from serious cultural engagement.
It is evident from the Adagio introduction with its strong dynamic contrasts that this will be a fairly tough reading. The Allegro spiritoso is, like Fekete’s, a tad slower than Blech’s but I must say I expected something a little more spiritoso[7] from a conductor who did comic operas by Cimarosa and Rossini so well – the second subject is neatly done, no more. Likewise, the nicely played Capriccio might have mapped out the various phases of the argument in the manner of Bruno Walter. However, I very much liked the minuet, which is perky at a steady tempo. The trio has nothing of the Ländler but, with the pairs of quavers well separated, it has considerable rustic charm – this, if you like, is spiritoso. The finale takes a fairly steady tempo, similar to Fekete’s, with plenty of vitality and spirito. The second subject does not sound like a Viennese polka, but it does arouse thoughts of opera buffa. This performance, then, assumes real character in the last two movements, for which reason I would prefer it to Blech, even though Caracciolo omits the repeats in the outer movements[8].
When I came to Italy in 1975, Caracciolo and the Naples orchestra were frequently to be heard on the radio – I never saw him conduct, unfortunately. His work by then had an air of unhurried authority and a firm character, though the orchestra was not always in such good shape as we hear on the 1955 disc. A live performance given on 21 October 1983 is not very different in essentials, except for the inclusion of the repeat in the finale. That in the first movement is still omitted. In reality, he finds that extra space to characterize the individual moments in the first two movements in the way I missed in the earlier version. The minuet is now agreeably relaxed but in the finale the performance really takes wing. Here, at last, is the dash this conductor found so easily when conducting opere buffe by Cimarosa, Rossini and so many minor figures of the scuola napoletana. But he might have used an Urtext edition by 1983.
The “Paris” Symphonies came in from the cold during the 1960s. Five complete cycles were issued, with a sixth following in 1971. By that time, too, at least one of the two complete cycles, the Dorati, had covered the “Paris” Symphonies[9]. And, of course, single symphonies continued to appear.
First off the mark was Ernest Ansermet and the Suisse Romande Orchestra, who set down no. 86 on 6 April 1962. It may be difficult today for younger listeners to realize just how omnipresent Ansermet was in the LP world of the 1950s and early 1960s. Somehow the image of good-hearted honesty hampered by a less than first rate orchestra has not encouraged large-scale reassessment – not even by myself, though the composers in whose works he was most famed – Debussy, Ravel and Stravinsky – are among my favourites too. He was equally anxious to set down his thoughts on the Austro-German repertoire, and here he met mixed reactions even in his own day. Typically, Robert Layton, reviewing a mid-price issue of Haydn Symphonies 82 and 86, allowed there was “a certain robustness and honesty about its approach that compensates for the lack of finish”[10]. I prefer to recall that I got to know Beethoven’s Ninth Symphony through Ansermet’s mid-price single-LP version and it has tended to remain with me although I have investigated most of the “greats” in the intervening years. When I go back to it, which is more often than you might expect, I am always impressed by the inevitability of its movement. Somehow, Ansermet seems to gauge the exact tempi at which the underlying harmonic processes of the music combine with formal structure so that everything happens in exactly the right place, almost without need for further intervention from the conductor. Perhaps the conductor’s early training in mathematics had a role here. Yet intervention there is, since Ansermet had a fine ear for texture. Not in the sense of creating ravishing sonorities – he did not aim to do this even in Debussy or Ravel – but in the sense of fine balancing. When a string line is suddenly doubled by a wind instrument, for example, the nature of the sound changes, as though an organist has pulled out a new stop. Thus far, I might almost be speaking of Klemperer, but this is not a parallel that is usually drawn. If Ansermet’s pursuit of the unvarnished truth led him in other directions, this probably goes back to his early association with Diaghelev and with ballet music generally, and also perhaps to his friendship with his compatriot Jacques Dalcroze, founder of the eurhythmics movement. For Ansermet, harmonic and formal structure were also bound up with rhythmic movement, indeed body movement. In his hands, symphonic structures unfold naturally and inevitably. That said, while phrasing and melodic shape are honed to fit into the unfolding structure, if you look for the sort of felicities present in almost every bar of Walter’s 1938 recording, Ansermet just does not speak that language. For myself, he can persuade me that what he does is more than enough, but I do see that others may find it penny plain.
His recording of Symphony 86 is an excellent example. The various events in the Adagio introduction are not specifically moulded and apparently are left to stand one beside the other. Yet the underlying steady tread provides the link between them. Ansermet has found just the tempo at which each new idea comes at the right time, in the right place. The Allegro spiritoso may seem lacking in flamboyance, indeed in wit, yet there is plenty of vitality in the forte passages. Moreover, Ansermet makes much of the second subject, which often passes for little, not because he shapes it with lavish affection, but because it finds its natural space in his tempo. Its sighing phrases register particularly well when it comes in the minor key in the development. The Capriccio is remarkable for the way in which the arpeggio figure that introduces each phase takes on the air of a sphinx-like riddle, to which the following music tries, unsuccessfully, to find an answer, till at the end the composer concludes with a simple forte chord, leaving the mystery unsolved. The minuet is sturdy, kept alive by Ansermet’s balletic sense of rhythm. The finale is steady but benefits from resilient rhythms. Ansermet plays the repeat here but not in the first movement. Elsewhere in these articles I have remarked that Günter Wand, in his performance of Symphony 103, raised non-interpretation to the level of genius. There is something of the same quality here. You think of Walter’s Haydn as “Walter’s Haydn” and very delectable it is. I wouldn’t think of Ansermet’s Haydn as “Ansermet’s Haydn”. He gives me something more abstract, something that takes me closer to the mind of Haydn himself. However … not so long ago my MWI colleague Philip Borg-Wheeler found that “Symphony 86 provides a good general illustration of Ansermet’s shortcomings. … the bland, matter-of-fact Ansermet does the composer a disservice”[11]. I think I have explained why I cannot agree with this, but I realize that many will hear this performance as Philip does.
Ansermet seems to have been pretty consistent between one performance and another. A live outing with the same orchestra given on 10 October 1960 is for much of the time interchangeable with the recording, except for the omission of the finale repeat. It would seem, though, that for the recording, Ansermet, or somebody acting for him, did some checking up on the score he was using. In the live performance, bar 69 of the trio is played twice. This curious reading is found in the 1907 Breitkopf score, which I therefore deduce Ansermet was using[12]. Curious, because it was not present in the 1857 Breitkopf score, let alone Haydn’s manuscript. Unsurprisingly, it is not present in either of Robbins Landon’s editions either. Other details suggest to me, though, that Ansermet is not using an Urtext edition, for example the 1950 Boston Haydn Society edition, which was readily available[13].
Max Rudolf and the Cincinnati SO set this symphony down on 17 December 1964 (American Decca DL 710 107, Brunswick AXA 4540 in the UK). One is immediately struck by the sweetness of the opening chord. Rudolf has brought the oboes well forward, and is fortunate in having two players with a well-rounded timbre. Care over internal orchestral dynamics is a particular feature of this performance. At bar 10 of the introduction, the fortissimo first violins suddenly drop to piano so that we hear clearly the rising scales in the lower strings. These are normally perceived rather than predominant, since they are left to fight it out against the rest of the orchestra – including trumpets and drums, which Rudolf also cuts back. Then, with the last three quavers of bar 11, the first violins make a sharp crescendo to resume their leading role. This is done again in the Allegro con spirito, from bar 37, for example, where the repeated syncopated notes in the violins are marked down to let the arpeggio figures in the lower strings through. Care is also taken over woodwind phrases. At bar 151 in this movement, for example, the bassoon and oboe get a little duet to themselves, beautifully dovetailed, as a backdrop to the violin melody. Similar adjustments are made in the finale. Some will feel that these are conductorial methods unavailable to Haydn himself, but who really knows? If melodic interest shifted to the cellos and an accompanying figure was obscuring it, did he just shrug and think “that’s the way the world wags”, or did he gesture to the violins to pipe down? Moreover, in pre-conductor days, orchestras were essentially enlarged chamber groups. A good chamber player, if he realizes that the melodic interest lies elsewhere, will automatically play more softly to give the other player a fair chance. Perhaps Haydn’s musicians had more intelligence and musicality that the HIP brigade tends to assume. In any case, it seems to me that Rudolf’s adjustments succeed in directing the ear rather than disturbing it. What is certain is that here and in the finale, the music spins along with a good deal of joi de vivre. Rudolf obtains a slightly cheeky effect in the enunciation of the theme in the finale by making the staccato crotchet at the end of each phrase very staccato and with a slight accent, as if to say “so there!” No repeat in either of these movements. In the Capriccio, he charts an unerring path between knowing elegance and bursts of drama. It would seem, from the absence of the crotchet b flat on the third beat of bar 24, that he is using the Robbins Landon edition. That being so, there is an oddity in the minuet, where the crotchets at the beginning of bar 8 are not played staccato – not slurred, as with Blech, but definitely not separated. Since the section is repeated three times, there is no question that this was not intentional[14]. The minuet and trio are nicely done but perhaps the least interesting part of a very fine performance[15].
Rudolf was evidently fond of Haydn 86. In 1958, he placed it as the opening item in his first concert as Music Director of the Cincinnati SO (followed by Till Eulenspiegel and Brahms 1). On 9 May 1970 he ended a much appreciated tenure by replicating the original programme. I cannot answer for the 1958 performance, if a recording survives. That of 1970 shows utter consistency with the recording – the presence or not of an audience does not seem to have affected Rudolf particularly. The one small difference is that, towards the ends of the outer movements, he – or the engineers – has the trumpets ring out more decisively. One small curiosity: he plays the first bar of the second movement very slowly, then adjusts the tempo during the second bar as though thinking, “oh dear, this is too slow”. Thereafter the movement proceeds as before.
In the course of 1965, Denis Vaughan set down eleven Haydn symphonies – the “Paris” set and the five filling the gap between these and the “London” symphonies – plus the Sinfonia Concertante with the “Orchestra of Naples”. This was really the Orchestra Alessandro Scarlatti di Napoli della RAI out of contract. Born in Australia, Vaughan (1926-2017) became a member of Beecham’s RPO in 1950 and was appointed chorus master of the Beecham Choral Society in 1954. In 1966 he moved to Rome for a time and recorded quite extensively with the Naples orchestra – Schubert and Mozart as well as Haydn. These records were issued by RCA Victor. RAI do not seem to have engaged him for the Naples orchestra’s concert seasons until the early 1980s, when he conducted three programmes of mainly very different repertoire, such as Elgar’s Introduction and Allegro and two short pieces by Delius. The Haydn set came out with an introduction by Robbins Landon and declaredly used his edition.
As with Rudolf, the oboes are well forward at the beginning of the symphony, though the sound is a little more tangy. There are no dynamic adjustments and, as the Allegro spiritoso started out at a steady tempo I was reminded more of honest-Ansermet’s plain-unvarnished-truth manner. Setting the two alongside one another demonstrates, however, that Ansermet had an indefinable something-or-other that raises his performance to another plane. A keener rhythmic sense, maybe, or a feeling for human movement. And if the first movement seemed decent but ungripping, in the third and fourth – the latter taken very steadily indeed – the impression is that the Naples players are rather cheesed off with their Australian maestro and disinclined to do more than the minimum indispensable. I have left the second movement till last, since Vaughan here takes a different line from the other conductors I have heard so far, with a notably slower tempo. At this pace, it is made to appear that Haydn, like Rossini later in the Stabat Mater, is using the opera buffa idiom to say something very serious. The Capriccio, then lies in writing one thing and meaning another. It is an interesting idea, but Vaughan would have needed to obtain more distinguished phrasing and rhythmic life to make it work. The first movement repeat is given but not that of the finale[16].
The following year, in July 1966, Leslie Jones recorded the “Paris” Symphonies with his Little Orchestra of London (Nonesuch HC 73011). The edition used is not declared, but everything conforms to the Robbins Landon. There is more life and vigour to his no. 86 than with Vaughan – he takes a mere handful of seconds longer while including all repeats. All the same, doubts I have expressed elsewhere in these articles surface here, too. Take the first forte passage in the Allegro spiritoso of the first movement. There is no attempt to pace and grade the music. With even accents the trumpets and drums seem to be tub thumping and the music does not go anywhere – it just sits noisily in the same place until the next piano passage comes along. And then take the second subject in this movement. It is actually rather nicely phrased, but the even “bump, bump, bump” of the accompanying chords bog it down. The Capriccio is more successful, since Jones’s orchestra had fine players to help him out, but the minuet had me thinking this movement sounded slower every time I heard it. The chronometer confirms that it is indeed the slowest I have heard so far. A nicely turned finale does not prevent me from finding this another disappointment.
A further cycle came the next year from Leonard Bernstein and the New York PO. Symphony 86 was set down on 7 March 1967. He plays all repeats. The edition is not named but everything here is as in Robbins Landon. After an introduction that underlines the contrasts, Bernstein provides an Allegro spiritoso that is zestful and elegant by turns. I have heard these performances described as “beefy”, but this one at least does not sound that way to me. While letting himself go on the one hand, Bernstein has evidently done a lot of preparatory work so that the strings sound lithe, the brass do not dominate and the wind are adequately forward. He does not adjust the balance in the sense of cutting back the violins to let the cellos through, as Rudolf did, but in long forte passages he grades the dynamics, sometimes dropping almost to piano in order to make a crescendo. In spite of his swift tempo, he manages a nicely shaped second subject. Much the same can be said of his finale, which is not as fast as you might expect, but breezes along with abundant vitality nonetheless. Thus far, I could be at least as happy with Bernstein as with anyone else.
The Vaughans and Joneses turn out to have their uses as preparation for listening to Bernstein’s middle movement. I remarked that Vaughan’s concept of the Capriccio was interesting but would need “more distinguished phrasing and rhythmic life to make it work”. Unsurprisingly, Bernstein, at a marginally slower tempo still, can provide this, and indeed does more. Rather as Giulini, when conducting Rossini’s Stabat Mater, can almost convince us that the composer is not using the opera buffa idiom at all, so Bernstein, with utter serious and grave pacing of the accompanying figures as well as expressive treatment of the melody lines, can almost convince us that Haydn is not using it either. But “almost” is the key word. While Bernstein held my attention, I wonder if the “capricious” element has been lost.
For the minuet, it is Jones who provides useful preparation. By a matter of seconds, Bernstein reverses the syndrome by which each performance I hear goes slower than the last and he proves that with keen rhythmic sense the music can just about remain afloat at this tempo – even when he goes slower still at the end of the trio, not just in the repeat but the first time too. But can Allegretto really mean this?
Bernstein’s recording session was preceded by a public performance on 3 March. More than anything, this proves what a lot of careful professional work had been done. By the time of the recording, the symphony had been rehearsed in detail and put to the test of a public hearing. In truth, differences are so minimal that CBS could have saved money by recording the live performance and issuing it as is. Indeed, the live performance perhaps has a more free-flowing air. If there are any orchestral slips that would have caused embarrassment, my feeble ears could not detect them, though doubtless Bernstein’s superfine ears would have done so. Either way, a notable performance but, on account of the middle movements, I would not wish it to be my only one.
The fifth set of “Paris” Symphonies of the decade was set down in 1969 by Günther Wich and the Süddeutsche Kammerphilharmonie of Stuttgart[17]. It offers excellent, vivacious string playing and lithe, transparent textures. This may be the issue for some. Whether the conductor told the trumpets and drums, and the wind generally, to pipe down or whether the engineers insisted on mixing the sound so as to have the strings predominate, we hear as much of these elements as we would if they were battling with a traditional string section of 60-70 players, in spite of the fact that the actual sound is that of a small body of strings. Given the chosen sound picture, it is all very nice. The introduction is more mobile than some and the Allegro con spirito spins along splendidly, though without the repeat. Coming after a couple of extra-slow versions, the Capriccio finds an excellent balance between elegance, expressivity and drama. The minuet is fast enough to set up an enjoyable lilt and the finale, with the repeat, goes with great brio. This proves to be among the more successful performances. The text is as per Robbins Landon.
The first thing that struck me about Kurt Sanderling’s performance, from his 1971 cycle of “Paris” symphonies with the Berlin Symphony Orchestra, was that this was as strongly string-based a texture as Wich’s. Indeed, for the first few pages, I wondered if somebody was pulling my leg and the two recordings were the same. Sufficient differences emerged later to set my mind at rest on this point and even during these first pages I realized that Sanderling has a larger string section, nimbly though they play. I heard Sanderling live only once, with the Scottish National Orchestra in the 1970s, and witnessed how he completely changed the sound of the orchestra by rigorous balancing. I think, then, that it was his decision, not that of the engineers, to keep the trumpets and drums from ringing out at the climaxes. He has a quite mobile introduction, after which the Allegro con spirito is swift but not very swift, with fiery articulation in the forte passages. The second subject is not played for charm but rather as a forlorn little march. This makes sense in view of the strangely plain chordal accompaniment, which sits oddly with the usual elegant approach. This theme then reveals its true nature when it appears in a minor key during the development. The Capriccio is a welcome corrective to several excessively slow versions. Like Ansermet, Sanderling appreciates the germinal nature of the persistent rising arpeggio and holds a fine balance – even more than Wich – between the opera buffa backdrop and the gentle expressivity alternating with drama which constitute the main business. The minuet has a fair lilt at a moderate tempo and Sanderling shows that the trio can be attractively droll without special pleading. The finale, like the first movement, is swift but not very swift, forthright and energetic. All repeats are played and the text corresponds to the Robbins Landon score. This is clearly among the best, if not quite as overwhelmingly so as I had expected. I wish he had allowed his trumpets to blaze out more at climaxes.
This could be expected from Märzendorfer, to come now to the two complete symphony cycles. His introduction is a little more rough-hewn than some, with very deliberate staccatos in the second bar and likewise throughout, stressing drama rather than continuity. Assertive trumpets and drums cap this stage of the symphony. The Allegro spiritoso spins along with splendid energy and fiery attack. As with Sanderling, there is more of wistfulness than charm to the second subject. His Capriccio seems to have similar aims to that of Vaughan, but succeeds, partly because, while among the slower ones, it is not quite that slow, and partly because there is the distinction of phrasing and the rhythmic alertness that was lacking there. I am not sure that I do not prefer Sanderling’s or Ansermet’s swifter tempo, but Märzendorfer makes his view work. His minuet is sturdy and he finds considerable bucolic charm in the trio, at the cost of taking it at a slower tempo. The trouble with taking trios more slowly than the minuets is not so much the drop in tempo when the trio starts, which can be quite fetching, but the jerk back to the original tempo when the minuet resumes. Märzendorfer is in top form in the finale, which has terrific dash. In the second subject he points to the Rossini anticipations and allows us to hear the burbling bassoon properly. All repeats are taken and the Urtext is used. This is not, I would say, one of the occasions where Märzendorfer startles us by offering an anticipation of radical authentic-brigade practices. Ranked alongside the “traditional” performances it stands up well. Compared with Sanderling, it has the smaller orchestra and the more forward trumpets and drums in its favour, as well as a sizzling finale. Comparisons go the other way in the middle movements. I am coming to find, though, that, while I was initially lukewarm over Blech’s recording, nobody else so far has played the minuet at a tempo I would realistically call Allegretto.
Dorati’s first two movements are remarkably similar to Märzendorfer. He, too, draws attention to the staccato quavers in the second bar and similar phrases throughout the introduction, and for once his trumpets and drums are well forward to bring this section to a fine climax. His Allegro spiritoso scuds along very briskly indeed, almost breathlessly at times, but it is certainly exhilarating. He makes a dramatic moment of the sudden stop before the coda (bar 176) and the typical Haydn hesitations as he tries one theme then another before brushing them aside in a final moment of brilliance. I think Dorati manages this moment better than anyone I have heard so far. His second movement, like Märzendorfer’s, follows the line of Vaughan’s but does it better. The droll bassoon countermelody after the introductory flourishes is nicely sketched in and he makes much of the almost religious-sounding moment (from bar 41) where the flute comments poignantly over the long phrases in the lower strings. It was nice to think that perhaps, at last, I would find a case where Dorati is preferable to Märzendorfer. Alas, after this comes what must be the slowest performance of the minuet ever, utterly dispiriting. By no stretch of the imagination could this be called Allegretto. Stern duty compelled me to hear it to the end, but never again. The finale begins very delicately and proceeds with much gusto at a quite steady tempo. Dorati’s forward trumpets and timpani have their downside in this context. Lacking the exhilaration of a swift tempo, as in the first movement, Haydn’s Allegro con spirito has a bit too much of the spirito militare for my liking. All repeats and Robbins Landon edition.
Rather unusually, almost all the live performances I had gathered together are by conductors who also made commercial recordings – Walter, Caracciolo, Ansermet, Rudolf and Bernstein – so have been discussed already. Just one live recording came to light by a conductor who did not take the work into the studio – by Carl Schuricht with the Norddeutscher Rundfunk SO in March 1961. We are back in the days of doubtful editions, and maybe empirical editing by the conductor, too. The first violin semiquavers at bar 34 of the first movement, and similar instances, are not slurred four to a bow as in the Urtext edition but played with separate bows, producing an effect of sizzling brilliance. If there is an older edition that omits these slurs, it is not either of the Breitkopf scores, and they are in the manuscript. Better, perhaps, to close the score and revel in the sheer vitality that Schuricht, in his eighty-first year, could conjure from the orchestra. He does not hang around in the introduction, bringing it to a forceful climax. There is a strange pause before the Allegro spiritoso, possibly a recording blip – it sounds unnatural. Then, after an innocent beginning, Schuricht fires the orchestra up as few could at a very fast tempo indeed. Yet it is not all a matter of “rage, rage against the dying of the light”. In the development section, as the minor key reminiscence of the second subject arrives, he reminds us that conductors of his generation had the ability to relax at such moments, rendering them parenthetical, without losing sight of the overall structure. The Capriccio is forward-moving. By the chronometer, it is exactly the same as Ansermet’s, a few seconds shorter than Sanderling. Here, however, Schuricht’s flexibility is slightly unsettling so, while thanking him for not dragging it out, I have to prefer the more poised readings of Ansermet and Sanderling. Schuricht’s minuet is a swinging Allegretto similar to Blech’s but, unlike Blech, he takes a slower tempo for the trio. The finale carries all before it. No repeat, while that of the first movement was played. The radio recording is excellent for the date. I suppose this comes in the guilty pleasure category. I only wish I had been there – Schuricht would have had me on the edge of my seat in those outer movements. No-holds-barred Haydn to remind us that Haydn was a no-holds-barred composer.
I began this article by enthusing over Bruno Walter. Undoubtedly, the 1938 recording is a remarkable testimony to a remarkable conductor, but in the light of later listening I have to conclude it was made to speak to another age. This and the amazing Schuricht’s late traversal had better be classed as guilty pleasures. If you want all repeats and Urtext, it is down to Sanderling and Märzendorfer, the latter with the smaller orchestra but not as radical as he usually is. If you want all repeats and are prepared to waive the Urtext, there is Blech. I expressed some disappointment that it was less characterful than usual with this conductor, but remarked that I was sure I would have heard “far worse” before the end of this article. In truth, I have not heard all that many better either, and Blech alone plays the minuet and trio at a tempo that could be called Allegretto. If repeats are not an issue, Ansermet, Rudolf and Wich seem particularly insightful.
| I | II | III | IV | |
| Walter 1938 | 06:29 without repeat | 07:18 | 04:47 2nd repeat of minuet omitted | 04:01 without repeat |
| Walter 1948 | 06:56 without repeat | 07:33 | 06:07 | 05:32 with repeat |
| Blech H | 08:18 with repeat | 07:26 | 05:02 | 05:42 with repeat |
| Fekete | 06:53 without repeat | 06:25 | 05:40 | 04:58 without repeat |
| Caracciolo 1955 | 06:52 without repeat | 06:52 | 06:11 | 04:29 without repeat |
| Caracciolo 1983 | 07:03 without repeat | 07:10 | 05:59 | 05:47 with repeat |
| Ansermet 1962 | 07:04 without repeat | 05:56 | 05:41 | 06:33 with repeat |
| Ansermet 1960 | 07:12 without repeat | 05:52 | 05:37 | 05:02 without repeat |
| Rudolf 1964 | 06:45 without repeat | 06:22 | 05:38 | 04:21 without repeat |
| Rudolf 1970 | 06:58 without repeat | 05:57 | 05:43 | 04:17 without repeat |
| Vaughan | 09:01 with repeat | 08:08 | 05:56 | 04:47 without repeat |
| Jones | 08:41 with repeat | 06:10 | 06:28 | 06:55 with repeat |
| Bernstein studio | 08:20 with repeat | 08:33 | 06:19 | 05:52 with repeat |
| Bernstein live | 08:24 with repeat | 08:23 | 06:24 | 05:36 with repeat |
| Wich | 07:04 without repeat | 06:52 | 05:46 | 06:27 with repeat |
| Sanderling | 08:28 with repeat | 06:02 | 05:55 | 06:29 with repeat |
| Märzendorfer | 08:23 with repeat | 07:23 | 06:17 | 05:33 with repeat |
| Dorati | 08:07 with repeat | 06:43 | 06:54 | 06:34 with repeat |
| Schuricht | 08:20 with repeat | 05:56 | 05:42 | 03:58 without repeat |
Christopher Howell © 2026
[1] The performances discussed below in which all repeats are played range from 26:28 (Blech) to 29:04 (Bernstein).
[2] See the article on Symphony 49 for a brief outline of his career.
[3] This matter proves to be rather complicated, more so than I can resolve with the materials at my disposal. I was struck particularly by the first two crotchets in the melody line of the minuet at bars 4 and 6, which Blech has elegantly slurred, whereas the Boston Haydn Society edition of 1950, which I have, edited by Robbins Landon, has them staccato. My assumption was that Blech was following an old edition but other performances, such as those of Walter, Fekete and Caracciolo, play these notes staccato while being at variance with the Urtext at several other points. I have consulted the material available on IMSLP and I find that the old Breitkopf edition has these notes staccato, as does the later edition (1963) edited by Robbins Landon. So the slurs, which considerably change the character of the music, were a whim of Blech’s? Maybe not entirely. IMSLP also has a reproduction of Haydn’s manuscript. Here, the crotchets at the beginning of bars 2 and 3 are staccato, as are the three crotchets following the “incriminated” slurs, but there is no marking over those at the beginning of bars 4 and 6. So it could be that Blech, or someone on his behalf, or the editor of an edition I have not seen, noted this point and deduced that Haydn wanted these notes to be played differently. The addition of a slur would seem to be going too far, but perhaps no more so than the addition of staccato dots. The 1950 Robbins Landon edition has a note stating that “Both sources clearly intend all quarter notes to be played staccato unless otherwise phrased”. Robbins Landon mentions two sources. One is the MS which can be seen at IMSLP, and I beg to suggest it is not at all clear that Haydn intended this. The other is a set of orchestral parts in the British Museum, in Elssler’s hand but with corrections by Haydn. Maybe the matter is clearer there. Since I have not seen these, I cannot take the matter any further.
Another signal point might be bar 24 of the Capriccio, on the third beat of which Walter, Fekete and Caracciolo (the versions I have heard at the time of writing this footnote) have a crotchet B flat in the cellos and basses, where the Robbins Landon editions have a crotchet rest. The crotchet B flat is present in the old Breitkopf edition. An unhallowed editorial interference, I supposed, till I consulted the manuscript and found the crotchet B flat clearly present. There is no explanation for this in Robbins Landon’s notes. Maybe the crotchet B flat was deleted in Haydn’s hand in the BM parts, but if so he might have told us. I would add that IMSLP also has two sets of parts, undated but clearly from Haydn’s time or not much later, by Imbault of Paris and William Forster of London. The crotchet B flat is present in both.
A further oddity comes at the end of bar 22 of the second movement. The last three semiquavers of the violin part are marked piano and slurred in the 1907 Breitkopf edition and in Robbins Landon’s editions and are so played by Blech. They have no piano marking and have added staccato dots (while still marked to be played in a single bow) in the 1857 Breitkopf edition and are so played by Walter, Fekete and Caracciolo. I therefore deduce that Walter, Fekete and Caracciolo are using the 1857 Breitkopf edition. Blech is probably using the 1907 Breitkopf edition with some added detective work (or else an edition I have not seen). I may add to this note in the light of further listening.
[4] E. Sackville-West, D. Shawe-Taylor, with A. Porter, W. Mann, rev. Ed. Collins 1955
[5] San Francisco Chronicle, 3. March 1950.
[6] CRQ Editions have issued a transfer of his recording of Das Klagende Lied and other songs by Mahler (CRQ083).
[7][7] Unlike the marking of the finale, Allegro con spirito, which can be translated “lively with spirit”, the Italian word spiritoso means “witty”.
[8] Caracciolo’s Haydn disc, which also has Symphony 92, has been transferred to CD by Forgotten Records, FR 1104
[9] While there is no doubt that the Märzendorfer cycle was complete and on the market – in a limited way – while Dorati and Decca still had some way to go, the whereabouts of all historical information (masters, session diaries, editing notes etc), if extant, is unknown. It cannot be assumed, therefore, that, for every single symphony, the Märzendorfer preceded the Dorati. Dorati’s “Paris” Symphonies were made in November 1970.
[10] Gramophone, October 1968.
[11] http://www.musicweb-international.com/classrev/2010/Jan10/Haydn_Ansermet_4801942.htm
[12] So far, this is the only performance to follow this variant.
[13] Use of a copyrighted Urtext edition can add considerably to the costs of a recording. Cheaper, perhaps, if you have time to go to the Paris Bibliothèque Nationale or can get a trusted friend to go for you, to take your old Breitkopf edition there, ask to see Haydn’s MS and make your own corrections. Up to a point, IMSLP now enables you to do this in your own home.
[14] I have already discussed this extensively in note 3. A propos Rudolf’s performance I will only add that if you take the absence of staccato dots over the first two crotchets of bars 4, 6 and 8 in Haydn’s MS to indicate that these notes are not to be separated (though Blech’s slurs seem to me to be overdoing it), logically you would apply this solution each time, not just in bar 8. The early sets of parts at IMSLP, by Imbault of Paris and William Forster of London, reproduce exactly the situation in the MS. As already remarked, there would seem at least some room – including musical interest, if this matters – to query Robbins Landon’s assumption that ”Both sources clearly intend all quarter notes to be played staccato unless otherwise phrased”.
[15] Editions of the EMG Art of Record Buying in the mid-1960s listed both Ansermet and Rudolf with their maximum two stars. They did not extend equal enthusiasm to all of Ansermet’s Haydn series.
[16] Given the premises of the set, I suspect the finale repeat was played but omitted by the engineers. At Vaughan’s tempi, the symphony stretched to 28 minutes even without the repeat.
[17] For brief biographical notes on this conductor, see my article on Symphony 85.
