Ouf course, that dubious combination wouldn't bring milk and butter against Lasker, but it would garantee lively play.

Part I

<Tarrasch’s comments, translated to the best of my ability, are in plain text; my/Fritz’s comments are in brackets.

Lasker plays the French for the only time in the match and gets an ugly position, though he is able to simplify quickly to an only slightly worse ending. As his notes reveal, however, even after simplification Tarrasch believes his advantage is all but decisive, and continues to grind forward. Lasker misses several good defensive moves, and gradually gets into a genuinely bad position. Near the second time control at move 45, Tarrasch misses first a beautiful win, then a prosaic one. A few moves later, the contestants agree to the first draw of the match. Lasker remains on top, 4-1.>

1. e4 e6 2. d4 d5 3. Nc3 dxe4

This variation of the French Defense, in which Black gives up the center, in my opinion is completely inadequate. With proper play Black never achieves a satisfactory development.

4. Nxe4 Nf6 5. Nxf6+

The tempo that Black gave his opponent must always be returned, for example after 5. Bd3 Nxe4 6. Bxe4 Nd7 7. Nf3 Nf6 8. Bd3. But with the text move White loses two tempi, namely in that, with his knight, which has made two moves already, exchanges itself for a knight that has moved only once, and furthermore, that the black queen is developed or the black g-pawn is brought to f6. Nevertheless the exchange here is correct, because the black queen is vulnerable on f6, so that Black must at least give back a tempo with …h6, or else, if the pawn retakes, the weakening of the black kingside gives White full compensation for the tempo sacrificed, especially since White intends to castle queenside.

5….Qxf6 6. Nf3 Bd7

Instead of this Black should prevent Bg5 with …h6. The black bishop aims at c6, to menace the square g2 in combination with the queen, but White can advantageously parry this threat with an attack of his own.

7. Bg5

One might expect 7. Bd3 here, when Black must respond …h6, because with …Bc6 he would lose his queen by the original combination 8. Bg5 Bxf3 9. Qd2!. White nevertheless rightly thinks that he can take immediate advantage of Black’s failure to play 6…h6.

7….Qg6 8. Bd3 f5?

Instead of this the queen should go to h5, whereupon he could after …f6 retreat the queen to f7, although the weakness of the e5 and e6 points would make it easier for White to break in here and in an endgame. The threat on g2 is perhaps the motivation for the text move.

9. h4!

Not only to protect the bishop (and so threatening to win the queen by Ne5) but also to drive the queen away from attacking g2 and to create a strong forward post on h5.

9….Nc6 10. Qe2

Here g4 came into consideration, but I thought this overaggressive move would fall flat. Equally mistaken was the idea of winning the queen with 10….Be2, because the black queen would with 10…h6 11. Ne5 Nxe5 12. Bh5 hxg5 13. Bxg6+ Nxg6 succeed in selling its life dearly.

10….h6

This this new weakening of the pawn formation, with the g-pawn left backward (which, however, is difficult to avoid) means Black must castle long. And after 10….Be7 11. Bxe7 Nxe7 12. Ne5 Qxg2 13. 0-0-0 Black has big problems.

11. Bf4

11. h5 is prohibited since of course comes the response 11….hxg5 12. hxg6 Rxh1+, etc. Instead of the bishop-retreat would some masters recommended d5; but after 11…hxg5 12. dxc6 Bxc6 13. Ne5 Qf6 White has lost a pawn and lacks a convincing followup, thus 14. Bb5 Bxb5 15. Qxb5 c6 16. Nxc6 bxc6 17. Qxc6+ Kf7 18. Qxa8 fails by Bb4+ winning the queen; and the move 14. h5, recommended by many, after 14….Bd6 15. Ng6 (or 15. Nxc6 bxc6 followed by …Kd7) 15….Rh6 with …0-0-0 and …Be8 comes to nothing.

11….0-0-0 12. 0-0-0 Bd6 13. Be5!

13. Bxd6 was of course bad, since White would strengthen the Black pawn center, especially the e5 square. Black sooner or later would with …e5 advance and get the better game.

<Fritz actually prefers 13. Bxd6, although it thinks Black is fine after 13….cxd6 14. h5 Qf6 15. Bb5 Rae8.>

13….Rhe8(?)

Black is embarrassed for a good move.

<Fritz thinks Black could equalize comfortably here with 13….Nxe5 14. Nxe5 Bxe5 15. dxe5 (after 15. Qxe5 Qxg2 is quite playable) Bc6. Obviously Tarrasch thought his position was better than it really was; I think Lasker may have shared Tarrasch’s opinion.>

14. h5

Causes the complete crippling of the kingside and prepares the later attack on the g-file.

14….Qf7 15. c3

To prevent …Nb4.

click for larger view

15….Nxe5

The black pieces stand unpleasantly crowded together, so Black looks to free his game by exchanges.

<Fritz’s view of the Black’s position remains far more positive than Tarrasch’s: 15…Nxe5 was better a couple of moves ago, but here it is still brings about near-equality.>

16. Nxe5 Bxe5 17. Qxe5 Qf6 18. f4

The text move is the best; White gives up the frontal attack on the pawn, but the pawn at e5 limits the black game and the main attack on the g-file will follow. Also, after 18…Rde1 (with the intention of retaking with the rook after …Qxe5), ….Qg5+ and …Qxg2, although White would maintain a positional advantage.

18….Qxe5 19. fxe5 Re7

If instead Black plays …Bc6, 20. Be2 could follow (with the intention of Bf3), because if 20….Bxg2 21. Rhg1 and Rxg7, when White has the much better game, now that the pawns on h6 and e6 are under direct attack and White controls the only open file.

20. Be2 Bc6 21. Bf3 Be8

If Black takes the bishop, so White’s cause is advanced, as Black achieves White’s goal of opening the g-file (which White would sooner or later achieve by g4).

<I thought this note was strange at first. Since White can indeed open the g-file himself, I couldn’t see any downside to Black getting rid of his bad bishop by 21.…Bxf3 22. gxf3. Although White is obviously better here, I didn’t see how Black could lose, since he can double rooks on his second rank and adequately defend g7 and e6. But Fritz found some semi-promising plans: 22….Rd5 23. f4 (White would like to keep this square free for the king, but needs to protect his spearhead at e5 from the threat of …c5) 23….b5 24. Kc2 c5 25. dxc5 Rxc5 26. Rd4 because of the weaknesses at e6 and on the g-file.

However, Fritz found an alternative to exchanging or retreating the bishop on move 21: 21….Bd5! Now there doesn’t seem to be any way for White to make progress. Bxd5 is a dead draw; b3 and c4 can always be met effectively by …b5.

In any event, Black’s position is perfectly defensible after 21….Be8.>

22. Rdg1

White’s plan is clear; but Black cannot prevent it. Black cannot change the position in any fundamental way.

click for larger view

<In fact, here and for the next several moves, Black can achieve complete equality with …c5, e.g. 22…c5 23. dxc5 Rc7 24. b4 b6! 25. Rd1 bxc5 26. Be2 Bc6 26. Ba6+ Bb7 27. Bxb7+ Rxb7 28. Rxd8+ Kxd8, etc.>

22….c6 23. Rh2 Kc7 24. Kc2

The king cannot cross the d-file, because Black would then have a chance to play the …c5 freeing move.

24….Kb6 25. b4(!)

Prevents the advance of the c-pawn. Lasker therefore said that he should have played …a5 before the king-move.

<With this move White begins to get a real advantage.>

25….Kc7 26. g4 fxg4 27. Rxg4 a5(?)

Black still seeks to open the a-file so that his rook can come into play. Of course White does not consider taking the a-pawn.

<But he should have; White has excellent chances for the full point after 28. bxa5!. If 28….Ra8? 29. Rhg2 Rxa5 30. Rxg7 White is winning; after the somewhat better 28….c5 29. dxc5 Bc6 30. Bxc6 Kxc6 31. Rb4 Rd5 32. Rb6+ Kxc5 33. Re2! Rd8 34. a4 Ra8 35. Rb5+ Kc6 36. Rd2 Fritz evaluates White as more than two pawns up.

In short, 27….a5, natural though it looks, is a big mistake. Also, Black can’t do anything with the a-file, as the game continuation shows.>

28. a3(?) axb4 29. axb4 Ra8 30. Kd3 Rd8 31. Be4

The bishop aims at g6.

31….b5

To assist in the later attack on the White pawns, by making the white c-pawn backwards (check). But this move is also the downfall of Black's c-pawn

32. Ke3 Kb6

click for larger view

33. Rg3

Here the game was adjourned for the first time, with White sealing the rook move.

33….Ra8 34. Rhg2 Ra3 35. Kd3 Bxh5(?)

If Black had guarded the g-pawn by ...Raa7, then Bg6 would follow. Whether Black then moved his bishop back to d7 or exchanged the bishops, so that the h-pawn moved to g6, the same continuation would follow: White doubles rooks on the f-file and binds the black pieces [by] Rf7 or Rf8, followed by doubling on the 8th rank. The decision in either case, in the end after suitable preparation – White’s pawn majority [is mobilized] by c4, cxb5 (?), and finally d5 then Kd4 brings it to glory[?] After that the passed e-pawn wins the game.

<This is more a sketch than a plan, so it is hard to evaluate…but I am quite skeptical. Fritz sees nothing too promising for White after 35….Raa7 36. Bg6 Bxg6 37. hxg6. Now if Black doubles rooks on his second rank, White doesn’t seem to be able to accomplish anything, even if he doubles rooks on the eighth. Look at the following position:

click for larger view

1…Kb6 2. c4 bxc4+ (of course) 3. Kxc4. Where does White go from here?

36. Rxg7 Rxg7 37. Rxg7 Ra2(?)

A fine move, and the answer is difficult to find. 38. Rh7 is weaker after …Rh2, when the h-pawn cannot be taken because of …Be2+. The obvious move Rg2 causes a draw, as Black trades rooks and then prevents the entry of the White king: 38. Rg2 Rxg2 39. Bxg2 Kc7 40. Ke3 Kd7 41. Kf4 Be8 42. Kg4 Be8 43. Kh4 Kf8. Now White cannot play Bf3 and Bh5, because Black after the bishop exchange wins with his passed pawn.

<Fritz finds an amazing hidden resource for Black here: Instead of 37…..Ra2, Black plays 37….Ra1, and if 38. Rh7 Rd1+ 39. Kc2 (39. Ke3? Re1+ 40. Kd3 Rxe4 ) 39….Rh1!!, since if 40. Bxh1 Bg6+. If 38. Rd7 as in the game, Black’s rook is active enough to maintain equality: 38….Rd1+ 39. Ke3 Re1+ 40. Kf4 Rf1+ 41. Kg3 Re1, and if 42. Bg2?! Re3+, while 42. Kf4 Rf1+ repeats.

On the other hand, after the move in the text it appears that Lasker is finally lost.>

click for larger view

38. Rd7(!)

The decisive move, which wins at least a pawn.

38….Be2+ 39. Ke3 Bc4 40. Rd6 Ra3

After 40….Bd5 41. Bxd5 followed by 42. Rxh6 wins easily.

41. Rxc6+ Ka7

click for larger view

42. Kf4

After I had played this very difficult game – [ver de vierte] – so well to this point, I missed the win here, a win however by a difficult and complicated combination, which I felt I did not have time to calculate precisely, and moreover it was not necessary to do so, since the game is still won simply with the text move. The very interesting and beautiful combination is as follows: 42. d5 exd5 43. e6 dxe4? 44. Rxc4 Kb6! (44….bxc4 45. e7) 45. e7 Ra8 46. Rxe4 Re8 47. Kf4 Kc7 48. Kf5 Kd7 49. Kf6 and wins. A better defense after 42. d5 exd5 43. e6 is 43….Ra1 (with the intention of stopping the passed pawn with …Re1+). Then White wins, however, by 44. Rxc4 bxc4! 45. Bf3, for if 45….Re1+ 46. Be2 follows, and after 45….Rg1 46. e7 Rg8 47. Bxh5, and if 45….d4+ winning is 46. Kxd4 Re1 47. Be4 Rd1+ 48. Ke5 Rd8 49. e7.

42….Rxc3 43. d5 exd5 44. Bxd5

Not at once 44. e6 because 44…dxe4 45. e7 Rf3+ 46. Kxc4 Rf1 followed by …Re1+.

44….Rc1

click for larger view

45. Bxc4?

But here I can win with the obvious 45. e6, for after 45….Rf1+ 46. Ke5 Bxd5 (46….Re1+ 47. Kd6 Rd1 48. Rxc4 or 48. Rc5) 47. Kxd5 the e-pawn must with the Black king (out of play) remaining, cost Black his rook. After 45. e6 Black cannot save the game by the exchange sacrifice 45….Bxd5 46. Rxc1 Bxe6. There follows first Rc6 and Rxh6; then White brings the king to c6, drives the black king off the a-file with a rook check to b8, and then plays his own king to b6. Black must then play …Kc8, whereupon White gives a check with the rook (from h7) on c7. If the black king then goes to d8, there follows Rc5 with an exchange sacrifice on b5, and the pawn queens. If the black king goes to b8, then White wins the bishop by Rc5 or else mates, for example 1. Rc5 Bf1 2. Rf1, or respectively 1….Be2 2. Re2 or 1….Bd3 2. Rd3, as in a well-known endgame study by Kling and Horwitz.

The text move provides Black with a passed pawn and new chances. It was the 45th move and I once again was short of time!

<Again, Fritz agrees that 45. e6 wins. >

45….bxc4 46. e6

In this position the arbiter, Mr. Otto Rosenfeld, the president of the Stuttgart Chess Club, thought by the pretty maneuver 46. Rxh6 c3 47. Rh2 c2 48. Rf2! to still force the win, but Black can bring his king to his own passed pawn and sacrifice his rook for the e-pawn: 48….Kb6 49. Kf5 Kb5 50. e6 Kxb4 51. Kf6 Kb3 and draws. The text move, with Ke3, retains some winning chances.

46….c3 47. Ke3 Kb7 48. b5 Re1+ 49. Kd3

Here the game was adjourned for the second time, with White sealing.

49….h5 50. Kxc3 h4

Mistaken was 50….Rc1+ 51. Kd4 Rxc6 52. bxc6+ Kxc6 53. Ke5 h4 54. Kf6 h3 55. e7 and wins, as the pawn would queen with check.

51. Rc4 Rxe6 52. Rxh4 Kb6 53. Kb4 Rg6 Draw.

My poor standing, 1-4, irritated me, but this game bothered me for a long time. That I could not win, after having played in exemplary fashion from the opening until almost the decisive moment -- indeed, from the 45th move (after so much work), any first-rate master would have won -- this vexed me, at least at first, and had a very depressing influence on me <einer niederdruckenden Einfluess aus>. I was dogged by misfortune: at the decisive moment, around the 45th move, I had to play quickly because of time pressure. A little earlier, at the 42nd move, I might have defended the c-pawn with Kd2, whereupon a rook-check would have surely followed, the king and rook would have repeated moves, and the game would have been adjourned. In the tranquility of my home, surely the decisive combination would have come with due deliberation. At that moment, I considered repeating moves, but I did not want to do it, precisely because it was the decisive moment, and I was not even in time pressure! Admirers <befruendeter Seite> reproached me for this severe sin of omission. It is a fact that, had I repeated moves, I would surely have won the game. And this reproach was not merely formal, but absolutely just, because, after a game of six hours' duration, I should have done it [repeated moves], as in my match with Marshall. But, as elsewhere, I must bow to my opponent, who did not permit the standings in the match to be put into question.

<Translating this passage was very difficult for me, and I am sure I made many mistakes. I include a link to the original at pp. 59-60 http://books.google.com/books?id=0C...

I would be grateful if <whiteshark> or <nescio> or anyone who really knows German would point out my errors.

Perhaps you are puzzled by the last word in <einer niederdruckenden Einfluess aus>: "aus" belongs to "übte" somewhat earlier in the sentence, and the whole is a conjugation of the verb "ausüben". In English the parts of a verb are never spread like this.

You wound me, Sir. :-) Of course Fritz is much stronger than I am, but I do not accept its conclusions blindly—see game 7 of this match for an example of me taking Dr. Tarrasch’s side against it. In every case, I have played over the games multiple times, and I generally try to come to my own conclusions before unleashing Fritz. For example, in this game, I spent a long time analyzing 21….Bxf3 22. gxf3. I evolved a slow plan of putting one white rook on g6 and keeping the other one on the d-file, maneuvering the white king to f4, and then playing c4 and d5. I gradually came to the conclusion that …Bxf3 was very bad for Black. But when I played over my plan with Fritz, it refuted it simply with …Rd5 and …c5, undermining the e-pawn. My idea of keeping f4 open for the king just didn’t work.

How trustworthy Fritz is depends on the position, of course. The more concrete calculation available, the better. At the ends of game 7 and game 11, when Fritz says White has a mate in so-many-moves, obviously that’s a totally reliable conclusion. A position like the one after 22. Rdg1 in this game is on its face less promising for calculation. But in fact the computer is able to find something to calculate: 22….c5! Positionally it’s a sensible, logical move—the kind of standard freeing move Black would like to play. If pawns are exchanged on d4 White is left with a backward pawn on an open file, and Black has the ideal d5 square for his pieces. If White plays dxc5, on the other hand, the d-file is opened and the e5 pawn is rendered a target. The only way dxc5 can be good for White is if he can keep the extra pawn or exchange it for another pawn (as in the lines after 28. bxa5). But as Fritz shows, White can’t accomplish either of those objectives after 22. dxc5.

I am the more willing to trust the computer when its judgment makes sense. When I look at the position after 22. Rdg1 (see diagram several posts above), if I forget everything I’ve read and put aside the fact that Lasker looked like a rank amateur for much of the opening, I don’t see much reason to think Black is in trouble. Yes, Black has a little less space, but there are just a few pieces left on the board. Yes, Black’s bishop is nominally worse, but in fact it’s got an enemy pawn (soon to be isolated) in its sights. Right now Black’s rooks don’t have targets, but they certainly will after …c5.

While there is plenty of interest in the game, what I found most fascinating was the discrepancy between the evaluations of the positions at various stages of the game by Lasker and Tarrasch as compared with that of Stockfish and Fritz. This key (for me) topic has already been broached by <keypusher> and <nescio> in their exchange on this site over 16 years ago. What I found astonishing were the number of times that one of both of the players concluded that White had a winning position while my silicon friends assure me that Black had a defensible position.

Who is right?

Stealing a page from <keypusher>, I think both Tarrasch/Lasker and Fritz/Stockfish are correct. My computer buddies are likely correct that with perfect (computer) play, many of the positions the players thought were lost for Black could in fact be held. But, Tarrasch and Lasker were also correct to the extent they recognized weaknesses in the Black game that should generally result--in a game between even the best of humans--in a win for White.

As always in my comments on this match, I have been guided by the superb commentary by <keypusher>. I have also--here and in other games in this match--generally (but not always) used the translations of Tarrasch's commentary so helpfully provided by <keypusher>.

Another human element to consider in playing over this game is the clock. As I will discuss, absent time pressure, Tarrasch would almost certainly have won the ending.

Anyway, here is the game with my comments.

1. e4 e6

The first five games of the match had all featured the Ruy Lopez. Lasker had won both his games as Black (Games 2 and 4) in that opening. But now he mysteriously shifted to the French Defense. Why?

Most probably Lasker thought to surprise Tarrasch for this one game. With a three-game lead in the match, Lasker could afford to mess around and try an opening in which--as the course of the match would soon reveal--neither Lasker nor Tarrasch felt comfortable as Black.

After getting outplayed in the opening in this game, Lasker returned to 1...e5 for the balance of the match. But in those games (Games 8, 10, 12, 14, and 16, Lasker varied from his play in Games 2 and 4, and Tarrasch managed to obtain a plus score in those encounter, winning twice and drawing twice before falling apart in the 16th and final game of the match.

Interestingly, and even after Lasker's debacle in the opening here, Tarrasch strangely decided to employ the French Defense in the remaining three games in which he faced 1. e4 by Lasker. This did not go at all well for Tarrasch, who managed only one draw against two losses.

Al in all, the use of the French Defense did not go very well for Black in this match.

2. d4 d5
3. Nc3 dxe4

The Rubinstein Variation, which yields strong center play to White. This is, as stated by Gligoric in his book on the French Defense, the main reason 3...dxe4 "is rarely seen in modern master chess." The drawbacks to the text must have been obvious to Lasker, this my hunch that Lasker was playing with Tarrasch's head in selecting his opening play here. By contrast, when Tarrasch played the French in Games 7, 9, and 11 of the match, he employed instead the MacCutcheon Variation which--whatever its overall merit--does give Black more counter-play.

4. Nxe4 Nf6

Playable, but 4...Nd7 is more usual and generally considered stronger. Once again, one gets the impression that Lasker was playing what he must have known was a difficult opening, effectively saying to Tarrasch that he could prevail with just about any variation that wasn't losing.

5. NxN+

This gives White a clear--though nowhere near winning--advantage:

click for larger view

5... QxN

5...gxN, though not without problems for Black, is--as noted by Gligoric--is probably superior. Lasker, however, seems content to carry on in an inferior position but without messing up his pawn structure.

6. Nf3

click for larger view

The text obviously threatens 7. Bg5, and thus 6. h6 (as recommended by Tarrasch) appears indicated. But Lasker (who must have known what was coming if he failed to play h6, had other ideas:

6... Bd7?

"A singular lapse on the part of Dr. Lasker to neglect the precautionary 6...h6." (Hoffer)

Not surprisingly, I can find no other game in which 6...Bd7 was played.

The text sure looks wretched, but my silicon buddies assure me it is playable. Apparently computers can see their way to defend the resulting horror show of a position for Black.

7. Bg5

This definitely looks like serious trouble for Black for us mere mortals. But Fritz and Stockfish like 7. Bd3 even more. As Tarrasch pointed out, after 7. Bd3, Black has to play 7...h6 (or maybe 7...g6--another ugly move) to avoid losing his Queen (e.g., 7...Nc6? 8. Bg5. But the silicon monsters think Black can survive fairly well in either case.

In any case, the text left:

click for larger view

7... Qg6

Plainly better than 7...Qf5.

8. Bd3

White's game seems to play itself.

click for larger view

Lasker surely must have foreseen this position when he played 6...Bd7. But my guess is that he had the following wild move in mind, perhaps in the hope of rattling his hyper-classical opponent:

8... f5?!

"?"--(Tarrasch)

"No doubt a weakening move, but compulsory." (Hoffer)

Nonsense. As Tarrasch pointed out (see notes here by <keypusher>), best for Black is 8...Qh5. The text, however, though seemingly zany, is--according to Fritz at least--playable.

Maybe so, but the backward e-pawn is now more than an eyesore. Maybe I'm reading too much into this, but it is hard for me not to suspect (since all of my misgivings about 8...f5 must have been obvious to Lasker) that Lasker was toying with his famous opponent. If he could draw (or even win) after the text, what would be left of Tarrasch!

9. h4

"!"--(Tarrasch)

"Not only to protect the Bishop (and so threatening to win the Queen by Ne5), but also to drive the Queen away from attacking g2 and to create a strong forward post on h5.

White has another idea here: 9. 0-0 with perhaps c4 to follow. In either case, White has much the best of the position, though Fritz still thinks Black can hold (Stockfish is not so sure).

In any case, after 9. h4, the position was:

click for larger view

Thus, even computers finally recognize how questionable Lasker's 8...f5 (and indeed his 6...Bd7 for that matter) were.

To the extent I was using this game to exemplify the differences between human and computer evaluations, I shall therefore have to rely on what follows in this game, though to my human mind White's edge (with Black's backward e-pawn and the desperate plight of his Queen) looks a good deal more profound than the less than a pawn advantage Fritz reveals even on a very deep search.

9... Nc6

Practically forced. After the text, Tarrasch had to decide how to proceed in order to exploit his advantage.

10. Qe2

This move, as noted by Hoffer, prepares to castle long and threatens d5. But while there was more than a bit of logic to support Tarrasch's choice, there were clearly more forceful ways to attempt to break down Black's defenses. 10. h5 and 10. Bb5 would have been very strong,

10. d5 immediately would allow Black to solve most of his problems after 10...exd5 11. 0-0 Bd6. White could of course now thwart Black's notion of castling Queen-side with 12. Re1+, but Black can probably then hold his own with 12...Kf8 and then if 13. c4 Black is at worst slightly inferior with either 13...h6 or 13...d4.

10. g4, as Lasker pointed out, goes nowhere after 10...h6 11. gxf5 and now not Lasker's suggested 11...Qf7 which runs into 12. Qe2! since 12...hxB 13. d5 (and even then Black is not so bad) but simply 11...exf5 12. Qe2+ Qe6 after which Black can castle long in a few moves with no real disadvantage.

Tarrasch, in his commentary, considered 10. Be2 (with the idea of winning the Queen) because Black survives just fine with 10...h6 after which White has nothing better than 11. Bf4 because Tarrasch's notion of 11, Ne5 (?) simply loses after 11...NxN 12. Bh5 (winning the Black Queen at ruinous cost) hxB 13. BxQ+ NxB 14. Qd2 gxh4leaving Black with three minor pieces (including two Bishops) plus an extra pawn for the lost Queen.

After 10. Qe2, the position was:

click for larger view

10... h6

This, as Tarrasch pointed out, creates a new weakness for Black (the backward g-pawn in addition to the already weak e-pawn). But, as Tarrasch also noted, the alternative: 10...Be7 was considerably worse for Black: 11. BxB NxB 12. Ne5 Qh6 (much better than Tarrasch's hazardous 12...Qxg2 ?! 13. 0-0-0 leaving Black in desperate straits.

After 10...h6, Tarrasch was again at the cross-roads:

click for larger view

11. Bf4

A prudent choice by Tarrasch. 11. h5? would have been awful for White after 11...hxB 12. hxQ RxR+ 13. Kd2 RxR after which White' attack via 14. d5 cannot compensate for the significant loss of material.

An interesting try, discussed by Tarrasch, Hoffer, and other commentators was 11. d5, but Black appears to survive in all variations: 11...hxB 12. dxN Bxc6 13. Ne5 Qf6 and now White should try Hoffer's 14. 0-0-0 (instead of Tarrach's suggested 14. Bb5 which seems to lose straightaway after 14...BxB 15. QxB+ c6 after which White has nothing better than 16. Qe2 since Tarrasch's 16. Nxc6 leads to catastrophe via 16...bxN 17. Qxc6+ Kf7 since now 18. QxR+ loses the Queen to 18...Bb4+ as pointed out by Tarrasch himself) with a small edge for White. (Actually, Hoffer's line after 14. 0-0-0 is entirely misguided and not worth the space to demonstrate its flaws, but that need not call in question the superiority of 14. 0-0-0.

11... 0-0-0

click for larger view

"Black is now completely safe." (Hoffer)

This is surely an overstatement, since Black (Lasker) still had problems with which to contend. But Black certainly seemed to have avoided catastrophe. But the true excitement had yet to occur, and Tarrasch had only begun his efforts to attempt to dismantle his hated rival.

12. 0-0-0 Bd6

click for larger view

13. Be5

"!"--(Tarrasch)

Tarrasch's enthusiasm for the text notwithstanding, leaving the g-pawn undefended could (as <keypushed has noted>) have allowed Lasker "to equalize comfortably: with 13...NxB 14. NxN BxN since Black can play 15...Qxg2 after 15. QxB and if instead 15. dxB Black just plays 15...Bc6.

The simplest--and I think best--line was 13. g3 leaving Black with all his weaknesses while White can build up his edge with 14. c4 and 15. Rde1.

Much less satisfactory is 13. BxB (which <keypusher> says Fritz prefers, but White then gets little, e.g., 13...cxB 14. h5 Qf6 15. Bb5 Rhe8 (as given by <keypusher>).

Tarrasch's move (13. Be5) left:

click for larger view

13... Rhe8

As noted above, Lasker could have equalized with 13...NxB. The text gave Tarrasch chances:

14. h5!

"Causes the complete crippling of the [Black] king-side and prepares the later attack on the g-file.

14... Qf7
15. c3

"To prevent ...Nb4." (Tarrasch)

click for larger view

White is undoubtedly better here. But Fritz and Stockfish report only a small edge; not the advantage Tarrasch (like most humans) perceive. All I can say to my silicone friends here is that, whatever the theoretical prospects, I would not enjoy trying to defend the Black position against so superb a strategist as Tarrasch. <keypusher> has made a similar point here.

15... NxB

"...Black looks to free his game by exchanges."

16. NxN BxN
17. QxB Qf6

click for larger view

18. f4

"!"--(Schroeder)

"There is no necessity to keep the King's file open, as he can direct his attention to Black's weak g-pawn." (Hoffer)

18... QxQ

Schroeder opined that this led to a winning endgame for White, but that declining this exchange would have allowed White to pursue "his choice of good plans."

So far as Fritz and Stockfish are concerned, either the text or 18...Rg8 are far from terrible for Black.

19. fxQ

click for larger view

19... Re7

"Providing at once a defence of the g-pawn--the fresh weakness--and a possible hold on White's h-pawn with Be8." (Hoffer)

Fritz and Stockfish recommend 19...Bc6 (which looks logical and probably superior to the text, Hoffer notwithstanding), but Tarrasch thought otherwise, suggesting that Black would then face difficulties after 20. Be2, since 20...Bxg2 would run into 21. Rhg1. That would indeed be bad for Black, but Lasker--never needlessly greedy--would likely have played the more sensible 20...Bd5 with decent chances to hold.

20. Be2 Bc6
21. Bf3

click for larger view

21... Be8

Tarrasch claimed that 21...BxB would be a mistake since it would open the g-file for White. The double-Rook ending does then seem bad for Black--but not clearly losing.

But <keypusher> found a much better idea for Black: 21...Bd5 after which he states that White can make no progress. But even then, White has prospect. Not 22. BxB which, as <keypusher> notes would be close to a "dead draw,: but instead a slow deliberate advance with 22. Kc2 after which the game might proceed: 22...a6 [or 22. c6 and now 23. BxB RxB 24. g4] 23. Rh3! b5 24. Rg3 and suddenly Black has to beware incursion into his position.

22. Rdg1

22. Rh4 with notions of rf4; g4, etc. might have been somewhat more promising.

"White's plan is clear; [and] Black cannot prevent it. Black cannot change the position in any fundamental way." (Tarrasch)

This all overlooks the idea of <keypusher> beginning with 22... c5! <keypusher> perhaps goes too far in claiming that Black could then "achieve complete equality." But 22...c5 does give Black counter-play which seems to be enough to save the day.

22... c6

Once again, Lasker missed a chance to play the much better 22...c5! here.

After the text the position was:

click for larger view

23. Rh2

"This move is not quite intelligible. Perhaps the intention is to give this Rook a wider range after the advance of the g and b pawns." (Hoffer)

The problem with 23. Rh2 is not that it is unintelligible, but that White had a much stronger option: 23. g4! and White now has pressure on several fronts.

23... Kc7

Although I have seen no criticism of this move, it in fact was the beginning of a poor plan by Lasker of going after the Queen's-side by advancing his King. Far better was 23...c5, the move Lasker (for reasons I cannot fathom) continued to avoid. But Tarrasch on the next move provided some clue as to his thoughts on the matter:

24. Kc2

"The King cannot cross the d-file, because Black would then have a chance to play the c5 freeing move." (Tarrasch)

In fact, Lasker had that very opportunity over the course of his last few moves. Meanwhile, Tarrasch continued to miss his most promising way to attempt to break through the Black defenses: 24. g4.

Tarrasch's actual move left:

click for larger view

24... Kb6?

Tarrasch immediately exploited this lapse by Lasker, who himself later said (correctly in my view) that he here should have played 24...a5. Tarrasch did not give him another chance:

25. b4!

"!"--(Tarrasch)

"Prevents the advance of the [Black] c-pawn." (Tarrasch)

"With this move White begins to get a real advantage." (<keypusher>).

I concur. Lasker was now, though not necessarily lost, getting in hot water:

click for larger view

25... Kc7?

With Tarrasch strangling on both sides of the board, Lasker was under great pressure. If, as he acknowledges, he should have played 25...a5 on his last move, he surely should have considered that move here. No doubt his position would still have been difficult after 25...a5, but the text--a retreat that was a tacit admission that he had gone awry--did nothing to heal his problems.

Now Tarrasch opened a front on the other wing with:

26. g4

click for larger view

26... fxg4

A painful necessity.

27. Rxg4

Hoffer claimed that 27. Bxg4 would have been better, but his analysis is unconvincing. Hoffer was quite correct that: (a) the (to my mind) unthinkable 27...Rf7?? would get crushed by 28. Bxe6; (b) that the doubtbul 27...g5? would lose after 28. Rf1;and (c) that the awful looking 27...g6? would be bad after 28. Kb3 [an initial move missed by Hoffer] and then 29. Rf1 But the two other two; only one of which was discussed by Hoffer in response to 27. Bxg4 tell a different story: (d) if 27...a5 then 28. a3--Hoffer notwithstanding--leaves Black various options to stay in the game: 28...b6 or 28...a4. And 27...b6 (a move not considered in Hoffer's analysis), appears to stabilize the Queen-side for Black and give him hope of being able to hold on.

27... a5?

"...a big mistake." (<keypusher>).

Tarrasch thought the text was good:

"Black still seeks to open the a-file so that his Rook can come into play. Of course, White does not consider taking the a-pawn."

But a look at the position shows that it is <keypusher> and not Tarrasch who understands the situation on the Queens-side, since Lasker's 27...a5? left:

click for larger view

28. a3?

"?"--(<keypusher>)

As only <keypusher> has recognized, Tarrasch had a likely win here with 29. bxa5. Black would then have had nothing better than 28...Kb8, a sign of just how poor Black's position would now be. The two moves for Black <keypusher> discusses would have been hopeless: (i) If no 29...Ra8 then 30. Rhg2! Rxa5 31. Rxg7 would be gruesome for Black; and if instead (ii) 29...c5 then White pounces with 30. Rhg2 (much better than 30. dxc5 as given by <keypusher>.

28... axb4
29. axb4

click for larger view

29... Ra8

The players were approaching the move-30 time control, and Lasker faltered.

29...b5 was much better, and perhaps the only true hope for Black to save the say.

30. Kd3

Tarrasch was in no hurry.

30... Rd8

Apparently changing his mind from his last move.

Black's position now seemed ripe for the taking:

click for larger view

31. Be4

"The Bishop aims at g6." (Tarrasch)

31... b5

"To assist in the later attack on the White pawns by making the White c-pawn backwards...But this move is also the downfall of Black's c-pawn." (Tarrasch)

click for larger view

32. Ke3

"The King moves seem ever so much waste. He could play 32. Bg6. If 32...BxB+ 33. RxB Rf8 34. Rhg2 Rf7 35. Ke3 followed by Rf2, and after exchanging one Rook Black would have to move his King, when White could break through with the Panws." (Hoffer)

I don't much understand this comment. While Hoffer's line seems to give White good winning chances, it was no improvement on Tarrasch's play. Tarrasch's true problems began with his next move.

32... Kb6

click for larger view

Here, according to Tarrasch, the game was adjourned. Had Tarrasch sealed 33. Rhg2, Lasker would have been hard pressed to survive: e.g., 33...Rdd7 34. Bg6 BxB 35. hxB after which White could go to work on the f-file and penetrate. But Tarrasch sealed an inferior move:

33. Rg3?

Now, suddenly, Lasker got counter-play:

33... Ra8!
34. Rhg2?

34. Bg6 was clearly stronger. Lasker now had very real chances to hold the game:

click for larger view

34... Ra3

Though not mentioned by any of the commentators, Lasker could have simply played 34...Bxh5 here: 35. Rxg7 RxR 36. RxR Ra1 with excellent drawing chances. The text was probably not a losing move, but now Tarrasch would have been better placed with 35. Rxg7; e.g., 35. ..RxR 36. RxR Bxh5 37. Kf4. But Tarrasch missed this chance and played:

35. Kd3

Now came one of the critical moments in the game, the position now being:

click for larger view

35... Bxh5!

"?"--(Tarrasch)

"I saw how I could lose surely (35...Raa7 36. Bg6); I therefore adopted a way of losing quickly with a bare possibility of drawing." (Lasker)

While <keypusher> claims otherwise, I agree with Tarrasch and Lasker here that 35...Raa7 36. Bg6 would be bad for Black.

The text looks plainly best, and--Lasker notwithstanding--need not have led to defeat with best play (though <keypusher> contends that it "brings Black to the brink of defeat"). In fairness to <keypusher>, he alone found the way Lasker could now have saved the game.

36. Rxg7 RxR
37. RxR

click for larger view

Matters now do seem desperate for Black. But, as <keypusher> reports, Fritz here found "an amazing hidden resource for Black." That resource is the remarkable 37...Ra1. This resource (see the analysis here by <keypusher>) demonstrates the differing views of the position by humans and computers. A human thinks Black is lost, but Fritz and Stockfish know better.

37... Ra2

Not surprisingly, Lasker over the board missed the subtle line our silicon friends zap out in a nanosecond. After the text, as <keypusher> says, "it appears that Lasker is finally lost." With this, Fritz and Stockfish agree.

38. Rd7

"!"--"The decisive move that wins at least a pawn." (Tarrasch)

click for larger view

38... Be2+

The beginning of a clumsy-looking plan by Lasker; but I see nothing better.

39. Ke3 Bc4
40. Rd6

click for larger view

The position towards which Tarrasch was undoubtedly aiming. It appears to win easily.

40... Ra3

If instead 40...Bd5, then--as Tarrasch pointed out--White wins "easily" with 41. BxB exB 42. Rxh6, leaving White a protected passed pawn to the good; not to mention Black's backward c-pawn.

Lasker's move offered the best chance to create complications in an otherwise simple endgame that one might expect Tarrasch to be able to win in his sleep.

41. Rxc6+ Ka7

click for larger view

Tarrasch's one problem here was the approaching move-45 time control.

42. Kf4

"?"--(Schroeder)

"...incomprehensible"--(Hoffer)

As Tarrasch, Lasker, and everybody else who commented on this game have shown, Black gets crushed after 42. d5! See the excellent commentary here by <keypusher> for details.

Hoffer says that 42. d5 "would have won without any difficulty whatsoever." But this all required time to "calculate precisely" (Tarrasch) which apparently Tarrasch did not have. Since the text also wins, though not as rapidly, Tarrasch's move had been unfairly trashed.

Given the time-trouble he was experiencing, Tarrasch's later suggestion of 42. Kd2 Ra2+ 43. Ke3 to save time would have been prudent. (But his further suggestion of continuing 43...Ra3 44. Kd2 Ra2+ would have forced him to find a square other than e3 for his King on his 45th turn to avoid a draw by triple repetition. In this line, after 43...Ra3 Tarrasch--if he still didn't fancy 44. d5; could have resorted to 44. Kf4, getting him closer to the safety of move 45.

42... Rxc3
43. d5

Now no deep calculation was required for Tarrasch to play this still winning move.

43... exd5
44. Bxd5 Rc1

click for larger view

Tarrasch now only had to make one more move to reach the time-control.

The killer here--as Tarrasch himself pointed out after the game, was 45. e6! If then 45..,Rf1+ then White romps with 46. Ke5 Re1+ [or 46...BxB 47. KxB with an easy win] 47. Kd6 BxB 48. KxB and the game would be over. Lasker might therefore have tried an exchange sacrifice with 45...BxB 46. RxR Bxd6, but then 47. Rc6 or 47. Rh1 should settle matters shortly.

Even if Tarrasch was too short of time to work any of this out, he could still have prevailed with the slow but relentless 45. Kf5.

But, with the clock ticking, Tarrasch got sloppy and threw away the win with:

45. BxB?

click for larger view

45... bxc4

"Provides Black with a passed pawn and new chances"--(Tarrasch)

click for larger view

46. e6

After the text, Lasker drew with ease. The only real question is whether Tarrasch would have done better with 46. Rxh6. The short answer is that, even as a practical rather than a theoretical matter, the Black c-pawn saves the day for Black:, e.g., 46. Rxh6 c3 47. Rh2 (the suggested line) leads to an easy draw for Black after 47...c2.

A more interesting line is 47. Ke3 Re1+ 48. Kd3 [not 48. Kf4??? Rf1+ (48...c2 also wins) followed by 49...c2 and now Black actually wins] 48. Rxe5 49. Kxc3 Re1 after which I see no way for White to make progress.

Back to the actual game:

46... c3

click for larger view

47. Ke3 Kb7
48. b5 Re1+

click for larger view

Here, according to Hoffer, the game was adjourned a second time (Tarrasch said it was after White's 49th move). In any case, and again according to Hoffer, Dr. Lasker then returned to the game with a "cut and dried" drawing variation.

49. Kd3 h5
50. Kxc3

click for larger view

50... h4!

The only way to draw. If instead 50...Rc1+, then, as noted by Tarrasch, White could win with 51. Kd4 RxR [if instead 51...Re1 White wins with 52. Kd5 h4 53. Kd6 h3 (53...Rd1+ is no better) 54. Ke7] 52. bxR+ Kxc6 53. Ke5! [the only winning move] h4 54. Kf6 [again, the only winning move] h3 55. e7 (yet again, the only way to win) since the White pawn now Queens with check.

All of this shows that even at this late moment is was still quite possible for Lasker to blow the win.

After 50...h4, the rest was simple:

51. Rc4 Rxe6
52. Rxh4 Kb6

Leaving a book draw:

click for larger view

53. Kb4 Rg6

Drawn