Just occasionally a composer felt more strongly about the matter. In Brahms's case, op. 121 was described in the title as "Four Serious Songs for Low Voice and Piano". By and large his wishes have been respected. In this performance the songs are put up a minor third. Oddly enough, this affects the piano more than the voice. By this time in her career Flagstad was no longer young and had retired from the operatic stage. Her voice, always powerful and solid, had a dark quality, something like a contralto but with a soprano range. As a result the voice does not sound "wrong" the way a more typical soprano lieder singer - such as Schwarzkopf or Seefried - would. We can still appreciate her fail-safe intonation and generally rock-steady delivery. Just occasionally the notes of her passaggio - Fs and Gs - seem a tad queasy, but above there the As and the one B flat are still splendidly firm.

And so to the piano. The problem is admittedly endemic to vocal chamber music. Imagine playing your favourite Brahms piano pieces a third higher than written and trying to give them the same body of sound they have in the original key. Or playing them a third down and trying not to make them sound any grumpier than they do as written. The pianist working in the lieder field is continually up against this problem. But somehow, maybe because we're not used to hearing these particular songs in a range of keys, maybe because Brahms himself gave freer rein to his pianistic fantasy than usual under the assumption that the music would not be transposed, these particular songs emerge with reduced impact. It doesn't help that the piano is a little backwardly placed, as mono recordings tended to be in those days. McArthur is playing splendidly but the backdrop of Brahmsian richness is missing.

In Mahler, as in Brahms, the problem proves not to be the voice. The richness and gravity of Flagstad's timbre actually convince the ear that it is listening to a lower voice. But then there is the orchestra. Quite apart from the work involved in copying out new parts, just imagine, also in this instance, your favourite romantic symphonies being played a third higher than written. The bass would be lightened and many of the wind solos would be shifted up from a "convenient" to a "difficult" register of the instrument. It says much for the Vienna Philharmonic that, apart from occasional signs of strain from the horns, they make the music convincingly sound as if it had been written that way.

The apparent eccentricity of carting Boult out to Vienna to record Mahler is belied by the remarkable artistic collaboration that emerges. We know from Boult's Beethoven, Brahms and Elgar that he was something of a master of the free-flowing slow movement, swift and un-indulgent alongside many of his colleagues. His Brahms "Alto Rhapsody" with Janet Baker is a celebrated case, one of the fastest on record. His Mahler is a less well-known factor. The tempi here are among the slowest I've heard. This helps to restore the gravity that the upward transpositions might have taken away. What I didn't expect was to find Boult so idiomatic in his handling of the sadly trudging bass-lines and the halting movement, the sensation of holding back the music and then releasing it as the harmonies change, so typical of Mahler. This goes beyond mere "good accompanying". Nor can it be explained away by saying the Vienna Philharmonic had this style in their bloodstream, for they notoriously didn't provide it for just anyone. The same may be said of the transparent textures and restrained yet tangy wind solos. In any case, listen to the many passages where the orchestra takes over from the voice, and vice versa, and you have to be impressed by the total unity into which singer, conductor and orchestra have been forged, and that can only have come from the rostrum.

As to the interpretations, I found "Kindertotenlieder" memorable for its sense of numbed desolation, while "Lieder eines fahrenden Gesellen" is dewy-eyed as befits youth, yet also with a sense of unease, of impending tragedy. For the reasons I have given, this cannot really be a first choice. That must be made from among the many fine versions that use the original keys. But equally, committed Mahlerians can hardly leave it out of the reckoning. The recording is excellent for the date.