After 32...Rd3! white is truly busted, all the black pieces come into the game with great power. 33.Qxe4?? is obviously easily refuted with Qe2 which wins a piece. Fritz thinks Euwe is stupid for playing 34.Qxe4+, but the game is already lost even if he would play 34.Qc2 Rc3 35.Qb2 Rh3! (the key move) 36.Qc2 f5 37.g3 Nf3+ and resignation is in order.

He deliberately allows Qxe4+ to divert it off e5 by f5. The queen must stay on the long diagonal or its mate. Another beauty was c6!, followed by Rxc3! with the same theme (I think every move there deserves a !! but obviously, the concept was simple).

< 38... Rxc1 39. h3>

Black now replied 39...Qg3 in a flash, and went on to win on the 56th move. Had it not been for the time trouble, < 39...Nf3+! > would not have escaped his attention

click for larger view

...is a cracker of a move! I'll bet that Keres kicked himself for not seeing it. Or perhaps he did see it but in time trouble thought it was safer to play any non-losing move to be certain of reaching the time-control.

This loss was the beginning of a disastrous tournament by Euwe, who ended with only one win and six draws in 20 games. This game set the pattern for what followed for Euwe in his effort to regain the crown. he got decent positions, but erred, often in time trouble as happened here.

An understanding of the drama of this first-round game (when Euwe still had hopes of success) is enhanced by the contemporary commentary by the players and others.

1. e4 e5
2. Nf3 Nc6
3. Bb5 a6
4. Ba4 d6

The so-called Modern Steinitz Defense (aka Steinitz Defense Deferred). It was employed several times at the 1948 World Championship Tournament by Keres and by Reshevsky.

"Keres has enjoyed much success both with and against this defence of recent years. Dr. Euwe was fully aware of this and had prepared himself theoretically to meet it, His handling of the opening stages is in fact admirable." (Golombek)

5. c3

5. c4, 5. 0-0 5. BxN+ (lauded by Keres in his commentary on this game) and even 5. d4 are all other reasonable variations for White.

How long did Euwe ponder before playing 5. c3? Accounts differ:

"A move played after long deliberation." (Euwe)

"Euwe made this move without long thought." (Keres).

5... Bd7

It was perhaps surprising that an attacking player such as Keres did not play 5...f5 here:

"A good alternative is 5...f5." (Euwe).

"Very interesting here is 5...f5, which was the continuation preferred by Capablanca." (Keres).

But the text has its points as well:

"Indicating his intention to preserved his centre intact and just sit tight till he sees what plan White is going to adopt. The other way of preventing BxN+, 5...b5, gratuitously weakens Black's Queen-side." (Golombek).

6. d4

He could equally wekk have played 6. 0-0.

6... Nge7

Another surprising (but definitely reasonable) choice from Keres. I would have guessed he would have played 6...Nf6. Much ink has been spilled on Keres' choice:

"Although basically sound this line is too passive for most players." (Kmoch).

"Steinitz' own move which aims at bringing the Knight to the defence of the e-pawn and so avoiding, for as long as possible, the exchange of this pawn for White's d-pawn. An alternative line is 6...h6 7. 0-0 Bg7 and if 8. Bg5 f6 9. Be3 Nh6, as played in the twenty-second match game between Alekhine and Bogoljubov, 1929." (Winter)

"Seldom seen nowadays, but pure Steinitz in that at the expense of time and at the cost of a temporarily constricted position Black brings a piece to bear on the vital squares f5 and e4." (Golombek).

"It is questionable whether the black king's knight is so much better positioned on g6 than on f6 that it's worth wasting a tempo on this mmove. But this entire variation by Black should be seen more as a search for less well-trodden pathes than as an attempt to find a flawless set-up." (Euwe).

"With the text[manoeuvre Black plans to develop his knight to g6, from where it protects the e5-point and later threatens to occupy the strong square f4 in several lines. The drawback of the move is the loss of time that is entailed with the knight development, and especially the weakening of the central square d5, where White soon places his knight. Even though many theoreticians consider the text-move to be inferior...I still believe that 6...Nge7 is a playable system of defence. If Black, for instance, continues with 6...g6 7. 0-0 Bg7, he will find himself facing much more unpleasant problems than in the game after the simple exchange 8. dxe5. In addition, 6...Nf6, which would lead the game into the so-called Kecskemet variation, would yiled Black a good game." (Keres)

click for larger view

Interestingly, this same position was reached (via transposition of Black's 5th and 6th moves) in the 11th round game between Smyslov and Reshevsky--a game won by White. Indeed, the games did not vary until Black's 12th move.

Keres and Euwe had played each other no fewer than nineteen times before the 1948 World Championship Tournament, with both players having seven victories (and Keres having come out one game ahead in their 14-game 1940 match). It was therefore doubtless a surprise that Keres demolished Euwe in the 1948 World Championship Tournament, winning four of five games and drawing the other.

Returning to the instant contest after 6...Nge7 (in the diagrammed position with which I ended my last post):

7. Bb3

"...threatens Ng5 and so compels Black's next." (Winter)

"After twenty minutes thought Dr. Euwe chose the line recommended by the Austrian theoretician, Becker. It is an attacking move that bears down on Black's weak point on f7, and so gives the opening strong resemblance to the Giuoco Piano. Normal development by 7. Be3 Ng6 8. Qd2 Nh4 gives Black equality.. [But White can keep an edge in this line with 8. d5 or 8. 0-0, and 8.,..Be7 or 8...Na5 are better than Golombek's 8...Nh4--KEG]. But 7. 0-0 Ng6 [7...h6 is better--KEG] 8. Re1 [better are 8. d5 or 8. Be3--KEG] keeps the initiative for White [but Black is OK in the above line after 8...b5]." (Golombek)

"This is stronger than 7. h4...Now Ng5 is threatening to become unpleasant." (Euwe).

"After Black has secured the e-point by Neg6, the light-squared bishop no longer has any prospects on a4, and for that reason its transfer to the a2-g8 diagonal is fully justifiable. Furthermore, it occurs here with the gain of a tempo, since Black has to play 7...h6 [7...f6 would not be awful, but 7...h6 is certainly best--KEG] in order to thwarts the threat of Ng5." (Keres)

Euwe had two good alternatives here: 7. 0-0 or 7. d5. The text is also good for the reasons outlined by Keres.

7... h6

"Against the threatened Ng5." (Golombek).

The position was now:

click for larger view

8. Nbd2

"An improvement [doubtful--KEG] on 8. Be3 which allows BLack easy equality by 8...Ng6 [White still seems better after 9. 0-0--KEG] 9. Nbd2 (citing the second 1942 Kashdan-Reshevsky match game." (Golombek)

"The value of this move depends on whether White subsequently has to fear the g5 advance. In the game Ahues-Rubinstein, San Remo 1930, White continued with 8. Be3 and after 8...g5 sacrificed a piece by 9. Bxg5 hxB 10. Nxg5 d5 11. exd5 Na5 [11...Nxd5 seems to win for Black here, i.e., 12. Nxf7 KxN 13. Bxd5+ Kg7--KEG] As later analysis proved, instead of the text-move 12. dxe5 [which appears to lose after 12...NxB--KEG] 12. d6! [surely best, but not leading to an edge for White--KEG] NxB 13. QxN Nd5 14. Nxf7 KxN 15. dxc7 [15. Qxd5+ Be6 16. Qf3+ Ke8 is no better for White--KEG] Qxc7 16. Qxd5+ [but isn't Black still better after 16...Be6--KEG]. But this whole line is ot very convincing...White lost an important tempo by the manoeuvre Bc1-e3 xg5 that he could have used for his development...instead of 8...g5 Black should have proceeded with the quiet 8...Ng6, after this White's bishop would simply be unfavorably placed on e3, blocking the path to White's knight on its way to d5 via e3. [But White seems somewhat better in this variation--KEG].

Smyslov's idea to play 8. Nh4 would also not have been dangerous..." (Keres).

Keres goes on to conclude that the text move (8. Nbd2) is "fully playable."

Perhaps best for White after 7...h6 was either 8. d5 or 8. 0-0 (or perhaps 8. a4).

8... Ng6

Golombek and Euwe suggest 8...g5 as a good alternative, but after 9. d5 [better than Keres' suggested 9. dxe5 which does not yield much more than equality for White--KEG] Na5 [better than Golombek's 9...Nb8--KEG] 10. Bc2 White seems clearly better.

9. Nc4

"White's plan is to bring this Knight to a decent post at d5." (Winter).

9. 0-0 and 9. d5 are both good alternatives.

The position was now:

click for larger view

9... Be7

Golombek here rightly condemned 8...Nf4 because of 10. BxN exB 11. d5 Ne7 12. e5 and White has much the better chances. But he was wrong in suggesting 12...Ng6 here for Black (Black is still in the game with 12...Nf5) and also wrong is suggesting 13. 0-0 in response to 12...Ng6 (13. e6 looks crushing).

Keres rightly rejects 9...h4 here for Black since White then gets the better game with 10. NxN (which is even better than Keres' 10. Ne3).

The text (9...Be7), however, was probably not best. 9...Qf6; 9...exd4; and 9...b5 all look better, but it was consistent with Keres' strategy of careful development and waiting for a chance to strike while keeping an eye on e5. The move (the only move even considered in MCO-13) was definitely not a significant mistake on any analysis.

10. 0-0

Better than MCO-13's 10. Ne3, but 10. d5 was most forcing and likely best.

10... 0-0
11. Ne3

The position was now:

click for larger view

11... Bf6

Reasonable and adhering to Keres' strategy. 11...exd4 (clarifying the center); 11...Re8 (suggested by both Keres and Euwe); and 11....a5 seem to be the main alternatives.

Several of the commentators, however, devoted their energy to considering 11...Nf4, which allows White to get the better game after 12. Re1. If instead 12. Nd5 (said to be the refutation by Winter and Golombek) Black is OK after 12...Ng6 (and not 12...Nxd5 13. exd5!)

12. Nd5

The game was now at a cross-roads:

click for larger view

This position was also reached in Smyslov-Reshevsky in the 11th round. Reshevsky played 12...Re8 and lost after 13. dxe5. While the commentators condemned Reshevsky's move, it does not look bad to me. When I get to the Smyslov-Reshevsky game, I will attempt to show that 12...Re8 was not the cause of Reshevsky's later woes.

Euwe here played:

12... exd4.

The move discussed in most of the commentaries was 12...Bg4. This move is probably inferior to the text, but hardly the disaster suggested in the sources I have consulted; u.e., 12...Bg4 13. h3 BxN 14. QxB and now after 14...exd4 [14...Bg5 looks even better--KEG] 15. NxB+ (given as best by Golombek, and Kmoch; Keres' 15. Bxh6 yields only a small edge to White after 15...Bg5--KEG] QxN 16. QxQ gxQ 17. Bxh6 Rfe8 18. Rfe1 (18. Bc2 or 18. f3 are even better) leaves White somewhat better.

13. Nxd4

"!"--(Kmoch).

As Keres, Winter and Euwe have noted, 13. cxd4 would have left White awkwardly placed after 13...Bg4

"White has a substantial edge"--(Kmoch). [which game was he watching?)

The best overall assessment at this point was that of Keres following Black's next move:

13... Re8

"After the exchange of central pawns, Black slightly relieves his cramped position, even though White still maintains a small spatial advantage. With the text-move, Black poses his opponent problems about how to protect the e4 pawn." (Keres).

The position was now:

click for larger view

14. NxB+

It must have been tempting to cut down material against a dangerous attacking player such as Keres, but at what cost! The commentators were not kind in discussing this move by Euwe.

"This gains the theoretical advantage of the pair of Bishops, but it greatly frees Black's rather congested position, and it is at least questionable whether White would not have done better to retain the Knight, at any rate, in the meantime." (Winter)

"Gaining the technical advantage of two Bishis, nevertheless dissipating most of his opening advantage. White's superiorirty has consisted in the beautiful position of his two centralized Knights, and in the fact that his King's Bishop is an active attacking piece as opposed to the passive defense of Black's Queen Bishop. He must therefore evolve a plan to profit from these advantages." (Golombek).

"Not bad but not strong either. White has an advantage in development and occupied territory and he should maintain it by protecting the e-pawn." (Kmoch).

"Too early." (Euwe).

"Sucn ah exchange is not recommended unless there is an urgent need, since first of all Black is even more freed, and secondly the position is further simplified and offers fewer attacking possibilities."

Okay, so if 14. NxB+ was not best, what should Euwe have played? Let's look at the possibilities:

A) 14. f3. This defends the e-pawn and seems the logical choice. It certainly looks best to me, but not to most of the commentators. One commentator who mentioned the move--Kmoch--dismissed it by saying "...the position calls for [an eventual] f4]." My only ally in opting for 14. f3 (apart from is Fritz) is Golombek, who contended that it would allow White to "continue to centralise his pieces."

B) 14. Bc2. This appears to have been the most popular choice:

"Correct is 14. Bc2!" (Kmoch).

Euwe also preferred 14. Bc2, but only considered the reply 14...Be6 (which allows 15. NxB)e6) RxN 16. f4 with a big advantage for White. But Black can seemingly equalize with ease by means of 14...NxN 15. cxN (if 15. NxB+ QxN 16. QxN with approximate equality) c5! with sharp play and interesting prospects for both sides.

Keres also chose 14. Bc2 as best, but only he recognized that after 14...NxN 15. cxN Bb5 (I think 15...c5! immediately is better, though Keres' move obviously had its points) 16. Re1 c5 Black obtains "adequate counterplay."

C) Re1 is another seemingly good alternative that was nonetheless rejected by every commentator who mentioned it:

"...not...14. Re1 because 14...Bg5 15. f4 Bh4 is annoying as 16. g3 weakens the King-side." (Kmoch). But after 16...Be7 in Kmoch's variation White has much the better game. Black's best chance to obtain equality after 14. Re1 (and the reason I do not advocate it) is because after 14...NxN 15. NxB+ (if 15. cxd4 Be6 and Black is fine) QxN 16. cxd4 Black's better pawn structure compensates for White's two Bishops and control of the center.

"...the move 14. Re1 has its drawbacks; Black would then play 14...Na5 (I prefer 14...NxN but most concede I did not even think of 14...Na5) 15. Bc2 c6 16. NxB+ QxN and now White is unable to carry out the f4 advance." (Keres). [White does, however, seem to have the better game at the end of Keres' variation--KEG].

C) A wild possibility is 14. Nf5?! simply abandoning the e-pawn for the attack, e.g., 14...Rxe4 15. Nfe3 Re8 16. Qh5. I'm not sure this is sound, but I wouldn't want to face this line if I were Black.

Returning Euwe's actual move: 14. NxB+, even if it was not best, it certainly did not lose the game. As Keres put is: "Although with the game continuation White makes some concessions, he still assures himself of a good position with the bishop pair and a slight initiative." This is surely the final (and best) word on 14. NxB+.

14... QxN

This left:

click for larger view

15. f3

Was 15. Bc2 better? Kmoch says yes. But the choice is close and other commentators disagree: Golombek thought that after 15. Bc2 Black would get the better position after 15...Re7, but White then gets some edge after 16. f3 or 16. Be3. Keres' commentary includes an interesting discussion of the merits of 15. Bc2, ultimately rejecting the move because Black can simply play 15...NxN 16. cxN Bb5 [16...c5 immediately looks stronger--KEG] 17. Re1 c5.

Keres also considered the wild attacking move 15...d5 after 15. Bc2. But after 16. NxN cxN (16...BxN 17. exd5 Bb5 looks better--KEG] 17. exd5 (17. f4 is stronger and appears to be the refutation of this variation--KEG] cxd5 18. Qxd5? Bc6 "with dangerous counterplay." But Keres overlooked the stronger 18. Qd4!which leads to an ending in which White has a small but persistent edge.

After 15. f3, the position was:

click for larger view

15... Nf4

"threatens 16...NxN 17. cxN c5" (Wade-Whiteley-Keene)

"Threatening 16...NxN 17. cxN Qxd4+ 18. QxQ Ne2+" (Winter).

But Euwe and Keres were critical of 15...Nf4:

"This will turn out to be ultimately a waste of time. Correct is 15...Na5 16. Bc2 c5" (Euwe)

But Euwe's line does not leave Black significantly better than the line actually played. The best critique appears in Keres' commentary:

"By this sally [i.e., 15...Nf4] Black, it's true, threatens Nxd4 followed by Qxd4+, but after the thwarting of this threat he just loses time and ends up worse off. A good plan instead was 15...Rad8 [Fritz and Stockfish agree], with the intention of answering 16. Be3 [16. NxN looks somewhat better--KEG] Na5 17. Bc2 [17. Re1 is probably best--KEG] c5 [strong, but even better would be 17...d5--KEG] with good counterplay."

16. NxN

"Black was threatening 16...NxN" (Euwe)

"White's reply is almost forced, for 16. Be3 can be met by 16...Na5." (Winter).

"If 16. Be3 Na5" (Kmoch)

While the text (16. NxN) was probably best,the criticism of 16. Be3 is over the top. Let's hear from Keres on this:

"16. Bee would be answered by 16...Na5. However, after the simple 16. Kh1 [17. Qd2 looks even better--KEG] Black would have had great difficulty in justifying his Knight sally." (Keres).

16... BxN

This recapture is passed over without comment in most commentaries. But Keres--who gives the text a "?"--looked deeper:

"Surprisingly, Black abstains from the favorable opportunity to play 16...bxB, which would after 17. Be3 c5 have deprived the white bishop of the strong d4-square, and in some lines enabled Black to apply pressure on the b-file. It seems that Black in this case could have achieved an equal game. But now White can develop his pieces unhundered and start a dangerous attack on the kingside, against which Black has difficult in finding a satisfactory defence."

But isn't White much better in Keres' line by playing 18. Qd2. But with 17...Ng6 (instead of Keres' 17...c5_ Black does indeed have approximately equal chances, thus supporting the general thrust of Keres' otherwise excellent comment.

The position after 16...BxN was:

click for larger view

17. Be3 Rad8

Retreating the Knight immediately by 17...Ng6 or 17...Ne6 were good alternatives, but the text also keeps White's edge to a minimum.

18. Qd2 Ng6

"The Knight is better advised to return to e6." (Euwe).

But Keres disagreed:

"Now the consequences of Black's 15th move become apparent. The Knight is forced to retreat, leaving Black in a completely passive position. No better was 18...Ne6 19. Rae1." [Perhaps 19. Rad1 or 19. Rfe1 would be improvements on Keres' line, but his central point is what is important: Black's 15th move has left the initiative solidly in White's hands.

The position was now:

click for larger view

19. Bd4

"A strong move after which Blaxk always has to reckon with possible sacrifices on g7." (Golombek)

"White still has a good game but he lacks a good plan. An action on the Queen-side starting with 19. c4, merely to anticipate a counter-action there, was indicated. The text-move has no positive value." (Kmoch).

I'm with Golombek in this debate. If 19. c4 Black has several good retorts all of which leave him in decent shape: e.g., 19...b6 or 19...Ne5. But either the text (19. Bd4) or 19. c4 leave White with wha initiative there is and a small advantage.

19... Qe7

19...Qg5 or 19...Qh4 were reasonable alternatives.

The position now was:

click for larger view

"Up to now White had taken some twenty minutes less on the clock than Black, the respective times reading: Dr. Euwe 1 hour 20 minutes, Keres 1 hour 42 minutes. But as the conduct of the game became increasingly difficult for White, he consumed more and more time from now on, becoming about level with his opponent by the 27th move." (Golombek)

20. Rae1

"White has still a very good position but is faced with the dilemma of whether to proceed with the attack on the Queen's side." (Winter)

Bronstein and Winter thought the text was best and would have given Euwe good chances if properly followed up.

20. f4? would be bad in light of 20...Qxe4 and if then 21. f5 [21. Rae1 may be best, but White would still be in trouble] Qe2!

Should Euwe should have played 20. c4, as claimed by Golombek, Wade-Whiteley-Keene, and Keres? Keres supplies the best argument for this:

"White's position is superior and, by correct play, probably won. [An overstatement--KEG]. In order to achieve the win, the following kingside attack had to be well prepared, and in so doing Black's only counter-chance, which lies in the d5 thrust, had to be liquidated. By the text-move [20. Rae1] White prepares the f4 advance, but this plan still looks premature and yields Black unnecessary counterplay on the queenside.

The correct continuation, in my opinion, was to play 20. c4! This move would have permanently deprived Black of all hopes of carrying out the liberating d5 thrust, and at the same time it would also thwart the possibility of ...Qd7 followed by ...Ba4, as occurred later in the game. In addition, the very strong 21. Qc3 would become a threat. It appears Black would thereafter been unable to defend himself against Rae1, Qc3, Bc2, and f4."

While Keres' proposed 20. c4 rules out ...d5, White is nowhere near winning. If 20...b6, White would probably play 21. Rae1. Golombek and Kmoch fault the text move, saying White's Rooks should be placed on the c and d files. But in Keres' analysis, and in the lines I have explored, the a1 Rook often ends up on e1.

My conclusion is that Euwe' move was OK and perhaps best, and had nothing to do with his loss of the game.

The position after 20. Rae1 was:

click for larger view

20... Qd7

"Economical and menacing. The Queen is removed from the line of White's Rook, so allowing a possible d5; at the same time Ba4 is threatened." (Golombek)

"Black now manages to exchange off one of White's strong bishops, since White can no longer prevent the positional threat of 21...Bb5 followed by ...Ba4 or d5. Even though several commentators still consider White's position to be clearly better [Fritz rates the position 0.32 for White before 20...Qd7 and 0.36 after; and Stockfish evaluates it as 0.28--in any case a small edge for White], I tend to believe that Black already has sufficient defensive resources, and that White spoilt his best chances on the previous move." (Keres)

21. c4

This move, though lauded by some on Euwe's turn, is faulted by nearly everyone. Let's look at the alternatives:

A) Bc2

"What was good on move 20 [i.e., c4] is bad on move 21, this being due to White's mistaken Rook move. He should play 21. Bc2." (Golombek).

But:

"Neither would 21. Bc2 give White an advantage: 21. Bc2 Bb5 22. Rf2 d5." [I am with Keres to this point, but 22...d5 allows 23. e5 giving White all sorts of chance. Either 22...c5 or 22...Ne5 are better and support Keres' evaluation. Keres goes on to say that after my suggested 23. e5 Black would play 23...c5 "with sufficient counterplay." But after 24. Bxc5 Rxe5 25. RxR NxR Bd4 White has all the chances].

B) 21. f4

"This time White could have gone forward: 21. f4 Bxe4 (21...Rxe4 is also possible) 22. Bxg7! It is doubtful, however, whether the liquidation KxB [overlooking the equalizing 22...Bxg2!--KEG] 23. Qd4+ Kh7 [23...f6 and 23...Kf8 would be better, though White would still have an edge--KEG] 24. RxB f5 [This courts disaster--24...c5 was essential--KEG] 25. RxR RxR...would offer White realistic winning chances. [But isn't White much better in Euwe's final position--KEG]. Nevertheless, White should have gone for this continuation, which entails no risk for him at all, since the text-move yields absolutley nothing." (Euwe)

Keres, who also prefers 21. f4, followed Euwe' line through 23. Qd4+, but then played 23...Kg8 24. Rxe4.

C) 21. a3

Keres exposed the error in this line: 21. a3 d5 (better than Euwe's proposed 21...Ba4) , but White's best, though not leading much of anywhere, would then be 22. exd5 rather than Keres' 22. e5 which gets White into "difficulties" after 22...Bb5 23. Rf2 b6.

D) 21. Kh1

The position after 21. c4 was:

click for larger view

To this point, despite the many criticisms and alternative moves suggested, play had been of a very high order. There had been no major mistakes, and neither side had been close to winning (or losing).

21... Ba4

"!"--(Golombek)

Keres, reasonably enpugh, decided to eliminate Euwe's white-square Bishop.

"With every step that he takes into the middle game, Keres' play increases its vigour and subtlety. White is not allowed time for Bc3, Qc3, f4 and an eventual e5." (Golombek).

22. BxB

22. Qb4 would be no improvement, as Golombek noted: 22. Qb4 BxB 23. QxB b5 24. Rd1 [better than Golombek's 24. c5 which leaves an even game after 24...d5] 23. Rd1 leaving White with a minimal edge at most.

The only good alternative to 22. BxB was 22. Qc3. This move only makes sense, however, if White is willing to respond to 22...BxB with 23. axB, accepting a pawn weakness in return for attacking chances.

On balance, the text was probably best.

22... QxB

This left:

click for larger view

23. Qc3 f6

"Black has almost completely equalized the game." (Golombek).

The position was now:

click for larger view

"An important moment. Can White sacrifice on f6?" (Euwe).

Winter said that White's best chance was 24. Bxf6?! Golombek said this was "not good." Let's see: 24. Bxf6?! gxB 25. Qxf6 Nf8 26. b3 [this move was recommended by both Winter and Golombek, but 26. Qh6 seems clearly better, leaving White with some compensation--three pawns for the sacrificed piece after 26...Qxc4 Qg5+ Ng6--KEG] and now simply 26...Qxa2 [instead of 26...Qd7 27. Qxh6 Qg7 as given by Winter and Golombek] with a likely win for Black.

So the answer seems to be that the immediate sacrifice 24. Bxf6?! would be an error. So what should Euwe play?

24. f4

"?"--(Kmoch)(Euwe)(Wade-Whiteley-Keene).

"This weakening of the white center is motivated by nothing in particular [a king-side attack?--KEG] and is the start of White's problems." (Euwe)

The consensus is that 24. b3 would have been better:

"This is Euwe's first mistake. He is playing for a King-side attack threatening 25. Bxf6...but actually he only weakens his King pawn and thereby cuts down the mobility of his Queen. 24. b3 directly is better." (Kmoch)

"Initially Euwe had planned to sacrifice the bishop by 24. Bxf6, but later considered this to be too risky and decided first to open up the third ran for his rook in order to support the attack. In fact, Black now manages to prevent the sacrifice and repulses White's attacking attempts. The f4 advance turns out to be just a weakening of the e4 point...it was better to abstain from a further weakening of the position and simply continue with 24. b3 followed by Rf2 with a roughly equal position." (Keres)

Keres has pretty much said it all, though I still think Euwe was for choice after 24. b3.

After Euwe's actual 24. f4, the position was:

click for larger view

24... Kh7

"Eliminating the possibility of a successful sacrifice on f6 and in turn threatening to win either the e or f pawn by c5." (Golombek)

"Preventing the sacrifice [on f6]." -- (Kmoch)

"Now Black's kingside is safe and he can commence operations against White's Queen's wing." (Winter)

The assumption in all the above comments is that Keres needed to take steps to avoid 25. Bxf6. But--as Fritz and Stockfish confirm, Black can begin his own counter-attack immediately with 24...Qc6 and have much the better game. If then 25. Bxf6? Black simply plays 25...gxB 26. Qxf6 d5 putting an end to any hopes of an attack by White and obtaining the better--and possibly winning--endgame after 27. QxQ (what else?) cxQ 28. cxd5 cxd5 29. e5 Ne7 with Black having lost only two pawns for the sacrificed piece.

After Keres' defensive 24...Kh7, White has at least equality.

To make one obvious point, Keres could of course not have played 24...Qxa2 because of 25. Ra1 winning the Queen.

25. b3 Qd7

Of course not 25...Qxa2 26. Rf2 and the Queen is lost as was noted by Golombek, Kmoch, and Winter

As on his previous move, Keres could and should have played 25...Qc6.

After 25...Qd7, the position was:

click for larger view

26. Qf3

"Black was threatening to win a pawn by 26...c5." (Winter)

But there was a better way to address the c5 threat: 26. Qd3. This was especially the case given that Euwe played Qd3 on his very next move.

"Black still threatened to win a pawn by c5, and the Queen move [i.e., 26. Qf3] is intended to support the e-oawn and prevent Black from playing Qg5. However, the latter threat is not at the moment important, and better is 26. Qd3." (Golombek).

"Better is 26. Qd3 at once." (Kmoch).

"Better is 26 a4 or 26. Qd3." (Euwe).

After 26. a4, Black gets at least equality with 26...c5.

Keres seems to have been on a different wavelength from the other commentators at this stage:

"White still hopes to create some attack, but this attempt turns out to be incorrect, and even forces him to take up a defensive position. If White didn't want to defend passively against the threat of 26...c5 by for example 26. Kh1, he could have chosen the interesting double-edged attacking continuation 26. f5 Ne5 27. Qg3 [this could lead to trouble--better seems 27. Re2--KEG] Qf7 [27...b5 or 27...Re7 are better--KEG] 28. Rf4 [This can also lead to problems, 28. h3 is better here--KEG]. However, from an objective point of view, this would not have given him any genuine winning chances."

Keres goes on to recommend 26. Qa5 as the way for White to play for "a roughly equal position,: but this could give Black chances with 26...c5!

My conclusion is that 26. Qd3 was the only way for White to play for an advantage, and the text (26. Qf3 was good enough for equality.

26... b5

"!"--(Golombek)(Kmoch)(Euwe).

This move by Keres was praised by all the commentators:

"By this admirably aggressive move, Black secures the initiative on the Queen-side." (Golombek)

"Black has the better of the game [Doubtful--KEG] and herewith takes the initiative." (Kmoch)

"A fine move which gives Black opportunities for stronb action on the Queens-wing." (Winter)

"The signal for the counter-attack." (Euwe)

While 26...b5 precipitated a crisis in which Euwe ultimately faltered, and while it must have been a difficult move to meet over the board with the clock ticking, as a theoretical matter it should not have yielded anything for Black. 26...Re7 was also good.

The position after 26..b5 was:

click for larger view