The obsession has taken a very precise, surveyor-like form: De Robertis has become an acknowledged expert on Van Gogh facts. Tied down by his day job in Milan, his research was at first slow, but then he won 100 million Lire (pounds 35,000) on a TV game show and spent it on a sabbatical. He has sifted painstakingly through catalogues and monographs, as well as the vast collection of frantic, detailed letters that Vincent wrote to his siblings between 1872 and 1890, in French, German or English as the spirit took him.
And from his sifting, he has deduced a thing or two. Namely, that there are perhaps as many as 300 works in the museums and private collections of the world labelled Van Gogh which aren't Van Goghs. And that the Dutch artist finished 450 works which have not been traced. Some of these were certainly destroyed, De Robertis says. Many more may be gathering dust in attics.
The research, published in the current edition of the art magazine Quadri e sculture, will force Van Gogh owners to re-examine their Vincents. It may also, De Robertis hopes, prompt buyers to think twice about taking dealers' appraisals at face value.
De Robertis has no delusions of artistic grandeur, and, though Van Gogh is his passion, he does not claim to know his works well enough to attempt stylistic attributions. "I don't deal in style. I would never be so rash as to attempt an attribution on the basis of brushstrokes," said De Robertis. "For me, the only way of establishing whether or not a work is by who it claims to be by is to study its history and documentation."
In 1993 he used his analytical method to prove that Japan's Yasuda corporation had thrown away $24.7 million on a fake Van Gogh Sunflowers at an auction at Christie's, London. A label from an exhibition staged in Van Gogh's lifetime, now on the back of that painting, was originally on another work, an Olive Orchard owned by the artist's brother, De Robertis says. Since then, he has been involved in a much grander project, matching up Van Gogh's catalogued works with those mentioned by the artist - in lengthy descriptions or fleeting remarks in his letters.
De Robertis' argument is as follows. In his letters, Van Gogh refers to 1,730 works. In Hulsker's catalogue of Van Gogh's output, considered the most complete and accurate by scholars, 2,125 are cited. Of the works - oils and sketches - mentioned by the artist, 450 feature nowhere in Hulsker's catalogue. The catalogue, on the other hand, includes 507 works which don't get a look-in in the letters.
Van Gogh was famously careless about his output: he lost works, gave them away to friends, creditors and prostitutes, sent them off to his beloved brother Theo and then ordered him to burn the lot (an order that Theo and his canny wife Jo Bonger never failed to disobey). De Robertis is prepared to accept that of the 450 untraced works mentioned in the letters, as many as 278 - 58 oils and 220 sketches on paper - may have been destroyed.
A little basic maths, says De Robertis, and you come up with 172 works - 100 oils and 72 sketches - mentioned by Van Gogh in the letters, which may be still in existence, but are nowhere to be found. Add to these some 150 early works done before the letter-writing period started, and you are looking at 322 Van Goghs which have yet to hit the art world.
On the other hand, if De Robertis's calculations are to be believed, that same art world has happily taken hundreds of fakes to its heart. "Of the 507 not mentioned in the letters, 203 are in the Van Gogh museum; they came straight from Vincent's brother's collection and are therefore unimpeachable. Of the remaining 304, well ... it's anybody's guess really. But it's fair to suppose that many of them are fakes."
The bulk of these were done shortly after Van Gogh's death by Claude- Emile Schuffeneker, a minor artist and a far-sighted collector who, De Robertis claims, saw the potential value of Vincent's work long before he was known to the general public.
"Vincent's sister-in-law sent many of the paintings in her possession to Schuffenecker, for restoration shortly after her husband died. She may have been aware that he was copying them too," De Robertis maintains.
Master faker Daniel Donde, whose academy of copyists in Cremona can run you up any old or modern masters - complete, naturally, with a certificate of inauthenticity - goes even further: "She was a very mercenary woman," he maintains. "There can be no doubt that she was in cahoots with Shuffenecker."
Donde is relieved that the issue of fake Van Goghs has been dragged into the limelight again. "I've been telling people for decades that there are hundreds of Vincent fakes hanging in galleries and collections around the world," and it's not surprising: "Van Gogh just isn't that difficult to copy. Any of my copyists could do you a perfect Vincent."
He agrees, however, that the bulk of Van Gogh fakes are historical. They are also, he says, very bad: "You can see from miles away that they're false, that they are not by the same hand. I find it incredible. Everyone knows that they're fakes, but no one wants to rock the boat."
De Robertis has promised a full breakdown of fake Vincents "in the next few months". For the time being, he has prepared his own favourite hit- list: "There are the Yasuda Sunflowers, of course, and the Jardin D'Aubigny in Basle. I'm pretty sure that the Self-Portrait with bandage in the Courtauld is a fake, as is the Alsacienne in the Metropolitan Museum in New York."
Thomas Hoving, a former director of the Met and keen fake-hunter, would beg to differ: "When all the Vincents in the Met came under suspicion when I was director, we examined all of them very rigorously and all passed muster," he says. "Even the one with the self-portrait on one side and the potato-eater lady on the other that every phoney fake-buster mumbles is no good."
Moreover, Hoving warns against putting too much faith in Van Gogh's letters. "Anyone who has dealt with the nettlesome issue of Van Gogh fakes knows that the letters to Theo are not particularly reliable machines for connoisseurship. The letters can be manipulated either way, just like statistics."
For Hoving, using documentation for attributions is dodgy: "It comes down to, as always, the eye."
Not so, says Donde. It's often a question of saving face which makes people turn a blind eye to tell-tale discrepancies. "If you're a museum director and someone called Rockefeller donates a painting to your gallery, you don't say 'are you sure this is genuine?', you just accept it gratefully. You don't want to call the donor's eye into question, or you may never get another picture; and you don't want to look ungracious yourself."
If you would rather buy, you can always turn to major dealers: "If a work of art comes with a certificate from Sotheby's or Christie's, whether or not it's authentic is immaterial. You've got that certificate, and that makes it as authentic as they come."
De Robertis is convinced that his research will shake up this ready acceptance of authoritative-sounding names. With his definitive list of fakes almost ready for publication, he senses "a certain amount of panic in the art-collecting world at the moment. The end result is that collectors are going to be much more cautious about what they spend their money on from now on."
"There are fewer and fewer Van Goghs on the market these days," says De Robertis. Despite this, those untraced works, as and if they appear, may have an ever harder time proving their credentials. !