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 1: 1927-1950
A pdf version of this article can be downloaded for offline reading and printing here.
I have around 60 recordings of Haydn’s last symphony lined up within my cut-off date of 1972. I have therefore decided, with this and other much-recorded symphonies, to divide my survey into several parts: in this case 1927-1950, 1951-1958, 1959-1972 (Märzendorfer and Dorati) and live/broadcast recordings. However, when a conductor of an earlier recording has made a later one, or when a live performance by this conductor is available, I shall deal with them together with the first recording, so my path will not be entirely linear. In this first article, I discuss recordings by Barbirolli (3), E. Fischer, Beecham (2) , Weisbach, Van Kempen, Dressel, Furtwängler, Dobrowen, Krips (2), Munch (2) and Scherchen (2). I shall not offer an analysis of this work, which is widely known.
Appropriately enough for the somewhat arbitrary title, the first two recordings were conducted by a Londoner, John Barbirolli. Maybe an explanation exists somewhere as to why the future Sir John set the symphony down for Vocalion with the NGS (National Gramophonic Society) Chamber Orchestra in November 1927 and then, with so many Haydn symphonies still unrecorded, repeated it for HMV with his own John Barbirolli Chamber Orchestra in March 1928. It would be interesting to hear the two transferred by the same person, applying the same sound criteria. As found on YouTube, the Vocalion was piercingly shrill until I tamed it, while the HMV was dull-sounding until I boosted the treble. In both cases, the almost total lack of surface noise seems unrealistic for a recording of that date and I wonder how much music disappeared along with it. All that said, the two performances seem fascinatingly different.
After the opening fortissimo gesture, the following piano passage is dominated, in the Vocalion version, by a seemingly doom-laden bassoon, against which the violin phrases, with their dotted rhythms and sforzato first notes, have a beseeching quality, almost as though Barbirolli wishes to draw a parallel with Tchaikovsky’s Pathétique symphony. In the HMV recording, the bassoon is barely perceptible and the effect is of mystery and restraint. In the Allegro, with the NGS CO, Barbirolli obtains fiercely precise articulation from what sounds like a very small string section. Though the wind playing is on a lower level, the conductor’s ability to combine speed and brilliance with singing phrasing suggests one reason why he was a likely choice to succeed Toscanini in New York many years later. With his own chamber orchestra, on the other hand, string articulation is no more than good average and there are touches of portamento that I did not notice in the earlier performance. On the other hand, the wind are better, so the effect is more homogenous and, perhaps, more conventionally driven. Neither performance plays the repeat.
In the Andante, with the NGS CO Barbirolli adopts a slow tempo, definitely four-in-a-bar, vindicated by pin-point control over the phrasing to produce a timeless, other-worldly effect. With his own band, he takes a minute longer still, spilling onto part of a third side. The violins’ portamenti are generous and sound anachronistic to modern ears. This performance suggests, at least as conveyed by the primitive recording techniques, that Sir John’s inclination in later years to over-egg the pudding was already latent. No repeats in either performance.
The minuet spins along brightly and breezily in both recordings, close to a scherzo,the high-spirited vigour all the more evident in the more abrasive Vocalion recording. The repeat of the second part of the minuet is omitted in 1927 but included in 1928. In 1927, the vigour of the minuet continues into the trio. In 1928, the repeated rising thirds that introduce the trio theme are subjected to a schmaltzy ritardando each time the theme occurs, which is surely three times too many even if you do not mind it that way the first time.
The differences are smallest in the finale, which has a good drive without special features in both performances. Strangely, the repeat is played in the Vocalion but not in the HMV. I suppose that, for the listener of 1928, and indeed for the next ten years since the symphony was not recorded again till 1938, the HMV was a safer guide to the London but today it is the 1927 which holds more fascination, suggesting that Barbirolli might have become an iconoclastic conductor in the Scherchen mould. With the proviso, however, that this impression may be largely down to the recording itself.
Barbirolli was not given many opportunities to display his Haydn on disc in later years. He gave several performances of the London with the Hallé but our only chance to hear how his interpretation matured is a live broadcast with the Houston Symphony Orchestra given on 18-19 March 1963. His violins in the introduction retain the beseeching quality of the early recordings but combine this with a noble restraint. The main body of the movement is not driven in the way it was earlier, expressing rather a generous enjoyment of the music. Pin-point articulation as such had long since ceased to be Barbirolli’s primary concern, though the orchestra copes well with the more brilliant passages. The same is to be said of the finale. On various occasions in these articles, I have noted the difference between real and perceived tempi. The stopwatch tells us he is slightly faster than before in both movements, but he does not seem so – a question, perhaps, of allowing the music to breathe in the tempo. No repeat in either. The difference in tempo is immediately noticeable in the second movement. He now takes it at a pace that could just about be described as two-in-a-bar – which is what Haydn wrote – and provides much delicate phrasing without seeking excessive profundity – the music flows naturally. Even with the first repeat played, his timing is shorter than either of the earlier performances. Only the minuet is slower than before, though still far from the pompous gait of many of his contemporaries’ minuets. There is an attractive lilt that carries into the trio without the affectations noted in the 1928 recording. It would be nice to have this performance more generally available.
The next commercial recording was set down in the Beethovensaal of Berlin on 11 April 1938. Edwin Fischer conducted the Edwin Fischer Chamber Orchestra, a group of Berlin musicians he had created in 1932 (Electrola DB 4615/7). Recording techniques themselves had made considerable strides over the previous decade and the sound is pleasantly euphonious, with good strings/woodwind balance but rather backward trumpets and timpani. Fischer takes the introduction very expansively, after which the main part of the movement strikes by its luminosity and light, transparent textures. Considering that Barbirolli was also conducting chamber orchestras, it would be interesting to know exactly what numbers were involved. More than different numbers, though, I think it is a matter of Barbirolli conducting a small band but thinking in terms of a large one. As with many musicians of Fischer’s generation, loud passages in his performance tend to go a bit faster than soft ones, something less apparent with Barbirolli, but in general the music maintains its momentum well. The repeat is omitted.
In the second movement, Fischer takes a plausibly two-in-a-bar tempo, and his treatment of staccatos may seem more in line with modern expectations. But, while on the one hand this is a beautifully elegant and poised interpretation, space is also found for a heartfelt inner warmth and, in the forte passages, passion. This is a remarkable performance. Neither repeat is played.
Fischer’s minuet dances along at a tempo similar to Barbirolli’s recordings. The effervescence continues into the trio – the contrapuntal passages after the double bar have a scampering quality that looks forward to Mendelssohn’s fairies. The second section repeat of the minuet is omitted. An scintillating, if repeatless, finale concludes a notable performance.
If Haydn lovers who were tiring of the primitive sound of the Barbirolli recordings greeted Fischer as the answer to a prayer, their loyalties were soon to be shaken by Sir Thomas Beecham’s first recording, made in the Kingsway Hall on 13 February 1939 with the London Philharmonic Orchestra (Columbia LX 856-8).
Beecham is not particularly slow in the introduction, but the chief interest here is the degree of dynamic shading he imposes. Haydn himself indicated only fortissimo and piano, with a few sforzati here and there and, at the end, pianissimo. You can argue eternally as to whether the musicians of Haydn’s day would have instinctively applied the sort of crescendi and diminuendi which Beecham applies liberally, or whether Haydn himself, sitting at the piano, which he apparently played only sparingly when things began to go wrong, could obtain with a nod here and a glance there the sort of expression a modern conductor might draw from the orchestra. Or whether it matters if you think Haydn benefits from a manner that would go unquestioned in César Franck. On its own terms, this is beautiful and richly expressive[1]. In the Allegro the sound is fuller than that of Fischer’s chamber orchestra, though certainly not heavy. I daresay the full complement of LPO strings was in attendance. There is an ebullience that looks ahead to Beethoven’s Second Symphony in the same key and Beecham keeps an even tempo between the loud and soft sections. No repeat.
The Andante is definitely four-in-a-bar, exquisitely shaded and with the first repeat observed. Coming after Fischer, I could not help feeling that, beautifully done as it is, the music was just not meant to be so slow. The minuet, too, is slowish, but how to resist the cheeky pomposity or the affectionate shaping – at the same speed – of the trio? The second section repeat of the minuet is omitted. The finale, without the repeat, spins along in the highest of spirits. Given the virtuosity of Beecham’s orchestra, there is no risk of producing brilliance for brilliance’ sake – all is pure joy. Nevertheless, if I had been a critic in 1939 I think that, albeit with many provisos and much soul-searching, I should have recommended my readers to go for the Fischer – the Andante, and maybe the minuet too, being the deciding factor.
Beecham returned to the symphony in the stereo age. His recording with the Royal Philharmonic Orchestra was set down in the Salle Wagram of Paris, presumably for fiscal reasons, on 10 May and 18 December 1958 and 2 and 8 May 1959. In the Beecham recordings discussed so far in these articles, I have noted often considerable differences between his various versions of the same work. Here, in the first two movements at least, he is fairly consistent. There are similar crescendi and diminuendi in the introduction and the Allegro bowls along splendidly. What I notice here is an increased naturalness to the music making. No special points, just relaxed but vital appreciation of the great music. In the second movement, a faster tempo by sixteen seconds may not amount to much, but Beecham often succeeds, as he did not before, in giving the impression of two-in-a-bar. I enjoyed this much more, perhaps because the improved recording techniques reveal just how exquisite his dynamic shading and phrasing was. After the music has faded slowly into nothing in the pause at bar 113, Beecham conjures the coolest, yet most expressive woodwind playing out of the silence with rare mastery. We could almost be listening to a Sibelius tone poem.
So far I have said nothing about editions, since in the years 1927-1939 conductors had to use what was available. By 1958-9 Beecham could have drawn on the latest musicological research, but he was not exactly alone among his contemporaries in not caring to do so. Actually, in this symphony there are no big issues. Following all these performances with the Robbins Landon score, I noted some extra notes for the second violins at bar 123 of the second movement which do not sound out of place, whoever wrote them, and some changed phrasing and added accents, which could partly be down to the conductor rather than the edition.
The two Beecham performances begin to part company from the minuet. It has all repeats this time and is markedly slower. I suppose he just about keeps it alert but I began to tire of it. He applies a valedictory quality to the opening notes of the trio, almost as if singing “Farewell, Farewell” to them, and for all the refined elegance of what follows this, too, outstayed its welcome at this speed. The finale is also a little slower. It is bright enough, and Beecham dwells lovingly on the gentle passage before the recapitulation, but for real joyous verve we have to go back to 1939. Somehow, we don’t talk about “late Beecham” in the way we talk about “late Klemperer” or “late Walter”, or for that matter “late Barbirolli” or “late Boult”, but he celebrated his eightieth birthday in the middle of these sessions and time was maybe catching up even with his indomitable spirit.
Beecham’s 1939 recording completes the pre-war discography of the London Symphony. Three recordings were made during World War II, all in Germany or German-occupied countries. First came Hans Weisbach, conducting the Vienna Philharmonic Orchestra on 12-17 June 1942 (Kristall 01466-8). Weisbach (1885-1961) had built up a fair reputation during the inter-war years in Düsseldorf and Leipzig. In 1932 he came to London to record Haydn’s 92nd and 97th Symphonies with the LSO. He moved to Vienna for the war years, conducting the Vienna Symphony Orchestra from 1939 to its temporary closure in 1944. Following denazification, he finished his career, from 1947, in Wuppertal.
The Vienna Philharmonic is picky over who conducts it and Weisbach appeared with it publicly just twice, neither occasion part of its famous subscription concerts. On July 4 1939, he shared the podium with Oswald Kabasta during the Greater Germanic Bruckner Festival. Weisbach conducted the Second Symphony, Kabasta the Sixth. On 11 May 1941, he conducted a programme for the Gesellschaft der Musikfreunde in memory of Max Reger[2]. It is unkind but inescapable to suggest that Weisbach owed the honour of recording a Haydn symphony with the VPO to the fact that some prime candidates – Walter, Kleiber – had been forced into exile. However, we have to deal here with the performance, not the events that surrounded it.
The introduction is broad, with added dynamic swells similar to Beecham’s. This is impressive. Impressive too, in their way, are the Furtwänglerian – i.e. not quite together – attacks to the fortissimo passages. The Allegro begins slower than any heard so far and is notable for its serenity. However, when the forte passage enters, the tempo increases to a more normal one, and so it goes on, back and forth. There is the danger, when this sort of thing happens and the conductor is Furtwängler or Walter, to praise the conductor’s flexibility while, when the conductor is little-known, to suppose this was simply the best he could do. I can only say I prefer here the tautness of Barbirolli or Beecham. I find, too, a certain woolliness to the tutti sound. This may be down to the acoustic or the recording itself, but I think it is due, also, to the fact that the various parts of the orchestra are not entirely together, so the cellos and basses follow a split second behind the upper strings. A split second may not seem much and this was an age, and perhaps a place, where precision was not the first priority but still, I found this had a cumulative effect. The repeat is omitted.
I have already mentioned the difference between real and perceived tempi. This raises its head again since my impression was that this was similar in tempo to Beecham’s last version – that is to say just about two-in-a-bar though fairly broad. For much of the time, later on, the effect is four-in-a-bar. The playing itself is often beautiful – the “Sibelius” passage noted in Beecham 1958-9 is almost as remarkable here. But the interesting thing is that the stopwatch shows this to be the fastest of any so far – the later Beecham would last 08:10 without the repeat (Weisbach plays neither). It is perhaps the frequent relapses into a four-in-a-bar feeling that make it seem slower than it is.
Coming after the last Beecham, the minuet has a refreshing lilt if not the young Barbirolli’s verve. The trio begins with a schmaltzy interpretation of the introductory rising third motif, but thereafter the VPO’s natural instincts in this sort of dance music make for attractive results. The second section repeats are omitted in both the minuet and the trio. The finale, like the first movement, begins serenely while moving forward in the forte passages. There is a touch of Brucknerian dawdling in the more gentle, chorale-like passages. Probably I have exaggerated the shortcomings of this recording, but it hardly equals the achievement of Fischer or Beecham.
In the occupied Netherlands, a much more famous conductor, Paul van Kempen, recorded sessions with the Concertgebouw Orchestra from 23 to 29 May 1943. As well as Haydn 104 (Polydor 68207), they set down Schubert 9 and Sibelius 5. The complete sessions were issued on CD by Tahra, though this is no longer available. Van Kempen has the reputation of an impulsive, passionate and wayward interpreter in the Mengelberg mould, but perhaps he was subject to mood swings. His conducting in the first Kempff Beethoven Concerto cycle has been admired and I recall – though from long ago – a Beethoven Eighth that was steady and unaffected yet characterful. He seems to have been in the latter frame of mind when he set down the London. The fortissimo phrases in the introduction are firm and strong without attempting romantic drama while the piano phrases are nicely expressive. There is a degree of crescendo-diminuendo but not the detailed shading we had from Beecham. The Allegro begins serenely in the manner of Weisbach, but Van Kempen holds the tempo, allowing the music to unfold naturally rather than drive it along. No repeat. The second movement is quite swift, suggesting a serenade, but with strongly expressed forte sections it is definitely not superficial. The first repeat is played. The minuet is not a quasi-scherzo in the early Barbirolli or Fischer manner, but it dances alertly. The trio has a nice lilt. The second repeat is missing from the latter. The finale raises again the difference between real and perceived tempi. It is ten seconds shorter than Barbirolli 1927, the only other so far to play the repeat, yet Barbirolli sounds faster. This is because we are conscious of Barbirolli driving the music along, and because he insists on brilliant string articulation. Van Kempen is more relaxed in his tempo. As in the first movement, the music is allowed to unfold naturally. The chorale-like passages are particularly effective here. Logically, this performance might be judged a tad “ordinary” beside Fischer or Beecham, but somehow it is more than the sum of its parts and I grew rather fond of it.
The 1944 version by the Munich Radio Orchestra under Alfons Dressel was originally intended for broadcasting purposes. However, the American company Mercury, in the early days before it embarked on its celebrated recordings in Minneapolis, London and other venues, created a catalogue with tapes obtained from European sources, including Bavarian Radio. This performance was issued in 1951 (Mercury MG 10050). Not much information emerges about Dressel (1900-1955) except that, after the war, he was conductor of the Nuremburg PO from 1948 to 1955. On this showing, he was energetic and punctilious over dynamics – piano and forte are well contrasted. The introduction has considerable strength, followed by an extremely fast Allegro, verging on the febrile at times. The repeat is played. The Andante is taken broadly, with both repeats, but to my surprise, this is the movement I enjoyed most. It is nicely phrased, if with no special distinction. The minuet, while hardly a scherzo in the Barbirolli-Fischer mould, has an attractive galumphing quality, but unfortunately the trio is merely humdrum. I find it strange that, having observed all repeats till now, Dressel omits that in the finale. Perhaps it was a requirement of the radio station that the symphony should fill a 30-minute slot including time for announcements – Dressel’s performance lasts just under 27 minutes. This finale is energetic at a good but not feverish speed, but fails to make much of the chorale-like passages. I could prefer this to Weisbach, but it is hardly among the essential versions.
Nevertheless, the performance has got around quite a bit. In 1952, the ultimate pseudonymous label Royale issued it as the work of the Berlin Symphony Orchestra under Kurt List (Royale 1401)[3]. Moreover, the radio tapes were among the many that travelled to the USSR after the war and the performance was issued by Melodiya in 1975 as the work of the Vienna PO under Furtwängler (Melodiya 33M-1037145-46). A little later, as east-west relations began to thaw, the tape was among a batch restored to the Berlin radio stations and was subsequently issued by DG as by the Berlin PO under Furtwängler. Duly alerted, DG withdrew the disc, but it can still be found with the false attribution on various YouTube channels. In the meantime, a genuine performance by Wilhelm Furtwängler turned up. This was given at the Teatro Colon, Buenos Aires, on 2 May 1950. It would properly be discussed in the section of this article dedicated to live and broadcast performances, but in view of the curious history related above, I will discuss it here.
The recording is crumbly and loses a bar or so near the beginning. The orchestra, if hardly the BPO or the VPO, copes well with some extreme demands and Furtwängler clearly has them under his spell. I at least was sufficiently riveted to set aside any shortcomings. It is soon clear that it was quite ridiculous for anyone to attribute the Dressel to Furtwängler. There is a tension in the introduction and the beautiful shading of the diminuendo in bar 6 shows there is a real conductor in charge. I commented on Weisbach’s approximations to a Furtwänglerian attack but here is the real thing – it seems part and parcel of the conductor’s concept. The Allegro is steady, close to Van Kempen, but with that extra feeling that the orchestra is stretched to give its all. No repeat. The second movement is not all that slow, but somehow tempo becomes irrelevant given the full-hearted expression. Every phrase leads somewhere. The first repeat is played. We know from Furtwängler’s Symphony 94 that he did not go in for pompous minuets and I shall be rather surprised if this article throws up anything faster than this one – he shaves time off the young Barbirolli and is around two minutes shorter than the second Beecham. It is exhilarating but a bit hectic at times – perhaps he really did need the BPO or the VPO to bring this off. The finale is a thrilling, no-holds-barred rustic festival, yet he finds space to express the chorale passages beautifully without loss of momentum. I shall be returning to this.
Returning to the sequence of commercial recordings, the first post-war traversal takes us back to London when Issay Dobrowen set the symphony down with the Philharmonia Orchestra – then less than a year old – on 27 June 1946 (HMV C3515-7). Dobrowen (1891-1953) was born in Nizhny Novgorod of Jewish parents but left Russian in 1922 and became a Norwegian citizen in 1929. He is mainly remembered as the conductor of Boris Christoff’s first recording of Boris Godunov and for some notable concerto accompaniments.
The introduction is very broad. It is impressive although there is a feeling, at times, that the music is being excessively spelt out. The Allegro is brisk and energetic in the fortes, but sometimes a shade didactic in the pianos. The repeat is omitted. By the stopwatch, the second movement is identically paced to Beecham’s later version, which I enjoyed much more. This is because Beecham, even at a slow pace, managed to create the impression of two-in-a-bar. With Dobrowen, we are increasingly conscious of the four beats ticking over, and as it goes on it verges on a tiresome trudge. The first repeat is played. The minuet is galumphing in the Dressel manner and the trio is nicely phrased, yet somehow static. The second section repeat of the minuet is omitted. The finale is by far the most successful movement. Dobrowen’s Mussorgskian leanings perhaps led him to notice that this could plausibly be a gopak and he plays it hell-for-leather, yet integrating finely the chorale-like passages. No repeat. In spite of this finale, I suspect a rediscovery of Issay Dobrowen would not start from his Haydn.
Still in London, it was time for Decca to enter the fray, with Josef Krips conducting the London PO on 5-6 April 1949 (AX 287-9). In the fortissimo phrases of the introduction, Krips concentrates on sonority rather than drama, while the piano passages have a hushed restraint. Hushed restraint also characterizes the first statement of the Allegro theme, but with a transparent orchestral sound and long, singing lines the effect is joyful rather than urgent, let alone rushed. No repeat. In the second movement we have at last a real two-in-the bar – with both repeats, this is the shortest timing of any so far, including the performances with no repeats at all. Krips vindicates this with a miraculous balletic lightness, while still finding strength in the forte passages. We are also made very much aware of the felicities of Haydn’s orchestral writing – the doubling of the theme with a bassoon at bar 17, for example, and then later with the flute. This may be a matter of the recordings themselves getting gradually better as we advance chronologically, but I suspect it is down to Krips’s ear and careful balancing as well. I did have some passing doubts where Haydn goes into dotted rhythms from bar 82 to 89. Here it does sound as if the music is supposed to be a tad slower, to allow these to register properly. The bar’s rest at bar 56 is omitted – I suspect this was where the 78 disc had to be turned over and the otherwise sonically decent YouTube transfer I have heard has stitched it together without realizing that a pause was needed. Krips is scrupulous over pauses elsewhere. The minuet has a midway tempo – neither a scherzo nor a pompous strut – and it has all the natural Viennese lilt that Weisbach failed to obtain from the VPO itself. This carries into the trio, which also benefits from Krips’s careful balancing and transparent textures – the contrapuntal writing is beautifully clear. The finale is not taking excessively fast – maybe Krips knew well that Spiritoso means “witty”, not “spirited” (which would be con spirito). There is certainly nothing of the gopak here – all is light and joy. No repeat. A lovely performance.
Krips returned to the London Symphony on 28-29 September 1962, when he set it down with the Royal PO for Readers’ Digest (later – 1966 – available on RCA Victor 630.629). Krips has clearly rethought the introduction, which is considerably faster. This avoids the feeling of eight-in-a-bar, but creates a problem for the double dottings. At this tempo, the demisemiquaver either practically disappears or gets blunted into a simple semiquaver. The Allegro, still without the repeat, is faster than before. Krips obtains clean playing and transparent textures – though with the trumpets and drums kept well in the background – but it does seem a little breathless. The second movement is now shorn of its second repeat and swifter still. There is much delicate grace, but this touch of extra speed makes several passages sound uncomfortably busy in a way they did not previously – a slight harbinger of the HIP performances of the next decade or so that chose tempi on principle rather than musical effectiveness and then rammed through regardless. The minuet is a tad slower but here the issue is not the tempo itself, it is that Krips fails to obtain the natural Viennese lilt he found in 1949. The trio consequently sounds somewhat inert, though the playing itself is good. The finale now has the repeat and a similar steady tempo, but there is not the joy of the previous performance.
It is difficult to know what has happened here. Maybe Krips’s act had staled. Or maybe the RPO, a year on from Beecham’s death, were disinclined to collaborate on a very different sort of Haydn from the Master’s. Or a bit of both. Maybe Krips’s statement that “everything I try to do as if it were by Mozart” gives us a clue[4]. There are times when one feels it would be better for musicians not to give interviews unless they really are as good at talking as they are at performing, for it is so easy for the “enginer” to be “hoist by his own petard”, as Shakespeare put it. Bringing a touch of Mozartian transparency to a Richard Strauss tone poem may be all to the good, but playing Haydn in the style of Mozart risks losing touch with what makes Haydn Haydn and consequently playing Haydn’s last symphony in the manner of a frothy piece of teenage Mozart. I exaggerate, perhaps, but the 1949 recording is the one to remember Krips by.
The year 1950 saw recordings from three distinguished conductors, not all of them noted for their Haydn. First of the mark was André Cluytens, who set it down with the Paris Conservatoire Orchestra on 5 April 1950 (Pathé PDT 241). This was actually one of four Haydn symphonies Cluytens recorded for Pathé, the others being nos. 45, 94 and 96. Perhaps we would associate him with this composer a little more if he had recorded them for EMI. He takes the introduction quite broadly, with nicely shaded dynamics that do not extend beyond the classical world. The entry of the flute in bar 12 is particularly telling – something I had hardly noticed before and the sort of detail that a fine interpreter of Debussy might be expected to seize upon. The Allegro is vital and ebullient in a pre-Beethoven manner, reminding us what a fine interpreter Cluytens was of the friendlier, even-numbered Beethoven symphonies. The repeat is omitted. I get the idea that, for Cluytens, the heart of the symphony lay in the Andante, which he takes quite broadly with both repeats. He nevertheless maintains a two-in-the bar feeling with cool but expressive phrasing. As the music get fuller of notes, everything falls into place without any sense of business, and the dotted rhythm section sounds just right, convincing me that the correct tempo must be about this. The minuet approaches, without quite matching, Krips’s Viennese lilt in 1949 and there is some lovely wind playing in the trio. The second section repeat of the minuet is omitted. The finale, without repeat, is exhilarating, again pre-Bethovenian rather than Dobrowen’s gopak (at a similar tempo). A splendid performance.
Five days later, on 10 April 1950, Charles Munch recorded the work with the Boston SO (RCA Victor LM-49). He begins with a Furtwänglerian attack and emphasizes his dramatic conception of the introduction, taken fairly broadly, by inserting a pause before the first piano phrase and treating the following phrase as a pianissimo echo. The Allegro is quite steady and punchy to begin with. There is real excitement in the development, by the end of which Munch has reached a notably faster tempo, which he maintains during the recapitulation. This sort of thing can happen in live concerts, especially with Munch conducting, but for a recording perhaps he should have repeated the exposition in his faster tempo. No repeat. The second movement, with just the first repeat, is pretty slow. It is beautifully phrased and the forte passages are powerfully expressed, but the persistent four-in-a-bar weighs it down. Tempo seems absolutely crucial to this movement – just a little too fast or slow and a great conductor on the podium will not save it. Munch’s minuet is not unlike Cluytens’s, and he too omits the second section repeat of the minuet. His trio is suave – the more characterful wind players of the Paris orchestra lend a particular attraction to the Cluytens version. The finale – repeatless – is thrilling. With the Boston orchestra in full cry it verges at times on a hoe-down. While I would place this below Cluytens and the 1949 Krips, I was surprised to read from a contemporary critic that the “benefits” of the recording were “unfortunately incapable of erasing the depression caused by the laisser-faire of Dr. Munch’s direction”[5].
Fans of Munch avidly collect his live recordings, since he could often leap into life before an audience in a way he did not always manage in the studio. In a performance given in Boston on 16 April 1955, at first the two performances seem identical, except that the timpani are very prominent in this radio recording and a typical Munch shout can be heard before the second fortissimo phrase. This performance also shows that Munch’s choice of repeats in the studio recording was not due to the side lengths available – he plays and omits the same ones here. In the Allegro, Munch manages his excitement without accelerating in the development this time. The performance is ruggedly impressive. I would have said that the second movement was slower, but the stopwatch says otherwise. Nevertheless, Munch seems to have all the time in the world to draw tender, delicate phrasing from the orchestra, and I enjoyed this much more. The minuet is a tad more pompous this time and, with emphatic trumpets and drums, there is an air of tub-thumping. The sinuously graceful trio is highly attractive, however. The finale is also just a tad slower. It is less of a hoe-down, but not necessarily to its advantage – a suspicion of tub-thumping enters here, too. During the first two movements, I thought this was the Munch performance to remember, but the latter two left me undecided.
The third recording made in 1950 – in September-October – was by the Vienna Symphony Orchestra under Hermann Scherchen. This is part of what became a celebrated, if controversial, series embracing all the London Symphonies and quite a few of the others. Several of these have already been discussed in these articles. Scherchen shows he means business with a long-drawn-out but beautifully shaded introduction. One feature of this recording is that, practically alone so far, it balances Haydn’s orchestra properly – the trumpets and drums are present, not relegated to the background in a string-dominated texture, nor are the trumpets allowed to blare as they sometimes do under Munch. The Allegro, without repeat, is taken steadily but with enormous strength to every note. The unhurried tempo allows Scherchen to point strongly the many sforzati, such as those on the third beats of bars 54 to 56 – but I wonder how he would have reacted if someone had told him that these were put there by the Breitkopf editor, not by Haydn himself? For all its inner fire, I found this a shade didactic. The second movement, with the first repeat, is enormously slow. It was part of Scherchen’s genius to choose tempi that are logically wrong, yet to perform them with such mastery that, at least for as long as the music lasts, the listener is left wondering if it is all the other conductors who are wrong. But I do not think he pulls it off this time. For all his exquisite shading, and despite the cataclysmic force with which he invests the forte passages, there are passages of glacial immobility where there is simply not enough music to fill the space. Ergo, Haydn did not expect it to go so slow. The minuet contrives a nice Viennese lilt, interrupted only by Scherchen’s holding back, in the trio, of the introductory rising thirds. Not quite to the extent that Barbirolli did in 1928, but why do it at all? The finale, without repeat, has plenty of energy at a not excessive speed. But, when the chorale-like passages arrive, he holds back the tempo and makes a further enormous rallentando at the end of each of these sections. Out-Furtwänglering Furtwängler, we might have said had we not discovered that Furtwängler himself did no such thing. Not one of Scherchen’s revelatory interpretations, I am afraid.
Unpredictable as Scherchen was, we need not expect to find his performance on 29 October 1953 with the French National Radio Orchestra – without audience so far as I can hear – a replica of the Vienna recording. One thing apparent is that the orchestral balancing is much the same – so this was Scherchen’s work, though the engineers could have vitiated it in either case. The introduction is spelt out very broadly and the Allegro seems a tad slower than before, but with longer lines and less concentration on sharp accents it conveys more enjoyment. I did find, though, that the forte passages have a somewhat baroque gait.Repeats in this performance are as before. The Andante is considerably faster, within reaching distance of two-in-a-bar, and even verging on the free-flowing at times. A lovely performance. The minuet has a fine, stomping lilt, but Scherchen once again holds up the rising thirds that introduce the trio theme, perhaps even more than before. The finale is mostly a riot, at a thrillingly fast tempo. But alas, he holds up the chorale-like passages and draws out the unmarked ritardandi at the end of each of them interminably. In conclusion, the eccentricities here and in the trio loom larger than ever, but much of this performance is considerably better, above all in the second movement, so I would direct seekers after Scherchen here.
Christopher Howell ©2026
| I | II | III | IV | |
| Barbirolli 1927 | 07:42 repeat omitted | 08:06 repeats omitted | 03:36 2nd section repeat of minuet omitted | 07:00 with repeat |
| Barbirolli 1928 | 07:20 repeat omitted | 09:07 repeats omitted | 04:16 | 05:19 repeat omitted |
| Barbirolli 1963 | 06:52 repeat omitted | 07:57 first repeat played | 04:56 | 04:42 repeat omitted |
| Fischer E. | 07:14 repeat omitted | 07:42 repeats omitted | 03:35 2nd section repeat of minuet omitted | 04:58 repeat omitted |
| Beecham 1939 | 06:58 repeat omitted | 08:52 first repeat played | 04:26 2nd section repeat of minuet omitted | 04:39 repeat omitted |
| Beecham 1958-9 | 07:02 repeat omitted | 08:36 first repeat played | 05:53 | 05:09 repeat omitted |
| Weisbach | 07:28 repeat omitted | 07:25 repeats omitted | 03:09 2nd section repeats of minuet and trio omitted | 05:04 |
| Van Kempen | 07:39 repeat omitted | 07:24 first repeat played | 04:29 2nd section repeat of trio omitted | 06:50 with repeat |
| Dressel | 08:16 with repeat | 09:04 with both repeats | 04:49 | 04:49 repeat omitted |
| Furtwängler (Colon) | 07:35 repeat omitted | 07:41 first repeat played | 03:58 | 04:50 repeat omitted |
| Dobrowen | 07:16 repeat omitted | 08:35 first repeat played | 04:03 2nd section repeat of minuet omitted | 04:32 repeat omitted |
| Krips 1949 | 06:40 repeat omitted | 07:21 with both repeats | 05:00 | 05:06 repeat omitted |
| Krips 1962 | 06:01 repeat omitted | 05:30 first repeat played | 05:13 | 06:48 with repeat |
| Cluytens | 06:43 repeat omitted | 08: 58 with both repeats | 04:08 2nd section repeat of minuet omitted | 04:42 repeat omitted |
| Munch 1950 | 06:52 repeat omitted | 08:54 first repeat played | 04:22 2nd section repeat of minuet omitted | 04:43 repeat omitted |
| Munch 1955 | 07:30 repeat omitted | 08:30 first repeat played | 04:26 2nd section repeat of minuet omitted | 05:01 repeat omitted |
| Scherchen 1950 | 08:06 repeat omitted | 09:56 first repeat played | 05:06 | 05:44 repeat omitted |
| Scherchen 1953 | 08:32 repeat omitted | 07:51 first repeat played | 05:04 | 05:26 repeat omitted |
[1] Some of these crescendi and diminuendi are found in the old Breitkopf edition, so we can expect to find conductors following them to greater or lesser degree for quite a while yet in this survey. The problem is when they are followed to greater degree and further embroidered, in which case the music risks sounding like Franck.
[2] A photograph found on Internet purportedly shows the VPO performing under Weisbach in Bucharest in 1941. Elsewhere, the same photograph is described as the Vienna Symphony Orchestra, Since no such concert shows up in the VPO Concert Archive, it is likely that the orchestra in question is the VSO.
[3] See research by Ernst Lumpe: https://soundfountain.com/allegro-royale/catalogue.html, viewed 19.6.2026.
[4] Quoted in John L. Holmes: Conductors, A Record Collector’s Guide, Gollancz 1988 p.153.
[5] C. G. Burke in High Fidelity, November-December 1952. He also thought the Krips “languid”. The Cluytens had evidently not been issued in the USA.
