This means Euwe could even have played 53. Rf2. Similarly Black can't drive White's king further away because after the manuever ...Kd3, ...Ke3, ...Rg1+ White has Rg2.

51.Kf3?
here Euwe allows his King to be driven a file further off, which he could avoid simply by 51.Rd2+ Kc3 52.Rh2 If 53...Kd3 53.Rg2 and the King has e5 if checked. And if 52...a3 then 53.Rg2 Rb8 54.Ke3 Rb2 55.Re2! drawing easily (Flohr)

51...Kd3 52.Rb2 Rf8+ 53.Kg3?
Here 53.Kg2 makes things easier. And it was just a matter of elementary logic. The text gives Black two ways of winning the pawn-Kg2 gives only one. See next note.

53...Kc3 54.Rb7 54.Rf1!
If the White's King were at g2, Black could only win the pawn by ...Rd8-d2, which would enable the said King to get two file nearer before the pawn fell, instead of only one. In such endings, the number of file between the defending King is the chief factor.[Purdy ignores 54.♖b7?]

55.Rb8 Ra1 56.Kf3 Rxa2 57.Ke3!
A usually reliable annotator calls this also a blunder and gives 57.Ra8? Actually, Ra8 gives black a comfortable win by 57...a3 and if then 58.Ke3 Ra1! 59.Rc8+ Kb4 Euwe's move must be the best because it brings the King nearer, and it should have drawn still.

57...Rh2

If now 57...a3 White draws by 58.Rc8+ and 58.Kd3 as Black's King will be driven in front of the pawn.[Wrong! Grigoriev gives a win here: 57...a3 58.♖c8+ ♔b2 59.♔d2 ♔b1+ 60.♔d1 ♖h2 61.♖b8+ ♖b2 62.♖c8 ♖b7 ]

58.Rc8+??
This seems to us the most incomprehensible mistake of all, yet we have seen no comment on it. It is so obvious that 58.Ra8! is better than the text since it gives black less choice. If then 58...Rh4. White simply checks until he drives the Black King in front of the pawn (the black Rook cannot afford to interpose) and then approaches with his King for a book draw. Likewise, if 58...Kb3, the 59.Kd3; and now 59...a3 allows Rb8+ and the same book draw results. 58...Kb2 59.Rb8+ The point is that 59.Kd3 is now useless because after 59...Rh3+ 60.Kd2 a3 Black can interpose his Rook to aheck in the file and thus avoid putting his King in front of the pawn.

59...Kc1! 60.Rc8+
White could resign now if 60.Ra8 a3!

60...Kb1 61.Rb8+ Rb2 62.Ra8 Rb3+ 63.Kd4 a3 64.Kc4 Kb2 65.Rh8 Rc3+ 0-1

-Cecil Purdy

Given the tablebase, the annotation should go as follows:

53. Kg3?! (53. Rf2 - the rook trade leads to a draw)

54. Rb7? (missing <Calli's> Re2 holding the second rank, with a draw)

57...Rh2? (missing ...a3 or ...Kc2 with a win)

58. Rc8?? (missing Ra8 with a draw, as explained in Purdy's annotation)

The win after 57...a3 is instructive.

White won't be able to drive black's king in front of the pawn after 57...a3 58. Rc8+ Kb2 59. Kd2 Kb1+ 60. Kc1 Rh2. Now if 61. Rb8+ Rb2 62. Rc8 Rb4, etc., or if 61. Rc1+ Kb2 62. Rc8 Rh3 Black's rook can shelter Black's king.

In the first note to 51.Kf3 (see my post on Jan 28), Purdy quotes Flohr who reaches the same drawn position. (55.Re2!) So he knew about it.

"What if 42...Kd4 instead?"

As to the earlier tries, Purdy makes the point that the game was adjourned overnight at move 40. Alekhine must have thoroughly investigated Kd4 and similar lines but found nothing. My theory is that he figured that Euwe looked at all the same lines so he decided on 42...g5+ in order to put his opponent on his own resources. It didn't seem to work at first because Euwe finds the key move 45.Rg2!, after which it is a draw as we have discussed. Euwe, however, seems not to be familiar with the ending and makes a lot of mistakes as Purdy says.

This seems to us the most incomprehensible mistake of all, <yet we have seen no comment on it>. It is so obvious that 58.Ra8! is better than the text since it gives black less choice. If then 58...Rh4. White simply checks until he drives the Black King in front of the pawn (the black Rook cannot afford to interpose) and then approaches with his King for a book draw. Likewise, if 58...Kb3, the 59.Kd3; and now 59...a3 allows Rb8+ and the same book draw results. 58...Kb2 59.Rb8+ The point is that 59.Kd3 is now useless because after 59...Rh3+ 60.Kd2 a3 Black can interpose his Rook to aheck in the file and thus avoid putting his King in front of the pawn. ***>

The complete post from which the above is excerpted is very interesting. On another occasion when I have more time available than at present, I intend to work through it carefully. For the present, and FWIW with respect to Purdy’s comment that there had been no comment on White’s blunder with <58. Rc8+?>, the recently published <Tragicomedy in the Endgame: Instructive Mistakes of the Masters>, by Dvoretsky, Mark, Russell Enterprises, Inc., ©2011, at page 28 says of the position after <57. … Ra2-h2?>: “<As Max Euwe noted>, he could have saved the game with <58. Ra8!>.”

So to summarize, there were three consequential mistakes according to tablebase.

56. Kf3? loses (56. Rf8 or 56. Rc8+ draw).

57...Rh2? draws (57...a3 or 57...Kc2 win).

click for larger view

Why does White play 54...Rc1? It doesn't seem to change anything, but neither does there seem to be any reason for it. If noting else, whey not 54...Ra1 as he does on the next move?

The answer may be a score error back on Black's 50th move. Back after <50.Kf4>

click for larger view

Now, instead of 50...Re1, the move <50...Re8> appears in several contemporary sources, such as the <London<London Guardian (November 12, 1935, p. 15), the American Chess bulletin> (November, 1935, p. 150) and <Chess Review>(February 1936, p. 29).

This makes the sequence following <51.Kf3 Kd3 52.Rb2 Rf8+ 53.Kg3 Kc3 54.Rb7 Rf1>. This avoids the question about ...Rc1. The rook had to spend an extra move to get to f1, so never had time to go to c1.

C.J.S. Purdy's book, 'Extreme Chess', published in 1999, includes his earlier 1936 publication, 'How Euwe Won', (Australasian Chess Review, Sydney 1936). In 'Extreme Chess', the moves were given in Algebraic notation: 50...Re8; 52...Rf8+ & 54...Rf1.

However, my Fritz database and Munninghoff's biography on Euwe, provide an Algebraic notation game score the same as given here on chessgames.com.