There are very few 'historic' recordings. We may like or admire 'great' recordings but 'historic' recordings give us unequivocally great music recorded in extraordinary circumstances, and Mravinsky's first recording of Shostakovich's Fifth Symphony undoubtedly counts. Or rather his first recordings - a film of a 1938 or 1939 performance was distributed both at home and abroad and though the soundtrack was not released at the time, it appeared on CD on BMG Melodiya Japan, BVCX- 8020[1].

The score was first published in 1939; this is particularly fascinating since the performers were probably using the same parts they had in the premiere. In this light I will discuss briefly one moment where the recording is at odds with the published score (I used Boosey and Hawkes' HPS 628).

The fourteen 78rpm sides (matrices 06820-33) were released on CD in 1998 by BMG Melodiya Japan (B0CC3; deleted), sourced from the Rodgers and Hammerstein Archive of Recorded Sound. Doremi do not reveal their source and claim that one side (the first 3'46" of the largo) comes from matrix 08755. As far as I can ascertain, no Shostakovich Fifth has used this number and the scant annotation (exemplified by the incomplete recording date for the Shostakovich), though mentioning problems in transferring the Pathétique, remains silent on this. BMG's restoration is minimal and changing levels of surface noise signal each side change. Doremi removed much of the hiss but took with it the music's upper frequencies. The sound is mellower but details such as the flute's doubling of the strings are inaudible. The timings vary slightly but not enough to alter the pitch.

So what of the interpretation? Obviously this is not exactly what the audience heard on 21 November 1937 but it is a fascinating insight into Mravinsky's general approach to the work, which we must presume was approved by the composer. One striking facet of the interpretation is the strong Tchaikovsky feel of it, particularly the bassoon and the very recognisable Russian vibrato-filled horns. Mravinsky seems to be concerned to place the work firmly in the tradition, and perhaps, even at that stage, the canon. If so, this reintegration may have been a factor in the composer's rehabilitation.

Mravinsky starts a little under the score's quaver = 76 but his tempi are endlessly flexible and he gently eases into the slow descending violin line. The new conductor- orchestra relationship was obviously working well, and Mravinsky appears happy to let soloists mould their own lines. The woodwind take particular advantage of this, and in the coda the piccolo's dotted notes (bar 306) have a more than usually jazzy, hopping quality. However the violins' high E in bar 17 (the start of the dotted-rhythm descent) is uncertain and wobbly. I was also surprised by the lack of ferocity in the march. It's not exactly genial but neither is it the grinding power-display that some have made it, and though initially it continues at the same tempo as the previous section, once it gets going, the slight quickening to something nearer the score's crotchet = 92 further undermines its implacability. At figure 38 Mravinsky moves the rushing orchestra's molto rit forward a bar so that he doesn't need to slam on the brakes so hard but the following climactic chord doesn't really qualify as con tutta forza, being less forceful than the previous one. Finally, after a very prominent glissando leading to figure 46 there's another rit for the actual end of the movement.

Just as the first movement's march is not as ferocious as it has since been made, so subsequent interpreters have turned the second movement into a Mahlerian Totentanz. But here death does not scrape his fiddle so violently. Still, it is hardly benign despite its decidedly steady tempo (it comes in at 5'38"). The coda drives straight through to the end without a rit.

Whatever Doremi's source for the Largo's opening, the sound is noticeably poorer. The score is peppered with espressivo markings and these are played for all they are worth, especially the first big one (bar 19-20) where the hairpin from piano goes a lot further than mezzo-forte. Perhaps it is this intensity that leads to the three-repeated- notes motif sometimes being blurred into one. Throughout the disc the side changes are handled well except for one. At figure 39 the cellos sink to rest before ethereal violins climb once again into the light, but Doremi have overlaid the violins' entry over the cellos' morendo, effectively cutting a bar, creating a rhythmic mess and destroying the last moment of repose. This is a transfer problem (it does not happen on the BMG disc or on the film soundtrack) but in the coda there is a change from the score when the harp is excluded from its famous duet with the celesta. Though unlikely, this could be a simple mistake, but there are other possibilities. Since the harp recorded particularly well on 78s it may have been deliberately dropped to avoid covering the celesta - it does play on the film where it would have been less of a problem. Another possibility is that it was added by Shostakovich after the first recording but before the film. More investigation is in order.

Since the revelation of the incorrect tempo marking for the finale's coda, it sometimes seems that this has become a lodestar for interpretation at the expense of the rest of the score. Overall timings vary from 8'35" (Koussevitzky, 1948) to 12'50" (Rostropovich, 2003), and there has been a slight tendency for them to become slower. For the interested, both Mravinsky's first attempts are in the middle; the studio version 10'55", and the film 9'47"[2].

But the overall timing is only part of the story and a lot also happens before we get to the coda's minefield. Mravinsky and Toscanini both gained (often undeserved) reputations for 'objectivity' and 'fidelity to the score'. However, in this case what you hear is remarkably close to what you read; pretty much exactly crotchet = 88, an accelerando in bar 8 to crotchet = 104 by bar 12. One downside is the timpani which, due to the engineers' fear of overloading the recording, are distant if not as disconcertingly 'ploppy' as Stokowski a year later. There is also some surprisingly scrappy string playing, especially around figure 101 (a recapitulation of the main theme), though a little later they have a marvellous glassy quality as they gently oscillate beneath the woodwind. One other oddity is figure 116-118 where, in the violins' repeated crotchets, majoring on an octave leap on a C, there is some uneven counting before the last crotchet begins to sound a little like two quavers.

So, to the big question - just how does Mravinsky take the coda? The answer is that from figure 127 (the beginning of the violins' stabbing A's) he employs a quite steady quaver = 184 (the score asks for quaver = 188), with a very small rit at the very end. Certainly not the grinding end á la Rostropovich, leaving us with a question: if the original tempo marking was a mistake why was it that Mravinsky did not 'correct' it in his first recordings? Perhaps the 'mistake' was more of a 'reconsideration', or perhaps they felt that to perform it that way in 1937 would have been suicidal! Whatever, it is an invaluable, if potentially ambiguous, piece of evidence.

No recording, however 'historic', and this certainly qualifies, is a holy grail of interpretation. It does not render all others redundant, but it is extraordinarily valuable and should inform, though not dictate others' approaches. Unfortunately Doremi's highly interventionist restoration makes me wonder just how close we are getting to the original. The famous weight and brightness of the orchestra's strings is dulled and details of ensemble lost. I realise that for many people high levels of surface noise are a black mark but Doremi have simply gone too far. Though the Japanese Melodiya disc may prove difficult to find, I recommend that for those who wouldn't mind the noise. Alternatively, wait for a transfer that steers a middle course.

The other items on this two-disc set have their own points of interest, including rare outings for Mravinsky with orchestras other than the Leningrad Philharmonic, and Felix Weingartner's orchestration of Weber's Invitation to the Dance, a pleasant alternative to the usual Berlioz.

[1] The film was shown on 26 November 1939 at the Academy Cinema in London. (Annual Report of the Society for Cultural Relations, 1940).

[2] A reel of 35mm is eleven minutes long so there would have been no technical constraint on the finale's length, though the first and third movements (respectively 14'48" and 15'48") would need a second reel. The film probably ran over five reels:

I Opening of first movement.
II End of first movement and Allegretto
III Opening of Largo
IV End of Largo
V Finale

John Riley
Top