Mahler Symphony No. 9. New York Philharmonic Orchestra, Leonard Bernstein
Recorded 16 December 1965, Philharmonic Hall, New York
CBS Masterworks M3K 42200 (3 CDs, with Symphony No. 7, Symphony No. 10 Adagio) (1986), no barcode (European edition?), 074644220025 (US edition):
Sony Bernstein The Royal Edition No. 50, SM3K 47585 (3 CDs, with Symphony No. 7, Symphony No. 10 Adagio) (1992), barcode 5099704758527 (European edition), 074644758528 (US edition):
Collected in “The Mahler Symphonies”, 12 CDs Sony SX12K 68304 (1995), barcode 5099706830429:
Sony Bernstein Edition SMK 60597 (1998), barcode 5099706059721 (European edition), 074646059722 (US edition)
Collected in “Bernstein Mahler Symphonies 1 à 10”, 12 CDs Sony SB12K 89386 (2000?), barcode 5099708938628:
Collected in “Gustav Mahler The Complete Symphonies”, 12 CDs Sony SX12K 89499 (2001), barcode 5099708949921 (European edition), 696998949928 (US edition):
Collected in “Carnegie Hall presents Bernstein Mahler The Complete Symphonies”, 12 CDs Sony 88697-45369-2 (2009), barcode 886974536925 (remastering Andreas K. Meyer, includes Das Lied and the “Gustav Mahler remembered” radio program):
SACD Sony Japan SICC-10060-2 (3 discs, with Symphony No. 6) (2007), barcode 4547366032055 (remastering Andreas K. Meier):
Collected in “Leonard Bernstein The Symphony Edition”, 60 CDs Sony 886976836528 (2010):
Collected in The Complete Mahler Symphonies, 12 CDs Sony 88697943332 (2012), barcode 0886979433328 (remastering Andreas K. Meyer, “original jacket collection”, with Kindertotenlieder but without Das Lied and “Gustav Mahler remembered”):
Collected in “Leonard Bernstein The Remastered Edition”, 100 CDs Sony 88985417142 (2017) barcode 889854171421 (see listing on Discogs.com for content details):
Collected in “Leonard Bernstein conducts Mahler The Complete Symphonies Das Lied von der Erde”, 12 CDs Sony 19439708562 (2020), barcode 194397085624 (remastering Andreas K. Meyer, same content as Carnegie Hall edition, without booklet):
The blueprint of better things to come
Originally posted on Amazon.com, 23 November 2011

Bernstein’s first recording of Mahler’s 9th, made with the New York Philharmonic on December 16, 1965 (but released only in October 1967, in the Columbia set collecting his complete traversal of the symphonies, 15 LPs GMS 765 ), was one of the significant milestones in the symphony’s recorded history, in a decade that was exceptionally favorable to the composition, with the versions of Bruno Walter (1961 – he had made the premiere recording, live in Vienna, in 1938; links will open new tabs to my reviews), Barbirolli and Kondrashin (1964), Karel Ancerl (1966), Klemperer and Solti (1967), Abravanel and Haitink (1969) to mention only the best. DG later published three more recordings by Bernstein, made live, one on DVD, from 1971 with the Vienna Philharmonic (released in 2005 on Unitel 00440 073 4092, barcode 044007340929 with Das Lied von der Erde and the Adagio of the 10th symphony), the next from 1979 documenting his only encounter with the Berlin Philharmonic (it was released only in 1992), and the last one in 1985 with the Amsterdam Concertgebouw Orchestra, part of DG’s complete Bernstein/Mahler traversal.
First, a note on the sonics: on the early 1986 remastering, which is the version I originally had (CBS Masterworks M3K 42200, see heading), the transfer was cut at a low level which required cranking up the volume a lot. Furthermore, a comparison with Walter’s recording from four years before, even in CBS’ early 1985 remastering, revealed how much more vivid and pungent sonics Columbia had afforded the older maestro, and listening to Bernstein’s three later recordings on DG made matters even worse: although it wasn’t something that struck me when I listened only to the New York version in that early transfer, jumping from New York to Berlin made me painfully aware how much got lost due to the comparative dullness of Bernstein’s sonics in 1965. I had skipped the later Bernstein Royal Edition (oh those water colors of Prince Charles!) as well as the Bernstein Century edition. But the sonic deficiencies of the older CD transfer were so bad and so detrimental to a full appreciation of Bernstein’s recording, that I decided to go ahead and invest in the recent “Carnegie Hall” remastering, “first-time remixed and mastered from original multi-track analog tapes”. Well, friends of Bernstein’s Mahler, that is the version you must get. Indeed it represents a major sonic improvement over the 1986 transfer and even, from all the comments I have read, the subsequent ones. Finally Bernstein’s Mahler 9th can be compared with the other versions on its own musical terms, instead of being heavily handicaped by its dulled sonics.
Already in 1965 Bernstein has a fine grasp over the symphony’s variegated moods, and with a few exceptions pays close adherence to Mahler’s myriad indications of tempo, articulation and dynamics. Columbia’s recording is detailed and lets you hear the various strands of Mahler’s complex polyphonic texture, although I would have liked more stereo separation between first and second violins in the first movement, essential with Mahler’s intricate antiphonal writing. On the other hand the presence and impact of the cellos is phenomenal, bringing heart-rending lyricism to their counter-melodies in the first movement. In the first movement, unlike Walter in 1961 or Klemperer two years later, Bernstein doesn’t downplay the contrasts of tempos and moods; but tempo changes are expertly shaped, and he never overdoes them either (which is a pitfall that Barbirolli and Kondrashin didn’t always avoid, to great dramatic impact but at the cost of some loss of structural integrity). Listen for instance to Bernstein’s acceleration to the first climax: the “fliessend” at 5:47 is flowing but not gushing, and you can feel that he keeps a lot in reserve – this is only the beginning of the movement after all. Likewise, his ensuing Allegro, at 6:32, moves forward but is not rushed. Or jump further: at 10:49, his 6-bar acceleration to the “mit Wut” (with fury) section of measure 174 in kept somewhat reined in – here, one of the minor but perceptible evolutions in Bernstein’s interpretive conception will occur: in his Vienna and Berlin recordings he is more unleashed; the effect in more dramatic but also slightly less organic. But consequently, his later “mit Wut” are more furious than in 1965 – although one abiding characteristing of Bernstein’s interpretation of that passage is that he doesn’t maintain the fury throughout, but instead, through expressive rubati, substituted spots of expansive grandeur. Likewise, his two bars of “stringendo” before the big crash at 18:44 aren’t quite the race to the abyss that Kondrashin had made them the year before, or Bernstein himself in 1971 and 1979 – and truth is, his crash isn’t either as devastatingly blasting as Walter’s, or his own with Berlin in 1979, and nobody has come even remotely close to Ancerl and Solti in the ensuing timpani thwacks. Nonetheless, the Carnegie transfer lends Bernstein’s much more impact than it had before. Bernstein’s relative reining in of tempo in those build-ups to the great climaxes may come at the cost of a small loss in immediate dramatic thrust, but its value is that the gear shifts sound more organic and not as abrupt as they later became. So the only spot where I’d want to take exception with Bernstein’s shaping of those accelerations in the first movement is in the unnotated and conspicuous cranking up at 21:47 (measure 365), the intensely passionate passage abruptly cut off by the wonderful dialogue of oboe-english horn-flute: I can see Bernstein’s point here, to heighten the passion and play up the abrupt contrast with the ensuing serenading mood, but I find that it needlessly spotlights what is already naturally in the notes. He maintained that personal touch in the later versions.
Conversely, while Bernstein’s “fast” is never too fast, his “slow” is slow, but always deeply felt. Just listen at how he has the New York violins sing the aching phrases of the beginning, enhanced by the great, outpouring countermelodies of cellos (same thing when the passage returns modified at 20:30, measure 347). Vienna here is very similar; in Berlin Bernstein brings even more aching lyricism to these beautiful opening violin phrases – but less soaring cello counter-melodies. Listen also how, like Walter, he markedly slows down for Mahler’s “plötzlich sehr mässig und zurückhaltend” at 8:05 (suddenly very measured and held-back, measure 130). Here, Bernstein the composer even adds his own invention, namely the marvellously eerie 2nd violin sul ponticello starting at 8:42 (measure 138); Bernstein must have thought that, if Mahler had lived to perform his symphony, he would have made that correction – and I agree. But go to the 1971 Vienna DVD and even more the 1979 live Berlin version and hear how much more snarling the effect is there and then; and that’s something the Carnegie remastering hasn’t changed: it was simply part of Bernstein’s conception in 1965, that he expounded things that he would only later develop much further. Likewise, in ’71 and ’79 Bernstein extended that sul ponticello effect to the return of the figure measure 255: in 1965 he didn’t dare (15:32). In general, a comparison between New York and Berlin shows how much more instrumental pungency there is in 1979; as another illustration, try the stopped horn “growl” at 5:26 in New York and beginning of track 2 in Berlin: it is (was in the 1986 transfer and remains in the Carnegie remastering) sepia vs wide-screen technicolor (but some listeners may find Berlin garish and over-the-top). But the funeral fanfares (14:38, 15:11, passage starting at 19:09) have all the weight of Mahler’s “schwerer Kondukt” (heavy funeral procession) and intone their sarcastic laughs like death invisibly mocking the living. Independent of its spots of softened instrumental bite, this must be, with Solti’s, the best-balanced and architectured “Andante comodo” recorded until the late 1970s (I haven’t yet heard Haitink’s).
In the Ländler the improvement brought by the Carnegie remastering is jaw-dropping. It’s the difference between banging on a can and banging on a can covered by a thick layer of foam. Like Barbirolli and Ancerl, Bernstein provides a welcome and valid alternative to Horenstein’s and Walter’s more pedestrian gait (to say nothing of Klemperer’s hippopotamic – but also uniquely characterful and fun – approach two years later), and one that I think realizes better the relationships between Mahler’s three basic tempos. Bernstein’s brisker opening tempo conveys less a sense of good-natured bonhomie and more impertinent braggadocio, and the second violins on their first entry at 0:16 have more biting energy than true clumsiness (“schwerfällig”) (of that entry heard on the 1986 transfer I first wrote that it had “earthy robustness”). Tempo II at 2:26 is powerfully vigorous (“powerfully” added to describe the effect of the Carnegie transfer) without being rushed, which conforms exactly to Mahler’s instruction, “POCO più mosso subito” (Walter adopts the same Tempo II, but because of his slower Tempo I the effect is not so “poco”). Again Bernstein is true to score on his hectic acceleration at 4:30 (measure 198, “flott”), which by contrast brings great sentimental charm to his ensuing Tempo III, “ganz langsam” (very slow), at 4:59. Bernstein is very consistent in his choice of these three basic tempos in his 1971 and 1979 performances. Those three finely-judged tempi allow him to observe with exactitude almost all of Mahler’s ensuing, constant tempo changes – and, more important still than exactitude, always with a feeling of evidence and naturalness. But Bernstein isn’t one to sacrifice expression to exact observance of the score: listen how, in the accelerating passage starting at 12:03 / measure 486 (“Noch etwas frischer / still somewhat more lively”, then “allmählig etwas eilend, doch nicht überhetzt / progressively somewhat hurried, though not over-excited”), he gets so worked up that he does exactly what Mahler instructs not to do: he gets over-excited; it is hugely effective, though. And he is not Harnoncourt either and one to sacrifice expression to architectural discipline: listen how, immediately after, he returns “subito” to Tempo I at 12:35, measure 523, for the coda, at a far slower pace than his opening one – it is Walter’s tempo, in fact. Other versions from that decade were more consistent here.
This is what I first wrote of the 1986 transfer: “All that is fine, but Bernstein’s vision is also marred a certain lack of instrumental pungency, and the Columbia engineers have really let him down in their pick-up of the horns especially, which badly lack presence. This may seem like a petty detail, but given Mahler’s instrumental writing, it is essential. For instance, right at the beginning, the “keck” (saucy) solo horn at 0:23 is barely audible, which really ruins it all. Just go to Walter in these opening bars and compare. In fact, one needs only to go to Walter, Klemperer or Solti (to say nothing of Bernstein in Berlin) and compare at any point to realize, their great sonics helping, how much more character they have throughout, and not just through their choices of tempi. With Mahler, instrumental pungency isn’t just a seasoning, it is the essence. Go to Vienna and even more to Berlin, and the difference jumps at you, especially in the faster Tempo II sections. They are taken at a slightly more deliberate pace, but with huge vigor, making New York sound in comparison dynamic, but leightweight”. Well, one thing the Carnegie remastering hasn’t been able to help is the covered “keck” horn at the beginning and each time the same figure reappears: it is almost inaudible (just compare with Walter and especially Solti). For the rest, I am happy to report that it has now made the previous observations invalid. The sonic advantage remains in favor of Walter and Solti, but only very marginally, and it is perceptible only on A-B comparison. As for Bernstein’s 1979 Berlin recording, despite its “kecker” horns, I am not so convinced anymore of its superiority in this movement, sonic or interpretive. Comparing with the Carnegie remastering, I now hear that its sound is somewhat cavernous and over-reverberant, that its approach is immensely powerful but slightly thick and heavy in the faster sections, and that New York has more vivid and characterful woodwinds, especially felicitous in the mocking serenades, at 5:29 (measure 230) and at the end, from the entrance of the ostinto cellos and basses measure 569 (14:03). My last comment about the Länder does remain valid, though: “And, ooops! for the anecdote, one New York clarinet enters one bar too soon at 1:56 (bar 71)”. This is engraved for eternity.
Bernstein takes his Rondo-Burleske at a very moderate but balanced tempo, again very similar to Barbirolli’s, Solti’s or Ancerl’s, and somewhere in between the mad rush of Leopold Ludwig (the first stereo version, from 1959, on Everest, and an unjustly neglected one) and the very deliberate one of Klemperer. Compared to Solti, Kondrashin or Ancerl, his attack seems somewhat mellow and lacking bite, but as the movement develops he makes up for that through the piling of dynamics and instrumental layers, and brings plenty of defiance (“sehr trotzig”, is Mahler’s character indication) and instrumental nastiness to the outer sections, while retaining, in some passages, a sense of the burlesque. And his coda is hair-raising, when he whips his Newyorkers to heights of frenzy in the passage’s double acceleration (“più stretto”, then “presto”). In the middle section however I find that he sacrifices too much tempo coherence and structural integrity to the needs of instant expressivity: he slows down markedly for the “mit grosser Empfindung” passage (with great feeling) at 7:08 (measure 394), to great lyricism undoubtedly, but so much so that he needs to brutally speed up at 8:02 (measure 424) where Mahler gives no such indication, and from there he constantly shifts tempo to accommodate for changes in expression, sometimes even within two bars (for instance 460 and 462, 8:46 and 8:49). Sure, none of this is unmusical, and without a score no one is likely to notice. But other versions (Walter 1, Horenstein, Ludwig, Walter 2 to mention only those before Bernstein, and Solti just after) have proven that you don’t need to play accordion with Mahler’s tempos to get the varied moods just right, and that expressing Mahler’s paroxysmal emotions musn’t be at the expense of discipline. The problem with letting your heart pour down on your sleeve is getting it back in its original place afterwards, and to me Bernstein’s constant shifts of pacing only show that there is something he didn’t get right about his basic tempo here. While there can be great validity in the interpreter bringing his own twists and doing more than the composer wrote (and here, there certainly is), to me the greatest versions are those that make sense, music and expression OF exactly what the composer wrote. Not just freedom of interpretation: freedom AND discipline, freedom based on discipline, and Bernstein is not quite there. Add to that that, even in the Carnegie remastering, Bernstein’s New York Phil has less presence and pungency than Walter’s Columbia Symphony Orchestra from four years before – to say nothing of Solti in London two years later. In fact, surprisingly, the Carnegie remastering does bring a touch of added clarity and vividness, but the improvement over the 1986 transfer isn’t as dramatic as in the other movements. Still, here again, I’m not so convinced any more that Berlin 1979 is sonically superior to New York 1965: the A-B comparison reveals here too that the Berlin sonics are somewhat cavernous and over-reverberant. Interpretively though, where Bernstein’s tempo in 1965 was middle-of-the-road, in 1979 he comes closer to the mad rush of Ludwig. But that’s nothing in comparison to Vienna in 1971: it is helter-skelter.
Heard on his own, Bernstein plays the finale with passion and intensity, and also refinement in the softer moments, at a tempo that, given one’s expectations with him, is surprisingly flowing – although this is arguably closer to Mahler’s intentions than Levine‘s (very beautiful) time-at-a-standstill approach in 1979 or Bernstein’s in Amsterdam in 1985. The Carnegie remastering brings it more intensity (with slightly more aggressive trebble), where the 1986 transfer sounds softer and sweeter in comparison. I do think there was scope for more of an adagio feeling at the beginning and especially at the end than what Bernstein brings to it in 1965; more broadly paced, the finales in 1971 and 1979 did catch that feeling. Compare the final page, the farewell-infused, other-worldly “adagissimo”: 2:56 in New York, 4:08 in Vienna, and 4:51 in Berlin. Possibly due to that relatively flowing tempo he adopts at the beginning, Bernstein in 1965 is only minimally responsive to Mahler’s various indications to play “straffer im tempo” (tighter) at 1:57, “fliessend” / flowing (3:07), “etwas drängend” / somewhat pressing (3:41). But he wasn’t more in 1971 and 1979, and there it had clearly to do with a desire to keep things agonizing and slow. He is also – again – not very consistent in his tempi, and when the opening tempo returns at 4:03, he is faster.
Then there are the tiny details that make a world’s difference. Why am I moved so much more and deeply by Barbirolli in the desolate section with tolling harp (opening measure 88, Barbirolli’s 12:14, Bernstein’s 11:53)? Not a matter of pacing since, remarkably, the full section has exactly the same duration with both conductors, 1:33. I had to go back and forth a number of times to try and analyse why. Tiny details: harp more vivid and resounding with Barbirolli (with Bernstein it is almost covered by the clarinet ostinato) and tempo held-back a breath more at the beginning, giving more the sense of a death knell, string tone more desolate. But while this is really subtle, there is no question that on the surge of passion that interrupts that section, measure 107 (Bernstein’s 13:26, Barbirolli’s 13:47), Barbirolli’s is incomparably more passionate, which has to do with better defined string tone, more fff outburst (“breaking out violently”, wrote Mahler), and faster tempo, in conformity with Mahler’s instruction (“fliessender doch nicht durchaus eilig”, more flowing though not throughout rushed), where Bernstein keeps it steady. And the Carnegie remastering has changed nothing to Bernstein’s disappointingly weak and indifferent string attack here.
And then, turn to Bernstein in Vienna in 1971 and even more in Berlin in 1979, and what you thought was “passion and intensity” in New York is suddenly revealed, for reasons that have to do even more with sheer intensity of playing than with sonics, to be 3 or 4 on a scale of 10… Of the versions from the 1960s, Solti’s is also more intensely passionate than Bernstein’s here (to the point that some listeners find it “impatient” – but not me).
Overall then, as good and significant as it was in its time and for years after, Bernstein’s first recording was not without flaws. It was badly betrayed by its sonics, even in comparison with the recordings of that decade, at least as documented in the 1986 transfer, but I’ve read that things weren’t so much better on LP. In that respect, the Carnegie remastering is a true revelation. To think that during all these years Bernstein fans have loved this version like Michelangelo fans loved the frescoes of the Sistine chapel: heard (seen), or rather blurred through a thick crust of patina. Yet, interpretively, the first two movements are great, the last two, when you consider the finer details, more problematic. Of all the recordings from that decade which I’ve heard so far, Solti’s is for me, unhesitantly, the best; all the others, Walter, Barbirolli, Kondrashin, Ancerl and Klemperer (that’s going by chronological order) , have their shortcomings. My mixed feelings about Bernstein’s first version here are motivated by the debilitating sonic shortcomings of the early transfer(s), partly but not entirely solved in the 2009 remastering, and the fact that Bernstein’s conception is much more fully and convincingly realized in Berlin in 1979. (And a post-script from March 2021:) Another snag about the Carnegie Hall remastering is that, while there have been multiple reissues of the complete Mahler symphonies (see heading), the symphonies are not yet, more than a decade later, available individually.
In many aspects Bernstein’s 1979 concert performance it very close interpretively to the New York version, especially in the first two movements, only the conception is realized with significantly more instrumental pungency, at least in the andante comodo (with the Carnegie remastering I’d rather give the edge to the New York Ländler); and, where it differs, in the last two, it is marginally (in the Rondo-Burleske) or markedly (in the finale) more thrilling and intense. Berlin too has its flaws, which have made it subject to much bashing by the anti-BBBs (as in “Bernstein-Berlin Beckmessers). It is a live performance and comes with various glitches, the most menial being various audience and stage noises including Lenny’s grunts, and the most glaring a terrible slip from the trombones who inexplicably forget to enter at the apex of the finale’s climax. Notwithstanding, it is a phenomenal performance, of harrowing intensity and breathtaking sonic presence. In consideration of that recording, the previous 1965 New York version, even in its Carnegie remastering, can be considered only as the blueprint of better things to come, things that will be presented already in a developped form with the Vienna Philharmonic in 1971 but that will be carried out to the full only in Berlin in 1979.