Kevitz v Capablanca, New York, 1931

Edward Winter


Capablanca played 36...f3

C.N. 2805 attempted to shed light on one of Capablanca’s most original and fortuitous victories by quoting the full text of a letter (undated) in our collection which was written to Bruce Hayden by the loser, Alexander Kevitz (1902-81). First, for ease of reference, the complete game-score:

Alexander Kevitz – José Raúl Capablanca
New York, 19 April 1931
Réti’s Opening

1 Nf3 d5 2 c4 c6 3 b3 Bf5 4 g3 Nf6 5 Bg2 e6 6 O-O Nbd7 7 Bb2 Bd6 8 d3 O-O 9 Nbd2 Qe7 10 Re1 e5 11 cxd5 cxd5 12 e4 dxe4 13 dxe4 Be6 14 Qe2 Rac8 15 Nf1 Ba3 16 Nxe5 Bxb2 17 Qxb2 Nxe5 18 Qxe5 Qa3 19 Ne3 Ng4 20 Nxg4 Bxg4 21 h3 Be6 22 Re2 Rfd8


23 Qb2 Qc5 24 Rd2 Rxd2 25 Qxd2 b6 26 Rd1 g6 27 Kh2 a5 28 Qe2 b5 29 f4 a4 30 bxa4 bxa4 31 Rd2 a3 32 g4 g5 33 Qf2 Qxf2 34 Rxf2 gxf4 35 Rf3 Ra8 36 Rf2


36...f3 37 Bf1 Rb8 38 Rxf3 Rb2+ 39 Kg3 Rxa2 40 Rc3 Ra1 41 White resigns.

The above is the score as it appears in most sources, including the tournament book by L. Eceizabarrena Gaba and R.A. Cela, whereas the tournament book by J. Spence had an inversion of White’s 27th and 28th moves (which is not in keeping with Kevitz’s remark below about ‘28 Q-K2??’). Both volumes specified the time used for the game: 2 hours 38 minutes by Kevitz and 1 hour 41 minutes by Capablanca.

kevitz capablanca

Below is the full text of the letter from Kevitz to Hayden:

‘351 W. 24 St.
New York 10011

Dear Mr Hayden,

Al Horowitz mentioned your interest in the game and your letter was expected and welcomed.

My game with Capa represents one of very many lost opportunities but after all these years the pain has been numbed and I think I can be objective to annotate it in its proper perspective.

Though the opening and defense had previously been played at London, Réti startled the chess world in the 1924 N.Y. Tournament with the ideas presented. All the grandmasters present were favourably impressed. As a budding chess player I was present and as a student of opening theory I was greatly influenced. For instance though the connection may be obscure the opening defense which has frequently been renamed in my honour viz 1 P-K4 N-QB3 is really an outgrowth of the same Réti theory.

Capa lost a beautiful game to Réti and his theory but in a later round Lasker managed to win with his version of the Lasker defense. His win was erroneously (I think) attributed to correct defense and Réti’s midgame errors were overlooked. Between 1924 and 1931 Capa frequently played W. in the Réti and always played Lasker’s defense as B.

I had studied Réti’s management of his system and with all due respect I thought I had improved over his handling of his own system. And the Capa game was my opportunity to demonstrate my improvement. So to the game.

The first nine moves arrive at the same Réti-Lasker position. By my present day knowledge the order of moves is wrong but that is tangential.

Though some books now recommend 10 R-K1 I know and knew it to be wrong. I played it because I was sure Capa would answer 10…P-K4. Correct is 10 PxP BPxP as Réti played it and then not 11 R-B1 but 11 P-K4, which I have since played with devastating results.

At move 14, …P-QR4 as Lasker might have played is met by 15 Q-N5.

Capa said his 15…B-QR6 was an oversight. In that case a strange one for him. But he also later said he couldn’t see how to meet the W. threats which are only superficially obscure. W. threatens on say 15…R-B2 16 N-R4 followed by implantation of a N. at KB5. As variant W. also threatens N-N5 B-N5 P-B3 B-R4 P-R4 etc. Therefore 15 P-KR3 or P-N3 is unplayable.

The play through the 22nd move seems to be errorless. Now the opinion of the consensus that says W. should win is based on the clarified position at W.’s 23rd move. W. is a P. ahead in an otherwise even position. The plan should have been simply 23 P-KB4 to start a pawn storming of the B. K. side position or a passed P. plus what K. attack develops. 23 Q-N2 was strategically and tactically unnecessary. Nor is 23 K-R2 necessary. Capa would probably have played 23 P-KB4 R-B4 24 Q-N2 QxQ 25 RxQ R-Q6 26 K-B2 and now even with 24…QR-B6 25 P-N4 W. could not be prevented from doubling Rs. on the K. file. The point is that B.’s demonstration with P-QR4 could not follow through because W. could arrive at a P-KB5 driving B.’s B. off the diagonal before B.’s threats could mount.

26 R-Q1 is the first of a new series of weak moves stemming from clock trouble. It is not addressed to the strategic needs. Correct is 26 K-R2 P-QR4 27 P-B4 P-N3 28 R-KB P-QN4?? 29 P-B5. Probably B. would have to play 28…Q-B7.

28 Q-K2?? is totally incomprehensible. 28 P-B4 continues to prevent …P-QN4 because of QxP.

31 R-Q2?? is again wrong preparation. Correct is 31 B-B3 and if 31…P-KR4 32 P-N4 or if 31…Q-R6 32 B-N4 or if the defensive ideas are found 31…P-R4 32 P-N4 Q-R6 33 P-B5 or R-Q2 first.

32 P-N4?? finally throws away the win. I overlooked the answer completely and in exasperation overlooked the draw as well. 33 P-B5 Q-K4+ 34 K-R R-B8+ 35 R-Q Q-N7 36 Q-Q3 BxRP 37 Q-Q8+ etc. First should have come some move like 32 B-B3.

Capa’s final combination starting with 33…QxQ was beautiful precision and included a problem-like move 36…P-B6.

I do not pretend these annotations are complete and some of them being present day offhand may be inaccurate or wrong. Please feel free to question them or to attack their validity.

In any case I would be pleased to hear from you further on this or any other question.

With best regards and good wishes.

Yours truly,

Alex Kevitz.’


In C.N. 2820 Gene Gnandt (Houston, TX, USA) commented on Kevitz’s annotations:

‘I think he is much too hard on himself. Even with Capablanca’s 36...f3!, White does not appear to be lost. After 36...f3! 37 Bf1 Rb8 …


instead of 38 Rxf3?? why not 38 Bd3 Rb2 39 Rxb2 axb2 40 Bb1 Bc4 41 Kg3 Be2, etc.?’

In that same C.N. item we noted that the second Chess Stars volume on Capablanca, published in 1997, also gives the above line (page 141), indicating (with symbols) that after 42 a4 White has counter-play and stands only slightly worse.

In C.N. 4193 Karsten Müller (Hamburg, Germany) informed us:

‘Capablanca’s 36...f3 was, in fact, a mistake, violating the important endgame principle “do not hurry”. Improving the position of the rook and king first and playing ...f3 later to disrupt White’s defensive coordination should win. For instance, 36...Ra4 37 g5 Kg7 38 h4 h6 39 gxh6+ Kxh6, and White is powerless against Black’s plan of bringing the rook to b2 and infiltrating with the king (e.g. 40 Bf3 Rb4, followed by ...Rb2).’

As regards the game-score, in October 2010 Olimpiu G. Urcan (Singapore) sent us the moves as they appeared in the New York Times, 20 April 1931, page 26 and the Brooklyn Daily Eagle, 23 April 1931, page 28. Both gave the move sequence as 27 Qe2 and 28 Kh2:

kevitz capablanca

kevitz capablanca

The game was not widely published in chess magazines of the time, although it appears in many collections of the Cuban’s greatest games, highly praised. For example, on page 238 of Capablanca’s Best Chess Endings (Oxford, 1978) Irving Chernev called it ‘a delightful gem with a piquant finish’. In Capablanca’s Hundred Best Games of Chess by Harry Golombek (London, 1947) 36...f3 was described as ‘a potent move that makes the game resemble an endgame study’ (or ‘a composed endgame’ in the 1996 Batsford/Nunn edition).

One of our copies of the original edition of Golombek’s book is inscribed to Bruce Hayden:

golombek hayden

We conclude with a photograph of Alexander Kevitz, from page 44 of the March 1929 American Chess Bulletin:


This article originally appeared at

Latest update: 13 June 2015.

To the Chess Notes main page.

To the Archives for other feature articles.

Copyright: Edward Winter. All rights reserved.