Guest Blog: Colin Matthews “Tempo Relationships in the Adagio of Mahler’s Tenth Symphony”

Tempo Relationships in the Adagio of Mahler’s Tenth Symphony; and two wrong notes

— Colin Matthews

(Originally published in The Musical Times)

There are many things that are misunderstood about Mahler’s Tenth Symphony, and most of them are understandable misunderstandings. However the one movement that is accepted by virtually everyone as complete, and – although still a draft – performable with minimal editorial intervention, is subject to what seems to me to be a misconception that has become so pervasive that it is necessary to put forward an alternative view.

My own perception of the state of completedness of the opening Adagio differs from the generally accepted opinion : I suspect it would have been substantially revised. One has only to compare the structural differences between the (comparatively more advanced) orchestral draft of the first movement of the Ninth Symphony with the fair copy (which itself would doubtless have been subjected to further revision had Mahler lived, although probably not as regards structure) to be aware of how radically Mahler was capable of reworking his music even at a comparatively late stage. The same applies to the second movement of the Ninth. Since the pivotal climax of the Tenth’s Adagio only reached its existing form at a very late stage, after at least three earlier attempts at finding the right material, its effect on the structure as a whole had not, I feel, been fully weighed up in Mahler’s mind when he came to write the draft score.

It’s not this aspect of the movement that I am concerned with however, although it’s one somewhat related to it: what I take to be the misinterpetation – in part owing to the movement’s unfinished state – of Mahler’s implied tempi. When we consider that there are 23 tempo directions in the Adagio of the Ninth Symphony (quite sparing compared with the practice of Mahler’s earlier years), but that there are only 7 in the Adagio of the Tenth (the entire manuscript has only 15 tempo markings, with none at all in the fourth movement), it is clear that Mahler’s practice – having no need to clarify what was self-evident to him at this earlier stage – was to leave the detailed indications of tempo to the fair copy.

The two main tempi of the Adagio, Andante and Adagio, are clearly set out at the beginning of the movement (bars 1 and 16) and repeated at the first reappearance of both (bars 40 and 49). The few subsequent directions are only modifications to these tempi : Rit – – – – A tempo (fliessend) over bars 79 – 81, Etwas zögernd in bar 184, and rit . . . A tempo in bars 216-17.

Although the International Mahler Society Edition (IGMG hereafter), edited by Erwin Ratz in 1964 and revised in 1968, adds no editorial tempo directions, no one, of course, interprets this as meaning that from bar 49 onwards there should be virtually no change of tempo. The reversion to the Andante tempo is implicit at bar 104, and the ‘fliessend’ of bar 81 is clearly also implicit in the first appearance of this material – the incorporation of the Andante melody into the main Adagio – at bar 32, and then later at bars 112, 153 and 172. The apocalyptic A flat minor episode (bars 194ff) needs its own independent tempo, as appropriate to music that is entirely unrelated to what has gone before, although the short linking passage before the build up to the dominant 19th chord (bars 199 – 202) is usually taken as reverting briefly to the ‘fliessend’ tempo. Although the final section consists for the most part of the Andante material, it is generally taken slower than fliessend, returning to Adagio by bar 238.

So much is straightforward. But what did Mahler mean by ‘Andante’ and ‘Adagio’? I don’t suggest that he intended by Adagio the literal meaning of ‘at ease’, since its use more or less exclusively for a slow tempo had become habitual by the the early twentieth century, and is the tempo marking of the finale of the Ninth. But Andante means ‘moving’, and that is what Mahler meant by it in the the second movement of the Sixth (Andante moderato), the fourth movement of the Seventh (Andante amoroso), the first movement of the Ninth (Andante comodo), and surely here as well. Yet these other three are invariably taken considerably faster than has become customary for the Andante of the Tenth, even though Andante is qualified in each case by an adjective that could be taken to imply ‘slower than Andante‘.

There seem to be two reasons for this : one, a change in what Andante has come to mean, from a moderate tempo to a relatively slow one (I have myself found the term misunderstood when I have used the marking più andante and found that performers think what is meant is ‘slower’ rather than ‘faster’). The second is perhaps brought about by an over-reverential attitude to the Tenth – ‘because it is Mahler’s last work, written under the shadow of death, it has to be slow and solemn’. But Mahler certainly didn’t regard it as his last work; and it was not written under the shadow of death, but of betrayal. The overall emotional world – and there is no point in trying to detach Mahler’s state of mind from his music, in the case of this symphony at least – is for the most part (excluding the second movement) one of anguish, only resolved in the final pages. What that opening viola melody ‘means’ is beyond the scope of this essay; but it seems to me to be clear beyond fear of contradiction that is intended to have a completely different character from the main Adagio melody; and that the whole thrust of the movement is concerned with the intertwining of these two different elements.

If this is a correct interpretation, then it is clearly a misapprehension for the two themes to be played at the same tempo. But this is precisely what has become the norm : the Andante is followed seemlessly by an Adagio which is almost exactly half as slow as the Andante – the minim of the Andante becomes the crotchet of the Adagio. Or worse, the minim of the Andante is actually slower than the Adagio. When the Andante becomes part of the fabric of the Adagio (the ‘fliessend’ passages), the reverse equation is adopted and the tempo becomes almost exactly twice as fast. (Yet is reasonable to assume that if Mahler had intended such a radical change of tempo, he would have marked these passages Andante.)

Even those conductors who begin the viola melody at a relatively faster tempo tend to slow down as they approach the Adagio; but there is no rit. indicated. It is interpretatively valid to slow down a little, perhaps, but here, right at the beginning of the movement it is likely that Mahler would have indicated a rit. if he had wanted one : the contrast between the two themes is only enhanced if they are clearly heard at different tempi. An interpretation that fails to differentiate between the two themes misunderstands the character of the unaccompanied viola melody, which is not, in my opinion, that of a grave and mournful lament, but that of a shadowy dance – a character which many conductors do allow it to assume in the ‘fliessend’ passages. But here there is a major anomaly, since the very first ‘fliessend’ section, bars 32 – 38, leads straight into the reprise of the viola theme, which in bar 40 is marked Andante come prima. Here the convention has become to be immediately slower in real terms than the preceding passage – so, say, crotchet = 58 at bar 32 (and many get faster in this short passage) becomes crotchet = 52 or less at bar 40. It seems evident that Mahler did not intend this ‘fliessend’ to be faster than the Andante, but to be a more flowing version of the Adagio. This carries with it the implication that the Adagio itself should probably not be as slow as it is often taken, or the fliessend passages, interpreted as the score indicates, become simply too slow.

There are indeed three main tempi at play in the movement. It would be presumptuous to be prescribe what these tempi should be, since the one thing that marks out a true Mahler performance is fluidity. (The nearest thing we have to the experience of hearing Mahler conduct has always seemed to me not the rather clumsy piano rolls of the last movement of the Fourth and first movement of the Fifth that Mahler made, but Bruno Walter’s remarkable performance of the Trinklied in his 1936 Vienna Philharmonic performance of Das Lied von der Erde, where no two bars feel the same.) But with a great deal of hesitation (on another day I might think quite differently) I would like to suggest some possible metronome marks for the Adagio that make sense to me. The character of the unaccompanied viola melody might best represented by a tempo of something like crotchet = 72-76; the main Adagio certainly not less than quaver = 63; and the ‘fliessend’ sections around crotchet = 48 – 52 (and frequently faster – but never quite approaching the Andante tempo).

[Postscript 2016 : it did not occur to me when I originally wrote this that no conductor adopts an unduly slow tempo when this theme reappears on horns in the Finale. Here it is almost invariably played at a tempo which approximates to my thoughts on how it should be played in the Adagio]




My involvement as co-editor of Deryck Cooke’s Performing Version of the Tenth inevitably means that I will for the most part defend his editorial decisions, which in the case of the Adagio are not of great consequence. However I have every sympathy with those who prefer to perform this movement as Mahler left it. But to play it exactly as drafted is in several instances (including the matter of tempo that Mahler neglected) manifestly wrong-headed.

That he was capable of making mistakes in copying from one sketch to another is evident from several instances in the manuscript of the Tenth Symphony. The surviving sketches show that Mahler employed a process of gradual elaboration from composition sketch through one or two short scores to ‘Partitur’, or draft orchestral score (the next stage would have been the fair copy). In the draft score of the Adagio there is a notable error in bars 170-1, where Mahler misread his short score sketch and copied what he had intended to be a cello part into the second violins, transferring the same notes from one clef to another without noticing that they did not really fit : an error which the IGMG score did not see fit to correct in either edition. To maintain that this simple mistake was intentional takes very special pleading. There is a similar instance in bars 58 – 62, where he wrote out the bassoon parts in the wrong bars: here, however, he realised his error, and adjusted the parts, although he still did not succeed in getting the passage quite right. The first movement also contains a number of mistakes in notation which arise from the complexities of writing intensely chromatic music in F# major : Mahler occasionally mistakes the pitch of a note to which he has applied a double sharp, and writes it a tone too high.

The other significant misjudgement on Ratz’ part is to leave the final statement of the Adagio theme, beginning at bar 178, with the trombones undoubled by lower strings, in contradiction of its every previous appearance. It is so evident here – from the untidiness of the handwriting apart from any other consideration – that Mahler was simply getting the music down as fast as he could, and omitting the strings for expediency only, that I find it incomprehensible that anyone could seriously think that this scoring was intentional. But Ratz’ ability to persuade people of the untenable – most notably manifest in the completely unfounded assertions that allowed the Sixth Symphony’s inner movements to be played in the wrong order for 40 years – holds sway here too, and the power of the printed score means that performances frequently follow the IGMG edition without question.*

A new and lesser, but still significant, error came to light some while ago, just after the bassoon passage mentioned above : in bar 74 Mahler misread a note which is clear in both of his earlier sketches, writing a B in the bass line (for 3rd trombone and double basses) where he had sketched a C#. Example 1 shows the passage as it appears in Mahler’s draft score (all three examples are exact transcriptions of the manuscript). In Example 2, the same music can be seen in the short score : the bass note here is placed a little ambiguously because of the rather clumsily written E#, and this led Mahler to misread it. In Example 3, the composition sketch, there is no horn note and no ambiguity at all about the clearly written C.

The harmony here has always seemed strangely unfocused, but I did not think to question it until in 1998 the conductor Justin Brown pointed out to me that the draft score chord is the only non-triadic accompaniment chord in the entire movement, and that the note in question is ambiguous in the short score. There can be no question that the reading of Mahler’s draft orchestral score, and of the three published editions of the Adagio, as well as Cooke, is wrong.

The note C sharp presents a further problem in bars 187-193. In Cooke the second violin notes are C sharps. However detailed examination of the sources makes it almost certain that they should be C naturals. When I discussing this problem with Deryck Cooke many years ago his view was that it was not soluble, and he simply felt that C sharps ‘sounded better’. But in the sketches, and in the recently released short score page in the Moldenhauer archive these are tied C naturals – two semibreves in bars 187-8 tied to a minim in bar 189, with, in the Moldenhauer sketch, the second tie crossed out and the D natural acciaccatura added, but without any accidental to the C. In bar 191 there is a semibreve C natural tied to a minim. In the draft orchestral score the semibreve in bar 187 is tied to a minim in 188 without a minim rest and there is a similar detached minim C natural without a rest in bar 191; then the part moves down to the next system with an arrow pointing to the minim C with no accidental. Since in the sketches Mahler was clearly thinking C naturals throughout in the 2nd violins it is difficult to believe that he would not have added C sharps if he had changed his mind when altering the note values in the draft score. In this instance the IGMG edition is correct.

* This is compounded by the the use throughout the IGMG score of whole bar rests, where Mahler has virtually none. In Cooke there are no whole bar rests that are not


COLIN MATTHEWS was born in London in 1946. He studied with Arnold Whittall and Nicholas Maw; in the 1970s he was assistant to Benjamin Britten, and worked for many years with Imogen Holst. His collaboration with Deryck Cooke on the performing version of Mahler’s Tenth Symphony lasted from 1963 until its publication in 1975. Over four decades his music has ranged from solo piano music through five string quartets and many ensemble and orchestral works. He was Associate Composer with the London Symphony Orchestra between 1992 and 1999, and Composer-in-Association with the Hallé from 2001-10, now their Composer Emeritus.