Edward Winter presents: Unsolved Chess Mysteries (2)

3/12/2007 – We bring you a further selection of intriguing chess mysteries from Chess Notes, including the origins of the Marshall Gambit, a game ascribed to both Steinitz and Pillsbury and the bizarre affair of an alleged blunder by Capablanca in Chess Fundamentals. Once again our readers are invited to join the hunt for clues.

ChessBase 14 Download ChessBase 14 Download

Everyone uses ChessBase, from the World Champion to the amateur next door. Start your personal success story with ChessBase 14 and enjoy your chess even more!

Along with the ChessBase 14 program you can access the Live Database of 8 million games, and receive three months of free ChesssBase Account Premium membership and all of our online apps! Have a look today!


Unsolved Chess Mysteries (2)

By Edward Winter

When did Marshall first play the Marshall Gambit? (C.N. 2332)

The oldest known specimen of the Marshall Gambit in the Ruy López is a game won by Carl Walbrodt (White) against four Cubans (Conill, Ostolaza, López and Herrera) in Havana, 1893. It was included in C.N. 1996 (see pages 151-152 of our book Kings, Commoners and Knaves) and is nowadays widely given in databases. But did Frank Marshall himself ever play the 8…d5 line before his famous meeting with Capablanca at New York, 1918? An oft-published game is Walter Frere v Marshall, ‘New York, 1917’ (see, for instance, pages 238-239 of 1000 Best Short Games of Chess by Irving Chernev), but we have yet to find it in a magazine or column of the time.

As mentioned in C.N. 2332, Marshall published the Frere game on pages 110-111 of his rarely-seen book Comparative Chess (Philadelphia, 1932):

Click to enlarge

It will be noted that Marshall offered no date or venue, merely announcing, ‘The following game was played some years ago, to test out my new defence in the Ruy Lopez.’

A further curiosity in Comparative Chess is that on page 104 it was 7…O-O, rather than 8…d5, that Marshall emphasized. Of 7…O-O he wrote (incorrectly), ‘This move of mine, I claim to be original’.

Click to enlarge

A game played by Pillsbury? (C.N. 3288)

This position comes from page 41 of Les échecs dans le monde by Victor Kahn and Georges Renaud (Monaco, 1952):

The co-authors stated that it was from a game between H.N. Pillsbury and E.F. Wendell in a simultaneous exhibition on 40 boards in Chicago, 1901, the finish being 12 Nxg5 hxg5 13 Qh5 Rxh5 14 Ng8+ Ke8 15 Bxf7 mate. However, the following score, with an identical finish, is on pages 319-320 of volume one of the second edition of Schachmeister Steinitz by L. Bachmann (Ansbach, 1925):

Wilhelm Steinitz – N.N.
London, 1873
(Remove White’s rook at a1.)

1 e4 e5 2 f4 Nc6 3 Nf3 Nf6 4 fxe5 Nxe4 5 d3 Nc5 6 d4 Na6 7 Bc4 Qe7 8 Nc3 h6 9 O-O g5 10 Nd5 Qd8 11 Nf6+ Ke7 12 Nxg5 hxg5 13 Qh5 Rxh5 14 Ng8+ Ke8 15 Bxf7 mate.

That game, with Steinitz as White, was published on page 230 of the October 1874 City of London Chess Magazine, but on what grounds has the same game (although not at rook odds) been attributed to Pillsbury?

Lasker and a composition (C.N. 2705)

In C.N. 145 Michael McDowell (England) mentioned a well-known photograph of Emanuel Lasker studying a composition. It appeared, for instance, on the front cover of the Dover publication Lasker’s Manual of Chess:

A diagram of the position (composer unknown) is given below:

At the time we believed that the solution was 1 Rg8 Rxg8 2 Rh8 Rxh8 3 g7 Rg8 (or 3...Rf8) 4 h7 and wins, and there the matter lay for 20 years. But then C.N. 2705 related that we had carried out a computer check in which, almost immediately, the Fritz program came up with a humdrum mate in five (i.e. one move faster): 1 Rf7 Rg8 2 Rhg7 (Or  2 g7.) 2...Rh8 3 h7 any 4 Rg8+ Rxg8 5 hxg8(Q) mate.

Information is still proving elusive regarding the first publication of the composition, the identity of the composer and the solution which he intended.

A pawn ending mystery: did Capablanca blunder? (C.N. 2011)

In Chess Fundamentals Capablanca published the following as ‘Example 8’:

Page 64 of Comprehensive Chess Endings, volume 4, by Y. Averbakh and I. Maizelis (Oxford, 1987) comments on the position as follows:

‘Here the only move not to win is 1 g5? in view of 1...g6. Curiously, in the 1st edition of his Chess Fundamentals, Capablanca asserted (he later corrected this) that 1 f5 also did not win in view of 1...g6 (he left the analysis of the variation to the reader). Capablanca gave the solution 1 Ke4 Ke6 2 f5+ Kf6 3 Kf4 etc. But it is precisely by 1 f5! that White wins more quickly.’

Critical analysis of this position appeared on page 268 of the November 1949 Magyar Sakkvillág, where Dr Jenő Bán claimed to correct Capablanca by demonstrating a win by 1 f5 g6 2 fxg6 Ke6 3 g5 (Not 3 Ke4 Kf6 4 g7 Kxg7 5 Kf5 Kf7, with a draw) 3...Ke7 4 Ke5 Kf8 5 Kf6 Kg8 6 g7 Kh7 7 g8(Q)+ Kxg8 8 Kg6 and wins. These moves were repeated by Fred Reinfeld on page 279 of The Joys of Chess (New York, 1961), in a chapter entitled ‘Boners of the Masters’. The ending was also discussed in Chess Life & Review in June 1971 (page 306), December 1971 (page 704) and November 1972 (page 549). The last of these was a contribution from Paul Keres:

‘Furthermore, I got interested in the position from Capablanca’s Chess Fundamentals, given in Dec./71, page 704. I have not got the English edition of the book, but in the German translation (1927 and 1934) the statement is that 1 P-B5 does not win because of 1...P-N3. The Russian translation of the book, on the contrary, states that 1 P-B5 also wins, Black’s best counterchance being 1...P-Kt3. My impression is that Capablanca originally made the mistake, thinking the position was a draw, but later, noticing his error, did make a correction. Of course that is only my impression – I have no facts to prove it.’

So did Capablanca commit a ‘boner’ by claiming that 1 f5 would not win and, if so, did he correct it? All editions of Chess Fundamentals published in the United Kingdom (including the 1921 original) seem to state: ‘In the above position White can win by 1 P-B5 [our italics]. Black’s best answer would be P-Kt3.’ The Cuban adds that White ‘cannot win by 1 P-Kt5, because P-Kt3 draws’ and later writes that ‘White can win, however, by playing 1 K-K4’. This is followed by a number of variations.

UK edition, 1921, page 12

G. Bell and Sons, Ltd. were the UK publishers, but the US edition was published by Harcourt, Brace and Company. The original 1921 US edition reads:

‘In the above position White can win by 1 P-B5. Black’s best answer would be P-Kt3 draws. (The student should work this out.) He cannot win by 1 P-Kt5, because P-Kt3 draws.’

US edition, 1921, page 14

The first two sentences obviously contradict each other, and in the second sentence the word ‘draws’, which is absent from the UK edition, is syntactically incorrect. Did the US typesetter accidentally add the word, perhaps led astray by the similarity to the way the final sentence ends?

To eradicate the contradiction, one of the first two sentences in the US version needed to be changed. Remarkably, the solution adopted – in the mid-1930s US edition, which was published with a new Preface by Capablanca dated 1 September 1934 – was to put: ‘In the above position White can’t win by 1 P-B5. Black’s best answer would be P-Kt3 draws.’

US edition, mid-1930s, page 14

To summarize, although it was suggested by Averbakh and Maizelis that Capablanca a) wrongly wrote that White cannot win with 1 f5 and b) later corrected the text to ‘can win’, in reality the change (made in the mid-1930s US edition) went in the opposite direction: ‘can’ was altered to ‘can’t’. Why and by whom?

Further complications arise concerning the German edition (published in 1927 by Walter de Gruyter & Co. under the title Grundzüge der Schachstrategie), from which Dr Bán was working.

For some reason it stated:

‘In der Stellung (Beispiel 8) kann Weiß nicht durch 1 f5 gewinnen; die richtige Erwiderung wäre 1...g6 (der Leser möge dies selbst durchdenken). Weiß kann auch nicht gewinnen mit 1 g5, weil Schwarz g6 antwortet.

German edition, 1927, page 8

The words nicht and auch have no equivalent in the original English-language editions. More confusion occurred when the 1979 reprint of the German translation amended the position to justify the nicht: White’s king was put on e3 and Black’s on e7.

Do readers know of any other pre-1934 editions, in any language, which state, ‘White can’t win by 1 f5’?

Submit information or suggestions on chess mysteries

Edward Winter is the editor of Chess Notes, which was founded in January 1982 as "a forum for aficionados to discuss all matters relating to the Royal Pastime". Since then around 5,000 items have been published, and the series has resulted in four books by Winter: Chess Explorations (1996), Kings, Commoners and Knaves (1999), A Chess Omnibus (2003) and Chess Facts and Fables (2006). He is also the author of a monograph on Capablanca (1989).

Chess Notes is well known for its historical research, and anyone browsing in its archives will find a wealth of unknown games, accounts of historical mysteries, quotes and quips, and other material of every kind imaginable. Correspondents from around the world contribute items, and they include not only "ordinary readers" but also some eminent historians – and, indeed, some eminent masters. Chess Notes is located at the Chess History Center.

Discussion and Feedback Join the public discussion or submit your feedback to the editors


Rules for reader comments


Not registered yet? Register