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. 104 in D major Hob. I/104 (1795)
Part 3: 1959-1972

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

This third part of the article takes the story from the advent of stereo to the appearance of the first two complete cycles. It also covers the period during which the Robbins Landon Urtext edition came out, in 1967, though within my cut-off date only two recorded performances, by Jochum and Dorati, had used it. Märzendorfer seems to be using an independently researched Urtext edition which abandons the accretions of the old Breitkopf edition but does not always agree in detail with Robbins Landon’s findings.  A final part of the article will deal with various live performances that have come to light. Recordings are discussed here by Faris, Georgescu, Boult, Karajan, Markevitch, Kertesz, Sanderling, Klemperer, L. Jones, Jochum, Märzendorfer and Dorati.

In 1959, the great and good flocked to the studios to record Haydn’s London Symphony. First off the mark was Sir Thomas Beecham who completed in March a recording, his second, which he had begun the previous year. I discussed this in the first part of this article, comparing it with his 1939 version.

The good were represented by Ulster-born Alexander Faris (1921-2015). Faris has been principally associated on disc with Gilbert and Sullivan and musicals, of which he recorded several for World Record Club. As a composer, he wrote for the stage and the screen. His waltz, The Edwardians, has become a light music classic. For World Record Club, he also made a very successful disc of string music by Dvořák and Grieg, as well as a Haydn coupling of Symphonies 92 and 104 with the Sinfonia of London (ST 28). This was issued in 1959 and  is available in a good transfer from Forgotten Records (FR 1743). It is unlikely that a conductor could do operettas and musicals really well if he has not first had a thorough grounding in the “classics” and it is worth noting that in 1956 Faris had obtained a Commonwealth Fund Fellowship to study at the Julliard School in New York.

A certain lack of unanimity in the opening fortissimo gestures suggests that his baton technique as such may not have been of the clearest, but thereafter the introduction has impressive gravity and is carefully phrased. If this is on the slow side, the Allegro is swift, but exuberant rather than fiercely driven. On a very positive note, Faris and the engineers between them handled Haydn’s orchestra better than many others. The Sinfonia of London seems to have a full complement of strings, but this is far from a string-dominated texture. The wind are well forward, and the orchestra could boast many fine players. Trumpets and drums ring out proudly without swamping the rest. No repeat in this movement. The Andante has a wistful grace at a slowish tempo that nevertheless maintains a two-in-a-bar feeling, allowing the dotted rhythms and the semiquaver triplets later on to sound natural. First repeat only. The minuet bounces along delightfully with a sinuous trio – this is one of the best. The finale, with the repeat, goes at quite a lick but, as with the first movement, there is no sense of driving, just a sense that the musicians are thoroughly enjoying themselves. The chorale-like passages are nicely shaped without loss of momentum. Faris may have been “only” a good conductor, but he certainly did some good Haydn.

Probably most of us outside Romania know too little of George Georgescu (1887-1964) to say whether he belongs among the great or the good. The writer of the unusually detailed Wikipedia article seems to have no doubt of his greatness, mentioning him in the same breath as Celibidache. His recorded legacy is small, and it appears that very little live material survives to supplement it. His Haydn 104,  set down with the George Enescu State Philharmonic Symphony Orchestra and issued in 1959 (Eletrecord ECE-042) has the sort of concentration and total conviction we associate with a great conductor, though I would say this makes for interesting rather than great Haydn. The introduction has an almost febrile warmth – this is one of the occasions when Haydn is made to sound unusually close to César Franck. The relatively full tone of the strings even in piano passages may be due to dynamic compression by the engineers, but there are occasional drops to a real piano – notably at the beginning of the finale, suggesting that Georgescu aimed for a vibrant warmth almost throughout. This makes the main theme of the Allegro unusually expressive, while the forte passages, even though the tempo is not especially swift, assume a Beethovenian drive and conviction. The repeat is played. In the Andante, Georgescu’s fullness leads to a four-in-a-bar feeling and I had the odd experience of thinking this slower than the Faris I had just heard, only to discover at the end that is was a few seconds shorter. However, Georgescu’s conviction holds the attention. The minuet has a nice lilt at a moderate tempo, but the trio did seem a little too fulsome. The final bars of the second part, complete with a notable string portamento, sound positively Tchaikovskian. The finale, without repeat and at a relatively steady pace, impresses by its Beethovenian strength, though Georgescu allows the momentum to drop in the chorale-like passages. Impressive if big-boned, Beethovenian Haydn is required.

The sole Haydn symphony recorded by Sir Adrian Boult was set down with the London PO for Concert Hall on 26-27 February 1959. This isolated case need not be taken as a lack of interest in the composer. A search in the BBC Programme Index, which is by no means a catch-all and in any case is limited to broadcast programmes, shows that Boult gave performances of symphonies 1 (1933), 13 (1932), 45 (1933), 49 (1950), 63 (1958, the original version with an introduction by Robbins Landon), 86 (1940), 88 (1938), 94 (1932), 95 (1932), 97 (1932), 98 (1941, 1942), 99 (1941), 100 (1932, 1950), 101 (1942), 102 (1942) and 104 (1941, 1969), as well as “The Seasons” (1933). Tapes of Symphonies 35 and 63 apparently exist and in a future article I shall discuss a later performance of Symphony 99. It seems, therefore, that he knew the Haydn symphonies better than we might have imagined, and not just the last twelve.  

The encounter with Robbins Landon over Symphony 63 the previous year might have inspired us to hope Robbins Landon’s advice would have been sought here, but the still-usual Breitkopf edition is used. Boult was nevertheless very much his own man. The introduction is gravely paced but, where many conductors separate the phrases, Boult leads the ear onward. The Allegro is swift but the principal theme is played so intimately one is hardly aware of the fast tempo. Moreover, Boult brings out the off-beat second violin line as no one else has done so far. Dare I suggest this brings an Elgarian tinge to the music? With the forte passage we realize this is actually a brisk, no-nonsense tempo, but by thinking in long lines (it almost swings along one-in-the bar) Boult avoids any feeling of haste. There is real tension in the development. The repeat is played. From what little we are able to know about Boult in Haydn and Mozart, when they wrote a second movement in two not four, whether Andante or Adagio, that is what he did. This is a delicately poised serenade-like interpretation. Does he manage to avoid the music sounding too busy in its later stages? Yes and no. He encourages the flute to sound perky rather than lyrical as it breaks into semiquaver triplets, but I’m not sure this works so well when the music becomes forte. First repeat only. We also know that Boult was not one to hang around in minuets. This is one of the fastest, but rather than turn it into a pre-Beethoven scherzo, he manages a bouncing Schubertian lilt. The finale, without repeat, enters stealthily, the theme barely emerging over the drone bass, but much of what follows is rip-roaring, while finding space for gentle nobility in the chorale-like passages. There is an interesting parallel here with Leslie Heward’s rather Schubertian performance of Symphony 103.

The Vienna Philharmonic made their third recording of the London on 27-28 March 1959, conducted by Herbert von Karajan. This is one of a series by Karajan and this orchestra that first appeared on RCA (LSC 2535), later reappearing, for contractual reasons, on Decca in the UK, London in the USA. With Karajan on the rostrum, there was no escaping a perfectly tuned unison in the fortissimo passages of the introduction, or total unanimity of ensemble. This introduction is roundly sonorous even when piano. The Allegro, with repeat, is quite swift, but does not actually seem so since Karajan, even more than Boult, concentrates on long lines and an almost one-in-the-bar rhythmic trajectory. Though the details are fast, the motion is slow, as though the scene were observed from an altitude. This is coruscated-brow Karajan and impressive in its way, though I was not sure how much substance lay beneath the grand upholstery. Karajan has the trumpets well evident, the woodwind somewhat less, but it is interesting to find that the relative balance between these three choirs is subtly changed from one movement to another. In the Andante, the strings are to the fore, the heavier instruments muted even in fortissimo interjections. The atmosphere here is of a delicate serenade, a delicately poised two-in-a-bar where the feet never touch the ground, everything floating ethereally as in a painting by Watteau. The tempo seemed to me quite fast, though avoiding business later on, but the timing shows me it is not so very fast – other conductors seem to need to go faster if they want to maintain the impression of two-in-a-bar. In its way, this is exquisite. Both repeats are taken. In the minuet, the wind are brought further forward, but not the trumpets. The tempo is fastish and the music galumphs in a refined, civilized manner with a sinuous trio. This orchestra is not allowed to indulge its famous upbeats with Karajan in the offing. The finale, without repeat, is swift with a wonderful clarity of texture – all the contrary motion scales and exchanges between first and second violins have pinpoint precision both of ensemble and of balance. Get the notes perfect and Haydn’s exuberance will take care of itself, seems to be the idea.

I am in two minds about this performance. One the one hand, I am a little repulsed by what seems a conscious intention to do it better than anyone else. Orchestrally, it is indeed better than anyone else so far and I hardly imagine that the later stages of my listening will produce anything more perfectly managed in this respect. On the other hand, while it is not as heartlessly brilliant as Szell, the later movements did not entirely dispel my feeling concern as to how much substance there is behind it – except, of course, Haydn’s.

Igor Markevitch set the symphony down with the Lamoureux Orchestra in sessions from 10-23 December 1959 (Philips A 02041 L). There is no pretending that ensemble and tuning are as tight in the introduction as with Karajan and the VPO, but there is a much greater sense of involvement, verging on the febrile towards the end of the introduction and in the strongly driven Allegro. Yet Markevitch also finds time for interesting details. From bar 65, we are made to hear how the melody is first doubled by the flute, then by the oboe – in most performances we hear only the strings. In the development, the bassoon pops through the texture from bar 150 to piquant effect. Before the recapitulation, he makes a rallentando which no one else has thought necessary, at least to such an extent. No repeat. The Andante is kept on the move, sounding chunky, almost droll, after Karajan’s silky-smooth rendering. The fortissimo interjections have real passion. In this sort of performance, it does not seem to matter if the music becomes “busy” later on – it comes over as slightly humorous. First repeat only. So far, so good, but the minuet, though at a good tempo, is punched out almost brutally at times. The trio is nice, with the pizzicato accompaniment more present than usual. The finale, repeatless, is a touch rabid at times, and was the big rallentando in the last bars necessary? I prefer to remember Markevitch’s Haydn by the 103 on the same disc.

The years 1950 to 1959 saw twenty-one recordings of the London Symphony. The years 1960 to 1969 saw only six, not all of which enjoyed wide circulation. István Kertész’s recording with the Bamberg Symphony Orchestra, for example, was set down in November 1960 and issued in 1963 by the German label Opera (St 1956), coupled with the Clock under Wallberg. I defined Kertész’s Philadelphia performance of Symphony 67 as “Haydn-lite”, so how does he fare in one of the composer’s undoubted masterpieces? The introduction is solid and sonorous, followed by an unhurried Allegro. I appreciated the ringing trumpets, but Kertész seems to concentrate on accurate and crisp articulation of the fortes on a bar-by-bar basis, letting the tension sag, even without actually slackening the tempo, in the pianos. No repeat.The Andante goes at a plausible tempo – broad enough not to become over-busy later – but is too rigid to hold the attention. First repeat only. The minuet is on the ceremonial side, with a bland trio. Others have brought off such a tempo, or even a slower one, but this is too strait-laced to swing. The finale, without repeat, is taken fairly steadily and here, too, the priority seems to be to get it all neatly and nicely played rather than to develop the notes into a larger line. The impression is that Kertész is consciously over-awed either by the recording process – I am not sure how many recordings he had made till then – or by the need to be on best behaviour for 18^th century music.

I mention briefly another red herring. Gray’s Classical Discography lists a version on Eterna by Rolf Kleinert and the Berlin Radio Symphony Orchestra. The record number shows that this is actually Symphony 101 – Kleinert’s only Haydn symphony recording.    

Kurt Sanderling set the symphony down with the Dresden Staatskapelle on 6-9 February 1962 (Eterna 8 20 322). The young István Kertész could have learnt a lot from this about how to make a grand approach with slowish tempi sound simply magnificent. In truth, it is not easy to identify the secret of Sanderling’s success, since he does not ostensibly offer special features of phrasing. Yet the introduction moves ahead slowly but inexorably and the Allegro seems to unfold in a single, unstoppable line from its serene beginning. The repeat is played. The Andante has a sort of chunky grace with, again, long lines that make considerations of whether this is two-in-a-bar or four irrelevant. First repeat only. Sanderling gives a lesson in how to make a ceremonial-style minuet sound glorious – phrase in long lines (I have to keep repeating this) and do not punch out each beat equally. The trio is warmly benign. The finale, like the first movement, unfolds steadily but unstoppably – not even a slight dawdling over the chorale-like passages seems able to hold up the movement Sanderling has set up. No one listening to this will doubt they are hearing a great symphony. It is a pity that the recording, at least as transferred at Internet Archive, is opaque in a boomy acoustic. Perhaps it could be brightened up.

Anyone reading this last discussion with the conductor’s name blanked out might suppose I am talking about Otto Klemperer. Klemperer himself recorded the symphony with the New Philharmonia on 12 and 14 October 1964 (Columbia CX1931/SAX2571). Unlike Sanderling, he gives all the repeats, taking 31:13 minutes, somewhat less than Ginzburg, who took exactly 33, and considerably less than the 35-odd Horenstein would have taken if he had played all the repeats – Klemperer’s first movement with repeat is actually shorter than Horenstein’s without. Statistics apart, what does it sound like? In the introduction, very similar to Sanderling. Klemperer is punchier and swifter in the Allegro but does not create the sense of inexorable unfolding that we get from Sanderling. It is as though, having set the tempo, style and phrasing he wanted in the first paragraph, the elderly conductor simply sat back and let the performance play itself. It occurred to me at times that you can plod even without being particularly slow. The Andante, though, is definitely on the slow side, without much warmth or grace to save it. It is respectably phrased but, again, after while seems to plod along under its own steam. The minuet is a surprise, with a galumphing gait and a nicely phrased trio. I did not expect to enjoy this movement most of the four. The finale begins sluggishly. There are times when Klemperer, towards the end of a symphony, can create the impression of drawing together the threads, so you end by wondering if the performance had not been much better than it seemed for most of its length. Fleetingly, I thought he was going to manage it here, but not quite. Just a little too late in his career, perhaps. I wonder if a live performance from the 1950s exists somewhere, though the Haydn symphony Klemperer most liked to conduct over the years seems to have been the Clock, of which earlier live versions survive. As it stands, if you like Haydn big and grand, Ginzburg and Sanderling seem the ones to go to.

Later in 1962 came the rather disappointing second recording by Josef Krips, which I discussed in Part 1 of this article together with his much better 1949 recording.

A recording said to be by the Nuremburg SO und Zsolt Deáky was issued in Fratelli Fabbri’s “The Great Musicians” series in 1965. Deáky (1933-2016) had a sketchy existence on disc. His accompaniments for rare romantic piano concertos – Brüll, Raff and probably others – played by Frank Cooper for the Genesis label are presumably genuine. The Dutch Wikipedia has a quite detailed account of his career – which mostly took place in the Netherlands after he fled Hungary in the aftermath of the 1956 uprising. He entered music lovers’ consciousness, while baffling them at the same time, as the conductor of a highly regarded version of the Emperor Concerto, initially available only in Woolworth’s, by the Japanese pianist Hanae Nakajima. This, like the romantic piano concertos, was made with the Nuremburg SO. Baffling them because, according to where and on what label you bought this Beethoven concerto, the conductor was named as Deáky or Räto Tschupp. Either way, the orchestra was the Nuremburg SO. Tschupp (1929-2002) was another real conductor who did extensive work in his native Switzerland, promoting much contemporary music. Najakima’s Beethoven recordings have been transferred recently by CRQ. They attribute the conducting to Tschupp. If any doubt remains, I would only point out that Deáky, rather than Tschupp, worked regularly with the Nuremburg SO. Warning bells should ring over the Haydn 104, which has also appeared coupled with a Clock supposedly conducted by the king of fakes, Alfred Scholz. A hearing of the first three movements rendered further discussion useless – it is a pirated version of the 1949 Krips. At least those who depended on Woolworth’s for their introduction to classical music got a good performance, though in a terribly shrill transfer.

Leslie Jones’s complete set of the “London Symphonies”, with the Little Orchestra of London, was issued in 1968 (Nonesuch HF-73019)[1]. This was the first time since Harry Blech’s 1954 recording that a declaredly “classical-sized” orchestra had been used for this symphony on disc. Jones took an interest in recent Haydn scholarship and the Robbins Landon edition of this symphony came out in 1967. Just too late for Jones’s sessions? To my surprise, this performance, like that of Wöldike, which has been similarly touted as using the latest research, has all the same variants with respect to the Robbins Landon score that have characterized every other performance heard so far, including the extra notes for the second violins at bar 123 of the second movement. Given that Blech’s chamber orchestra version is not discernibly different textually and that both he and Jones play all the repeats, does Jones offer anything new except stereo sound? Well, Blech’s wind section was well forward but Jones goes further, encouraging the oboes, in particular, to play with an almost rasping sound. If it could be said that Blech used a chamber orchestra but thought in terms of a big one, Jones, with his laissez-faire approach to orchestral balance, less-than-tight ensemble and small dynamic range (this may be down to the recording), produced a lusty, merry sound, albeit a little raucous, that came as a revelation to many ears. He anticipated the earlier HIP versions in their seemingly deliberate attempt to recreate the presumed inadequacy of Haydn’s own band – a presumption I find it difficult to share. It is certainly lively, but the outer movements are often just noisy and the Andante, while going at a plausible pace, is four-in-a-bar in its effect, proving that you can plod even in a faster tempo. It is interesting that Jones’s minuet goes at exactly the same pace as Klemperer’s, yet it is Klemperer who manages a springier step.

I have so far looked on Eugen Jochum’s set of the “London” Symphonies, with the London Philharmonic Orchestra, as just beyond my cut-off date – it was issued in 1973. But no. 104 was one of the first to be recorded, on 28 October 1971. If nothing else, it can proudly claim to be the first to use the Robbins Landon edition. So we hear that the legato bowings in the first forte passage of the first movement Allegro really do sound effective way instead of punched out staccato and that the sforzatos on the third beats of bars 54 and 55 are unnecessary. Quite a few phrases in the Andante sound strikingly different in their correct phrasing and, above all, we hear that bar 123 without the second violins’ interpolation is not as empty as it looks on paper, since the horns sustain their chord till halfway through the bar. This recording also joins those that give all repeats.

I have given the highest praise to Jochum’s Haydn in several earlier articles. Doubts tend to be raised over the recorded sound in this cycle, and I have to agree. Apart from the string-based balance, even when we can hear that the trumpets are in fact ringing out boldly somewhere way back, the engineers seem to have wished to wrap the music in cotton wool, distancing the listener from the raw impact the performance might have had – and which other Jochum performances recorded elsewhere certainly did have. The beginning is impressive, Jochum carefully characterizing each dotted-rhythm phrase in the piano parts of the introduction without any need for the hairpin crescendo-diminuendos in the old Breitkopf edition. The Allegro is fleet but it is here that I began to regret the small-scale effect that the recording has – in all probability – imposed on the performance. Likewise the finale. The Andante has an air of wise innocence but even here, the fortissimo passages lack impact. The minuet has a nice lilt but the delayed upbeat before the long trills is surely too mannered for repeated listening. By the end of the finale, too, I wondered if Jochum himself was stretching the players to the fullest extent. Did he not feel at home with the London players as he obviously did with those of Berlin or Dresden?   

Time now for the two complete cycles.

The first question raised by Märzendorfer’s recording, with the Vienna Radio Orchestra as usual, is what edition he is using. No hairpin dynamics in the introduction, so evidently the Breitkopf edition has been discarded. But the violins in the first forte outburst of the Allegro  separate their notes smartly instead of joining them with legato bowing as indicated in the Robbins Landon edition. However, there is a surprise in store. In the recapitulation, Märzendorfer has these same phrases bowed legato. Time to look at the holograph score again. In fact, Haydn put no legato slurs over these phrases in the exposition, but did so in the recapitulation. So Märzendorfer, or whoever prepared the scores for the project, took Haydn at his word, treating the slurs in the recapitulation as a variant introduced by Haydn at that point. Whereas Robbins Landon appears to have assumed Haydn would have been consistent, and so applied the slurs retroactively to in the exposition[2]. Märzendorfer is at variance with Robbins Landon at several points in the Andante, too, and the reason can usually be found in the holograph. But not always. The flute solo starting at bar 131 has, in the Robbins Landon edition, the first two notes of each semiquaver triplet slurred, whereas Märzendorfer’s flute has the entire passage played with a very dainty staccato. The holograph has squiggles that I take to be slurs under the first two semiquavers of each group in the first bar of the passage, then nothing. Robbins Landon doubtless felt it was in line with the practice of the day to mark slurs the first couple of times and leave the player to assume that the entire passage is to be played that way. I should be inclined to agree. Märzendorfer’s flute, though, does not even slur the first two. Only a musicologist with all the historical materials to hand could sort out this riddle.

That apart, the performance, with all repeats, is magnificent. The hard timpani sticks make a fine impression at the beginning and Märzendorfer uses his very slow tempo in the introduction to create a sense of profound unease. In the Allegro, Märzendorfer has the wind well forward and creates a sense of open-air rustic celebration. The tempo is not all that fast, but it has splendid vitality all the same. The Andante is taken broadly and tenderly – apart from the fortissimo outbursts – yet retains a two-in-a-bar feeling. This is little short of sublime. The minuet and trio are bright and perky – yes, the trio too – while the finale is a whirl of excitement. Märzendorfer often takes first movements a tad slower than his colleagues and then lets rip in finales, and so it is here. At this tempo, there is no risk of the chorale-like passages holding things up. In many ways, this performance seems to have the same aims as Jones’s. It likewise removes the veneer of classical respectability that hovers over even the best previous performances, with Wöldike the nearest exception. But is succeeds splendidly where Jones seems simply to be having a good try.

Dorati’s performance, with the Philharmonia Hungarica, came in sessions from 27 November to 2 December 1972. With Robbins Landon as editorial advisor, the Robbins Landon edition is unsurprisingly used and all repeats are played. In other respects, Dorati seems to take a step backwards in the evolving world of Haydn interpretation. The sound is string-based, with the wind and timpani audible but kept in their places unless specifically brought forward for a particular phrase. This latter point is symptomatic. Dorati exerts a degree of conductorial control that aligns this performance with those of an earlier age, while lacking the sense of line that informed the best of them. At the beginning, the fortissimo phrases have their last note tapered away in a long diminuendo, something I had not heard tried before. In the piano section, while Dorati studiously avoids hairpin dynamics at the point where the old Breitkopf edition marked them, he lards the rest with little dynamic nudges. The Allegro has, on the face of it, a good drive, but there is a fussiness in the way the repeatedly hammered-out four-crotchets-plus-two-minims phrase which dominates the development gets an internal hairpin dynamic every time. The Andante is delicately done but again, the leaning on every sforzato and harmonic appoggiatura becomes tiresome. At bar 123, while Dorati obviously does not have the added phrase in the second violins, he has the horns taper away rather than hold their chord the full length, so there is a hole in the music that did not happen with Jochum or Märzendorfer. The minuet is another case of leaning on accents to create an atmosphere of rococo simpering and curtseying. Delightful if you like it, no doubt. Another consistent feature of Dorati’s performances are a tendency to take first movements briskly and breezily and then fail to drive home the finales exuberantly enough – the opposite of Märzendorfer. This finale is no more than decently lively and Dorati lingers on the chorale-like passage the last time round.

The conclusion has to be that, if you want an Urtext edition, Märzendorfer is on his own, both the versions using the Robbins Landon edition being flawed in some way. Landmark recordings along the way were those by E. Fischer, Ginzburg, H. Blech, Rosbaud, Wöldike and Sanderling, but this may seem unfair when so many of the others have a good deal to offer.

Christopher Howell © 2026

[1] There may be two versions by Jones. Prior to the issue of the Nonesuch set of 93-104, a coupling of 101 and 104 was issued by Oryx in the UK and Checkmate in the USA. Since neither Gray’s Classical Discography nor CHARM list any recording of this symphony by Jones, I cannot say whether this was a separate recording. The none I have heard is the Nonesuch.

[2] But readers should see my comments on the Wöldike recording in Part 2 of this article. It may be that the original orchestral material consulted by Robbins Landon shows that Haydn adjusted the parts in favour of consistency when it came to performance.