The Bolshoi is coming to London with a 'reconstructed' Pharaoh's Daughter. But how authentic can a ballet be?

July in Moscow is hot and dry, and the Bolshoi Theatre, a palace in ice-cream colours, rises beckoningly out of the dusty grey roadworks around it. Strawberry, pistachio and lemon paintwork reflect the escapist flavours offered within for more than 200 years.

I am here to throw myself back through the swirling sands of time to the stage debut of the greatest genius in classical ballet. The Bolshoi Ballet is coming to London, and among its offerings of timeless classics and Bolshoi specials is an old ballet with the intriguing and unfamiliar name, The Pharaoh's Daughter.

Created in 1862, its plot is splendidly Indiana Jones, telling of an English aristocrat-archaeologist who falls in love with a mummy, and incorporating pyramids, jungle lions, underwater ballets, and a cobra scene that leaves one faint with laughter.

The Pharaoh's Daughter was the career-making first success of Marius Petipa, the choreographer who over the next 40-odd years created most of the great Russian classics: Don Quixote, La Bayadere, The Sleeping Beauty and, in part, Swan Lake. The Tsar's mistress, Mathilde Kchessinska, danced the lead in all her Romanov Faberge diamonds, but the Soviet era declared it of "negligible artistic value" and abandoned it in 1928.

The Bolshoi Ballet's project to recover this fascinating curiosity started five years ago, paralleling the attempted reconstruction by Russia's other great company, the Kirov, of two Petipa masterpieces, The Sleeping Beauty and La Bayadere, as he produced them.

But is reconstruction of such old ballets possible? And what are its consequences for the art? Revolutionary, potentially.

Other arts have it easier. Music or drama has an author's text, and "authentic" movements have hugely benefited performance practice. Years of varnish can be cleaned off a painting. In ballet, however, the classics really have become timeless, their origins all but lost in the accretions of generations.

Writing down even the most grandiose ballets was not necessary for performance. For these collaborations between choreographers, designers, composers and dancers, watching and being told what to do was easier than issuing textbooks. Although dance notation systems have existed since 1700, training and custom largely ignored them, and it has long been accepted that leading artists make personal adjustments - a different spin combination here, a new solo there - which others may modify in turn. Effectively, Petipa has become a ground-bass for dancers' and directors' own riffs. Can people really dig below the sands of time to his original thoughts?

My first meeting is with Russia's senior dance historian, Elizabeth Souritz, who tells me that the issue is causing much bitterness. "There is a big clash," she says sadly, telling me how she fought for the Kirov's reconstructed Sleeping Beauty to win Russia's top ballet prize. "We managed, but many of my colleagues in St Petersburg are very, very angry about it."

Much of the upset is embedded in the Soviet attitude that established the Kirov and Bolshoi companies' reputations. Around 1900 the elderly Petipa ordered the notation of all his productions in St Petersburg; these precious texts were smuggled out of Russia by his chief balletmaster, Nikolai Sergeyev, in the 1917 Revolution. He brought them to England (the Royal Ballet's productions are based on them), whence they ended up at Harvard University in the US.

The Soviet choreographers, not considering Petipa's Tsarist ballets as in any way sacrosanct, felt quite at liberty to modernise his classics. The Swan Lake being performed at Covent Garden is, for example, a revision by Bolshoi former chief Yuri Grigorovich. And it is normal everywhere for companies to produce their own versions - not complete new choreographies, but simply tinkering a bit with Petipa.

Souritz regrets that Russia - the cradle of so many 19th-century classics - isn't conducting this crucial debate at a higher level. She fears that "spitefulness" is behind much of the current criticism of the Kirov's reconstructions from Sergeyev's notebooks. "When Sergeyev emigrated and took all these notebooks, it was an important reason for Soviet historians to denigrate him. He robbed us, they say. Not me. I think it was good thing."

But she also stresses that these old ballets cannot just be read from the notes straight on to the stage. For one thing, the notes are often patchy. The Pharaoh's Daughter has one of the fullest records, with more than 250 pages, but the Bolshoi's reconstructor Pierre Lacotte says they were mostly unusable, and much of what we will see is his own choreography in tribute to Petipa.

For another thing, 19th-century dancing was much less flamboyant and emotionally naked than today. Unlike singers, who approach Rossini and Verdi in different styles, dancers are schooled to bring the latest technical style to whatever they dance. The hip-splitting looseness required for a William Forsythe ballet, say, will be transferred pitilessly by ballerinas into modest Giselle - spectacular, yes, but jarring its aesthetic style.

Then there's the logical consequence of acknowledging an "authentic" classic. Must all other productions of that classic be thrown out, and with them the integrity of the serious artists who made them?

These are deep waters, as I find stepping into the dressing-room of the Bolshoi's ballerina, Svetlana Zakharova, 25, who leads The Pharaoh's Daughter and Swan Lake in London. She tackles my questions thoughtfully. She loves Lacotte's work on Pharaoh's Daughter, enjoying its intricate knitting of fast, precise footwork - but she doesn't consider it an authentic reflection of 1862 technique, and she thinks we should all be very glad. "Lacotte's production is beautiful because he took in consideration the ability of today's dancers and united contemporary technique in a rich, marvellous old production," she says.

But why not dance it in 1862 style? "Ballerinas were smaller then," she replies, "how would one say, quite soft, even cuddly - and if someone appeared like that in the old technique on the stage today no one would get it. We have seen so much, we are so spoiled, that people watching today don't give a damn what was going on in the 19th century. We have to amaze them. It's our duty."

With the massed ranks of Russian ballerinas past and present ranged behind this flawed argument, it's hard to see any rapid headway by the "authenticity" camp. And yet reconstructions, some very dubious, of Nijinsky, Fokine and early romantic ballets are ongoing around the world alongside the ambitious Kirov Petipa reconstructions. The Bolshoi Ballet's director Alexei Ratmansky would even like to recover early Soviet ballets.

Modern choreographers' practice, too, suggests that Petipa might be turning in his grave at what's happened to his works - virtually all ballets made in the past 50 years are videoed and notated with their creators.

I go to Paris to meet Lacotte, 72, the world specialist in reconstructing early 19th-century ballets. His recreation at the Paris Opera Ballet of the first romantic ballet, Taglioni's 1832 La Sylphide, is acclaimed, but there is criticism that Pharaoh's Daughter for the Bolshoi could have been far more faithful to its original had he stuck to Sergeyev's extensive notes.

Lacotte says his year's research on Pharaoh was unexpectedly frustrating; only three solos were noted well enough to be reproduced as "real" Petipa. "I'm sorry, if I think the note is not correct, I don't want to destroy the memory of Petipa by using a variation [solo] that I think is poor," he says. "I don't want to lie."

Two other solos he was shown by Petipa's now elderly dancers. That was all, in a four-act ballet. Hence his careful rechoreographing, "based on motifs from the ballet of the same name by Marius Petipa", as runs the cumbersome but respectful attribution.

I press him whether the ballet world is willing to establish the authentic works of its past geniuses - those that can honestly be recovered - and enjoy them for themselves, rather than go on adulterating until no original eventually remains. It doesn't help that "reconstructors" are often trading attacks on each others' methods.

Lacotte agrees. "We have so many possibilities with notation and technology now to try to fix the choreography and style of the past. We have all to get together and agree nicely that at least we must reconstitute originals to document them, even film them, before it is too late. I'm sure there are a lot of ballets very ripe for reconstitution that we don't know. We must be the testing ground for people to do it after us."

And if someone said to Lacotte that they could produce a more authentic Pharaoh's Daughter? He laughs. "Do it!"