Learn from the Legends

Chess Champions at their Best

Mihail Marin
Quality Chess
www.qualitychessbooks.com
Contents
Bibliography 4
Forewords 5

1 Akiba Rubinstein’s Rook Endings 9

2 Alexander Alekhine and the Fourth Phase of the Game 69
3 In the Patriarch’s Footsteps 95
4 Tal’s Super Rooks vs. Two Minor Pieces 123
5 Petrosian’s Exchange Sacrifices 149
6 Bobby Fischer’s Pet Bishop 199
7 Karpov’s Opposite Coloured Bishops Endings 233
8 Viktor, the “Non-Existent” Hero 263

Biographies of Our Heroes 317

Index of Games 330


Akiba Rubinstein’s Rook Endings
 Chapter One: Akiba Rubinstein’s Rook Endings 11

When, at the age of 18, it became clear that I had no other choice but to join the glorious Romanian
army, some older friends warned me that the main thing I had to avoid was dying of boredom.
As a future student of the Polytechnic Institute I had been assigned to a technical division, which
meant that I would spend most of the daytime in a classroom where nothing special was going to
happen. (Indeed, most of my colleagues-in-arms used this time to sleep, with their heads on the tables. A
few others would chat in low voices while even fewer would write letters to their beloved young ladies).
Following my friends’ advice, I decided that I would spend the time available learning the Russian
language; even then I already had a good collection of Soviet chess books, but could not take full
advantage of it, since the linguistic barrier was rather difficult to overcome. I took with me a pocket
dictionary and a carefully selected book (the criteria were: it had to be not too thick, in order to be easy
to carry and hide; to be printed in hard cover in order to survive possible accidents; finally, it could not be
one of the best books in my library, to avoid endless regrets in case of deterioration, loss, or confiscation
by one of my superiors).
The main plan ended in total success: nine months later when, much to my relief, I became a civilian
again, I felt able to read and understand most of my Russian chess books. And yet, there was something
that I had not foreseen: the “carefully selected book” which I had carried so many times from the
dormitory to the classroom and back, hidden under the military robe, very close to my heart, had become
my chess Bible. A book I would open again and again to see for the nth time one game or another.
Some time ago, wishing to remember the good old days, I opened again that book: Akiba Rubinstein,
written by Razuvaev and Murakhvery, and containing a biography and selected games of my classic idol.
I was immediately struck by the same old feelings, but I also became curious: would the book stand up
to the analysis of a more mature and critical eye?
Much to my disappointment, it did not. I discovered that Razuvaev’s analysis contained countless
mistakes or omissions and very little original work. But when I admitted this as a fact, I realized it had
nothing to do with Rubinstein’s games: they provided me with the same feeling of clarity, fluency and
logic as 20 years ago.
I took it as my duty to publish my own commentaries on some of his games, and the obvious theme
was his rook endgames: a great Akiba specialty.
However, this is a book dedicated mainly to World Champions. Therefore, the reader might wonder
why it opens with a chapter about the games of a player who did not even play a match for the supreme
title? I shall try to explain that there is more to my decision than subjective memories.
Until Alekhine’s death in 1946, the World Champion had the personal right to choose his challenger
for the title.

Was Akiba ever the best player?

Akiba reached his peak of form in the years preceding the First World War. In 1909 he obtained one of
his best results ever, sharing first place with Lasker in St Petersburg, 3½ points ahead of the field. This
was the moment when public opinion started considering Rubinstein as the most likely challenger for
the title.
However, his magical year was 1912, when he won every single strong tournament he played in.
These were long tournaments, with about 20 rounds to be played. Therefore, proving such absolute
superiority was far from easy and, in fact, had no precedent since Morphy. Akiba’s performance should
be put at the same level as Kasparov’s domination in the tournaments played around the turn of the
millennium.
Much to Akiba’s misfortune, the reigning champion in the pre-war years was Emanuel Lasker. Besides
being a very strong player, Lasker knew how to take full advantage of the right to choose his opponent.
12 Learn from the Legends – Chess Champions at their Best

For instance, he practically forced Schlechter to play a match under scandalously unfair conditions
(although it should be mentioned that even so he only retained his title with luck). Little wonder that
he reigned for more than a quarter of a century.
Although there is not much historical evidence about it, we can suppose that Lasker also did his best
to delay the seemingly inevitable match against Akiba. After long negotiations, the match was scheduled
for October 1914 but then the war came and chess life was practically frozen. Akiba’s nerves seem to
have been seriously affected by the conflagration and, although he remained a fearsome opponent until
his last important tournament (Prague Olympiad 1931), he was never the same player again. As a
consequence he ceased to be, according to public opinion, a plausible challenger.
However, since during one specific period Akiba was clearly the strongest player in the world, I feel
entitled to include his masterpieces in this book.
Rubinstein’s name is closely linked with the main lines of such openings as the Nimzo-Indian, the
Queen’s Indian and the Tarrasch Defence. He invented several set-ups for Black that are still topical in
the French Defence and the Ruy Lopez. He was also the first to play the modern Meran variation of
the Semi-Slav defence.

Opening expert or endgame virtuoso?

Why is it then that I have focused on the endgames played by such a great opening expert?
Rubinstein was probably the first great player to use the so-called long plans, lasting through the
whole game, from the opening till the endgame. His opponents didn’t always understand Akiba’s
monumental play and were only concerned with parrying the immediate threats. Only after the game
was over, did it become clear that Rubinstein had planned the contours of the ensuing endgame from
a very early stage.
You will notice that more than once I have started analysing a game at an earlier moment than the
endgame itself. There is also a complete game in this chapter. The reason for doing this is to illustrate
the idea of the “long plan”.
I have divided the material in accordance with the number of rooks present on board. As will soon
become evident, this is more than a formal classification. Pure rook endings (one rook each) tend to be
rather technical. Four-rook positions present far more tactical possibilities, although naturally they can
also transpose to a pure rook ending.
Rook endgames are not only the most common in practice, but also the most difficult to analyse. The
rook is such a strong and mobile piece that it allows countless tactical possibilities and makes over-the-
board calculation especially difficult. I have tried to point out in my annotations the moment when one
of the players missed a win or a draw, but the probability of mistakes in my comments is quite great,
precisely because of these difficulties.
Rook endgames are well known for their drawish tendency. A material advantage of one or, in some
extreme cases, two pawns is frequently difficult or even impossible to convert into a win. This does not
really mean that Tartakower’s axiom “all rook endings are drawn” should be taken literally, however.
The technical purpose of this chapter is to highlight those elements that mark a clear advantage for one
of the players, with an elevated probability of resulting in a win. (I made the specification “technical”
because another purpose, this time of a sentimental nature, is to pay tribute to Akiba).
After a thorough study of Rubinstein’s games, I have developed a rough method for preliminary
evaluation of rook endings. For each favourable element such as an extra pawn, spatial advantage, the
more active rook or an important weakness in the enemy camp, one point is awarded. If the difference
between the two sides is one point (scores such as 1–0, or 2–1) the position is clearly better but not
necessarily winning. This marks a significant difference compared with pawn, same-colour bishop, or
knight endgames where the score 1–0 (equivalent to an extra pawn) is usually sufficient for a win.
 Chapter One: Akiba Rubinstein’s Rook Endings 13

With the exception of some extreme cases (for An ideal situation for the stronger side is to have a
instance, the ending with f- and h-pawns) a score minimal material advantage and some other kind
of 2–0 or 3–1 should guarantee a win. of positional advantage. This is a specific example
The reader should understand that this system of the score 2–0.
is only a guideline. It can help the practical player Lasker had sacrificed (or rather lost) a pawn in
(and definitely helped me throughout the years) the opening, but his position looks rather active.
to choose one or another ending when simplifying In view of the threat …¦xe3, it would seem that
from the middlegame, but should not be treated he would have no problem regaining his pawn.
as an infallible rule. Some concrete advantage Unless...
might weight much more than just one point, 16.¦c1!
or on the contrary, have no real significance. We Akiba simply continues his development,
should take into account that a greater number cutting off most of the black attack’s energy.
of pawns present on the board would usually 16...¦xe3
increase the winning chances, while a simplified The prophylactic 16...¢b8 would give White an
position would normally help the defending side. important tempo to develop his initiative: 17.¦c5
This is an element that cannot be easily quantified. £f4 18.d5 ¦xe3 19.£c1! Just like in the game,
Besides, if chess were pure mathematics, it would this elegant move puts an end to Black’s hopes for
be much less interesting. counterplay. 19...¦e4 20.dxc6 bxc6 21.£c3 with
At the same time, if a rook ending is winning it a clear advantage for White, according to Lasker.
does not necessarily mean that the player with an 17.¦xc6† bxc6 18.£c1!
advantage will automatically win it. We shall see It is worth mentioning that with the same
from the selected examples that good, sometimes intermediate move Rubinstein would later defeat
even intricate, technique is needed. another (this time, future) World Champion:
We shall make a further division between the
situations where the stronger side has a material Akiba Rubinstein – Jose Capablanca
advantage and those where the superiority is only San Sebastian 1911
of a positional nature.

Converting a material advantage I   
Akiba Rubinstein – Emanuel Lasker   
St Petersburg 1909   
    
       
     
      
       
     
     Black has built up strong pressure against the
f2-pawn. In case of the timid 15.e3 he would
   consolidate his centre with 15...¦ad8, obtaining
   a perfectly viable position. Noticing the slightly
hanging position of the black bishop and the

14 Learn from the Legends – Chess Champions at their Best

c8-rook, Rubinstein initiated a thematic tactical 

operation.
15.¤xd5!? £h6?    
Capablanca tries to solve by simple means a   
position that is basically quite complicated.
Obviously, 15...exd5? loses material to 16.£xd5†    
¢h8 17.¥xc8.
Recently, a Russian amateur chess player named
    
Sorokhtin discovered that 15...¥xf2† would have     
allowed Black to stay in the game, for instance
16.¢g2 £e5!. This is Sorokhtin’s improvement
    
over the variation 16...£f7? 17.¤f4 given by   
Kasparov. For instance 17.¦xf2 ¦xf2† 18.¢xf2
¦d8 19.¤e7† ¢h8 20.£b3 ¤xe7 21.£xe6 £d4†    
22.¢g2 ¤d5 when the weakness of the white
kingside as well as better piece coordination offers

18...¦xd4
Black a reasonable game. 18...¦e5 19.£xc6† ¢b8 20.dxe5 £xe5 21.¦c1
16.¢g2 ¦cd8 also looks winning for White due to Black’s
This was the move Capa relied on. 16...¥xf2 weakened king’s position.
would be less efficient now because of 17.¤f4 for 19.fxe3 ¦d7
instance 17...¦cd8 18.£a4 when White will win 19...¦d6 20.¦xf7
the e6-pawn. 20.£xc6† ¢d8
17.£c1!
An elegant multi-purpose move. The queen 
escapes the unpleasant pin along the d-file,
attacking the c5-bishop at the same time. By
    
offering to be exchanged for the black queen,  
it also undermines the e6-square, making
17...¦xd5 impossible.    
17...exd5     
Black would lose quickly after 17...£xc1?
18.¥xe6† ¢h8 19.¦axc1, or 17...¦xd5? 18.£xh6     
gxh6 19.¥xe6† with a huge material advantage
for White in both cases.
    
18.£xc5 £d2 19.£b5 ¤d4 20.£d3 £xd3   
21.exd3
with a safe pawn up for White, who went on
   
to win the game. 
1–0 21.¦f4!
“A remarkable concept. White threatens to
decide the game with a direct attack against the
king: 22.£a8† ¢e7 (22...¢c7 23.¦c4†! also looks
bad) 23.¦e4†. In order to avoid this, Black has
to exchange queens, entering a lost endgame.”
(Lasker). By means of this remarkable rook lift,
Let us now return to Rubinstein – Lasker. White defends the e3-pawn and gains access to
(Position after 18.£c1) the e-line and to the queenside.
 Chapter One: Akiba Rubinstein’s Rook Endings 15

21...f5 a-pawn to a6 and then transfer the rook to b7.

Relatively best. Black takes the e4-square under This is, however, not necessary for the moment:
control. He would lose after 21...£a5 22.£a8† White should first strengthen his position on
¢e7 23.¦e4† ¢f6 24.£c6† ¢g5 25.h4†. The the other side. The only thing left for the World
counterattack 21...¦d1† 22.¢f2 ¦d2† 23.¢e1 Champion was to wait for the execution.
£xg2 would fail to 24.¦d4†! (overloading the 28...¢f8 29.e4 ¦c7 30.h4 ¢f7 31.g4 ¢f8
rook) 24...¢e7 25.£d6† and White wins. Both 32.¢f4
lines were indicated by Lasker. The attack on the kingside is massive. There
22.£c5 £e7 would be little sense in keeping one of the pawns
Now, 22...¦d1† 23.¢f2 ¦d2† 24.¢e1 £xg2 back, since it could be attacked at a later time.
would simply lose the rook to 25.£a5†. White is not in any hurry, since Black has only
23.£xe7† waiting moves at his disposal.
Actually Fritz considers 23.£c3 to be equally 32...¢e7 33.h5
strong, but that move would lead us to the next
chapter.
1222222223
23...¢xe7 24.¦xf5 ¦d1† 25.¢f2 4 + + + +5
Too passive would have been 25.¦f1 ¦d2
26.¦b1 ¦e2 with drawing chances for Black. 4O T L Oo5
25...¦d2† 26.¢f3 ¦xb2 4r+ + + +5
 4+ + + +p5
     4 + +pKp+5
    4+ + + + 5
     4p+ + + +5
    4+ + + + 5
     7888888889
    33...h6
This move weakens the g6-square, but letting
   White advance all his pawns to the fifth rank was
     also a bit scary. In his old endgame book Lisitzin
gives a nice winning method (I am sure it is not
 the only one): 33...¢f7 34.¢f5 ¢e7 35.g5 ¢f7
27.¦a5! ¦b7 36.e5 ¢e7 37.g6 h6 38.¦e6† now Black faces a
An important moment. Besides his extra pawn, difficult choice:
White has the more active rook. Black has nothing a) 38...¢d7 would allow the incredible
to compensate for White’s trumps. Therefore, we 39.¦f6!!. A remarkable example of how a space
can safely attach to the position the score of 2–0. advantage in an apparently blocked position can
White’s win will require some accuracy but will allow tactical solutions based on modification
never be put in any doubt. of the pawn structure. It happens all the time
28.¦a6 in modern lines of the King’s Indian. 39...¢e8
A typical method in rook endgames. With the (After 39...gxf6 40.g7 ¦c8 41.exf6 Black cannot
last two moves, White has restricted both Black stop the pawns, since the king cannot cross
pieces’ activity. In principle, if everything else the eighth rank.) 40.¦f7 ¦xf7† 41.gxf7† ¢xf7
failed, White could at some moment push his 42.e6† and White will stalemate the black king,
16 Learn from the Legends – Chess Champions at their Best

forcing g7-g5, hxg6 and mate within two moves. It is, however, not always possible to have an
For instance, 42...¢e8 43.¢e5 ¢e7 44.¢d5 extra pawn and the better-placed rook. If the
¢e8 45.¢d6 ¢d8 46.e7† ¢e8 47.¢e6 a5 48.a4 enemy rook is more active, then an extra pawn
g5 49.hxg6 h5 50.g7 h4 51.g8£ mate. guarantees only practical chances, but not a clear
b) 38...¢f8 39.¦d6 ¢e7 40.¦a6 ¦b7 (If win. Here is a typical example.
40...¢d7 then 41.¦f6 is again winning or
40...¢f8 41.¢e6 ¢e8 42.a4, planning a5, ¦d6, Akiba Rubinstein – Aron Nimzowitsch
a6, when Black is too passive to resist) 41.¦c6 Gothenburg 1920
¦d7 42.¦c8 once the rook captures the g7-pawn
the game will be over.

34.¢f5 ¢f7 35.e5 ¦b7 36.¦d6     
The only purpose of the following moves was
to reach the 38th move where the game would be   
adjourned.
36...¢e7 37.¦a6 ¢f7 38.¦d6 ¢f8 39.¦c6 ¢f7
  
     
        
       
       
       
    
Compared with the previous endings, it will
     soon be the defending side that has a more active
rook. As compensation for White, there are more
    pawns on board and, besides, within just a few
     moves Rubinstein will obtain a very favourable
configuration on the kingside. Therefore, a
 preliminary evaluation, based on the score 2-1,
40.a3 suggests that the position should be placed
Black is in zugzwang: 40...¢f8 (40...¦e7 is somewhere on the edge between a draw and a win
well met by 41.e6† ¢g8 42.¢g6 ¦e8 43.e7 for White. The further course of the game will
followed by ¦d6-d8 and White wins) 41.¢g6 support this evaluation. Faced with Rubinstein’s
¦b3 42.¦c8† ¢e7 43.¦c7† ¢e6 44.¦xg7 strong and consequent play, Nimzowitsch will
winning. make just one significant mistake, but this will
Ironically, this nice win over the World be enough for the balance to swing decisively in
Champion didn’t help Akiba’s cause. Although White’s favour.
from a formal point of view Lasker acted as a 32.g4
gentleman and praised his opponent’s play in his In principle, Black would like to play ...h5, in
annotations, he also understood that he would order to defend his pawns more easily. Only the
have a tough (if not impossible) job defending f7-pawn would need permanent care from the
his title against such a strong player. As is known king, but this pawn is close enough to the centre
Akiba never got a title match with Lasker... to avoid the black monarch’s decentralization.
1–0 After 32.g4, Black is at a crossroads: shall
he allow g4-g5 or not? It is easy to say, already