"Recently, the Dutch master Gerard Welling pointed out to me that a very famous game, a classic case of audacity, has in fact never been played. [...]

"With such a game, who would care that White's stroke of fantasy is far from perfect - for one thing, 25...Rd8 gives White an - immediately decisive - advantage for the first time in the whole game; 25...Bc8! forces the repetition Bb6+, Ba7+ etc.

"Also, instead of 20...f6 which allows this draw, 20...Rd5 is winning, as was demonstrated by Shipley in a match game against Morgan in 1891. The Steinitz Gambit was subject to extensive analysis in those days, which had already made me wonder whether Steel - NN (the nameless victim adds to the suspicion) couldn't have been a brilliant analytical discovery instead of a real game. And indeed, as Gerard Welling demonstrates convincingly, the game has never been played.

"In the first issue of Steinitz' 'The International Chess Magazine' (January 1885), under the heading 'Chess in Calcutta', there is a game between 'two well known strong amateurs of that city'. The 'elegant and instructive game' Steel - Ross which then follows is, except for a transposition of moves 22 and 23, the same as Steel - NN. At that point, it deviates with: 24.Bc5 Rd8 25.Ra7+ 'and draws by perpetual check'. "The existence of this game Steel - Ross alone, makes it unlikely that Steel could have played almost exactly the same game one year later, again in Calcutta, where the chess community must have been small. But when he browsed further, Gerard Welling found more. In the January 1888 issue of Steinitz' magazine there was, under the caption 'A variation of the Steinitz Gambit', a reader's letter by Steel himself, dated: Calcutta, November 1st, 1887.

" 'I have been amusing myself,' Steel writes, 'with analyzing further a remarkable variation of the Steinitz Gambit, of which you published an illustration in the first number of Vol. I of your Magazine, which had occurred in a game at this opening between Mr. Ross and myself. That game was drawn, but I am inclined to think I had a forced win. The position is one of extraordinary interest and difficulty, and I think the analysis will please your readers, many of whom enjoy fireworks.'

"Steel's forced win is based on two variations. The first is (see above, in Steel - NN): 23...Qxb2 24.Bc5, when Black will have to give his Queen for the Bb5, after which White has, according to Steel, a won ending with Bishops of opposite colors. And against Ross' 23...Qxh1, Steel had an improvement over his 24.Bc5 against Ross, which he notes almost offhandedly: 'If 23...Qxh1; 24.Ba7+ Kc7 25.Bc5 Rd8 26.Ka7, and wins.' "That remark is the crown witness: a move given by Steel as analysis in 1887, cannot have been played by him in a game in 1886.

"The Honourable Robert Steel (1839-1903) was one of 15 members of the 'Council of India', the body that governed the colony. He must have done much to further British chess, but to him, his immortal game was only a remark in his own analysis.