MUSIC / Mix well, brew for 50 years: What makes a great recording? Stephen Johnson weighs up the ingredients

People soon tell you what doesn't do it. The digital age studio session, with its countless retakes and terror of intrusive noise, is - they will say - a hostile environment for authentic music-making. On top of that there's the producer, who, in a ruthless quest for perfection (very much a dirty word in today's specialist magazines), manages to purge the playing or singing of whatever life it may have originally possessed.

Two reactions follow. First comes the surge in popularity of 'classic' recordings - old masters revitalised, paradoxically, by the very technology that devotees often claim to despise. 'Classic' begs the question, but leave it open for now. The second is evident in the rise of a new buzz- word, 'live'. More and more performers are insisting that companies record them in concert, rather than pack them off to the studio, where there is no audience to engage with and an invisible producer threatens at every moment to cut them off in mid- flow.

If the implication is that a great recording is simply a great performance overheard by a discreetly placed microphone, experience suggests otherwise. One of the oddest aspects of concert recordings is that the special quality one felt listening in the hall sometimes refuses to transmit itself to disc. It isn't just the absence of concert atmosphere - the feeling that all around, others are responding similarly. The very features of the performance that made it alive at the time - the risk-taking, the sudden inspirations - can sound eccentric, affected, or just plain irritating when one puts on the disc. Even stranger, the reverse can happen: the session that dragged painfully, after which everyone left cursing everything from the composer's ashes to the producer's trigger finger, somehow gains that elusive specialness when it's all stitched together.

There are times when the process seems as inscrutable as alchemy. But one thing can be safely said: a recording is not a performance - even when there was a single performance behind it. For the singer or player, 'all is always now': the response of a definite number of people at a specific time in a specific place is everything. For the producer, and for the record-buyer, what matters is repeatability - adaptability to different moods and conditions. As Hans Keller was fond of pointing out, a repeatable performance is a contradiction in terms.

There is another important difference. A concert performance is a collective experience: composer, performer and audience make music together. The recording on the other hand is aimed, increasingly, at that utterly modern phenomenon, the solitary listener - a listener who may, with the dubious aid of headphones, have chosen to shut out everything around him or her (is that what really irritates us about public Walkman wearers?). One begins to see why this kind of hearer should have appealed to that obsessive introvert, Glenn Gould - and why he should have come to prefer him or her to the tribal group in the concert hall.

If any artist can be said to have validated that form of musical experience - to have shown that repeated solitary listening can be musical, it is Gould. Recent reissues, especially his 1955 mono version of Bach's Goldberg Variations, have been given a royal welcome - without question deservedly. But it's fascinating to hear how a little tag like that '1955 mono' is affixed to classic recordings, as though it validated the contents. Show the true enthusiast Bruno Walter's stereo version of Mahler's Ninth Symphony, and you can predict the response: 'Yes, but have you heard the 1938 mono version with the Vienna Philharmonic?'

Granted, that is a revealing recording, despite primitive sound, sometimes scrappy playing, and nagging doubts as to what the Viennese musicians thought of the long, difficult and unfamiliar symphony by a Jewish composer in the year of the Anschluss. But it's hard to avoid the suspicion that records improve with age - that they mature. Otto Klemperer's Beethoven symphonies were admired in their time, but gradually, and especially since Klemperer's death, they have been transformed into a sacred object. In place of the fleeting, transitory performance, dependent on fallible memory, there is something fixed, enduring - something one can go back to and explore, like the score, but unlike the score, intelligible to just about anyone.

Even the flaws can become part of this sacredness. Admirers tell the story of how, during a recording of Brahms's First Piano Concerto, Adrian Boult asked the pianist, Artur Schnabel, if he'd like to take a certain tricky passage again - perhaps he could do it better? 'It might be better,' Schnabel replied, 'but it wouldn't be as good.' For some listeners now that moment of inspired clumsiness is a special revelation - a point of contact with the artist himself. When Maria Callas's recordings began to reappear on compact disc there were near-ecstatic remarks about how much better she sounded - audible breathing, and all those little creaks and rustlings producers today go through agonies to remove.

Yet there is also the factor of distance - a distance which, one suspects, adds significantly to the enchantment. The fact that one cannot see the performer is one element - in the modern recording as much as in the remastered 78 - but for many the effort to hear the 'classic interpretation', through surface noise like a sizzling saucepan, with sound so lacking in bass or badly balanced that important details are virtually inaudible, is part of the attraction. The vaguer the outline, the less distinct the colours, the greater the challenge to the imagination to round them out. Once more, as in the concert hall, the listener becomes an essential participant in the musical experience.

But there are still those recordings which, irrespective of all questions of sound quality, age or historical background, remain uniquely alive however often they are replayed. In some of the Busch Quartet's 1930s recordings of the Beethoven String Quartets, insight, poetry and feeling combine to give an almost numinous quality. I am not the first to wonder whether we will ever hear Beethoven quartet playing of this quality again, in the concert hall or on record. But should you listen to the Op 130 quartet in Sony's excellent CD transfer, remember that what you are hearing - a continuous Busch Op 130 - never existed. In those pre-tape days, the Busch would have had to record that 45-minute work in chunks of no more than three minutes at a time - surely no modern producer could dream up conditions more hostile to inspired music- making?

Could it be that in some cases the very adversity of those conditions was a valuable stimulus - a challenge to achieve the impossible? The more one hears the stories behind acknowledged great recordings - the horrors that attended Britten's recording of his own War Requiem, for instance - the more plausible that seems. But in contrast there are occasions when, mysteriously, everything just goes right. The weather is perfect, the environment friendly, the performers and the production team get on, and everyone seems able to concentrate effortlessly. I remember being at the chamber group Domus's sessions for the two Faure Piano Quartets and agreeing at the end that something magical had happened. The disc went on to win the Gramophone Chamber Music award the following year - confirmation that the magic had not died on the producer's operating table. Mention of a record like that in the company of such names as Gould, Walter, Schnabel, Klemperer, Callas and the Busch Quartet will seem absurd to many. But is it possible that a 21st-century collector, confronted with perfect, virtual-reality Faure Quartets in an auditorium you can choose yourself, might murmur affectionately, 'Yes, but have you heard the 1985 stereo Domus version?' I'd like to think so.