Tax deduction for fakes?

The Kansas City Star had a piece on an interesting income tax dispute involving Santa Fe dealer Gerald Peters and the so-called "Canyon Suite," a group of 28 paintings that at one point had been attributed to Georgia O’Keeffe but later turned out to be fakes. In the early 90s Peters had sold two dozen of the paintings to Kansas City banker Crosby Kemper for $5 million, but ended up refunding the money. But he had also donated four paintings to the Kemper Museum of Contemporary Art, taking a $1.1 million tax donation in the process. The IRS disallowed the deduction, saying that the works, having been shown to be inauthentic, were worthless. (The IRS has also made the alternative argument that “there was no true donation because the donation was an integral part of the sale of the 24 other paintings to Kemper.”) Peters has brought suit in U.S. District Court in New Mexico, seeking $692,000 in taxes and penalties he paid. It's an interesting question. The first reaction is: why hasn't the case been dismissed? If the works were fake, they were worthless, and why should it matter what anyone thought at the time of the donation (the news story describes Peters's position to be that "his deductions were fair and reasonable at the time of the donations, which was almost three years before they were publicly tainted and judged worthless")? It shouldn't matter that I reasonably believed I was giving away a genuine O'Keeffe if the fact is I didn't and the museum didn't receive one. On further reflection, though, maybe there is an argument that what the museum received was a kind of lottery ticket -- that at the time of the donation it wasn't clear whether or not they were genuine O'Keeffe's, but that four works with a chance of being O'Keeffe's were still worth something. You might not pay full value for such works, but you'd still pay something greater than zero. If I donate a lottery ticket to a charitable institution, with say a one in 10 chance of winning $1,000, and we later find out that it was a losing ticket, isn't it nevertheless the case that I gave something of value to the institution? In any case, I don't think that's the argument Peters is making, since he's pretty clearly seeking to deduct their full value (he's claiming a $1.1 million valuation for the four works when he sold the 24 similar works to Kemper for $5 million -- although I suppose it's possible to claim the price to Kemper reflected the same discount for the possibility they would turn out to be inauthentic).