Prévia do material em texto
Dvoretsky I Yusupov · Secrets of Endgame Technique
Pgress iifCtiess
Volume 24 of the ongoing series
Editorial board
GM Victor Korchnoi
GM Helmut Pfleger
GM Nigel Short
GM Rudolf Teschner
2008
EDITION OLMS
m
Mark Dvoretsky and Artur Yusupov
Secrets of Endgame
Technique
School of Future Champions 3
Edited and translated
by Ken Neat
2008
EDITION OLMS
m
4
Books by the same authors:
Mark Dvorelsky, Artur Yusupov, School of Future Champions
Vol. 1 : Secrets of Chess Training ISBN 978-3-283-00515-3
Vol. 2: Secrets of Opening Preparation ISBN 978-3-283-00516-0
Vol. 3: Secrets of Endgame Technique ISBN 978-3-283-00517-7
Vol. 4: Secrets of Positional Play ISBN 978-3-283-00518-4
Vol. 5: Secrets of Creative Thinking ISBN 978-3-283-00519-1
Mark Dvoretsky, School of Chess Excellence
Vol. 1 : Endgame Analysis ISBN 978-3-283-00416-3
Vol. 2: Tactical Play ISBN 978-3-283-00417-0
Vol. 3: Strategic Play ISBN 978-3-283-00418-7
Vol. 4: Opening Developments ISBN 978-3-283-00419-4
Bibliographic information published by Die Deutsche Bibliothek
Die Deutsche Bibliothek lists this publication in the Deutsche
Nationalbibliografie; detailed bibliographic data is available in
the internet at http://dnb.ddb.de.
© 2008 Edition Olms AG
Willikonerstr. 1 0 · CH-861 8 Oetwil a. S./Zurich
E-mail: info@edition-olms.com
Internet: www.edition-olms.com
Available
Available
Available
In Preparation
In Preparation
Available
Available
Available
Available
All rights reserved. This book is sold subject to the condition that it shall not. by way of trade or
otherwise. be lent. re-sold, hired out or otherwise circulated in any form of binding or cover
other than that in which it is published and without a similar condition including this condition
being imposed on the subsequent purchaser.
Printed in Germany
Editor and translator: Ken Neat
Typeset: Arno Nickel · Edition Marco, D-1 0551 Berlin
Printed by: Druckerei Friedr. Schmucker GmbH, D-49624 Lbningen
Cover: Eva Konig, D-22769 Hamburg
ISBN 978-3-283-00517-7
5
Contents
Preface (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
PART I ENDGAME THEORY
How to Study the Endgame (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
The Theory and Practice of Rook Endings (Mark Dvoretsky, Artur Yusupov) . . . . . . . . . . . . . . . . . . . 38
From the Simple to the Compl icated : The Theory of Endings
with Opposite-colour Bishops (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
The Arithmetic of Pawn Endings (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
PART II ENDGAME ANALYSIS
Typical Endings with Connected Passed Pawns (Vladimir Vulfson) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Adventures on Adjournment Day (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Solo for a Knight (Artur Yusupov) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 02
More about the 'Monta ign ian ' Knight (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 08
PART I l l TECHNIQUE
Converting an Advantage (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 2
Technical Procedures i n a Grandmaster Battle (Artur Yusupov) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 53
Lessons from One Particu lar Endgame (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 58
Grandmaster Technique (Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 65
Analysis of a Game (Artur Yusupov, Mark Dvoretsky) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 99
PART IV
From Games by Pupi ls of the School (Artur Yusupov) .... ....... ........... ............. ................. 2 1 2
Index o f Players and Analysts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
6
Mark Dvoretsky
Preface
Perhaps, dear reader, you are a l ready
fami l iar with the fi rst two books, based
on material from the Dvoretsky-Yusupov
school for talented young players (Secrets
of Chess Training and Secrets of Opening
Preparation). Then you wi l l a l ready know the
main principles by which we are gu ided . We
have held severa l thematic sess ions of the
school , devoted to a particu lar aspect of
improvement in chess. The a im of the
sessions was not to convey specific infor
mation - for this we had too l ittle t ime. It was
far more important to d iscover the deficien
cies in the pupils' play and to help them to
eradicate them, to demonstrate the most
effective ways of studying chess , and to
acquaint them with general mechanisms,
ideas and methods of play.
You now have before you the th i rd book
(there are five in a l l ) . I t is based on our work
at the th i rd session of the school , devoted to
the problem of improving endgame and
technical mastery.
During recent years the regu lations for
tournaments and matches have changed
sign ificantly - now games are hardly ever
adjourned . Previously, after taking play into
an end ing , you could investigate its subtle
ties in home analysis, whereas now you
have to act d i rectly at the board . Without an
excellent knowledge and, what is even more
important, understanding of the laws of the
endgame, it is not easy to cope with this
task, especial ly if account is taken of
accumulated fatigue from the preced ing
battle. And yet endgame mistakes are the
last in the game - it is no longer possible to
repair them! It is clear that today the
importance of endgame and techn ical mas
tery has grown sharply.
I n my bookcase there are numerous weighty
tomes devoted to the theory of endings . Is it
conceivable to assimi late and remember a l l
the information conta ined in them? It turns
out that it is not at al l necessary to do th is .
After read ing the fi rst part of this book, you
wi l l see that 'your ' system of endgame
knowledge can and should be compact,
easy to assimi late and remember, and you
wi l l learn how to develop it, by making the
acqua intance of certa in important sections
of the theory of rook and minor piece
endings.
The second part of the book is devoted to an
analysis of compl icated practical endings.
Such an analysis helps to deepen and
consol idate endgame knowledge, and aid
the development of tra its of character and
th ink ing necessary to every player.
The techn ique of converting an advantage is
a stumbl ing-block for many players. To ra ise
your techn ica l mastery requ i res developing
in yourself certa in important ski l ls i n seeking
and taking decis ions, moreover not purely
chess , but, so to speak, 'psycholog ical
chess' decis ions. The question of how to
improve your techn ique is examined in the
th ird part of the book. Here are both the
'theory ' of this question , and.
1 .:.b1 1 ! c:Ji;a2
2 .l:te1 !
3 c:Jl;e7
aS
c:Ji;b3
3 . . . a4 is hopeless : 4 c:Ji;d6 a3 5 c:Ji;c5 c:Ji;b2 6
l:.e2+ (6 c:Ji;b4 a2 7 l:te2+ c:Ji;b1 8 c:Ji;b3 is a lso
good , or 7 . . . c:Ji;c 1 8 l:.xa2 c:Ji;d 1 9 c:Ji;c3)
6 . . . c:Ji;b1 (6 . . . c:Ji;b3 7 l:.xe3+) 7 c:Ji;b4 a2 8 c:Ji;b3.
4 c:Ji;d6!
Only not 4 lbe3+? �b4 5 c:Ji;d6 a4 6 l:te4+
c:Ji;b5! with a draw (again a 'shoulder
charge' ) .
4 . . . a4
4 . . . c:Ji;b4 5 �d5 a4 6 c:Ji;d4 a3 7 l:.b1 + .
5 c:Ji;c5 a3
6 l::txe3+ c:Jl;a4
6 . . . c:Ji;b2 7 c:Ji;b4 a2 8 l:te2+ c:Ji;b1 9 c:Ji;b3 .
7 c:Jl;c4
8 l:te1
9 c:Ji;c3
6. J. Peckover ( 1 960) .
1 c:Ji;d5 ! !
Everyth ing else loses :
1 :b8+? c:Ji;a5 2 llg8 c:Ji;a6;
a2
'it>a3
1 c:Ji;e3? c:Ji;c5 2 .l:tc8+ c:Ji;b6! 3 c:Ji;f2 g 1 'iW+ (or
3 . . . c:Ji;b7) ;
1 !tg7? c:Ji;b3! 2 c:Ji;e3 �c4 3 .l::l.c7+ �b5 ! .
1 • . • c:Ji;b3
2 l:r.g3+ c:Jl;a4
3 l:tg4+ c:J;as
4 .l:.g8 �b5
5 l:tg7! c:Ji;b6
6 l:.g6+ �c7
7 l:r.g7+ 'it>d8
8 c:Ji;d6 c:Ji;c8
8 . . . c:Ji;e8 9 l:te7+ and 1 0 l:.e 1 .
9 c:Ji;c6 c:J; b8
1 0 l:lg8+ c:Jl;a7
11 l:.g7+ c:Ji;a6
12 l:.g8 c:J;as
1 3 c:Ji;c5
The black king is u nable to escape from the
pursuit .
7. V. Pachman ( 1 960/6 1 ) .
1 .l:tf1 ! c3
2 lih1 + ! !
I t i s very important to place the rook on g 1
with ga in of tempo. White loses after 2 c:Ji;f??
c:Ji;h6 3 �f6 c:Ji;h5 4 c:Ji;f5 c:Ji;h4 5 c:Ji;f4 �h3 6
c:Ji;e3 c:Ji;h2 ! , when he ends u p i n zugzwang: 7
l:ta 1 �g2 or 7 l:td 1 c2 8 c:Ji;xd2 cxd 1 'ii'+ 9
c:Ji;xd 1 e3 (9 . . . c:Ji;g2) 1 0 c:Ji;c2 c:Ji;g2 1 1 c:Ji;c3
c:Ji;f1 ! 1 2 �d3 c:Ji;f2 .
2 . . . c:Ji;g6
2 . . . c:Jl;g7 3 .l::l.g 1 + �f8 4 l:tf1 + c:Ji;e8 5 l:.h 1 .
3 Itg1 + �h5
4 c:Ji;fS �h4
5 'it>t4
6 'it>e3
7 .l:.f1 1
c:Ji;h3
c:Ji;h2
The same zugzwang position has now
arisen with Black to move.
7 • • • c:Ji;g2
8 l:la1 ! c:Ji;g3
9 .:.g1 + c:Ji;h2
9 . . . c:Ji;h3 1 0 .U.h 1 + c:Ji;g2 1 1 .:.a 1 ! or 1 0 . . . c:Jl;g4
1 1 l:.g 1 + c:Ji;f5 1 2 .:.f1 + c:Ji;e5 1 3 l:td 1 .
1 0 l:.f1 !
Black is unable to win .
8. V. Hortov ( 1 982) .
Which pawn shou ld be advanced? This
question can only be solved by a deep
How to Study the Endgame ctJ 37
calculation of the variations.
1 g7! l::rb8
2 'it>g 1 'it>g3
3 'it>f1 'iii>f3
4 'it>e 1 'it>e3
5 'it>d 1 'it>d3
6 'it>c1 .l:.c8+
7 '>t>b2 .l:b8+
8 'it>a3
8 'it>a 1 ? 'it>c2 .
8 . . .
9 'it>a4
1 0 'it>a5
11 'it>a6
1 2 'it>a7
1 3 a4
14 'it>b6!
'it>c3
�c4
'it>c5
'>t>c6
.l:tg8!
'it>d6
I f 14 �b7?, then 14 . . . 'it>e6 1 5 a5 ( 1 5 '>t>b6
l:'tb8+ 1 6 We? l::rg8 ; 1 5 Wc6 .l:!.c8+) 1 5 . . . 'it>f6
1 6 a6 �g6 1 7 a7 Wxh6.
14 a5?, hoping for 14 . . . �e6? 1 5 'it>b6 l:tb8+
16 'it>c7! .l:tg8 1 7 'it>c6! lieS+ ( 1 7 . . .f6 1 8 h7)
1 8 'it>b 7 l::rg8 1 9 a6, also does not work .
Black repl ies 1 4 . . . �c6! 1 5 a6 .l:te8(d8) 1 6 h7
.l:te7(d7)+ with perpetual check.
1 4 . . . ltb8+
1 5 Wa6! 'it>c6
1 6 Wa7 l:tg8
1 7 a5 'it>d6
1 7 . . . 'it>c7 1 8 h7 .
18 'it>b7!
1 9 a6
20 a7
21 a8'if
'it>e6
'it>f6
'>t>g6
The black king was only just too late.
With the pawns on g6 and h7 the king is able
to attack them a move earl ier.
1 h7? .l:tb8 (or 1 . . . 'it>g3 2 Wg 1 .l:Ib8) 2 'it>g 1
'it>g3 3 'it>f1 Wf3 4 'it>e 1 'it>e3 5 Wd 1 'it>d3 6
Wc1 l:tc8+ 7 'it>b2 .l:i.b8+ 8 'it>a3 'it>c3 9 'it>a4
Wc4 1 0 Wa5 Wc5 1 1 'it>a6 'it>c6 1 2 �a 7 l:th8!
1 3 a4 Wd6 14 'lt>b6 l:ib8+ 15 'lt>a6 Wc6 1 6
'it> a 7 llh8 1 7 a 5 'it>d6 1 8 'it>b 7 We6 1 9 a6 'it>f6
20 a7 'it>xg6, or 1 8 'it>b6 I:Ib8+ 1 9 Wa6 'lt>c6
20 'lt>a7 .l:.h8 2 1 a6 .l:.e8(d8) ! .
38 cj{
Mark Dvoretsky, Artur Yusu pov
The Theory and Practice of Rook Endi ngs
Mark Dvoretsky
Of a l l the types of endings, it is rook
endings which demand the most inten
sive study. Why is this?
Fi rstly, they occur more often than other
types. A good half of a l l the endings that
occur in practice are rook endings.
Second ly, here there exists a fa i rly deta i led
theory of positions with a smal l amount of
materia l (for example, rook and pawn
against rook) , which may a lso be repeated
in our games. This theory should be mas
tered .
I n other types of endgame the situations
with a min imal number of pawns are either
qu ite simple, or not very important. This
means that there a knowledge of exact
positions is hardly ever requ i red - it is l i kely
that you wi l l never need it. I t is sufficient to
know the typical ideas and methods. But in
the rook endgame you cannot get by without
studying a considerable number of exact
positions.
I offer for your attention one of the sections
of rook endgame theory - endings with a
pawn on the rook's fi le . As usual , we wi l l
beg in our analysis with the s implest cases .
And in genera l , we wi l l not delve too deeply
into theory - we wi l l merely pick out the most
important positions and the ideas involved
with them.
1) Stronger side's king in front
of the pawn
A draw is inevitable. The only way to try and
free the king from imprisonment is by
playing the rook to b8. But then the black
king wi l l stand guard in place of the rook.
1 l::th2 'it>d7 2 l::th8 �c7 3 .l::i.b8 �c1 (of
course, 3 . . . l::th 1 is a lso possible) 4 �b2 �c3,
and White cannot strengthen h is position .
Let us move the king and rook one fi le to the
right.
The Theory and Practice of Rook Endings ctJ 39
Now White wins, s ince the black king does
not succeed in reach ing c7.
1 .l:!.h2 c5 . With the black king at d6 it is
no longer possible to escape via c5 , and a
d ifferent route has to be found .
3 .Mb8 .Ma1 4 'it>b7 .l:!.b1 + 5 'it>c8 l:!.c1 + 6 'it>d8
l::!.h1 7 .U.b6+ �c5
This is the only subtle moment. I t is
hopeless to play 8 .l:!.e6? .l:!.a 1 or 8 .l:!.a6?
.Mh8+ 9 Wd7 l:th7+ 1 0 �e8 l:th8+ 1 1 �f7
J::!.a8 with a d raw.
8 l:!.c6+! �b5 (8 . . . �d5 9 lia6 .Uh8+ 1 0 �c7
l:.h7+ 1 1 'it>b6) 9 l::!.c8 lih8+ 1 0 'it>c7 lih7+ 1 1
'iiib8
2) Stronger side's rook in front of the
pawn ; pawn on the 7th rank
This is a standard defensive scheme:
Black's rook is beh ind the enemy pawn, and
h is k ing is on g7 (or h7) . The wh ite rook is
tied to the pawn and cannot move from a8. I f
1 'it>b6 , then 1 . . . ltb1 + . The king has no
shelter from the vertical checks. After driving
it away, the rook returns to a 1 .
I should mention that other, more compl i
cated and less rel iable systems of defence
also exist: the black king may hide ' i n the
shadow' of its wh ite opponent (say, at c3) ,
or, with the black rook on the 7th rank - ' in
the shadow' of its own rook. We merely
mention these ideas, but we wil l not study
them. Sometimes they a re sufficient for a
d raw, sometimes not.
Let us add a wh ite pawn on h5. Noth ing has
changed . B lack does not pay any attention
to it . I t is also a d raw with a wh ite pawn on
g5.
But with a pawn on f5 White wins. After 1 f6+
�f7 ( 1 . . . '1t>xf6 2 :fa+ ; 1 . . . �h7 2 f7) 2 .l:!.h8
Black loses h is rook.
I t is no accident that I have 'chewed over'
these elementary cases in deta i l . You
should have a very clear impression of
them , and should a lways remember and
make use of them when considering more
compl icated positions.
Khaunin - Fridman
Len ingrad 1 962
In the game there fol lowed 1 . . . hxg3 2
hxg3? g4+! 3 fxg4, when a d raw was
inevitable , s ince White was left with a
knight's pawn (whether one or two is of no
particu lar importance) .
He cou ld have won by 2 h7 3 h4!
40 � The Theory and Practice of Rook Endings
gxh4+ 4 �h3 �g7 5 f4 , when the f-pawn
advances with decisive effect.
3) Stronger side'srook in front of the
pawn ; pawn on the 6th rank
The ma in d ifference between this position
and the previous ones is that the wh ite king
now has a shelter against vertica l checks -
at a7. It heads for there, in order to free the
rook from having to defend the pawn .
But the black king is unable to run to the
queenside: 1 . . . ci;f7? 2 'it>e4 (2 a7? 'itg7 is
prematu re) 2 . . . cJ;;e7 3 a7! 'itd7(f7) 4 .l:!.h8.
2 . . . .l:!.a5 ( instead of 2 . . . �e7) is a lso hope
less: 3 �d4 'it>g7 4 'itc4 �f7 5 'it>b4 .Ua 1 6
'it>b5 l:tb 1 + 7 'itc6 .l:!.a 1 8 'it>b 7 .l:!.b 1 + 9 'it> a 7
'ite7 1 0 l:tb8 �c1 1 1 'it>b7 (but not 1 1 .Ub6?
'itd7) 1 1 . . .l:tb 1 + 1 2 'it>a8 .l:!.a 1 1 3 a 7 , and a
situation that is fami l iar to us arises: the
black king does not reach c7 in time.
I n view of this analysis , Siegbert Tarrasch
considered this position to be won . But later
(in 1 924) a saving plan was found . I t is
based on the fact that the a6-pawn provides
the king with a shelter against vertical
checks, but not against horizontal checks.
The rook must be transferred to f6 .
1 . . . .Uf1 +! 2 'it>e4 .Uf6!
It is important to attack the pawn , in order
not to release the rook from a8. The
result ing posit ion is known in endgame
theory as the 'Vancura posit ion' (from the
name of its d iscoverer - Joseph Vancura) .
What can White do? If a6-a7 there a lways
fol lows . . Jla6 (of course, the black king wi l l
not move from g7 and h7) . I f the pawn is
defended by the k ing, there fol lows a series
of checks , and then the rook returns to f6 .
For example: 3 'it>d5 l:tb6 4 �c5 l:tf6! (the
best square for the rook ! ) 5 �b5 .l:!.f5+! etc.
I f in the d iagram we move the wh ite king to
f4 , we obta in a position which was analysed
in 1 950 by Pyotr Romanovsky. 1 . . . .l:i.f1 +? 2
'it>e5 .l:lf6 is now bad because of 3 .l:!.g8+ ! . But
a l l the same there is no other p lan - only the
switching of the rook to the 6th rank .
Therefore let us p lay 1 . . . l:tc1 ! . If 2 'it>e5 there
fol lows 2 . . .lk6 - and we reach Vancura's
d rawn position . Wh ite must take the oppor
tun ity to remove h is rook from a8: 2 l:tb8
.l:!.a1 3 .l:r.b6 (weaker is 3 .Ub7+ 'it>f6 4 a7
'it>e6).
With the rook on a8 Black's king was t ied to
the kingside, but now it can head towards
the pawn . But this must be done cautiously:
he loses after the hasty 3 . . . �f7? 4 .'it>e5 �e7
5 . .l:!.b7+ 'it>d8 6 .a7 . Correct is 3 . . . .l:r.a5! 4 We4
'it>f7! 5 'it>d4 (if 5 .l:!.h6, then 5 . . . 'it>g7 ! , but not
5 . . . cl;e7? 6 a7 Wd7 7 l:th8 ! ) 5 ... 'it>e7 6 'it>c4
�d7 7 ci;b4 Ua1 , and the draw is obvious .
Note that the k ings had a race to reach the
queenside. I f the wh ite king had been closer
to the pawn , Black's king might not have
arrived in t ime. This means that the attempt
to switch the rook to the 6th rank cannot be
delayed - this p lan must be carried out as
soon as possib le.
The system of defence examined by us is
very important. In particular, th is is how the
defence should be arranged when the
opponent has two extra pawns - 'a' and 'h ' .
The Theory and Practice of Rook Endings ltJ 41
The h-pawn does not help White - the draw
is just as elementary as in the previous
example. If 1 'it>b5 there fol lows 1 . . . .l:.f5+.
After driving away the king , the rook contin
ues its watch of the 6th rank . If the wh ite
pawn is repositioned at a5, the black rook
would be placed on the 5th rank, and so on .
Now let us analyse a position with a- and g
pawns.
1 . . . 'it>h7!
2 �h5
Threatening 3 .l:!.a7+ �g8 4 g6 and 5 'it>h6.
2 . . . .l:!.h6+!
3 'it>g4
3 gxh6? - stalemate!
3 . . . .l:!.b6
Strangely enough , in books on the endgame
this posit ion is not ana lysed . I t had to be
stud ied independently. White wins. The
main reason is that the black rook does not
have the important f6-square , and the 6th
rank proves too short .
4 �5
Again threatening 5 l:Ia7+ .
4 . . .
5 �6
6 'it>e5
6 'it>f7 .:tb7+! is pointless.
�b5+
l:f.b6+
6 . . . .l:!.c6
Of course, B lack does not have time to
captu re the g5-pawn: 6 . . J:tb5+ 7 'it>d6 (7
'it>d4) 7 . . J:txg5 8 .l:!.e8 .l:!.a5 9 .Ue 7 + 'it>g6 1 0
a7 . He also loses qu ickly after 6 . . . 'it>g7 7
'it>f5! .l:!.b5+ 8 �g4 .l:tb6 9 'it>h5 and 1 0 .l:la7+.
7 'it>d5 l:f.b6
8 'it>c5 l:te6
8 . . Jig6 9 .l:!.a7+ 'it>g8 1 0 'it>d4 .
9 .l:.a7+! �g6
In the event of 9 . . . 'it>g8 the white king returns
to the kingside.
1 0 'it>b5 .l:!.e5+
1 1 'it>c6
1 2 'it>c5!
.U.e6+
42 � The Theory and Practice of Rook Endings
The decisive zugzwang!
Instead of 9 l:!.a7+ White also wins by 9 'it>bS
.l:!.eS+ 10 'it>c6 .l:!.e6+ 1 1 'it>cS! (but not 1 1
'it>c7? .l:.g6 1 2 a 7 .l:!.g7+ ! with a d raw). Here
too Black is in zugzwang! The variation
1 1 . . . lieS+ 1 2 �d6 .UxgS 1 3 .Ue8 is a l ready
fami l iar to us, while after 1 1 . . . 'it>g7 the rook is
deprived of an important square , from where
it cou ld g ive a check. Now the wh ite king
boldly advances: 1 2 'it>bS lieS+ 1 3 Wc6
lie6+ 1 4 �c7 'it>h7 (there is no longer the
reply 1 4 . . . .l:!.g6) 1 S a7! l:!.a6 ( 1 S . . . .l:!.e7+ 1 6
'it>d6) 1 6 'it>b7 and wins.
Artu r Yusu pov
A practical player should be able to
onfidently find h is way in typical rook
endgame positions. See how, making use of
the ideas we have just examined , I was able
to save a difficult ending against an ex-world
champion.
Karpov - Yusupov
Linares 1 99 1
I t i s Karpov to move. What possib i l it ies does
he have?
Black must seriously reckon with 'it'c3-c4
( immediately, or after the preparatory 4 1
a6). But, after exchang ing queens, h e can
give a check on d1 and place his rook
beh ind the passed pawn - this is a very
important defensive resource, typical of rook
endings.
On prophylactic g rounds it makes sense to
remove the king beforehand from the fi rst
rank : 4 1 'lt>g2!? . Now after the exchange of
queens the black rook does not come to the
rear of the passed pawn . Even so, B lack
ga ins sufficient counterplay, by continu ing
4 1 . . . cS 42 �c4 'ifxc4 43 .l:!.xc4 �c7 fol lowed
by . . . �f7-e6-dS, or 42 a6 lia7 43 'i!VaS (43
l:!.a4 f4) 43 . . .'ir'c6(d6) .
41 a6 'iia2
The pawn has to be halted . In the g iven
instance it is not the rook that is placed
behind it , but the queen . I thought for a long
t ime about the possib i l ity of continu ing the
fight in the middlegame, but I d id not find
anyth ing convincing and I decided not to
avoid the exchange of queens.
42 'ii'c4+
Anatoly Karpov d id not th ink for long over
this move. He had to reckon with the threat of
a counterattack by 42 .. J�d 1 + and 43 . . . 'i!Vb1 .
42 . . . 'iixc4
43 .l:.xc4 .Ud 1 +
Of course, the rook i s switched to the rear of
the passed pawn . This very important
device is merely a particu lar instance of the
general principle of rook endings, which
says that the rook should be active .
44 Wg2 lia1
45 l:.c6
When a pawn is attacked from the rear, it is
The Theory and Practice of Rook Endings 43
usual ly preferable to defend it with the rook
from the s ide, rather than by stand ing in
front of the pawn . The rook on c6 is very
active - it is contro l l ing the 6th rank and
attacking the c?-pawn .
45 . . . �f8
Sooner or later the wh ite k ing wi l l try to
break through on the queenside. B lack
beg ins a counter-plan - he plays his k ing to
d7, in order to activate his c7-pawn or
ach ieve the exchange of severa l pawns.
46 f4
I f 46 'it>g3, then 46 . . J:i.a4 .
46 . . . .Ua3!
Subsequently every tempo may prove deci
sive - therefore the wh ite k ing's passage to
the queen side must be h indered as much as
possible.
47 �f1
48 �e1
49 �d1
l:ta2
'it;e8
�dB!
An accurate move . 49 . . . Wd7 suggests itself,but I was concerned that after 50 .Ug6 the
g7 -pawn would be captured with check. Of
cou rse , 49 . . . .Uxf2? was premature in view of
50 a7 l:ta2 51 l:txc? , and with his k ing cut off
along the 7th rank , B lack loses qu ickly. But
now the capture on f2 is th reatened .
50 .l:!.g6 c5
Not 50 . . . .Uxf2? 51 l:txg7 l:ta2 52 l:tg6 .
Therefore Black activates h is passed pawn .
51 Wc1
In such situations one somewhere has to
stop making common sense moves and ,
after accurately calculat ing a way to d raw,
force events . Such a moment has now
arrived .
51 . . .
52 .Uxg7+
53 l:tg6+
54 .l:i.c6
'it>c7!
'it;b6
Wa7
Karpov a ims to e l iminate as many black
pawns as poss ible. In the event of 54 l:txh6
l:txf2 fol lowed by 55 . . . l:txf4 the draw is
obvious.
54 . . .
55 .l:!.xc5
.Uxf2
.Uxf4
A text-book position with a- and h-pawns is
reached . Of cou rse, 55 . . . Wxa6 was possi
ble, but it was more method ical to play 'by
the book' , especia l ly s ince I was short of
t ime on the clock.
56 l:tc6
57 �d2
58 .Uxh6
59 �e2
.l:!.g4
�g5
f4
f3+
Black does not need th is pawn . If you know
for certa in that a position is d rawn , you
should try to ensure that extraneous detai ls
( l ike a 'non-theoretica l ' pawn) do not acci
denta l ly h inder you .
6 0 �xf3 .Uc5
61 l:th8
Here , just in case, I adjourned the game. To
my surprise, Karpov turned up for the
resumption and made a few more moves.
44 � The Theory and Practice of Rook Endings
61 . . .
62 'it>e4
63 'it>f4
64 'it>e5
65 'it>e6
66 �f7
�g5
l:!.c5
�c4+
.Uc5+
.Ug5
.Uc5!
In such positions the c-fi le is the best place
for the rook. If now 67 h6 l:c6 68 h 7, then the
rook should be placed beh ind the pawn , but
the immediate 68 . . . l:!.h6 loses to 69 'it>g7 .
Therefore Black must fi rst g ive some checks:
68 . . . l:tc7+! 69 '1t>f6(e6) .l:lc6+ with a draw. But
on the d-fi le the rook would be too close to
the king , and after 68 . . . .l:ld7+ 69 'it>e6 Black
would lose.
67 l:!.h7 �xa6
Only now, when the rook has gone to h 7,
can the a6-pawn be taken . But with the rook
on h8 it should be ignored .
68 h6 l:!.c7+
Draw.
Mark Dvoretsky
1\ lthough the ideas that we have been
./"\d iscussing are elementary, by no means
a l l players are fami l iar with them . Even
grandmasters sometimes 'flounder' in stand
ard theoretica l endings. Here is a tragic
comic example.
Szabo - Tukmakov
Buenos Aires 1 970
For White it is sufficient simply to wait,
keeping the a5-pawn under fi re , in order not
to release the rook from the a-fi le . For
example, 66 l:!.b5 �d6 67 l:!.f5 .l:i.a 1 68 'it>h2!
a4 69 l:!.f4 ! a3 70 l:!.f3 ! 'it>c5 (70 . . . a2 71 l:!.a3)
7 1 l:!.b3 �c4 72 .l:lf3 �b4 73 l:!.f4+! etc. When
you know the plan of defence, the moves
make themselves - here there is noth ing
cunn ing .
However, the h igh ly experienced grandmas
ter Laszlo Szabo had no idea of how to play
these types of endings, and he lost a
completely d rawn position . Apparently Vladi
mir Tukmakov also d id not know them, s ince
he commented on the course of the play as
fol lows: 'Theory considers this endgame to
be drawn , but I seemed to win qu ite
convincingly. '
6 6 'lt>g2?!
67 '.tf2?!
68 'it>e1 ?
Wd6
l1a2+
68 'it>g 1 ! would sti l l have led to a draw.
68 . . . .l::ta 1+ !
69 'it>e2
69 Wd2 l:i.h 1 ! 70 .l:!.xa5 h3 7 1 l:!.h5 h2 and
72 . . . l:i.a 1 .
69 . . . a4
70 l:!.h6+
70 l:!.xh4 a3 7 1 .l:!.a4 a2.
70 . . .
71 .Uh5+
72 �f2
'it>e5
'it>f6
a3
The Theory and Practice of Rook Endings l2J 45
73 'it>g2
74 l1a5
White resigned .
.Uc1
l:Ic3
I found another, s imi lar example i n the
magazine New in Chess, i n an article by
Tony Mi les about the 1 989 U SA Champion
sh ip . He analyses the ending of a game by
the winner of the championsh ip , Stuart
Rachels, against g randmaster Dmitry Gure
vich . It would appear that none of them,
includ ing the commentator, was fami l iar with
the ideas of the g iven endgame.
Rachels - Gurevich
USA Championsh ip , Long Beach 1 989
Mi les writes: 'S ince rook and a- and h
pawns vs rook is often drawn , it is hard to
bel ieve that White should win th is . The
defensive task is not s imple , though . '
O f course, i t is not essentia l to g ive up the
pawn , but from the practical point of view
this is the best cou rse. Subsequently you no
longer have to think, s ince you wi l l be acting
in accordance with theory. Remember: this
is how Yusupov acted in his game against
Karpov. Otherwise you wi l l be forced to act
independently in a position which , although
drawn , is unfami l iar, and it wi l l be easy to
make a mistake .
48 . . .
49 h5+
'it>g6
�f7
A clear demonstration of B lack's ignorance
of theory: Gurevich , l i ke Szabo in the
previous example, incorrectly moves h is
k ing to the opposite wing .
50 Wf4 �c6
51 'it>e5 'it>e7
52 'it>d5 .Uh6
53 Wc4 �h8
54 l:te5+ 'it>f6
55 l:tc5 rt;e7
56 rt;b4 .l:!.h6
57 �a5 Wd6
58 l:tg5 �c7
59 a4 Wd7
60 .l:!.g7+ Wc8
61 .Ug5
Here the game was adjourned . In home
analysis it is important to look i n a book and
fami l iarise yourself with the theory of the
endings that may arise during the resump
tion - in the g iven instance, with the theory
of the endgame with two extra a- and h
pawns. But Gurevich d id not do th is .
61 . . . c.t>d7
62 .ti.c5 Wd8
63 .ti.d5+ rt;e7
Black changes h is p lan of defence and
returns h is k ing to the kingside. I n the event
of 63 . . . b4 We6
65 �c5 �e7
66 l:tg5 �f7
67 �d5 �f8
68 .Ue5 Wf7
69 We4 l:.c6
70 g7! 72 .l:.xa6 'it>h7 would have forced a
draw. I n principle, delaying this is now not
without its dangers - after a l l , B lack has to
reckon with the fol lowing p lan : the wh ite
pawn goes to a5, the rook defends every
thing along the 5th rank, and the king heads
for b7.
72 'it>gS
73 .U.d5
74 aS
75 �d7+
76 l::i.a7?
'it>g7
�c4
l::i.c6
'it>g8
After 76 �e7! Wh ite's position is apparently
now won . For example: 76 . . . �c5+ 77 'it>g6
.U.c6+ 78 'it>f5 l::i.c5+ 79 �e5, or 76 .. .'it>f8 77
.l:!.b7! (with the deadly threat of 78 .l:l.b6)
77 . . . l:tc5+ 78 'it>g6 .l:!.c6+ 79 'it>h 7.
76 . . . �d6?
I t was essentia l for B lack to exploit h is
opponent's mistake, by playing 76 . . . �c5+
77 'it>f6 �c6+ (neither 77 .. Jbh5? nor
77 . . .lba5? is possible, in view of 78 'lt>g6)
78 'it>e7 l:k5 (or - as recommended by M i les
- 78 . . . .Uh6) 79 l:txa6 .l:txh5 (or 79 . . . 'it>h7) with
an obvious draw.
77 'it>fS
78 'it>f6
.l:!.dS+
.l:.d6+
79 'it>eS .l:tc6
80 l:.d7
Here M i les makes an amusing comment: ' I f
80 . . . 'it>f8 , then 8 1 l::i.d6 l:!.c5+ 82 'it>f6 'it>g8 (or
82 . . . '1t>e8 83 h6} 83 �a6 wins . ' But 83
.U.xa6?? 'it>h7! leads to an immediate draw,
whereas 83 'it>g6! wins.
80 . . . 'it>h8 cou ld have been tried , i n the hope
of 81 .l:.d6 l:tc5+ 82 .l:!.d5 (82 'it>f6 'it?h7! )
82 . . . �c6 83 'lt>f5 'it>g7 84 .l:.e5 (with the threat
of 85 l::i.e7+ and 86 .l:!.e6) 84 . . . 'it>f7 ! . But the
subtle move 81 �e7 ! enables White to win :
after 8 1 . . . '1t>g8 82 'it>f5! �c5+ (82 . . . 'lt>f8 or
82 . . . �h6 - 83 �e6! and 84 'lt>g6) 83 .l:.e5!
(only not 83 'it>f6? l::i.xh5 84 'it?g6 'it?f8 ! ) he
wins. This position occurred later i n the
game .
80 . . .
81 'it>fS
.l:th6
The sealed move . Here the game was again
adjourned .
81 . . . �c6
In the event of 81 . . . 'it>f8 !? (not a l lowing 82
.Ue7) 82 c.t>g5 l:tc6 White wou ld have won by
83 �b7 !.
82 .Ue7 !
83 l:i.e5
l:tc5+
.Uc1
After 83 . . . l:i.c6 84 �e6 �c1 White has a
pleasant choice between 85 c.t>g6 and 85
The Theory and Practice of Rook Endings 47
l::txa6 �f7 (unfortunately, 85 . . . 'it>h7 is not
possible because of 86 .l:.e6) 86 l:!.a7+ 'it>f8
87 .!::!.a8+ 'it>f7 88 a6 �c5+ 89 t>e4 �c6 90 a?
l::!.a6 91 .l:!.h8. The flank attack on the rook's
pawn only works if the king is on g7 or h?.
84 'it>e6!
The king heads towards the a6-pawn. B lack
can no longer save the game.
84 . . . Wg7 85 'it>d6 Wh7 86 l:!.c5 l:!.b1 87 Wc6
'it>h6 88 l:!.d5 .Ub2 89 l:!.d7 l:tb5 90 .Ud6+
'it>h7 91 .Ud5 l:!.b1 92 l:!.d7+ Wh6 93 l:!.b7
l::ta1 94 'it>b6 t>xh5 95 'it>xa6 'it>g6 96 l:!.b5
00 97 'it>b7 .Ue1 98 a6 l:!.e7+ 99 Wb6 l:!.e6+
100 �a5 Black resigned .
In these last two examples both the moves
and the comments made by grandmasters
make a comic impression , for one s imple
reason - they were not sufficiently fami l iar
with the basic theory of rook endings.
Our next step should probably be an
analysis of positions that are closely l i nked
with the type of endgame a l ready stud ied -
namely, endings in which each side has two
or th ree pawns on the kingside and one has
an extra passed pawn on the queenside
(usual ly a rook's pawn ) . Such a situation
often occurs in practice . But th is is a l ready
another topic, and here we wil l merely
mention it. I wi l l restrict myself to one
example, in which the same idea was used
as we saw in the Karpov-Yusupov game,
but d id not see in the Rachels-Gurevich
end ing : the sacrifice of a pawn to transpose
into a theoretical ly d rawn position .
Bakul in - Dvoretsky
Moscow Team Championship 1 974
Reckoning that if 1 8 . . . �e6 the reply 1 9 tt:Jc5
was unpleasant, I wanted to play 1 8 . . . tt:Jc6.
But my sense of danger operated and I
began to have doubts about the position
ar is ing after 1 9 c4 d4 20 tt:Jc5 b6 21 tbd3.
Wh ite creates a pawn majority on the
queenside, he securely blockades the d4-
pawn , and he contro ls the e-fi le . The
advantage is on his side: perhaps not a
great advantage, but an enduring one.
When I showed th is ending to Rafael
Vagan ian , an expert on the French Defence,
he assessed the position as unfavourable
for Black.
By accu rately defending , one can probably
avoid defeat, but th is is a d ifficult and
thankless task. Being an active player, I
usual ly avoided this type of passive de
fence , and endeavoured to fi nd a way of
sharply changing the course of the play, of
forcing events , either with the aim of
clarifying the situation , or, on the contrary, of
compl icating the play as much as possible.
Returning to the rook move to e6, I qu ickly
found the forcing variation which occurred in
the game.
1 8 . . . l:!.e6!?
19 tt:Jc5 �xe1 +
48 � The Theory and Practice of Rook Endings
20 l:ixe1 �c8!
20 . . . lt:Jg6 21 lt:Jxb7 .l::tb8 22 lt:Jc5 �xb2 23
l:te8+ lt::lf8 is not worth considering - the p in
on the kn ight is extremely dangerous. For
example, Wh ite can p lay 24 g3 f6 25 lt:Jd7
�f7 26 l:txf8+ 'it;e7 27 �b8 .
21 l:l.xe7 'it>f8
As you see, combinative vision sometimes
helps even when playing 'ted ious' end
games.
22 .l:f.xb7
In the event of 22 l:ie2 l:l.xc5 the position is
roughly equa l .
22 . . .
23 c3
24 �f1
24 l:tb3?? d3 25 'it>f1 .l:te5 ! .
24 . . .
25 bxc3
26 l:.xa7
.l:txc5
d4
dxc3
�xc3
Uc2
I knew for sure that th is was a draw, and a
fa i rly simple one, and so without hesitation I
went in for the exchanging combinat ion. Of
course , had I not made a previous study of
this type of endgame, I would hard ly have
decided to give up a pawn . Who knows how
the game would have ended after 1 8 . . . lt:Jc6 ,
whereas as it was I easily made a draw.
Thirty years later I realised that endings of
this type were by no means as harmless as I
then thought. In 2003 in the theory of rook
endings with an extra pawn on the wing a
revolution occurred, and positions which
had seemed completely drawn proved to be
won or at the least very dangerous. You can
find these new ideas in my articles or those
of grandmaster Carsten Muller in the ar
chives of the internet site Chesscafe. com,
and also in my book Dvoretsky's Endgame
Manual from the same publisher. If I had
known about the future discoveries, I would
perhaps have rejected the combination (in
favour of 1 8 . . . l:l.e8!? followed by 19 . . . 'it>f8),
and at the least I would have played the
endgame more carefully.
However, my decision to sacrifice a pawn
was taken not only on purely chess grounds
- it was i nfluenced , as wel l as the subse
quent play, by certa in extraneous factors . At
that time I was teaching in the chess
department of the I nstitute of Physical
Cu lture . We were p lann ing to hold a tourna
ment on the Scheveningen system : stu
dents (candidate masters) against masters,
and I had to find some opponents for our
students. The Moscow Team Champion
ship, held in the Centra l Chess Club ,
attracted nearly a l l the Moscow masters ,
and th is was the most convenient place to
hold d iscussions . After transposing i nto a
comparatively s imple ending soon after the
start of play, I ga ined the opportun ity, by
making my moves without much thought, to
rush to the other boards and press players
who were stro l l ing about to take part in our
tournament.
Because my attention was d ivided , some of
my moves were not the most accurate , but I
nevertheless gained a draw.
27 g3 g6 28 g7 29 'it;t3 h5 30 h4 'it>f6
31 'it>e3 �c3+
31 . . . We6 or 31 . . . '>i>e5!? was s impler, keep
ing the f-pawn under attack.
The Theory and Practice of Rook Endings ctJ 49
32 �e4 �c2 33 f3 .l:I.e2+
With the a-pawn sti l l on its in it ia l square ,
Black could have permitted h imself 33 . . .lk4+
34 �d3 �c1 , intending to attack the g3-
pawn by 35 . . . �g 1 .
34 �4 Ub2 35 .l:!.a6+ 'it>g7
The confin ing of the king at g7 is rather
unpleasant for Black. However, i n such
situations he has ava i lable qu ite a good p lan
of counterplay: . . .f7-f6 and at the fi rst
conven ient opportun ity . . . g6-g5. For exam
ple, 36 a3!? .l:I.b3 37 'it>e4 f6 !? . And it is not
easy for the opponent to decide on 36 a4 in
view of 36 . . . .l:!.b4+ 37 'it>e3 .l:I.b3+ 38 'it>e4
.l:tb4+ 39 'it>d5 l:tb3, a lthough sometimes
(but, I th ink , not in the g iven instance) the
result ing compl ications favour Wh ite .
36 .l:!.a3 'it>f6 37 .l:ta6+ 'it>g7 38 .l:!.a4 'it>f6
(38 . . . f6 !? ; 38 . . . .l:l.f2 !? ) 39 g4 hxg4 40 fxg4
�f2+ 41 'lt>g3 .l:I.c2 42 .l:tf4+ (42 g5+ 'it>e5 is
not dangerous for Black) 42 . . . 'it>e6 43 a4 (43
.l:I.f2 .l:I.c3+ 44 Wf4 f6) 43 . . . f5 44 gxf5+ gxf5
45 .l:I.f2 .l:l.c4 46 .l:l.a2 .l:I.c3+ 47 'it>f4 lk4+ 48
'l.t>g3 (48 'l.t>g5 .l:l.g4+ 49 Wh5 'it>f6 50 a5??
.l:l.g8) 48 .. . .l:l.c3+ 49 'it>g2 .l:l.c4 50 h5 .l:!.h4
Draw.
Artu r Yus u pov
I shou ld now l ike to show you several
extracts from my games, in which practica l ,
rathe r than purely theoretica l rook endings
arose . However, in it ial ly the fi rst example
does not resemble an endgame at a l l .
Ljubojevic - Yusupov
Linares 1 99 1
2 0 . . . .i.f5!
Black carries out a tactical exchanging
operation , involving a positional pawn sacri-
fice - a procedure which has a l ready been
mentioned here severa l t imes. In the g iven
instance the pawn is g iven up for the sake of
activating his own forces.
21 .i.xf5
22 tt::lxf5
23 'ifxc5
24 �xc7
25 1t'd6
26 'i!Vxa6
tt:Jxf5
'i!Vxf5
.l:!.xd2
.l:!.c8
.l:I.xc3
.l:!.cc2
Of cou rse, the extra passed a-pawn is
dangerous, but B lack correctly ca lculated
that the pressure of h is rooks along the 2nd
rank would enable him to mainta in the
ba lance.
27 ifb6
I f 27 �a7 , then 27 . . . d4 28 'iia8+ 'it>h7 29a6
(29 1i'f3 �xf3 30 gxf3 Ua2 is roughly the
same as that which occurred in the game)
29 . . . .l:txf2 30 a7 �xg2+ 31 ifxg2 .l:I.xg2+ 32
'it>xg2 �g4+ with perpetual check.
27 . . . d4
28 iVd8+
29 'ii'h4
'it>h7
g5
50 � The Theory and Practice of Rook Endings
As Dvoretsky pointed out, 29 . . . iixe5 30 a6
l:ta2 was simpler, s ince if 31 a7?! Black has
31 . . J:txa 1 32 l:txa 1 .l:i.a2 ! , and he remains
with an extra pawn.
30 'i¥h3
I n the event of 30 il'g3 Black would have
continued 30 . . J:td3 31 f3 l:td2 with the
dangerous threat of 32 . . . '1Wf4 .
30 . . . iixh3
31 gxh3
Things seem to be bad - after al l , the wh ite
rook is positioned beh ind the passed a
pawn . However, thanks to a tactica l subtlety
Black nevertheless succeeds in stopping
the pawn from beh ind .
31 . . . .tr.a2
32 a6
33 l:txa2
Forced .
33 . . .
34 .tr.xf7+
35 .l:.d7
35 .Uf6 d3.
Draw.
35 . . .
36 .l:.xd4
37 h4
.l:.xf2!
.l:!.xa2
�g8
l:txa6
'i;f7
The fol lowing example is also devoted to
rook activity. Genera l ly speaking , the main
principle in rook endings is that the rook
should be active!
Yusupov - Barbero
Mendoza 1 985
Wh ite has a sl ight advantage thanks to the
fact that h is rook is more active , and also as
a result of the rather unusual position of the
black king at h6.
Now the most natural try seems to be the
activation of the black rook by 28 . . . l:td8 29
l:te7 b5 30 l:txa7 .U.d2 31 b3 c4 32 bxc4 bxc4 .
If the king were not at h6 , Wh ite would
immediately have to agree to a draw in view
of the unavoidable exchange of the queen
side pawns. But here he can sti l l play for a
win by 33 h4! c3 34 'lt>g2 c2 35 .l:!.c7 c1 'ii
(35 . . . g5 is s impler, obta in ing a theoretica l ly
d rawn ending with h-pawn against f- and g
pawns) 36 llxc1 .l:.xa2 37 .l:.c7 with the th reat
of 38 g4.
Gerardo Barbero carried out an operation
which a lso makes sense. Exploit ing the fact
that the pawn endgame is satisfactory for
Black, he decided to secure the 7th rank for
h is rook.
28 . . . l:tg8
The Theory and Practice of Rook Endings lZJ 51
29 �1
30 'it>e2
31 h4
l:.g7
l:r.d7
The immediate 31 l:.e5 came into considera
tion .
31 . . .
32 l::te5!
33 l:te6
33 . . .
'it>g7
b6
'it>f7?!
Black should have deprived the wh ite rook
of the i mportant c6-square . After 33 . . Jk7!
34 l::td6 Wf7 35 h5 'i;e7 the position would
have remained roughly equa l .
34 .l:lc6
The white rook is very wel l p laced . I t cuts off
the enemy king along the sixth rank and
p revents the advance of B lack's queenside
pawns.
34 . . . 'it>e7?!
I t was better to take active measures to
divert White from h is p lanned offensive on
the kingside: 34 . . . l:.e7+ 35 Wd3 l::td7+ 36
'it>c3 l::te7, intending . . . .l::!.e2 .
35 h5! 'i;f7?
Again passively played . 35 . . . gxh5! 36 .l:!.h6
'it>d8 was essentia l . After playing his king to
c7 , Black could then have advanced his b
and c-pawns.
36 hxg6+
37 f4
hxg6
White now has a serious advantage. B lack's
king is tied to the g6-pawn, and his rook has
to defend the 7th rank - its activity wi l l lead
to the loss of a pawn . And against passive
defence White is free to strengthen h is
position .
37 . . .
38 b3
39 'it>e3 'it>g7
40 'i;e4 'it>t7
41 'i;f3
4 1 We5 is also good .
41 . . . .l:!.e7
42 'i;g4 l:r.d7
43 Wh4
Threaten ing 44 g4 and 45 'it>g5, when after a
check on the 5th rank there fol lows f4-f5 or
'it>h6.
43 . . . l:.d2
Black decides to play actively, but it would
have been better to do th is a few moves
earl ier.
44 .l:!.c7+
45 .l:!.xa7
45 . . . Wf5 was more tenacious .
46 .l:!.c7?!
Wf6
b5?
After 46 .l:la5! 'itf5 47 '>t>h3 a second pawn
would have been lost.
46 . . .
47 I!.c6+!
c4
cJ;ts?
47 . . . Wf7 was more tenacious, but th is too
would not have saved the game: 48 bxc4
bxc4 49 a4 ! .l:!.d4 50 'it>g5 .l:!.d5+ 51 'it>g4 l:Id4
52 l:tc5! 'it>f6 (52 . . . c3 53 a5 .l:!.a4 54 'i;g5) 53
a5, and B lack is in zugzwang (53 .. J::te4 54
.l:!.c6+ 'it>f7 55 'it>g5) .
48 l::tc5+
49 l:!.xb5
50 l::tc5
'i;e6
c3
c2
52 � The Theory and Practice of Rook Endings
51 b4
52 fxg5
53 'it>h5
54 l:txc2
Black resigned .
g5+
.l:.d4+
l:txb4
I n it ial ly the ending was almost equal . What
was the reason for B lack's defeat? F i rstly,
he defended very passively, and was too
late in activating h is rook. And secondly, he
was caused a mass of problems by the
excellent position of the wh ite rook. Note the
intermed iate move 32 .Ue5 ! , wh ich secured
the ideal square c6 for the rook. From here it
tied down l iteral ly a l l the opponent's pieces
and pawns.
Yusupov - Tseshkovsky
Moscow 1 98 1
A typical situation : soon I wi l l have to g ive u p
m y rook for the c-pawn and an endgame
with rook against pawn wi l l be reached .
Every tempo may have a decisive influence
on the outcome.
The d i rect 43 .. . hxg3? (43 . . . 'it>d3? 44 l:.f2 ! or
44 l:tg2! comes to the same th ing) 44 'it>xg3
'ifi>d3 45 l:.a2 al lows White to save the game.
The main variation is qu ite instructive .
45 . . . c3 46 h4 c2 47 l:txc2 Wxc2 48 'it>t4!
Of course , 48 h5?? l:td4 ! is bad for Wh ite ,
but he also loses after 48 'it>g4? 'ifi>d3 49 h5
'it>e4 50 'it>g5 'it>e5 5 1 'it>g6 'it>e6 52 h6 l:f.g 1 + .
From f4 the wh ite king 'shoulder-charges'
B lack's, not a l lowing it to approach the
pawn .
48 . . . 'itd3 49 h5 .Uh1 50 'itg5 'it>e4 51 h6
'Wte5 52 'itg6 'Wte6 53 'lt>g7 ! (but not 53 h7?
l:tg 1 + 54 'ith6 cJilf7 55 h8lt:J+ 'it>f6 56 'ith 7
l:tg2 , and Black wins) 53 . . . 'it>e7 (53 . . . l:.g 1 +
54 'it>f8) 54 h7 l:tg1 + 55 'it>hB! with a d raw.
Unfortunately, my opponent found a much
stronger possib i l ity.
43 . . . .l:.f1 + !
44 'it>g4 hxg3
Now after 45 'it>xg3 'it>d3 46 :ta2 c3 4 7 h4 c2
48 l:1xc2 'it>xc2 White can no longer save the
game, s ince h is king cannot go to f4 .
45 .l:td2+ 'it>e3
46 l:tg2
46 l:tc2 would not have helped in view of
46 . . . l:r.f8 ! (46 . . . l:tf4+! 47 'it>xg3 l:td4 fol lowed
by 48 . . .d3 is no less strong) 47 'it>xg3
l:!.g8+ , and the king on the h-fi le is extremely
badly placed . For example, 48 'it>h4 'it>d3 49
l:t.a2 c3 50 'it>h5 c2 51 lla 1 'it>d2 52 h4 c H i
5 3 .:txc1 'it>xc1 5 4 'it>h6 'it>d2 5 5 h5 'it>e3 56
'it>h 7 l:tg 1 57 h6 'it>f4 58 'it>h8 'it>g5 59 h 7
'itg6. Or 48 c;i;>h2 'it>d3 49 .l:.a2 c3 50 h4 c2 51
l:ta 1 'it>d2 52 Wh3 c 1 'ili' 53 l:txc1 'Wtxc1 54 h5
'it>d2 55 'it>h4 'it>e3 56 h6 Wf4 57 'it>h5 .l:Ig5+ -
in both cases Wh ite loses.
46 . . .
47 'iitxg3
48 h4
49 l:.c2
.Uf4+1
c3
.l:Ic4
$>d3
But now the fact that the wh ite king is cut off
along the 4th rank proves decisive.
50 l:lc1 c2
51 h5 'it>d2
52 l:th1 c1'ji'
53 J:txc1
White resigned .
'lt>xc1
The Theory and Practice of Rook Endings ltJ 53
Two ideas , typical of such endings, were
clearly seen during the course of the play:
1 ) the cutting off of the king along the 4th
rank - thanks to th is , B lack won the game;
2) the 'shoulder-charge' - Wh ite was hoping
to save himself by employing this device ,
bu t with an intermed iate check Vita ly
Tseshkovsky disrupted my p lan .
I n the fol lowing, more compl icated ending ,
simi lar motifs occurred .
Yusupov - Timman
Candidates Match , 5th Game, Ti lburg 1 986
The evaluation of the position is not in
question - White has a big advantage. The
log ical move was 38 a4! - it is important to
advance the passed pawn as soon as
possible. B lack's passed pawn is not dan
gerous - to . . . e4-e3 there is a lways the
reply 'it>f1 .
·
How could the game have developed?
38 . . . l:i.d3 39 a5 c3 40 bxc3 e3 (threatening
41 . . . l:i.d 1 + 42 'it>g2 e2) 41 'it>f1 .l:l.xc3 42 a6
:t:l.a3 43 .l:l.b6 � (43 . . . g5 is bad: 44 Ug6+
rttf7 45 .l:i.xg5 l:i.xa6 46 .Ue5 l:i.a3 4 7 'it>e2
:t:l.a2+ 48 'iin3 l:th2 49 'it>g3 l:.e2 50 'it>f4) 44
g5.
If Black stays passive, he ends up in
zugzwang (for example , 44 . . . .l::!.a2 45 h4 ) .
He is forced to exchange pawns: 44 . . . e2+
45 'it>xe2 .Uxh3. Now, after s l ightly improving
the position of h is rook in a typical way: 46
l:i.f6+! 'it>g7 47 l:i.c6 (threatening 48 l:i.c7+
'it'f8 49 a7) 47 . . . 'it'f7 , White plays 48 'it'd2.
I f 48 . . . l:i.a3 there fol lows 49 'it'c2 lla5 50
'it>c3 .l:!.xg5 51 'it>b4 .U.g 1 52 'it'c5 g5 53 a7
l:i.a 1 54 'it'b6 g4 55 Wb7 and wins (the black
king is cut off from its passed pawn ) .
And if 48 . . . .Ug3, then 49 a7 .Ua3 50 l1c7+
'it'e6 51 'it>c2 Wf5 52 '1t>b2 l1a6 53 'it'b3
'it>xg5 54 Wb4 (th reatening 55 .l:lc5+ and 56
.l::!.a5) 54 . . . 'it>h6 (the on ly defence) 55 'it'b5
.l::!.a 1 56 'it'b6 .
Look in Mark Dvoretsky's book School of
Chess Excellence 1: Endgame Analysis -
there in the chapter 'Rook against Pawns' a
very s imi lar position is analysed . The best
defence - 56 .. Jib1 + (if 56 . . . g5 the most
accurate is 57 .l::!.c8 ! ) 57 'it>c6 .l::!.a1 58 'it>b7
IIb1 + 59 'it'c8 .l::!.a 1 - al l the same does not
help : 60 'it'b8 'it>g5 61 a81i' I:Ixa8+ 62 Wxa8
'it'f4 63 .Uf7+! (a typical i ntermed iate check
to gain a tempo; the hasty 63 Wb 7? g5 leads
to a draw) 63 . . . 'it>e4 64 .l::!.g7! 'it>f5 65 'it>b7 g5
66 'it>c6 g4 67 �d5 ..t>f4 68 'it>d4 ..t>f3 69
Wd3 g3 70 .l::!.f7+ and 71 'it'e2. Or 60 . . . g5 6 1
a81i' l:txa8+ 6 2 'it'xa8 Wh5 (62 . . . g 4 6 3 .Uc5!
- cutting off the king ! ) 63 'it'b7 'it'g4 64 'it>c6
54 � The Theory and Practice of Rook Endings
'it>f3 65 .l:if7+! We3 66 .l:ig7! 'it>f4 67 Wd5 etc.
Incidental ly, now you wi l l see why on the
46th and 47th moves White replaced h is
rook at c6 - in order to free the b6-square for
his king. Such 'trifles' sometimes decisively
affect the result of the game, and on no
account should they be disregarded .
Unfortunately, in a t ime scramble I commit
ted a sign ificant inaccuracy.
38 Wf2 .:d3
Of course, the king must not be a l lowed to
go to e3.
39 a4 c3
39 . . . .l:ixh3? would have lost: 40 l:!.c5 l:!.b3 4 1
.l:ixc4 .l:ixb2+ 4 2 We3 . After the inevitable fa l l
of the e4-pawn White ach ieves an ideal
construction - his rook defends both pawns
along the 4th rank , not a l lowing any
counterplay. The king can calmly approach
the a-pawn.
There was also the interesting move 39 . . . Wf7
with the idea after 40 .l:ic5?! 'i.t>e6! of
supporting the passed e-pawn with the king .
40 a5 c3 4 1 b4 ! is stronger (but not 4 1 bxc3
.l:ixc3 with a draw). After 4 1 . . J::td4 42 \t>e3!
l:.c4 43 l:.c5 l:.xb4 44 .l:ixc3 l:.a4 45 .:c5 the
same winning position , as occurred later in
the game, is reached .
40 bxc3
40 . . . l:!.xc3?
On the last move before the t ime control Jan
Timman makes a decisive mistake. He
thought that he wou ld always be able to
advance his pawn to e3, but he d id not take
account of the strong impeding reply 41
.l:ie5 ! .
If Tim man had seen th is , then , even without
delving i nto variations, s imply by the method
of comparison he would have preferred
40 . . . e3+ ! 41 'i.t>e2 .:xc3 . Here the black rook
is s l ightly more active , and the wh ite king is
s l ightly further from its kingside pawns than
in the game. After 42 g5 l:!.a3 43 a5 '>t.>f7 44
l:!.e5 '>t.>f8 the position wou ld apparently have
been drawn . For example : 45 l:!.e6 .l:i.xa5
(45 . . . l:!.a2+ fol lowed by 46 . . J::txa5 is even
more precise) 46 I!.xg6 .l:te5 4 7 h4 'it>f7 48
Itf6+ 'it>g? 49 .l:.f3 .l:!.e4 50 h5 .l:ih4 51 h6+
Wg6.
41 l:.e5!
Here the game was adjou rned . Analysis
showed that White wins without d ifficu lty.
41 . . . l:!.c4
The situation ar is ing after 4 1 . . .l:!.xh3 42
.l:ixe4 has a l ready been d iscussed . In the
event of 42 . . . '>t>f7 (with the idea of bring ing
the king to g5) the strongest is 43 g5 ! .
42 a5 .l:i.a4
43 'it>e3 �g7
44 g5 'it>f7
45 h4
46 'it>f4
'it>g7
Also good is 46 .l:!.e 7 + 'it>f8 4 7 .l:!.e6 .l:!.a3+ 48
�f4 (or even 48 �xe4 .l:!.xa5 49 .l:.xg6 .l:!.a4+
50 '>t>f5 .l:.xh4 51 .l:!.a6) .
46 . . .
47 .l:.b5
d5 49 .l:!.xg6
or 49 a6, whi le if 47 .. :.t>g7 there fol lows 48
l:!.b7+ 'it>f8 49 .l:i.b6 .l:i.xa5 (49 . . . '1t>f7 50 a6) 50
.l:i.xg6 with an easy win . I n the last variation
The Theory and Practice of Rook Endings LtJ 55
we clearly see the d ifference in the position
of the black pawn - with the pawn on e3
there would have been no win .
47 . . .
48 Wxe3
e3+
�e6
The only active chance - in reply to the rook
check the king can now go to f5 .
49 l:tb6+ �f5
50 a6 'it>g4
What would have happened after the cap
ture of the h4-pawn? Of cou rse, 51 .l:!.b5+
and 52 .l:!.a5 - how can one not exploit an
opportun ity to place the rook behind the
passed pawn!
51 l:f.xg6
51 .Ud6? was less good in view of 51 .. .h5
and 52 ... .ttxh4 .
51 . . .
52 �d3
53 .l:!.c6
54 'it>c3
55 �b3
56 'it>c4
57 �c5!
�xh4
�h5
'it>xg5
'it>f5
l::ta1
�e5
The final touch - a 'shoulder-charge' .
57 . . . I1a2
58 �b6
Black resigned .
The 1998 No. 5-6 issue of the magazine
Shakhmaty v Rossii published an extensive
article by Alexey Kuzmin, in which the
grandmaster pointed out a number of mis
takes in the analysis of the Yusupov
Timman ending. It turns out that 40 ... e3+
would not have saved Timman. White suc
ceeds in breaking through decisively with his
king: 41 ..ti>f3! 1hc3 42 l:!.e5 l:ta3 43 �f4!
.!:!.xa4+ 44 �g5, and the resulting ending with
two pawns against one is won. This means
that there is no reason to criticise his choice
on the 38th move.
Moreover, the alternative 38 a4?! would in
fact have thrown away the win. The variation
considered by Yusupov - 38 ... 'iJ.d3 39 a5 c3
40 bxc3 e3 4 1 �f1 l:f.xc3 42 a6 .l:!.a3 43 l:!.b6
Wf7 44 g5 e2+ 45 Wxe2 11xh3 46 .l:tf6+ 'it>g7
4 7 .l:i.c6 �f7 48 �d2 l:!.g3 49 a 7 'iJ.a3 50 .l:!.c 7 +
'it'e6 51 Wc2 Wf5 52 Wb2 etc. is unconvinc
ing. Instead of the incorrect 5 1 ... Wf5? Black
should hold his ground: 5 1 ... �d6! 52 �g7
Wc5 53 �b2 l:f.a6 54 Wb3 .:ta1 , and White is
unable to strengthen his position. He in turn
could have successfully forced events ear
lier: 4 1 a6! (instead of 4 1 �f1) 4 1 ... '1J.d1 + 42
Wg2 e2 43 a7 l:f.a1 (43 ... 'iig 1 + 44 �f3) 44
l:f.e5 l:f.xa7 45 l:f.xe2. But this happened only
as a result of Black's mistake 40 ... e3?.
According to Kuzmin's analysis, by playing
40 ... '1J.xc3! 4 1 a6 'iJ.a3 42 .l:!.b6 Wf7 43 h4 e3
he would have gained a draw.
I n conclusion , here is a very compl icated ,
purely analytical end ing .
Yusupov - Mestel
Esbjerg 1 980
This interesting position with an unusual
wh ite pawn configuration on the kingside
arose immediately after the adjournment.
Only a win offered me chances of taking fi rst
place in the tou rnament and achieving the
56 c3
White has an extra piece in play - his k ing.
This factor wi l l tel l , for example, if Black
plays 'accord ing to the ru les' (but in fact
routinely): 43 . . . .l:!.a8? - placing h is rook
behind the passed pawn . In this case there
fol lows 44 l::td2, then .l:!.a2 and 'it>xc4, and the
a-pawn wi l l be lost. Black cannot defend in
this way - he is clearly too late with his
counterplay.
The best chance was 43 . . . e5! . Jonathan
Mestel d id not play this , because he was
afra id of 44 .l:td2 'i¥te6 45 'it>xc4 . But after
45 . . . .l:i.c8+ 46 'it>b5 l:ta8! (46 . . . a3? loses to 47
'.ta4 .l:!.c3 48 b5) 47 'iilc6 (47 l:ta2 'iild5 ! 48
.l:i.xa4 .l:!.b8+) 47 . . .a3 48 .l:!.a2 .l:i.c8+ 49 '.tb7
.l:!.c3 50 b5 'iild5 51 b6 'it>c4 Black saves the
game. 52 '.taB 'it>b3 53 .l:!.xa3+ 'iilxa3 54 b7
.l:!.b3 leads to a drawn pawn ending , while if
52 Wa6 , then 52 . . . 'it>b3 53 b7 Wxa2 54 b8'ik
.l:!.b3 55 'ii'xe5 'it>b1 56 'iVe1 + 'it>b2 57 �e2+
'lt>b1 58 Wa5 a2 59 'it>a4 .l:.xf3 ! (59 . . . .l:.b7 !?
60 'it'e4+ 'lt>c1 6 1 'il¥c6+ 'it>d2 is a lso
possible) 60 'ii'd 1 + 'it>b2 6 1 'Yi'xf3 a 1 'ti'+ 62
'it>b4 iVa? with a drawn queen end ing .
I was intending 44 .l:!.a7 ! , but then Black
activates h is rook by 44 . . J!d8! . After 45 b5!
(weaker is 45 'iilxc4 .l::td2 46 b5 l:tc2+ ! )
45 . . . 'iile6! (45 . . . .l:i.d3+? is bad : 46 Wxc4 l::txf3
47 b6 .l:!.xf2 48 .l::txa4 , or 47 . . . .l::tb3 48 b7 Wf5
49 'it>c5) 46 'i¥txc4 (46 b6? .l:.b8) 46 . . . .l::td2 47
b6! (47 .l:.xa4 .l:.c2+ ! ) Black would have had
to make a difficult choice between 47 . . . .Uc2+,
47 . . . Uxf2 and 47 . . . '1¥td6.
1 ) 4 7 . . . .l:.c2+ 48 'i¥tb5 .l:i.b2+ 49 '.tc6 .l:!.c2+ 50
'it>b7 .l:!.xf2 (note that Black has employed a
typical procedu re - he has fi rst d riven the
king onto the square in front of its own pawn ,
and only then captu red a pawn) 5 1 'i¥ta8
.l:!.xf3 52 b7 l:!.b3 53 llxa4! (a l l the same the
pawn has to be captu red , so it is better to do
this immediately, in order to halt the passed
e5-pawn, even if only for a �oment) 53 . . . f5
54 b8'ii' .l:.xb8+ 55 'i¥txb8 .
The result ing sharp position would appear to
be won , for example: 55 . . . e4 56 .l::ta5! (it is
important to cut off the black king ) 56 . . . h6
(56 . . . e3 57 .Ua3 f4 58 gxf4 'it>f5 59 llxe3) 59
.l:.a6+! '>td5 60 llxg6 e3 6 1 l:!.g8 '.te4 62 'iilc7
'it>f3 63 'it>d6 e2 64 l:te8 'it>xg3 65 .l:!.xe2 f4 66
'it>e5 or 66 h5.
2 ) 47 . . . .l:!.xf2 48 l:!.xa4 (48 b7 llb2 49 'i¥tc5
does not work in view of 49 . . . a3 ! ) 48 . . . Wd7?!
(what happens after the strongest move
48 . . . '1¥td6! wi l l be seen in the analysis of the
fol lowing variation) 49 .Ua7+ 'it>c6 50 .l:.xf7
'i¥txb6 5 1 'i¥td5, and White must win , s ince
the black king is too far away from the
kingside pawns.
3 ) 47 . . . '1¥td6 48 llxa4 (48 :Xf7 a3 49 .l:.xh7
a2 50 l:!a7 .l::txf2 leads to a d raw) 48 . . . .l:.xf2
Black loses after 48 . . . 'it>c6? 49 .l:!.a7 f5 50
.l:!.xh7 'i¥txb6 51 .l:!.h6 .l:!.xf2 52 .l:!.xg6+ 'it>c7 53
'i¥td 5 .l:!.xf3 54 'iilxe5.
49 .l:!.b4 l:!.c2+ 50 'iilb5
The Theory and Practice of Rook Endings ltJ 57
I n it ial ly I thought that the pawn ending
arising after 50 . . . 'it>d71 51 b7 'it>c7 52 l:.c4+
l:!.xc4 53 'it>xc4 'it>xb7 54 'it>d5 was won in
view of 54 . . . 'it>c7 55 'it>xe5 'it>d7 56 'it>f6 'it>e8
57 'it>g7 h5 58 g4.
But Black can defend more strongly: 54 . . . f6 ! .
Now it does not help to play 5 5 'it>e6 'it>c6 S6
�xf6 'it>d5 57 'it>g7 'it>d4 58 'it>xh7 'it>e3 59
'ittxg6 'iitxf3 60 h5 e4 or 57 g4 'it>d4 58 h5
gxh5 (58 . . . 'it>e3 is also possible) 59 gxh5 e4 !
60 fxe4 'it>xe4 6 1 'it>g 7 'it>f5 62 'it>xh 7 'it>f6 !
with a draw. If instead 55 g4, then 55 . . . h5!
( 55 . . . 'it>b6? 56 g5! or 55 . . . 'it>c7? 56 'it>e6 'it>c6
57 g5! ) 56 gxh5 gxh5 57 'it>e6 'it>c6 58 'it>xf6
�d5 59 'it>g5 e4! , and again it is a d raw.
As you can see , to find the narrow path
enabl ing Black to hold on is extremely
d ifficu lt, even in home ana lysis. In any
event, Black was obl iged to play 43 . . . e5! .
The continuation in the game loses without
a fight.
43 . . .
44 'it>xc4
45 lla7
46 b5
l:tb5?
l1d5
l:td2
In essence, B lack has s imply lost a tempo.
In the analogous posit ion , wh ich we have
a l ready ana lysed , the pawn stood at e5 and
the king cou ld be brought i nto play with
. . . 'it>e6. Now this resource is not avai lab le ,
and therefore Black has no defence. I f , for
example , 46 . . . l:!.xf2 , then 47 b6 l:.b2 48 b7
a3 49 'it>c5 a2 50 'it>c6 .
46 . . . .l:tc2+
47 'it>b4 e5
47 . . . l:!.xf2 48 l:!.xa4 ltxf3 49 b6.
48 b6 l:txf2
49 b7 'it>f5
49 . . . l:.b2+ 50 'it>c5, threaten ing 5 1 l:ta6+ and
52 l:.b6 .
50 g4+!
51 l:txa4
Black resigned .
This game also g ives an opportun ity for
d iscussion about the techn ique of playing
the endgame. Every tempo, even a seem
ing ly ins ign ificant one (such as . . . e6-e5! ) ,
can have a s ign ificant and possib ly decisive
influence on the outcome of the game. You
should a lways choose carefu l ly the most
accurate way of putt ing your plans into
practice .
58 � The Theory and Practice of Rook Endings
Exercises
1 . White to move 2. B lack to move
3. White to move 4 . Wh ite to move
The Theory and Practice of Rook Endings
5. White to move
7 . Black to move
9. Black to move
How would you evaluate the position?
6 . B lack to move
Is 59 . . . �g4 possible?
8 . White to move
1 0 . B lack to move
Is 1 . . . a2 good or bad?
CZJ 59
60 � The Theory and Practice of Rook Endings
Sol utions
1 . A . Rinck ( 1 906) .
1 f6 .llxe2
1 . . . 'it>b5 2 �h8 .l:i.d7 3 �e8 , or 1 . . . lld4 2 l:!.e7
�e4 3 .lle8 and wins.
2 .l:th5+!
2 .l:!.h8? .l:!.f2 3 �f8 'it>b6 4 f7 'it>b7 5 'it>c4 .l:tf5 .
2 . . . �b6
3 .l:tf5 ! !
2. Trabattoni-Barlov (Valetta 1 979) .
It is not enough to find a good idea ; you must
also choose the most accurate way of
carrying it out. Wh ite has two moves: 1 �e6
and 1 �g5. Which of them is correct?
1 :C.e6! �g2+
2 'it>h1 ! .Uxg3
3 .l:!.xg6!
The game went 1 :C.g5? :C.g2+ 2 �h1 .l:tf2!
(2 . . . .l:!.xg3? 3 :C.xg6! ) 3 'lt>g1 .l:tf6. White found
h imself in zugzwang and resigned a few
moves later.
3. P. Romanovsky ( 1 950) .
The rook must be switched to the 6th rank,
but how can this be achieved? 1 . . . �f1 +? 2
�e5 .l:!.f6 3 l:tg8+ is not possib le, and
1 . . J1b1 ? 2 .l:ta7+ 'it>h6 3 .l:!.b7 .l:!.a 1 4 a7 a lso
loses .
1 . . . �a5+!
2 'it>e6
If 2 'it>e4 , then 2 . . . l:!.b5 3 .l:!.a7+ (3 .l:!.c8 lla5 4
:C.c6 'it>f7) 3 . . . 'it>g6 4 .l:!.b7 .l:!.a5 5 a7 'it>f6 6
'it>d4 'it>e6 7 'it>c4 'it>d6 8 'it>b4 Ita 1 (or
8 . . . 'it>c6) with a draw.
2 . . . .l:!.h5! !
The only way! 2 . . J::tb5? is bad in view of 3
.l:ta7+ and 4 l:!.b7 . If instead 2 . . . .l:!.g5?, then 3
.l:ta7+ 'it>g8 4 �f6! .l:!.a5 5 .l:ta8+ �h7 6 'it>e7.
3 'it'd?
3 .Ua7+ �g8 4 l:!.f7 �a5 5 .l:ta7 l:th5 ! .
3 . . . l:th6!
4 �c7 �f6!
The draw becomes obvious, for example: 5
a7 .l:!.f7+! (5 . . . .l:!.a6? 6 'it>b7) 6 'it>d6 .l:tf6+ 7
�e5 l:ta6 .
4 . Vaisser-Martinovic
(Vrnjacka Banja 1 984 ) .
1 'it'd 1 ! ! gxh4
2 �xh4+ 'it>xf3
3 .l:!.h5
3 .l:!.h3+ �g2 4 l:i.h5! l:.f2 5 'it>e1 is a lso
possible.
3 . . .
4 l:Ib5
5 We1
'it>g4
f5
And the game soon ended i n a d raw.
All other king moves lead to a loss:
A) 1 'it>d3? gxh4 2 .l:!.xh4+ 'it>xf3 3 .l:!.h5 'it>g4
and 4 . . .f5 . The wh ite king is stuck on the
' long side' of the pawn .
B ) 1 'it>e1 ? 'it>e3! 2 'it>d 1 gxh4 3 .l:txh4 f5 ! 4 f4
.l:!.a 1 + 5 �c2 .l:!.f1 6 .l:!.h3+ .l:!.f3 7 l:th8 .l:!.xf4.
C) 1 'it>f1 ? 'it>xf3 2 'lt>g 1 (2 'it>e 1 .Ua 1 + 3 'it>d2
gxh4 4 .l:!.f5+ ..t>g3 5 .l:!.xf7 h3 6 .l:!.g7+ �f4 7
.l:tf7+ �e5 8 llh7 h2) 2 . . . .Ug2+ !
Not 2 . . . gxh4? 3 .l:!.f5+! (3 .l::i.xh4? Wg3)
3 . . . �g3 4 .l:txf7 or 2 . . . g4? 3 .l:!.f5+ �g3 4 h5!
.l:!.g2+ (4 . . . .l:!.a 1 + 5 .l:tf1 ) 5 �f1 l:th2 6 �g 1
with a d raw.
3 �h 1 (3 'it>f1 tlh2 4 l:i.xg5 .l:!. h 1 + 5 .l:!.g 1
l:::txg 1 + 6 �xg 1 �g4) 3 . . . g4 4 l:!.f5+ 'lt>g3 5 h5
(or 5 .l:!.xf7) 5 . . . .Uf2 ! , and Wh ite can resign .
5. Dorfman-Kholmov (Saratov 1 98 1 ) .
43 .l:!.d5! .Ue4
The Theory and Practice of Rook Endings l2J 61
43 . . J1b8 44 .l:txd7.
44 .Uxb5
And White retains excellent winning chances,
for example: 44 . . . d5 45 .Ub7 ! , cutting off the
king on the 8th rank , or 44 . . :;itt? 45 �e2 !
and 46 Wd3 .
The game continuation 43 l1xd7? was amistake because of 43 .. J1e4 ! . The point is
that i f 44 l:td4 Black can go i nto the pawn
end ing: 44 . . . l:Ixd4! 45 exd4 )!;>f7 , and if 46 d5
rt:ie7 47 'it>e3, then 47 . . . Wd7 ! 48 d6 (48 Wd4
rt:id6) 48 . . . 'it>c6 ! 49 Wd3 c.t>d7! 50 'it>d4 'it>xd6.
White's comparatively best chance is to go
into a queen endgame by 46 )!;>g2 ( instead
of 46 d5} 46 . . . )!;>e6 47 �h3 )!;>d5 48 )!;>h4
rt:ixd4 49 'it>g5 'it>c4 50 'it>xg6 �xb4 51 'it>xf5
rt:ic4 52 'it>g5! (52 We5 b4 53 f5 b3 54 f6 b2
55 f7 b1'iV 56 f8'ii' 'ife1 +) 52 . . . b4 53 f5 �d5 !
54 f6 'it>e6 55 �g6 b3 56 f7 b2 57 f8'if b 1 'iV+
58 \t>g5 (58 Wg7 it'b2+ 59 �g8 'ir'f6 ! ) - in
th is case the opponent is sti l l requ i red to
defend accurately, although objectively the
position is a d raw.
44 g4 fxg4 45 l1d4 g3+! 46 'itxg3 (46 Wf3
g2) 46 . . . l1xe3+ 47 'itg4 l:r.b3 48 f5 � 49
rt:Jf4 gxf5 50 'it>xf5 'ite7 51 )!;>e5 l:Ib1 52 )!;>d5
.!:!.c1 Draw.
An interesting attempt to play for a win was
suggested by Viorel Bologan : 43 �f3?! I1e4
44 g4 ! .l:txb4 (44 . . . Wf7 45 .Uxd7+ We6 46
.!:!.g7 �f6 47 l1b7 .l:txb4 48 g5+ We6 49 l:Ig7)
45 gxf5 gxf5 46 .l:txd7.
(see diagram)
Black can apparently hope to save the game
after 46 . . . l:!.c4 47 .ti.d5 .l:Ic7 48 .ti.xf5 (48 .ti.xb5
.M.f7) 48 . . . .ti.b7, for example: 49 l:td5 b4 50
J::i.d2 b3 51 .l:tb2 Wf7 52 'it>e4 )!;>e6 53
'.t>d4(d3) 'it>f5 , or 49 'ite2 b4 50 )!;>d2 b3 51
rt:ic1 .l:te7 52 .l:tg5+ 'it>f7 53 l:tg3 Wf6 .
6. Portisch-Petrosian (Cand idates Match ,
1 2th game, Palma de Mal lorca 1 974) .
I n the game Black preferred the cautious
59 . . . We6 and after 60 'it>c5 he made a
decis ive mistake : 60 . . J1c2+? (60 . . . Wd7!
was necessary, with good drawing chances) .
There fol lowed 61 c.t>b5 Wd6 62 'it>a6 'itc6
63 l1a1 l:Ic4 64 b7 l1b4 65 l1c1 + �d7 66
lieS B lack resigned .
As was shown by Igor Zaitsev, the active
king move wou ld have secured a draw, but
only if B lack had found a far from obvious
defensive idea .
59 . . .
60 I1a4!
Threatening 6 1 Wc3+ .
'it>g4!
60 . . . c.t>h3 ! !
6 0 . . . Wg3? i s hopeless: 6 1 Wc5 f5 6 2 l:!.b4
.l:tc2+ 63 c.t>d6 lk8 64 b7 .l:tb8 65 Wc7 .l:th8
66 b8'iV .l:txb8 67 .l:txb8 �xh4 (67 . . f4 68 Wd6
f3 69 We5 f2 70 .l:tf8 Wxh4 71 We4) 68 )!;>d6
'lt>g3 69 )!;>e5 h4 70 Wxf5 h3 71 l:tb3+ Wg2
(7 1 . . . 'it> h4 72 �f4 h2 73 .l:tb 1 ) 72 'it>g4 h2 73
lib2+ 'itg 1 74 'itg3 , or 68 . . . �g4 69 )!;>e5 h4
70 �d4! (70 .l:tb4+? f4! 71 I1xf4+ Wg3)
70 . . . h3 (70 . . . f4 7 1 'it>d3 'it>f3 72 l1h8 ! ) 71
)!;>e3 h2 72 .l:tg8+ �h3 73 �f2 h 1 4J+ 74 Wf3 .
61 'it>c5 f5
62 .l:tb4 .l:Ixb4!
62 � The Theory and Practice of Rook Endings
63 �xb4
64 b7
65 bB'iV
And White cannot win .
7. Petrosian-Karpov
f4
f3
f2
(44th USSR Championsh ip , Moscow 1 976).
51 . . . 'it>h6? 52 f7 �h7 (52 . . J1a 1 53 'it>g8) 53
h6 'lt>xh6 (53 . . . .l:!.a 1 54 .l::txc2 a lso does not
help) 54 'lt>g8 leads to a position from
Emanuel Lasker's famous study. Wh ite wins
by gradual ly pushing back the opponent's
king: 54 . . J:ig 1 + 55 �h8 .i':lf1 56 .i':lc6+ Wh5
57 'it>g7 l:r.g 1 + 58 'it>h7 .l:!.f1 59 l:lc5+ 'it>h4 60
'it>g7 .l:!.g 1 + 61 'it>h6 l:tf1 62 .l:!.c4+ 'it>h3 63
�g6 l:tg 1 + 64 Wh5 l:tf1 65 l:tc3+ �h2 66
l:txc2+.
51 . . .
52 f7
53 'it>e7
'it>h8!
tl.a1 !
I n the game there fol lowed 53 l:i.xc2 .l:!.a8+
54 'it>e7 .l:ta7+ 55 'it>f6 .l:ta6+ 56 'lt>g5 .l:ta5+
57 Wg4 .l:ta4+ 58 'it>g3 .l:ta3+ 59 Wg2 'lt>g7
60 .l:tf2 'it>fB 61 .l:!.f5 l:ta6! (6 1 . . . .l:la7? 62 h6
.l:!.xf7 63 h7 or 62 . . . l:ta6 63 .l:!.h5) 62 �g3 .l::th6
63 'it>g4. Draw agreed in view of 63 . . . .l:lh7.
53 . . . l:te1 +
54 'it>f6 tl.f1 +
55 'it>g6
56 'it>h6
57 .l:!.xc1
.Ug1 +
c1 "ii'+!
.l:lg6+!
Either capture leads to stalemate.
8. Makarychev-Vasyukov
(Vi ln ius 1 980/8 1 ; variation from the game).
Black wil l win if he can manage to advance
the h-pawn just one step more . After the
obvious 1 .l:la 1 ? �h5 (th reatening 2 . . . h3) 2
l:th 1 l:th6! Wh ite ends up in zugzwang : 3
'it>e7 'it>g4 4 .l:lg 1 + �f4 with the th reats of
5 . . . 'it>xe5 and 5 . . . h3 , or 3 tl.h2 'it>g4 4 l:tg2+
'it>f3 5 .l:!.h2 �g3 and 6 . . . h3 .
Wh ite needs to obta in the same position , but
with Black to move .
1 l:ta2! ! Wh5
1 . . . 'it>g5 2 l:tg2+! 'it>h5 3 .i':lh2, or 2 . . . Wf5 3
.llf2+ 'lt>e4 4 .l:!.f6 ! .
2 .l:th2 .l:!.h6
3 .l:lh1 !
Now it is B lack who is in zugzwang . He is
unable to win .
3 . . . 'it>g4 4 .l:!.g1 + �f3 5 .l:tf1 + �g2 6 .Uf6 .l:!.h8
7 l:txe6 h3 8 l:tg6+ 'it>f2 9 l:lf6+ We2 1 0 .l:!.g6!
h2 11 .l:!.g2+ 'it>f3 1 2 l:lxh2 .l:txh2 1 3 e6.
9 . Larsen-Kavalek
(7th match game, Sol ingen 1 970) .
Wh ite wants to play .l:.c4 . I f B lack is forced
total ly onto the defensive with . . . .l:!.a7, then
White's extra pawn together with the pas
sive black rook should ensure him a
stra ightforward win .
That is what happened in the game:
1 . . . 'it>g7? 2 .l:!.c4 .l:ta7 (2 . . . .l:tb3 3 l:lxa4 l:txg3
4 tl.g4+) 3 'it>c3 h5 4 Wb4 'it>g6 5 .Uc6+ 'it>g7
6 tl.c5 Wh6 7 'it>b5 l:le 7 (otherwise 8 .l:!.c4) 8
'it>xa4 l:te3 9 g4 hxg4 1 0 hxg4 .l:te4+ 1 1
�b5 .Uxg4 1 2 a4 .l::!.g 1 1 3 a5 l:lb1 + 1 4 Wc6
.l:la1 1 5 'it>b6 l:tb1 + 1 6 .l:tb5 .l:lf1 1 7 a6 tl.f6+
1 8 'it>a5 :tf7 1 9 .l:lb6+ 'it>g5 20 tl.b7 .l:lf1 21 a7
'lt>h6 22 .Ub6+ 'it>g7 23 .l:la6 Black resigned .
The Theory and Practice of Rook Endings ctJ 63
1 . . .Wf7, suggests itself, i n order to meet 2
:!:ic4 with the counter-attack 2 . . . �b3 ! . But
White plays 2 g4! , intending 3 h4 and only
then 4 l'ic4. After 2 . . . �e6 3 h4 'it>d5 White's
threat is parried , but 4 g5! creates a new
threat: 5 �g3 fol lowed by 6 �g4 or 6 h5 (rook
beh ind the passed pawn) . B lack's posit ion
becomes critica l .
1 . . . h5! !
The only way to save the game! I n the event
of 2 �c4 .ti.b3! 3 .UXa4 .l:t.xg3 there is no check
on g4 - the position is d rawn . I f 2 h4, then
2 . . . .l:l.g7 and 3 . . . .ti.g4, if there is noth ing better.
The rook on g4 is very active - it attacks the
white pawn , defends its own and restricts the
mobil ity of the wh ite king . F ina l ly, if 2 g4 there
is the reply 2 . . . h4! , fixing the white h3-pawn
as a target for a counter-attack along the 3rd
rank (in the event of .l::tc4 ) .
1 0 . Moiseev-Bagirov (Moscow 1 956).
Sooner or later Black wi l l have to play . . . a3-
a2 (the march of the king to the a2-square is
obviously unreal ) . The only question is
whether at the same time he can obta in a
second passed pawn on the f-fi le .
There was a stra ightforward win by 1 . . . g5! 2
hxg5 'it>g6 3 .l:Ia7 a2! 4 'it>h2 'it>h5! 5 'it>g2 g6
6 �a4 'it>xg5. Then Black moves h is king
and plays . . . g6-g5 and . . . f5-f4 , achieving
his a im .
I n the game he chose a different, far less
successfu l move order.
1 . . .
2 .Ua6!
a2?
'it>h6
I n the event of 2 . . . g5 3 hxg5 the king cannot
approach the g5-pawn.
3 �h2 'lt>h5
Can anyth ing be done against the threat
ened . . . g6-g5 ?
4 .Ua4!
5 l1a6!
6 .Ua4!
7 g4+ ! !
'it>h6
'it>h5
g5
This is the whole point - Black can no longer
obta in a passed pawn on the f-fi le .
The game concluded : 7 . . . 'lt>xh4 8 gxf5+ g4
9 'lt>g2 .l:!.b1 1 0 llxa2 .l:!.b4 1 1 .i:tc2 g3 1 2 .Ua2
'it'g4 1 3 llc2 .l:!.f4 1 4 llc8
After 1 4 .Uc7 .Uf2+ 1 5 'lt>g 1 .Ua2 White
cannot play 1 6 .Uxg7? 'it>f3 . It was for the
sake of such variations that Black left the f5-
pawn al ive .
1 4 . . . .Uf2+ 1 5 'it>g1 .l:!.e2 1 6 .Ua8 'it>f3 1 7 .l::!.a3+
.Ue3 18 .Ua1 g2 ( 1 8 . . . 'it>g4 1 9 .Ua8) 19 'lt>h2!
( 1 9 .l:!.b 1 ? 'it>g3 20 .l::!.a 1 .Uf3 2 1 .Ub 1 l'if1 +)
19 . . . 'it>f2 20 .l:!.a2+ .l::!.e2 21 .l::!.xe2+ 'it>xe2 22
'it>xg2 Draw.
64 �
Mark Dvoretsky
From the Sim ple to the Com pl icated
TheTheory of Endi ngs with Opposite-colour Bishops
When trying to master any type of
endgame the most important thing is
to lay a firm foundation : to pick out the most
important theoretical positions, ideas and
techniques, which underl ie our notions of
the endgame in question . As a rule, the
necessary basic knowledge is made up of a
small number of rather s imple positions, but
they must be understood in every deta i l and
fi rmly remembered .
A successfu l ly developed system of basic
endgame knowledge provides a rel iable
guide in the analysis of more compl icated
situations and helps them to be more
successfu l ly understood . How this occurs , I
wi l l show using an example of endings with
opposite-colou r bishops.
Connected passed pawns
Let's d iscuss in detai l the fol lowing elemen
tary ending.
after a bishop check) , then 'it>e5 and f5-f6 .
I n order to counter this p lan , B lack must take
control of the e6-square. But from where , d 7
or b3? We wi l l consider both poss ib i l it ies.
After 1 . . . i..b3? the position is lost. F i rst
Wh ite g ives a verifying check with his
bishop, i n order to determine the position of
the enemy king . I n so doing it is important
that the bishop should prevent the king
from wedging itself between the pawns
after White plays e5-e6. Hence , 2 i..g5+ ! .
Next the white king makes a by-pass
manoeuvre to the aid of the e-pawn, on
the opposite side to where the oppo
nent's king has moved. For example:
2 . . . 'it>f7 3 'it>d4 i..a2 4 �c5 i..b3 (4 . . . i..b 1 5
e6+ and 6 f6) 5 �d6 and 6 e6+ . Or 2 . . . 'it>d7 3
�f4 i..a2 4 i..h4 i..f7 5 �g5 �e7 6 �h6+
�d7 7 �g7 i..c4 8 'it>f6 and 9 e6+ . After the
pawns have reached e6 and f6 , even if the
threat of f6-f7+ is parried , Wh ite repeats the
same procedure : a verifying check with the
bishop and a by-pass by the king .
It is incorrect to play 2 il.b4+? �f7 ! (Black's
only hope is to provoke a prematu re e5-e6+
and wedge h is k ing between the pawns) 3
�d4? i..c2 ! 4 e6+ �f6 5 e7 i..a4 with a draw.
As soon as the pawns are blockaded on
squares of the same colour as their
bishop, the draw becomes obvious.
Thus, with h is bishop on b3 Black loses. On
the other hand , 1 . . . i..d7! 2 i..g5+ �f7 1eads
to an easy draw. Subsequently B lack waits,
moving his bishop between c8 and d7. I n
order to prepare e5-e6 , Wh ite wou ld need
Wh ite is threatening to play e5-e6 (perhaps to make a by-pass with his k ing from the left,
From the Simple to the Complicated l2J 65
but this is impossible, s ince the king is tied to
the defence of the f5-pawn.
The fol lowing ru le suggests itself: the
bishop should be positioned such that,
while preventing the advance of one
pawn, it simultaneously attacks the other.
We wi l l use the ideas from the basic position
just examined for an analysis of other
positions. F i rst some comparatively s imple
ones.
Let us sh ift a l l the pieces one rank forward .
What has changed?
After a l l , with the king on f8 White's only plan
- a by-pass with the king from the r ight - is
impossible: the edge of the board prevents
it.
But it is Wh ite to move , and he shuts the
opponent's king in the corner by 2 .il..c4 ! and
then carries out the standard manoeuvre -
the by-pass with the king from the left: 'it>h5-
g4-f5-e6-f7.
After 1 . . . if..f8 ! the by-pass is no longer
possib le , but what about the threat of
zugzwang? I n order to put the opponent in
zugzwang, Wh ite must deprive the king of
the g8-square by playing 2 if..c4. But after
In the event of 1 . . . .i.b4 there is no d iffer- 2 . . . .i.xh6! 3 �xh6 th ings end in sta lemate .
ence . White wins in exactly the same way (a
check and a by-pass by the king); moreover,
here , as it is easy to see , both checks at b6
and b5 are equal ly good .
After 1 . . . .1i.d8 2 if..g6+ (or 2 .i.b5+) 2 . . . �f8 3
Wf5 Black loses because of zugzwang - in
contrast to the previous position , he no
longer has a waiting move with h is bishop.
(see diagram)
The proximity of the edge of the board
introduces new features into the evaluation
of the position . Let us verify 1 . . . if..b2. I f B lack
could also manage to play 2 . . . Wg8 and
3 . . .'it>f8 , the draw would become obvious . I n a l l the situations examined earl ier the
66 � From the Simple to the Complicated
weaker side was a 1m1ng to g ive up h is
bishop for the two pawns . Here, of course,
this fami l iar plan of defence wi l l no longer
save him. Does this mean that B lack is
doomed? It turns out that he is not - wing
pawns can sometimes be stopped without
resorting to the bishop sacrifice .
1 . . . i.d1 ! 2 �h4 (otherwise g4-g5 cannot be
played ) 2 . . . � 3 g5 �e6! 4 g6 'itf5! White
cannot advance either his king (the edge of
the board prevents th is) , or h is h-pawn . And
if 5 g7, then 5 . . . .ib3 and 6 . . . .ig8 , with a
secure l ight-square blockade of the enemy
pawns.
The fol lowing example is much more d iffi
cult.
M. Henneberger
1 9 1 6
The black bishop i s not i n its best position
(the place for it is at e7 or dB). In the basic
theoretical position , with which we began ,
against such a bishop White won easi ly. I f
we reason logical ly, on ly one factor, d istin
guishing the g iven posit ion from the basic
one, can prevent the implementation of the
standard winning plan - the proxim ity of the
edge of the board . Let us see!
'Accord ing to the ru les' Wh ite should give a
check on h5 , to control the g6-square . The
black king should move to e7, forcing the
wh ite king to make a by-pass to the right,
where there is l i tt le space for manoeuvring.
1 i.h5+ �e7! (after 1 . . . 'it>g7? 2 �e4 there is
noth ing to prevent the by-pass by the king
from the left) 2 'it>g4 .ib2 3 .ig6 (otherwise
the king cannot advance, but now the
important g6-square is inaccessible to the
k ing) 3 . . . .ic3 4 �h5 (threaten ing 5 �h6, 6
.ih5 and so on) 4 . . . ..tg7 ! 5 .ih7 �f7 ! 6
.ig6+ �e7, and Wh ite has been u nable to
ach ieve h is a im - to prepare f5-f6+ .
As we know from the basic position , the
check from the other side a lso does not
ach ieve anyth ing : 1 .ic4+ �g7! 2 �e4 .id2!
3 f6+ �g6 .
Even so , t he resources for playing for a win
are not yet exhausted . The black king can
fi rst be lu red to g7 , and on ly then the bishop
switched to the e8-h5 d iagona l , preparing a
by-pass by the king from the left.
1 �g4 .ib2
2 �h5 'it>g7!
The threat was 3 �h6; 2 . . . .ig7? is bad in
view of 3 .ic4+ and 4 'it>g6.
3 .ib5
4 i.e8
i.c3
.id4
4 . . . �f8 5 .ig6 �g7 is equal ly good .
5 .ig6
In the event of 5 'it>g4 (threatening 6 .ih5, 7
'it>f3, 8 �e4 and so on) the black king
succeeds in switch ing to e7: 5 . . . �f8 ! 6 i.h5
�e7, transposing i nto the fi rst of the
variations we examined .
5 . . . .ic3
6 �g4
White's plan appears to have triumphed:
6 . . . �f8 7 f6 is bad for Black, and otherwise
Wh ite plays 7 i.h5. But at this moment the
black bishop succeeds in switch ing to its
lawfu l place.
From the Simple to the Complicated ctJ 67
6 . . . i.a5! !
Because of the position of the bishop on g6 ,
7 f6+ i s not possib le.
7 i.h5 i.d8
Black has set up the d rawn position which is
basic to th is type of endgame.
At one of our tra in ing sessions Sergey
Dolmatov and Vad im Zviagintsev tried to
solve a study by Jan Timman, composed in
1 989.
1 'iite2 ( 1 i.e2? e4 or 1 . . . d4 is bad for White)
1 . . .e4 2! .
2 . . . 'iite5 3 i.e2!
The attempt to keep the bishop on the
queenside is incorrect: 3 i.a6? d4 4 i.b5 d3
5 .ia6 c5!
(zugzwang) 8 i.b7 e3, or 8 f6
After 4 . . . d3 5 i.g6 f5
5 . . . f4
8 i.h7 d4 1 0 i.h7 e3 1 1
i.xd3 'iii>xd3 - sta lemate!
The stalemate defence is pretty, of course,
but how necessary is it? Dolmatov and
Zviagintsev had doubts about the evaluation
of the position ar is ing after the capture of the
d2-pawn . Let's set it up with colours re
versed , so that it wi l l be easier to draw an
analogy with ideas that are a l ready known to
us .
The wh ite b ishop occupies an ideal position ,
whereas on the h3-c8 diagonal the black
bishop is not it its best place. Without the e2-
pawn the win would not be i n doubt, but here
White constantly has to reckon with the
threat of the d ivert ing sacrifice . . . e2-e1 'ii',
after which for an instant the bishop loses
control of important squares in front of its
pawns. The question is whether or not Black
is able to make use of th is resource .
68 w From the Simple to the Complicated
1 �e3
Threatening both the capture of the e2-
pawn, and the march of the king to f6 . After
1 . . . il..g4 (1 . . . il..h3? 2 �xe2) 2 �f4 B lack is
unsuccessfu l with 2 . . . e 1 'ik 3 il..xe1 il..h5
(with the threat of 4 . . . il..f7 ) 4 e6+ �d6 5 �e4,
whi le if 2 . . . il..h3 3 �g5 e 1 'ik 4 il..xe1 il..g2 ,
then 5 e6+ �d6 6 il..b4+ �xd5 7 e 7, and the
pawn queens. But what else can he do?
1 . . . �c7 ! !
The key idea of B lack's defence! I t i s
important that the pawn should not advance
to e6 with check (for example, after 2 �xe2
Si.e4 ). Without the e2-pawn White would
reply 2 e6, but here this leads to an
immed iate draw: 2 e6 il..xe6 3 dxe6 e 1 'ik 4
il..xe1 �d6 (Black's moves can also be
interposed) .
I n the event of 2 �f4 the s implest is 2 . . . il..d3
(2 . . . il..h3 also does not lose) 3 �g5 il..c4 or 3
e6 e 1 'ik 4 ii.xe1 �d6 . F ina l ly, after 2 il..a5+
�d7 3 �f4 the main defensive idea in such
endings proves possible - the switching of
the bishop to f7 : 3 . . . Ji.g6! and 4 . . . il..f7 (4 e6+
ct>d6) .
After moving the pieces around for a short
whi le, we decided that the endgame was
drawn and hence that Timman's study was
incorrect, s ince it conta ins a second solu
tion .
Later, when I was on my own , I again set up
the pieces and found another attempt to play
for a win , based on zugzwang.
2 il..e1 'it>d7
3 il..a5!
Now it is bad to play 3 . . . e7? 4 il..b4+ �f7
(4 . . . �d7 5 �xe2) 5 d4, when there is no
defence against the march of the king to d6
(active counterplay is too late: 5 . . . il..c8 6
'itc5 ct>g6 7 e6 'it>f5 8 �d6 il..a6 9 �c6! �f6
10 Si.c3+ �e7 1 1 il..e 1 ! , or immed iately 1 0
i.e1 ).
3 . . . .Jli.g4
3 . . . il..h3 4 �f4 (4 ii.b4 �c7 ! ! ; 4 �xe2 .Jli.g2 5
e6+ �e7 ! 6 il..b4+ �f6) 4 . . . �e7 ! 5 il..b4+
�fl comes to the same th ing .
4 �4 ii.h3
Here the idea of playing the b ishop to f7 no
longer works : 4 . . . il..h5? 5 e6+ �d6 6 �e4.
5 �g5 e7!
Otherwise �f6 cannot be prevented (as we
a l ready know, 5 . . . Ji.g2? is bad : 6 e6+ �d6 7
ii.b4+ �xd5 8 e7) .
6 il..b4+ �f7
6 . . . �d7? 7 �f6 .
7 �4
White has managed to l u re the opponent's
k ing to f7 and now his k ing heads in the
opposite d i rection - a by-pass to the left.
I phoned Zviagintsev and told h im about the
p lan I had found . Ha lf an hour later Vadim
phoned me back and reported that the
position was nevertheless d rawn !
7 . . . ct>g6 ! !
The on ly chance of saving the game is the
bold manoeuvre of the king to f5 . After the
incorrect 7 . . . ii.c8? 8 'it>e4! B lack unexpect
edly ends up in zugzwang and loses:
8 . . . 'it>g6 9 e6 or 8 . . . ii.h3 9 'it>d4. It is curious
that the zugzwang here is mutua l ; if it is
Wh ite to move he cannot win - 9 �d4 �g6!
transposes into the main variation , analysed
From the Simple to the Complicated CZJ 69
below, wh i le in the event of 9 ii.e1 ii.h3 1 0
'it>d4 the black king returns to the queenside:
1 0 . . . 'it>e7 1 1 ii.b4+ �d7 1 2 �e3 �c7 ! ! etc.
8 'iite4
Threatening 9 e6.
8 . . .
9 'iitd4
1 0 'i.t>c5
11 'i.t>d6
12 e6
ii.f5+!
.tc8!
�f5
.ta6
.tc4
Or 1 2 . . . ..ltb5 - White cannot win .
As you see, the analysis proved to be rather
difficult, and fu l l of by no means obvious
manoeuvres by both sides. But even so, at
the basis of the analysis were ideas which
we derived from the basic theoretical posi
tion .
Separated pawns
General ly speaking , the further apart the
pawns are, the more d ifficult the defence.
When I was young I learned a humorous
rule for assessing such endings: if you can
reach both pawns with the fingers of one
hand , then the position is d rawn ; if you can't
(the distance between the pawns is too
great) the position is won !
Alas, such a gu ide is too imprecise to be
trusted . In fact, here there exist many
different situations which it is not at a l l
necessary to study and remember. The
outcome usual ly depends on the possibi l ity
of a breakthrough by the stronger side's king
to the pawn being stopped by the bishop, i n
order to queen it.
But the fol lowing ending should defin itely be
included in our system of basic knowledge.
Berger - Kolterman
Arkhangelsk 1 948
1 �e2
2 'it>d1
3 .th7
4 ii.g6
b3
'i.t>b4
'iita3
If now 4 . . . b2 (with the threat of 5 . . . 'it>a2) ,
then 5 ii.b1 ! �b3 6 'it>e2 .
4 . . . 'i.t>b2
5 .tf7!
The threat was 5 . . . �a1 and 6 . . . b2 . By
attacking the b3-pawn, White foresta l ls the
opponent's plan .
5 . . . 'it>a2
6 ii.e6 �a3
With the threat of 7 . . . b2 8 ii.f5 �a2 .
7 .tf5!
Draw.
Let us consider a more compl icated ending .
70 � From the Simple to the Complicated
Yu. Averbakh
1 954
I n the fi rst edition of the monograph on
chess endings ed ited by Averbakh, the
analysis of this endgame conta ined a seri
ous mistake - it was d iscovered by Yusupov,
when I i nvited h im to try and solve th is
position . However, Averbakh h imself cor
rected the mistake in a subsequent ed ition .
1 'it>c3 i.f1 2 'it>d4 .lte2 3 �e5 'it>d7
Now it is wrong to play 4 'lot>f6 i.d3 5 a6?
i.xa6 6 �xg6 'it>e8 , when the fami l iar
Berger-Kolterman ending is reached (with
reversed colours) .
The correct plan is to p lay for zugzwang.
From d3 the bishop defends the g6-pawn
along one d iagona l , and along the other it
prevents the advance of the a-pawn ; there
fore it has no moves. The wh ite king must
not be al lowed to reach e 7 - this means that,
apart from d7, the black king also has two
other squares: e8 and d8. The first can be
taken away from it by placing the white king
on f7, and the second by moving the bishop
to c7.
4 .ltc5 i.f1 5 i.b6 .te2 6 .tc7 i.d3 7 'it>f6
'it>e8 8 'it>g7 'it>d7 9 'it>f7 , and Black is i n
zugzwang.
But if it is Black to move he saves the game
- he can prevent the opponent's k ing from
going to the kingside and set up a secure
defence on the queenside.
1 . . .
2 'it>c3
3 �d4
4 'it>c5
5 �b6
6 a6
'it>d7!
�e6
.ltb7
'it>d7
i.f3
�c8!
The threat was 7 'ita? 'lot>c8 8 d7+! �xd7 9
'it>b8
7 'it>a7
Now the threat of 8 d7+ must be parried by
the bishop, but from which square , c6 or g4?
7 . . . .ltg41
After 7 . . . i.c6? 8 i.b4 Black ends up i n
zugzwang : 8 . . . i.d7 9 'it>b6 .lli.f5 1 0 d7+!
'it>xd7 1 1 'it>b7 or 1 0its practica l
appl ication - both by a critica l analysis of
games played by young players , and at the
very highest grandmaster level .
By trad ition , the fina l part is devoted to an
analysis of games by the pupi ls of the
school .
Th is book was fi rst publ ished in the 1 990s.
During the preparation of this new edit ion I
checked a l l the games and endings on a
computer, which , natura l ly, gave rise to
Preface
numerous improvements and addit ions. I n
addition , a qu ite large chapter has been
added , one which was written many years
after the fi rst edition was publ ished . I n it
some instructive examples of the successfu l
or unsuccessfu l solving of compl icated
technical problems are analysed - they wi l l
help you to understand more deeply the
technique of converting an advantage, a
problem which is exceptional ly important for
every player.
Practica l ly a l l the players whom I have
tra ined have possessed good technique and
an excel lent understanding of the endgame.
This means that the working methods
described in this book have stood the test of
t ime. I hope that they wi l l a lso prove su itable
for you .
8
PART I
Endgame Theory
Mark Dvoretsky
How to Study the Endgame
Many young players 'flounder' when it
comes to playing endings. They would
not be averse to improving their endgame
mastery, but they don't know exactly how to
do this. I n chess l iterature practica l ly noth ing
is said about methods for the independent
study of endgame theory. We wi l l now
endeavour to partly fi l l this gap .
Two main ways of improving in the endgame
can be d isti ngu ished :
I . The study of theory (development of
erudition , en richment of the store of end
game knowledge) .
I I . Improvement in the genera l technique of
endgame play.
It stands to reason that these two d i rections
are closely inter-connected , and progress in
one of them invariably leads to progress in
the other. However, let us nevertheless
consider them separately.
I . THE STUDY OF THEORY
To expand your store of knowledge you
need to make a systematic study of various
types of endings. Here the tradit ional d ivi
sion by material is qu ite appropriate. By
successively examin ing , for example, pawn,
kn ight and queen endings we ass imi late the
specific featu res of these types of endgame.
Al l endgame positions can be arb itrari ly
d ivided into 'exact' and 'problematic' . Posi
t ions which are fami l iar to us , i n wh ich we
know beforehand the evaluation and correct
plan of action , we cal l 'exact' . Note that they
are fami l iar to us, and not to the theory of
endings in genera l . Different players have
d ifferent stores of exact positions.
Al l rema in ing positions belong to problem
atic. I n them we do not demonstrate our
knowledge, but fight, seek the best moves ,
and ca lculate variations - in short , we play.
A na"ive opin ion is prevalent, identify ing the
mastery of endgame theory with a knowl
edge of numerous exact positions. But is a
large store of specific knowledge real ly
necessary? After a l l , exact positions (apart
from the most elementary) occur rather
rarely in practice . More often a player has to
fig ht in problematic situations. He should
study the general endgame laws which
apply in them and the most common
regu larities, p laying methods and typical
evaluations . Al l this, together, of course,
with the most important exact positions , is
what comprises the integra l system of our
endgame knowledge.
I must once again emphasise: the store of
positions which you need to know exactly is
How to Study the Endgame CLJ g
comparatively smal l . Only in rook endings
do you need to have a fi rm knowledge of
several dozen specific positions; in other
types of endings - much fewer. When
studying them it is often not necessary to
delve into compl icated analyses - it is
sufficient merely to remember the main
conclusions.
Take, for example, rook endings with f- and
h-pawns. They occur qu ite rarely, but never
theless they do occur, so that it would be
useful to obta in some impression of them.
However, i t is hard ly advisable to study the
entire theory of this type of endgame - it is
just too compl icated . What, then , from this
theory should the practica l player add to his
armoury?
Above a l l , the information that such endings
are normally d rawn . I t is usefu l to examine a
practical ending , demonstrating the main
defensive ideas.
Gl igoric - Smyslov
Moscow 1 947
The black rook is excel lently positioned on
the 5th rank , preventing the wh ite king from
advancing. If 1 f5 there fol lows 1 . . . .Ub1 ,
threatening a series of checks from the rear.
1 .Ug6+ 'lt>f7!
1 . . . Wh7 would a lso not have lost, but in this
case the defence would have been much
more d ifficult.
2 l:Ig5 .l:!.b1 !
A typical rook move in such situations - from
here it reta ins the possib i l ity of checking the
enemy king both along the fi le , and along
the rank .
3 l:!.c5
If 3 h6 it is wrong to play 3 . . . l:tg 1 +? 4 Wf5
.i:.h 1 5 l:tg7+ , when Black's king is forced
back onto the 8th rank and this leads to a
loss. He is saved by the waiting move
3 .. Jla 1 !, for example: 4 .Uh5 (4 'it>f5 .Ua5+ ; 4
h7 .l::i.g 1 + 5 Wf5 .l::i.h 1 ) 4 .. . Wg8 5 f5 '>th7.
3 . . . '>tf6
4 .l:!.c6+ �g7!
The main danger for Black is having h is king
forced back onto the 8th rank . This would
have occurred after 4 . . . Wf7? 5 Wg5 .U.g 1 + 6
'it>f5 .l::i.h 1 7 .Uc7+ .
5 Wg5
6 'lt>f5
7 .l:lc7+
7 .l:.g6+ Wf7 .
7 . . .
8 .l:!.e7
9 .Ue8
1 0 l:!.e5
1 1 .Ud5
.Ug1 +!
.Ua1
Wh6
.l::i.b1
'it>g7
l:!.a1
l:!.f1
Not a bad move , although it was qu ite
sufficient to keep the rook in the corner.
1 2 :d4 .U.a1
1 3 l:!.d6 :as+
1 4 �g4 l:!.a1
14 . . . l:!.b5 is a lso possib le , return ing to the
posit ion with which we began .
1 5 l:!.e6 .l::i.g 1 +
1 6 'it>f5 .l:!.a 1
1 7 h6+
1 8 .l::i.d6
'lt>h7!
l:!.a2
1 0 � How to Study the Endgame
19 'iiig5 .l:.g2+
20 'iiif6 'iiixh6!
21 'it>e7+ 'it>h7
22 f5 .l:!.e2+
23 l:l.e6 .l:ta2
24 f6 !:. aS!
25 'Ot>f7 '.t>h6
An important theoretica l posit ion has arisen ,
one which should have been included in our
system of precise knowledge at an earl ier
stage - in the study of endings with rook and
pawn against rook.
26 .l:.e1
27 .l:!.e7
l:!.a7+
l:!.a8
It is simplest to keep the rook on the eighth
rank, not al lowing the wh ite k ing there. But
also possible is 27 . . . l:l.a 1 28 Wf8 'iiig6 29 f7
'iiif6 ! 30 g8 .l:tg 1 + ! with a draw.
Draw.
28 .l:td7 'iii h7
29 .l:td1 .l:!. a7+
30 e6 .l:.a6+
31 l::!.d6 .l:!.a8
32 .Ud4 'Ot> g8
33 .l:!.g4+ 'it>f8
An examination of such an ending helps us
to draw certain general conclusions. We
now know where Black should place h is
rook. And the king, as was shown by l lya
Maizel is , is best kept at f7 unti l there is a
danger of it being driven onto the back rank .
Then i t can stand at g7 and subsequently
even at h6, attacking the wh ite pawn.
I t stands to reason that by no means a l l
positions with f- and h-pawns are drawn .
The most important exception has a l ready
been mentioned several times - when
Black's king is cut off on the back rank , he
normal ly loses.
1 f6 l:!.a1
2 l:l.g7+ 'it>h8
2 . . . 'it>f8 3 h6 fol lowed by the u navoidable
h6-h7 .
3 'it>g6
4 'Ot>f7
5 .l:!. g8+
6 .l:!.e8
7 f8
.l:!. g 1 +
.l:.a1
'Ot> h7
l:!.a7+
The next move will be 8 f7 (the h5-pawn
deprives the black k ing of the g6-square) .
I t is sufficient to play through th is variation
just once on the board - there is no need to
memorise it, especia l ly s ince White a lso has
other ways to win .
This is probably a l l that the practical player. . . .txd7 1 1 a7 .
8 'it>b6 i.f3!
8 . . . 'it>d7? 9 'it>b7.
9 'it>c5
1 0 Wd4
And White cannot win .
'it>d7
'it>e6!
Endings with many pawns
After studying for many years the theory of
endings with opposite-colour bishops, I
observed several ru les which , as it tu rned
out, apply in nearly a l l such end ings and
greatly help in confidently find ing your way
in them.
Before tu rn ing to a description of my theory
of endings with opposite-colour bishops, I
wi l l show an example i n which , despite its
s impl icity, nearly a l l the rules that we wi l l be
ta lk ing about are d isplayed .
From the Simple to the Complicated 71
Textbook example
If it is White to move he saves h imself by
1 c5! i..xc5 2 �b3 e5 3 �e6 'l,;c7 4 'it>e4.
Later he s imply p lays h i s bishop up and
down the h3-c8 d iagona l .
1 . Drawing tendencies. This i s perhaps
the best-known property of endings with
opposite-colour bishops. Here it is some
times possible to save the game when you
are 2-3 pawns down (as, for instance , i n the
example just examined ) . And remember the
endings with two connected passed pawns
- in what other type of endgame may such
an enormous material and positional advan
tage prove insufficient for a win?
The consequences of this rule a re obvious :
the stronger side should be extremely
careful both when transposing into an
ending with opposite-colour bishops,
and when playing such an ending - here
it is easy to run into a drawing counter
chance. And for the weaker side, trans
posing into an ending with opposite
colour bishops is sometimes a last
resort - here the chances of a draw are
sharply improved.
I I . Fortress. A fortress is a system of
passive defence, consisting in the construe-
tion of an impregnable position , in which it is
sufficient to stick to wait ing tactics since
everything is securely blockaded and de
fended .
The main theme in endings with oppo
site-colour bishops is the theme of the
fortress. The weaker side aims to con
struct a fortress, while the stronger side
aims to prevent its construction or (if it
has already been constructed) find a way
of destroying the opponent's defences.
I n the textbook example the conclud ing
position constitutes a fortress. Wh ite does
not seek any active counterplay, but s imply
wa its , and the opponent is u nable to do
anyth ing .
When playing end ings , an abi l ity to analyse
positions logical ly, by th inking in p lans and
schemes, is very important. The role of
logical th inking is especia l ly great in endings
with opposite-colour bishops. In the majority
of cases they should not be 'p layed ' , but
'constructed' - fi rst look for the arrangement
of pieces and pawns which makes the
position impregnable, and only then verify
by calcu lat ing variations whether it is possi
ble to achieve the p lanned set-up and
whether it is i ndeed impregnable.
The fol lowing mechan isms constitute either
the most important genera l methods of
constructing and destroying a fortress , or
featu res of the most typical and frequently
occurring types of fortresses.
I l l . Arrangement of the pawns. There is a
wel l-known principle which prescribes that
pawns should be placed on squares of the
opposite colour to those on which you r own
bishop moves. In endings with opposite
colou r bishops th is pr inciple remains val id
for the stronger s ide ( it is especia l ly impor
tant with regard to connected passed pawns).
But, contrary to the general rule, the
weaker side should keep his pawns on
From the Simple to the Complicated
squares of the colour of his own bishop
- in this case it is usual ly possible to ensure
that they are securely defended . I ndeed , a
pawn defended by the bishop can be
attacked only by the enemy king , wh ich
means that it remains invul nerable. In other
types of endings such a pawn may be
attacked not only by the king, but a lso by
another piece (knight or l i ke-colour bishop) .
I n the textbook example the weaker s ide's
pawn is on a l ight square - the colour of its
own bishop, and this factor ensures the
sol id ity of the fortress constructed by White .
I n the in it ial position the stronger side, with
his dark-square bishop, has only the one
pawn on e6 correctly placed on a l ight
square. If B lack were able to approach it
with his king , he would then play . . . f6-f5 and
easily convert h is materia l advantage. The
only way to draw is to force the e-pawn to
move onto a square of the colour of its own
bishop.
IV. Nuances in the position are more
important than material . I n endings with
opposite-colour bishops the number of pawns
on the board is often of far less importance
even than seemingly insign ificant changes
in the placing of the pieces or pawns.
Therefore in endings with opposite-col
our bishops, positional pawn sacrifices
constantly occur. Thus in the textbook
example Wh ite happi ly sacrifices a th ird ( ! )
pawn in order to achieve a 'trifle' - sh ift the
black e-pawn one step forward .
V. Principle of one diagonal . Both for the
stronger, and the weaker side it is very
important that the bishop should defend
its own pawns and restrain the enemy
pawns 'without being torn', along one
and the same diagonal. In the concluding
position of the textbook example the bishop
on the h3-c8 d iagonal defends the h3-pawn
and stops the two enemy pawns on f6 and
g5.
But in the Averbakh position analysed
earl ier the bishop defends the g6-pawn
along one diagonal and restra ins the passed
a5-pawn along another. Such a situation is
unfavourable for Black. In the solution and
the false trail you saw two typical ways of
exploiting the defects of a 'torn ' bishop:
zugzwang and diversion.
VI. Pawns 'under attack' . A typical
defensive procedure is an attack on the
opponent's pawns by the bishop. I n th i s
way either they are forced to move onto less
favourable squares of the colour of their own
bishop (as in the textbook example) , or the
opponent's king is tied to the defence of the
pawns (as in the basic position with two
connected passed pawns or the Berger
Kolterman ending) .
End ings very often occur where the stronger
side has a passed pawn . I t must be blocked
by the king (fi rst system of defence) or the
bishop (second system of defence) .
VII . F irst system of defence - the weaker
side's king blocks the opponent's passed
pawn , and the bishop defends its own
pawns. This is the basic and usual ly the
most rel iable method of defence.
Attempts to destroy the first system of
defence always involve creating a sec
ond passed pawn, often by means of a
pawn breakthrough.
VI I I . Second system of defence - the
bishop stops the passed pawn (or some
times two , along the same d iagona l ) , whi le
the king , expressed in footbal l language,
engages i n 'zonal defence' - it protects its
pawns and restricts the activity of the
opponent's k ing .
Attempts to destroy the second system
of defence always involve breaking
From the Simple to the Complicated ttJ 73
through with the king to its passed pawn
(sometimes after a preparatory diver
sionary attack on the opposite wing).
We wil l now do some tra in ing in the
employment of this theoretical foundation
for the analysis of specific endings. We wi l l
try t o approach them in a logical way: we wi l l
point out which system of defence has been
employed or should have been employed by
the weaker side and in what way it may be
possible to try and destroy this fortress,
whether the pawns are correctly placed ,
whether it isn 't possible to put the oppo
nent's pawns 'under attack' , whether, in
order to carry out some idea, it is possible to
sacrifice a pawn or two , and so on .
Fuchs - Kholmov
Dresden 1 956
Black wi l l probably obta in a passed pawn on
the queenside, but itwi l l be blockaded by
the opponent's king (fi rst system of de
fence). The only winn ing chance is to create
a second passed pawn . For this B lack
needs to play . . . f7-f6 , . . . 'it>f5 and g6-g5,
then exchange on h4 and win the h-pawn. In
t he game Ratmi r Kholmov successfu l ly
carried out th is p lan and won .
I n a book h e wrote on the endgame, N ikola i
Krog ius considered th is outcome to be
perfectly logica l . I n fact the position is , of
course , d rawn - this is clear at fi rst g lance, it
being sufficient only to remember the draw
ing tendencies with opposite-colour bish
ops.
How can one expla in such a bad mistake in
eva luation , made by a player who at one
t ime was qu ite a strong grandmaster? I n my
view, by a change of profession : one by no
means fine day Krog ius decided to ' re
qual ify as a manager' , fi rst in h is native
Saratov, and then in Moscow - he became
head (and , it should be mentioned , a very
nasty head ) of the Chess Admin istration of
the USSR Sports Committee . Apparently
Caissa is a jealous woman who seeks
vengeance when she is betrayed .
43 . . . f6!
44 'it>d2
White's objective is to defend the kingside
with h is bishop and not a l low the opponent
to create a second passed pawn there. For
the moment the move in the game does not
yet spoil anyth ing , but it was simpler to play
44 d5! 'it>xd5 45 'it>d3(d2) fol lowed by i.e3-
b6-d8 (the f6-pawn 'under attack' ) . The
d raw would then be obvious - after moving
h is k ing to f5 and playing . . . g6-g5 , B lack
would be u nable to make any further
progress.
44 . . . 'it>f5
45 .if4?
Now the opponent inevitably obta ins a
passed pawn on the kingside. Meanwh i le
the 'pawns u nder attack' procedure cou ld
a lso have operated successfu l ly here: 45
.ih6! g5 (45 . . . '1t>g4 46 .ig7 f5 47 .ih6 or 47
d5) 46 i.g7 ! , preventing 46 . . . '1t>g4 . 45 d5!
i.xd5 (45 . . . g5 46 d6 i.c6 47 .id4) 46 i.d4
or 46 i.b6 g5 47 .id8 was a lso good . I t is
evident that procedures in the playing of
endings with opposite-colour bishops, such
74 � From the Simple to the Complicated
as the sacrifice of one's own pawns or
attacks on the opponent's , were unknown
both to the national master Fuchs, and to
grandmaster Krog ius.
45 . . .
46 i.c7
47 i.d8
48 gxh4
49 i.xf6+
50 'it>e3
51 i..e7
g5
'it>g4
gxh4
'it>xh4
'it>g4
i.d5
b5
White resigned - in the opin ion of Krogius ,
because of the variation 52 i..d8 h4 53 f3+
i.xf3 54 'it>f2 h3 fol lowed by the switching of
the king to the queenside (on the h3-c8
d iagonal the black bishop defends its own
pawn and restra ins the wh ite d-pawn) . I n
fact 53 . . . i..xf3? i s a mistake , of course, i n
view of 54 i.xh4 ! ; Black should play
53 . . . 'it>g3! .
Meanwh i le , even after 4 5 i..f4 a draw is sti l l
possible. As was shown by grandmaster
Sergey Shipov, by continu ing 47 'it>e3!
( instead of 47 i.d8) 47 . . . gxh4 48 gxh4 'it>xh4
49 'it>f4 'it>h3 50 i.d8 White would have
saved the game.
And also later, just two moves before
capitu lation, it was possible to gain a draw
by choosing 50 �c3 ! ( instead of 50 'it>e3?)
50 . . . h4 51 'it>c4 h3 52 i..e5 �3 53 d5 'it>xf2
54 'it>c5 i..f3 55 d6 i..c6 56 'it>b6 �g2 57 d7
(Carsten Mul ler, Frank Lamprecht).
And yet White's 45th move was a funda
mental mistake: instead of find ing and
erecting a secure fortress, he a l lowed the
opponent to compl icate the play advanta
geously.
Bogoljubow - Ed. Lasker
New York 1 924
White should win thanks to h is powerfu l pair
of connected passed pawns. I t was s implest
to bring the king to the centre : 36 �f2 .
Apparently Efim Bogoljubow was striving to
play as safely as possible - he wanted to
prevent . . . a7-a5 and with this a im he
decided to exchange the rooks. I n the game
h is p lan proved justified .
36 '/J.c7 �f7 37 '/J.xe7+ �xe7 38 i.d2!
(foresta l l i ng Black's counterplay on the
queenside) 38 . . . We6 39 'it>f2 Wd6 40 We3
�c5 41 i..a5 , and White won easi ly.
Remember the need to exercise caution
when transposing into an ending with oppo
site-colour bishops , in view of the inherent
d rawing tendencies. As was shown by
Alexander Alekh ine, B lack could have saved
the game.
36 'fJ.c7?
37 i..xc7
.Uxc7!
b4!
The sacrifice of a pawn is a common
phenomenon in end ings with opposite
colour bishops.
38 axb4
39 d4
i.a6!
i.d3!
This is a lso a standard defensive procedure
- an attack on the enemy pawns. They are
From the Simple to the Complicated ttJ 75
forced to move onto squares of the colour of
their bishop, where they completely lose
their strength , s ince they can easily be
blockaded .
40 e5 ii.c4
41 �2 a6
When defending, pawns should be kept on
squares of the colour of the bishop.
42 'ite3 'itf7
43 �4 h5
The position is d rawn .
Taking into account the principle 'nuances in
the position are more important than mate
ria l ' , we should a lso check 38 'it>f2 !? ( instead
of 38 axb4 ) , in order not to a l low a blockade
of the central pawns. However, after 38 . . . bxa3
there is no win for White - the a-pawn
diverts the bishop from its control of the
squares in front of the connected passed
pawns. Here is an approximate variation ,
suggested by Igor Bondarevsky: 39 t>e3 a2
40 .te5+ �f7 41 ii.b2 �e6 42 d4 ( 42 �f4
h6) 42 . . . 'it>d6 43 d5 h6 44 'it>d4 ii.a8 45 e5+
�d7 46 'it>c5 (46 e6+ 'it>d6) 46 . . . ii.b7 47 e6+
�e7, and White is not able to strengthen h is
position .
Kharlov - Khenkin
Copenhagen 1 993
The game concluded : 44 .. . a6? 45 'ifa7+
'it>h6 46 'i!Ve3+ 'lt>g7 47 �g5! (Black under
estimated the strength of this move) 47 . . . 'ikd4
48 c7! ii.xg3+ 49 �xg3 Black resigned .
It is not my intention to g ive a deta i led
analysis of the end ing . I wi l l merely show
one way (I would not assert that it is the only
one, but in my view it is the s implest) of
ga in ing a draw. Why not immediately e l im i
nate the main enemy - the c6-pawn?
44 . . . ii'xc6!
45 "it'xa7+
Noth ing is g iven by 45 'i!t'f7+ 'it>h6 .
45 . . . "iVc7!
The bishop and the g3-pawn are attacked ,
and therefore the exchange of queens is
practica l ly forced .
46 'ii'xc7+ ii.xc7
Transposing into an ending with opposite
colour bishops is an important defensive
procedu re , with the help of which one can
sometimes save a d ifficu lt position , and
therefore , of cou rse, the suggested plan
deserved serious consideration . Grandmas
ter Igor Khenkin was afra id that the end
game was lost, s ince White has two extra
pawns. I n fact it is a s imple draw, and in
establ ish ing th is we are helped , apart from
the genera l gu ide ( 'drawing tendencies' ) by
a qu ite specific one. If Wh ite g ives up h is g3-
pawn, we obta in the wel l-known d rawn
situation from the Berger-Kolterman game.
But if he advances it to g4, Black rep l ies
. . . g6-g5, and blocks all the enemy king's
approaches to the upper half of the board .
Here is an approximate variation :
47 �xh3 �6
48 �g4 ii.d6
While there is t ime, it is usefu l to force the
opponent's pawn to stand on a square of the
colour of its bishop.
49 b5 i..c7
50 �d5 ri;e7
76 � From the Simple to the Complicated
51 �c6
52 �f3
53 g4
54 'it>e3
55 �d3
56 �c4
f6
�e7
g5
�b6+
'it>d6
�e5
The draw is obvious - there is nowhere for
the wh ite king to break through .
Vakhidov - Timoshchenko
Tashkent 1 982
For the moment B lack is not threatening to
play 1 . . . g3 in view of 2 g5! and 3 �g8 - he
is plann ing 1 . . . g5! and only then 2 . . . �g3.
Now White has to decide how he wi l l defend
his kingside pawns and which piece wi l l hold
back the enemy passed pawn on the
queenside.
In the game he chose the fi rst system of
defence: he switched his bishop to the
defence of hispawns, and kept his king on
the queenside.
1 .i.b3 g5! 2 .i.d1
2 h4 gxh4 3 g5 does not work: 3 . . . �g3! (but
not 3 . . . 'it>xg5? 4 .i.e6 f4 5 .i.f5 'it>g3 6 .i.h3
with a draw) 4 ..tg8 'it>xg2 5 ..txh7 ..tc5 ! , and
Black wins.
2 . . . a6 3 ..tf3 b5 4 axb5 axb5
I t appears that White is out of danger - his
bishop has defended his pawns, and h is
k ing is blocking the opponent's passed
pawn . But the bishop is real ly very passive
soon it wi l l not have a single waiting move. If
the king can a lso be deprived of its mobi l ity,
a zugzwang situation could result . This aim,
strangely enough , is qu ite ach ievable: the
wh ite king is g radual ly pushed to b3 and the
black king wi l l occupy the d3-square , from
where it continues to tied down the oppo
nent's bishop and at the same time threat
ens to support its passed pawn . But if the
pawn were sl ightly further away - on the a
fi le - a d raw would become inevitable.
5 ..td1 'it>g3 6 �f3 'it>f2 7 'it>c2 b4 8 �b3
..tc3 9 �c4 h6 1 0 �d3 ..te1 1 1 �c4 ..td2 12
�d3 �c3 (zugzwang ! ) 1 3 �c4 �e3 (again
zugzwang! ) .
I f 14 �b3, then 14 . . . �d3 (the decisive
zugzwang ! ) 1 5 �a2(a4) �c2 and wins.
The game went 14 'it>d5 b3 and White
resigned .
Now let us try to set up the second system of
defence - use the king for the defence of the
kingside. But this plan too is not a ltogether
rel iable - after a l l , the bishop wi l l have to
perform two tasks: not only conta in the
enemy passed pawn , but also defend its
own e4-pawn , and along a d ifferent d iago
na l . This means that here too a zugzwang
position is qu ite l i kely.
1 ..td5 g5! 2 'it>e2 h6 3 ..tb7 ..tc5 4 ..td5 a6 5
..tc4 (5 ..tb7 b5 6 a5 b4 7 ..txa6 �xe4 is no
better) 5 . . . b5! 6 axb5 a5 (Black happily
sacrifices a pawn for the sake of creating a
passed pawn ) 7 ..td5 a4 8 ..tc6 a3 9 ..td5
..tb6. Zugzwang! The wh ite bishop has no
moves, s ince it is 'torn ' between two
d iagonals. I n the event of 1 0 Wf1 (e 1 ) 'it>e3
the black king breaks through to its passed
pawn , whi le if 1 0 'it>d3 , then 1 0 . . . Wg3 1 1
'>t>e2 (as i t i s easy to see , 1 1 'i!i>c3 'it>xg2 1 2
b3 'i!i>xh3 1 3 .i.e6 also does not help)
From the Simple to the Complicated ltJ 77
1 1 . . .'it>xg2 1 2 ii.e6 �xh3 1 3 �f3 'it>h4 1 4 ii.f7
ic? (again zugzwang because of the
bishop being 'torn ' ) 1 5 ii.e6 h5 1 6 gxh5
�xh5 17 'it>g3 'it>g6 1 8 �g4 'it>f6 1 9 ii.d5
�e7! 20 'it>xg5 �d6, and Black final ly carries
out the main idea for destroying the second
system of defence - the breakthrough with
his king to the passed pawn .
Is the in it ial position rea l ly lost for White?
Let's use our knowledge of opposite-colour
bishops to guess where a saving l ine might
nevertheless be concealed .
First of al l one should usual ly check the
basic system of defence - the fi rst . But how
to securely defend the kingside with the
bishop, and prevent there the creation of a
second passed pawn? The manoeuvre of
the b ishop to f3 solves this problem , but it
inevitably leads to zugzwang. I s there no
other way? Remember the procedure 'pawns
under attack' and for the sake of implement-
ing it let us be prepared to sacrifice a pawn !
1 g5! ! �xg5
2 ii.g8! h5
2 . . . h6 comes to the same th ing . 2 . . .t>f6 3
�c4 also does not achieve anyth ing .
3 ii.f7 h4
4 'it>c4
The d raw is obvious, s ince now the bishop
easily copes with the defence of the king side.
White's moves can be transposed : 1 �g8 h6
2 g5! ! �xg5 3 iLf7 .
I n conclusion I offer a few exercises , in the
solving of which you wil l tra in yourself in the
practica l appl ication of your theoretical
knowledge. I advise you a lso to look at the
instructive endings with opposite-colour bish
ops, analysed i n my book School of Chess
Excellence 1 - Endgame Analysis.
78 w From the Simple to the Complicated
Exercises
1 . Black to move 2. White to move
3. White to move 4 . B lack to move
From the Simple to the Complicated LtJ 79
Sol utions
1 . S . Tarrasch ( 1 92 1 ) .
I t is not possible to prevent the advance of
the pawns to the 5th rank (for th is the bishop
would have to be switched to c6) . But how
should the black pieces be deployed against
pawns on the 5th rank? Obviously, bishop
on f7(g8) and king on d7. I t is th is set-up that
must be prepared .
1 . . . Si.c4!
1 . . ..ib5? is incorrect: 2 Si.b4+ ! (but not 2
ig3+? 'it>e7! 3 d5 Si.e8 4 e5 Si.f7) 2 . . . �c7 3
d5 .ie8 4 e5 jLf7 5 e6 - Black is one tempo
short. Or 2 . . . 'it>e6 3 d5+ 'it>e5 4 Si.c3+ �d6 5
litd4 .ie8 6 e5+ , and the bishop has not
managed to reach f7 .
2 i..g3+ �c6!
Of course , not 2 . . . �e6? 3 'i!.>d2 and 4 'it>c3 .
3 �f4 .i.g8
4 �es 'it>d7
5 d5 ii.h7!
'Pawns under attack' - Black does not a l low
the opponent's king to go to f6 . However, the
less accurate 5 . . . Si.f7 6 'it>f6 We8! 7 Si.f4 3i.g8
was also sufficient for a d raw.
6 �f4 ii.g6
7 e5 .il.f7!
A basic drawn position has been reached .
2. V. Chekhover ( 1 950) .
If Wh ite should succeed in winn ing the d7-
pawn , th is wi l l lead to a fami l iar position from
the Berger-Kolterman game. But if he
doesn 't? Then he must at least force the b
pawn to take a step forward , onto a square
of the colour of its bishop, so that the black
king wil l be unable to break through via b2 .
1 i..e8!
'Pawn under attack ! '
1 . . .
2 'it>e2 !
2 3i.f7? d5 .
2 . . .
'it>c6
�c1
While Wh ite is tied down , the black bishop is
switched to a better position .
3 �d1 3i.b2
4 'i!.>e2
5 �d1
.il.d4
'it>d6
5 . . . �c7 is answered in the same way.
6 .il.f7!
Again attacking a pawn!
6 . . . b2
7 i..g6 �cs
8 'it>e2 dS
9 i..f5 �b4
1 0 3i.g6 �a3
11 3i.b1 ! 'it>b3
12 'it>d1 'it>c3
1 3 'i!.>e2 ii.c5
14 'it>d1 d4
1 5 �e2 'it>b3
The last hope : 1 6 �d 1 ? d3 ! is bad for Wh ite .
1 6 �d3!
Black cannot make any progress .
3. A. Norl in ( 1 922) .
The typical p lan is to march the k ing to the
pawn which is being stopped by the bishop,
i .e . to f8 . But then Black wi l l advance h is a
pawn , d iverting the bishop from the defence
of the c 7 -pawn.
The only winn ing chance is to switch the
bishop to a5, from where on the same a5-d8
d iagonal it will defend its own pawn and stop
the opponent's. But fi rst the c7-pawn must
80 � From the Simple to the Complicated
be defended with the k ing, without a l lowing
. . . a7-a5-a4. If the black pawn should reach
a4, the position will become drawn , for
example: 1 �c5? a5! 2 �b5 a4 3 �b4 �c8 .
1 �c3! Jl..f7
2 �b4 Jl..e6
3 Jl..e5!
I t is important to vacate the d6-square for
the king beforehand. 3 �c5?! is inaccurate
in view of 3 . . . Ji..b3! with the threat of 4 . . . a5.
3 . . . �c8!?
I f 3 . . . JI..f7 , then 4 �c5 Ji..b3 (4 . . . a5 5 'it>b5) 5
�d6 (threatening 6 �d7) 5 . . .'it>c8 6 Jl..c3 ! , or
4 . . . �c8 5 �c6! (threatening 6 Jl..c3) 5 . . . JI..e8+
(5 . . . a5 6 d6 Jl..f7 7 Jl..c3! and 8
Jl..a5.
4 �b5!
The variation g iven by the author is sl ightly
longer: 4 �c5 Ji..b3! 5 'it>b5! 'it>b7 6 c5 .
4 . . . 'it>b7
5 'it>a6 was threatened .
5 'it>c5 Ji..b3
6 �d6 �c8
7 Jl..c3
The next move wi l l be 8 .Jta5, after which the
king wi l l fi nal ly be able to win the bishop for
the g-pawn .
4. Nimzowitsch-Tarrasch
(Kissingen 1 928).
Black must decide how to combat the
threatened attack by the king on h is kingside
pawns. The 'active' 39 . . .f4? is hopeless: 40
Jl..g5 e3 (40 .. .f3 41 g4) , and Wh ite has a
pleasant choice between 41 fxe3 and 4 1 f3
e2 42 Ji..h4 fol lowed by �g 1 -f2 . F i rst let us
see what happened in the game.
39 . . . c4?
Moving the pawn onto a square of the colour
of its own bishop is , in genera l , a sound
positional idea ( imaginethat White were to
play c2-c4 , b2-b3 and a2-a4 - then the b6·
pawn would be transformed into a serious
weakness) . The move made by Black is not
bad in itself, but for the reason that it does
not help to solve the main problem of the
position - the defence of the kingside
pawns.
40 'it>g3 �c8 41 'it>f4 'it>d7 42 j;_b4 �e6 43
Jl..c3 Ji..d7
I f Black keeps his bishop at g6 and uses his
king to stop the future passed pawn on the
queenside (fi rst system of defence) , at an
appropriate moment White will attack the
bishop by h3-h4-h5 and obta in a second
passed pawn. For example, 43 . . . ii.g6 44
g5 �d5 45 g3 b5 46 h4 �c6 4 7 b3 cxb3 48
cxb3 'it>b6 49 a4 bxa4 50 bxa4 'it>a6 51 a5
�b5 52 h5 ii.e8 53 �xf5 ii.xh5 54 �xe4 with
an easy win . Therefore Black leaves his
bishop on the queenside. Unfortunately for
h im , h is king cannot s imultaneously defend
the h7- and f5-pawns, and therefore his
bishop wi l l be 'torn' between the defence of
the f5-pawn and the struggle against the
opponent's passed pawn .
44 g3 b5 45 �g5 �f7 46 h4 Jl..c8 47 �h6
�g8 48 b3 cxb3 49 cxb3 f4
This is a l ready desperation in a hopeless
position . If 49 . . . j;_d7 Aaron N imzowitsch
gave the fol lowing variation : 50 ii.b2 ii.c8
(50 . . . JI..e8 51 �g5 j;_d7 52 'it>f6 , and the
wh ite king breaks through on the queen side)
51 a4 bxa4 52 bxa4 j;_d7 53 a5 ii.c8 54 ii.a1 ,
and Black i s i n zugzwang (54 . . . ii.a6 55 �g5
ii.c8 56 'it>f6) .
50 gxf4 j;_d7 51 �g5 'it>f7 52 f5 j;_c6 53 �f4
(the standard p lan : the king heads towards
the passed pawn wh ich is being combated
by the bishop) 53 . . . e7 54 'it>e5 ii.e8 55
c5 Jl..e8 59 Jl..e5 ii.d7 60 �b6 f7 61 f6
Jl..e8 62 f4 e6 63 'it>a6! Wf7 64 b4 �e6
65 a4 bxa4 66 b5 Black resigned .
From the Simple to the Complicated 4J 81
As usual , we shou ld fi rst look for a possib i l ity
of setting up the fi rst system of defence -
leave the king on the queenside and ensure
the defence of the pawns by the bishop. I f
the principle 'pawns under attack' is remem
bered , the correct solut ion (pointed out by
Averbakh ) does not seem at all d ifficult .
39 . . . .ib5!
40 'it>g3
40 g4 fxg4 41 hxg4 i.e2 42 'it>g3 i.f3 .
40 . . . .if1 !
41 h4 h5!
42 �4
Otherwise it is not possib le to strengthen the
position .
42 . . . .ixg2
The black bishop easily copes with the
defence of the kingside pawns.
82 �
Mark Dvoretsky
The Arithmetic of Pawn Endi ngs
A rapid evaluation rule
Positions with a pair of blocked rooks'
pawns and an outside passed pawn for
one of the sides occur qu ite often in practice.
Therefore it is usefu l to be able to evaluate
them quickly and accurately. The winning
plan is obvious: march the king over to the
rook's pawn . The opponent has to e l iminate
the pawn on the other wing and then rush
with his king to the corner, in order to stop
the rook's pawn . I n which cases does he
manage to do this?
Here White wins: 1 'it>d5 'it>f6 2 'it>c6 'it>xf5 3
'it>b6 'it>e6 4 'it>xa6 'it>d7 5 'it>b 7.
Now let us shift the queenside pawns back
by one rank.
I t is easy to see that the position has
become drawn : 1 '>t>d5 'iM6 2 'it>c5 'it>xf5 3
'it>b5 '>t>e6 4 'it>xa5 'it>d7 5 'lt>b6 '>t>c8.
I f, say, the kings and the f-pawn are sh ifted
one rank down or to the left, Black again
loses . But what happens if the queenside
pawns are a lso sh ifted down?
Of course, if you have the position in front of
you , it is easy to g ive an answer to any such
question . But i n practice such situations
often arise at the end of lengthy variations,
which you have to calculate , and to lengthen
the calculation by several more moves may
prove d ifficult. I t would be desirable to learn
to determine the evaluation of the position
immed iately, on fi rst looking at it.
A s imple method of rapid evaluation was
suggested by Walther Bahr in 1 936. To me
this ru le seems not a ltogether conven ient,
and besides it does not extend to cases
where the king is not to the side of the
passed pawn , but in front of it. I n connection
with th is I should l i ke to offer a somewhat
The Arithmetic of Pawn Endings lLJ 83
different method of rapidly eva luating such
positions.
1 ) The fi rst rule coincides with Bahr 's
analogous ru le : if the rook's pawn of the
stronger side has crossed the middle of
the board, the position is always won. It
follows that from a s ingle g lance at the fi rst
diagram it may be concluded that the
position is won .
2 ) We wil l ca l l the position in the second
diagram 'norma l ' . This is what makes i t
such :
a) between the queenside pawns there
passes the invisible demarcation l ine , sepa
rating the upper and lower halves of the
board ;
b) the black king , which is a iming for the c8-
square, reaches there without loss of time.
Th is happens because the passed pawn
has already crossed the key h3-c8 d iago
nal , or is on this d iagona l .
A 'normal' position is drawn.
3) Each sh ift ing of the kingside pawn one
square down from the h3-c8 d iagonal is
equivalent to a tempo in favour of White . For
example, the pawn on f4 g ives one tempo in
favour of White , and the pawn on e4 g ives
two. One further tempo for the stronger side
may be given by having h is king not to the
side of the passed pawn, but in front of it .
But each shifting of the queenside pawns
one square down compared with the 'nor
mal ' position g ives the defending side a
tempo. With the pawns on a3/a4 B lack has
one tempo in h is favour, and with the pawns
on a2/a3 he has two .
White wins only if the sum of tempi,
calculated in this way, is in his favour.
The formulation suggested by me looks
rather compl icated and cumbersome, but if
you learn it thoroughly you will find it very
easy to use.
With White to move , it is a win : 1 a5! (the
pawn has crossed the middle of the board ) .
With B lack to move , i t is a d raw: 1 . . . a5! , and
a 'normal ' position arises.
Wh ite wins: he has two tempi (the g2-pawn
is two squares lower than the g4-square) ,
whi le B lack has only one. But if the
queenside pawns are sh ifted down one
rank , the score becomes 2-2 and the
position is now d rawn .
84 \t> The Arithmetic of Pawn Endings
Here, of course , it is White to move (with
Black to move he has to concede the
opposition and White can queen his f-pawn) .
Wh ite wins, s ince the score is 3-2 in h is
favour (two tempi are g iven by the pawn on
f3 and another one by the position of the
k ing in front of the pawn) . He wins by 1 'ite4!
'ite6 2 'itd4(d3). It would be a blunder to
play 1 'ite3? 'ite5(f5) , s ince then a position
with a tempo ratio 2-2 is reached (the wh ite
king is no longer in front of the pawn , but to
the side of it) , and this means a d raw.
One more usefu l deta i l . Let us suppose that
White's passed pawn is a rook's pawn , and
his k ing is in front of it , but the enemy king is
confin ing its opponent to the h-fi le . This
situation is equivalent to the one in
which the king is to the side of the pawn.
Accord ing to the afore-mentioned ru le, this
should be a d raw. And i ndeed : after 1 . . . �5
2 'ith5 (2 'itg3 'itg5 - a 'normal ' position)
B lack does not play 2 . . . 'iii>f6? 3 'itg4, when
White acqu i res an extra tempo, since his
king is in front of the pawn , but 2 . . . 'itf4! 3 h4
(3 'itg6 'itg3) 3 . . . 'itf5 4 'ith6 'itf6 5 �h7
'itm(f5) 6 h5 'itt6 ! etc.
Now let us examine some more compl icated
endings, i n which a mastery of the rule
suggested by me sign ificantly eases the
calculation of variations.
Privorotsky - Peterson
Riga 1 967
Black's positional advantage is obvious. His
p lan is clear: . . . 'itg6-f5-e4 and then an
attack by h is bishop or king on the queenside
pawns. This plan can be prevented by
offering anexchange of bishops, but this
demands precise calculation .
1 ..td4! ..txd4+
1 . . . ..th6 2 'iii>f2 ..tc1 3 'itxf3 ..txb2 4 a4 with
equal ity.
2 cxd4
3 'itf2
4 d51
'iii>f5
'ite4
Otherwise 4 . . . 'itxd4 5 'itxf3 'itd3 .
The Arithmetic of Pawn Endings l2J 85
4 . . .
5 'it>xf3
6 'it>e2
'it>xd5
'it>d4
c3
If 6 . . . h4 7 'it>d2 a5 there fol lows 8 'it>c2 or 8
a4 , but not 8 'it>e2?? c3 9 bxc3+ 'it>xc3 , and
Black wins.
7 bxc3+ 'it>xc3
8 h4! !
The only way! Otherwise Black h imself
would have played 8 . . . h4! , then picked up
the a3-pawn and won , s ince his pawn on the
other wing has crossed the m iddle of the
board . But after the move in the game a
'normal ' position arises, and th is means a
draw.
8 . . .
9 'it>d3
10 'it>c3
'it>b3
'it>xa3
We have reached the last of the textbook
examples that we analysed .
Draw.
1 0 . . . a5
11 'it>c4! 'it>a4
12 'it>c3 'it>b5
13 'it>b3
I n the calculation of th is end ing , d ifferent
vers ions of this type of position arose. I f
White does not evaluate them ' mechani
cally' , by using the ru le g iven above , but
tries to work out the variations to the end, he
each time has to calculate some ten more
moves, and this is not at a l l easy.
The players in the fol lowing ending faced
even more compl icated problems.
Matanovic - Botvinn ik
Belgrade 1 969
In h is notes Mikhai l Botv inn ik analyses two
courses of action for Wh ite : 43 .l:td5 and 43
l::td6+ 'it>e7 44 l:f.a6. In fact there is a lso a
th i rd : 43 'it>f2 ! , for example, 43 . . . exd3 44
'it>e3 l::.a 1 (44 . . . .l::!.g 1 45 'it>f2) 45 l::.xd3 .l::i.xa4
46 l:td6+ fol lowed by 47 l::.a6, and White
should ga in a d raw.
But let's forget about th is possib i l ity and try
to choose the more accurate of the two
possible rook moves.
F i rst, after analysing some short variations,
we must try to d isclose the d ifference
between them, and compare their virtues
and d rawbacks .
I n the event of 43 .l:td6+ 'it>e7 44 .l::i.a6 a clear
d raw resu lts from 44 .. J1xd3 45 .l::i.xa5 or
44 . . . .l:!.d2+ 45 lt:Jf2 e3 46 'it>f3 ! (46 l:txa5
.l:!.xf2+ 47 'it>g 1 is also possible) 46 . . . e2
(46 . . . exf2 47 'it>g2) 47 l:1a7+. However, the
capture on d3 with the pawn is unpleasant:
44 . . . exd3! 45 .l:!.xa5 'it>d6. Now 46 'it>f2? is
bad in view of 46 . . J1g 1 ! ; Wh ite is forced to
play 46 .l:!.a8 , a l lowing the black king to
approach its passed d-pawn . I s this rook
ending lost or d rawn? You can't say immedi
ately, and th is means it is t ime to cut short
the ca lcu lation and switch to a verification of
the alternative poss ib i l ity.
86 � The Arithmetic of Pawn Endings
43 l:td5! .l::td2+!
Now the rook ending arising after 43 . . . exd3
44 .UXa5 is not dangerous for White : 44 . . . d2
45 l:Id5, or 44 . . .'it>e6 45 .l::te5+ �d6 46 �f2
d2 (46 . . .l:tg 1 47 .l::te3) 47 f4 with a draw.
50 gxh4+ �xh4
The Arithmetic of Pawn Endings lLJ 87
Which of the two natura l moves, 5 1 'it>e3 or
51 '.t>e2 , should be made? Let us refer to the
ru le g iven above . After Black wins the h3-
pawn , accord ing to our arithmetic he wil l
have one extra tempo, s ince the f-pawn is
one square higher than the key c1 -h6
diagonal . Wh ite wil l d raw only if he can force
the pawn to advance to f4 .
It becomes clear that 5 1 �e2? loses :
51 . . .'it>g3 (zugzwang) 52 h4 (52 �1 'lt>xh3
53 'it>f2 'it>g4 , and Black has even two extra
tempi) 52 . . . �xh4 53 �f3 �g5 54 'lt>g3 'it>f6
55 '.t>f4 'it'e6 56 �f3 'it>d5 and so on .
51 'it>e3 ! 'lt>g3
52 'it>e2
Now it is Black who is in zugzwang and he is
forced to advance h is pawn .
52 . . . f4
53 �1
54 �2
55 �g2
�xh3
'lt>g4
A 'normal ' d rawn position has ar isen .
Alexander Matanovic d id not manage to
calculate the pawn ending exactly and he
preferred to reta in the rooks. Let us see
what th is led to .
43 .l::!.d6+? 'it>e7 44 l:!.a6 exd3! 45 �xa5 'it>d6
46 l:ta8 �c7 (Black repeats moves, to ga in
t ime for thought) 47 l:ta5 'lt>c6 48 l:ta8 �c5
49 'it>f2 l:ta1 ! 50 l:td8
The only saving chances were offered by 50
�e3 !? .l:!.g 1 ! 51 g4 fxg4 52 hxg4 l:txg4, and
now, probably, 53 f5 .
50 . . . '1t>c4 51 'it>e3 I:.e1 + (51 . . . .l:!.g 1 ? 52
l:.d4+) 52 'it>f2 �e2+ 53 'it>f3 .l:!.e6! 54 a5 'it>c3
55 �c8+ 'it>d2 ! (only not 55 . . . �b3? 54 a6! d2
55 .l:!.d8 'lt>c2 56 a? or 54 . . . lixa6 55 �e3 .l:!.d6
56 �d2) 56 h4
Accord ing to ana lysis by Botv inn ik , Wh ite
would also have fa i led to save the game by
56 Uc7 h5 (56 . . . Ue1 ? 57 a6 l:!.a 1 58 a?) 57
'it>f2 �d 1 58 �f3 d2 59 �f2 l:te2+! 60 'i!tf1
lie3 61 a6 (61 �f2 l:ta3 fol lowed by . . . Ua 1 -
c1 ) 6 1 . . . �xg3 62 a ? Ua3 6 3 'it>f2 h 4 64 'it>f1
l:ta4 65 'it>g2 'i!te2 66 lie?+ '1t>d3 67 l:!.d7+
'it>e3.
56 ... .l:!.e1 ! 57 a6 lia1
Now if 58 ltc6 Black decides matters with
58 . . . '1t>e 1 59 l:.e6+ 'it>f1 60 l:td6 (60 We3
l:te 1 +) 60 . . . d2 61 l:txd2 �a3+ , and White is
mated ! 58 l:ta8 'i!te1 59 a? d2 60 l:!.e8+ 'it>f1
61 .l:!.d8 .l:!.a3+ leads to the same fin ish .
58 lic7 �e1 59 'it>g2 l:txa6 60 l:te7+ 'it>d1 61
�xh7 l:ta2+ 62 Wf1 d2 63 .l::!.c7 l:ta1 64 'it>f2
lic1 White resigned .
bpk
88 �
PART I I
Endgame Analys is
Vlad i m ir Vu lfson
Typical Endi ngs with
Con nected Passed Pawns
I should l i ke to show you the rather
compl icated analysis of an ending of one
of my games. After studying it you wi l l have
a better understanding of the theory of rook
endings with connected passed pawns.
Zlotnik - Vulfson
Moscow 1 983
The adjourned position ; it is my move . The
national master Anatoly Donchenko sug
gested an excellent idea for Black. Usual ly
with an advantage you are recommended to
avoid pawn exchanges, but this is an
exception to the ru le.
1 . . . g4!
Black wants to tie the enemy rook to the
defenceof the g3-pawn . 2 l::th4 is obviously
hopeless, and therefore I stud ied , in the
main , 2 hxg4 .l:i.xg4 . Things are difficult for
White : noth ing is g iven by 3 .l:i.h6+ l:i.g6, and
so he is forced to play 3 l:th3, but here the
rook is exceptional ly passive .
But here Boris Zlotnik unexpectedly made a
move which I had overlooked in my analysis.
2 'it>b2
The idea is clear - to avoid the capture of the
g3-pawn with check.
The move is an interesting one, and duri ng
the game it seemed to me to be very strong .
But after a thorough analysis I began to
have doubts about its strength . The point is
that when Black e l iminates the g-pawn and
obta ins a position with connected passed
pawns, the basic method of defence is to try
and wedge the king between the pawns and
blockade them. But here the king , in solvi ng
a partial problem ( involving the g3-pawn),
voluntari ly moves away from the queenside
pawns.
2 . . .
3 �xh3
4 l::th8
gxh3
.l::tg4
With h is rook on h3 White , natural ly, cannot
hope for success, and so he activates it.
Typical Endings with Connected Passed Pawns CZJ 89
Black's reply is forced , s ince if 4 . . . .Uxg3?
there fol lows 5 .l::!.a8 with an immed iate d raw.
4 . . . aS
Passed pawns must be pushed .
S .l::!.c8
Aga i n it is not possible to capture on g3
because of 6 .l:!.c6+ fol lowed by 7 l:tc5 . I a lso
reckoned with 5 .l:!.a8 , to force one of the
pawns to advance and a l low the king more
quickly to wedge itself between them. But I
th ink that in this case too White would not
have been able to save the game.
5 . . . .l:f.g5
The rook defends the pawns from the side.
I n such situations this is the best place for
the rook . Now the black king is free to go
where it wants.
6 g4!?
Now the capture of the pawn leads to a
typical drawn position with connected passed
pawns , one which occurs qu ite often :
6 . . . .!::!.xg4? 7 .l:!.c6+ rJ; e 7 8 .l:!.c5 l:tb4+ 9 'itta3
a2 . Black cannot strengthen h is position ,
since his king has n o shelter from the side
checks.
6 . . . �e6?
6 . . .cj;>e5! suggests itself. Why d id I reject th is
move? The reason was a psychological one.
My opponent was the national master
Zlotn ik , a chess teacher in the I nstitute of
Physical Cu lture . I g reatly respected h im ,
and to me he was an expert. When you play
such a person , a defin ite complex appears ,
you beg in to fear everyth ing , and therefore i t
can be d ifficult to make an active move.
Besides, I did not consider that the position
was one where every tempo counted , I
thought my king would a lways be able to go
over and capture the g4-pawn , and for the
moment it would not be bad to help the
queenside pawns .
What would have happened after 6 . . . rJ;e5 ?
Let us try to provoke the advance of one of
the pawns: 7 l:!.a8 . Black repl ies 7 . . . a4 , and if
8 �a3(c3) , then s imply 8 . . . .l:txg4. The king
has not managed to reach b4 and after 9
l1b8 l1g3+ 1 0 �a2 .l:!.b3 Black wins.
I f 8 �b8 ( instead of 8 �a3) , then 8 . . . �d4 9
�a3 �c5 1 0 lk8+ �b6 1 1 llb8+ �c6 1 2
lk8+ �b7 and 1 3 . . . I:txg4 with a win . The
fact that his king is cut off along the 6th rank
does not concern Black - his rook wi l l free
the king by . . . l:tc4-c6 .
Thus, 6 . . . 'it>e5 was a very good move, but I
played differently.
7 l:tc1
White wants to place his rook behind h is
passed pawn .
7 . . . �d6?
7 . . . �d5 was far stronger.
8 .l:f.g1
I should l i ke to dwel l on this position in more
deta i l .
(see diagram)
White has succeeded in sign ificantly activat
ing h is rook. If h is pawn were on g5 , he
would undoubtedly be able to draw. But with
the pawn on g4 h is rook has not so many
squares for manoeuvring . Black now has
two plans for playing for a win :
90 � Typical Endings with Connected Passed Pawns
1 ) play his king to the help of the queenside
pawns;
2) fi rst capture the g4-pawn with the k ing,
and only then return to the queenside.
We wil l fi rst examine the simpler p lan ,
involving 8 . . . 'it>c5. I t is obvious that if Black
can place his pawns on a4 and b4 he wil l win
easi ly. Therefore White's objective is to
hinder the advance of the pawns, lu re the
rook away from g5 as soon as possible and
begin advancing his passed pawn .
First let us analyse 9 Wb3 . If 9 . . . b4 (with the
threat of 1 O . . . 'i!i>b5) there fol lows 1 0 'it>a4
'it>b6 1 1 .l:!.f1 .l:!.xg4 1 2 .l:!.f5 ! with an immediate
draw.
Let us verify 9 . . . a4+ 1 0 'it>a3 'it>b6! 1 1 'it>b4
(preventing 1 1 . . . 'it>a5) 1 1 .. Jie5! . Against the
threat of 1 2 . . . l:te3 White has two defences:
12 l:tg3 and 12 'it>c3 ( 12 'it>a3 .l:te2 is
unsu itable, since the k ing remains in a
mating net).
After 1 2 .l:!.g3 .l:!.e4+ 1 3 'i!i>a3 'i!i>a5 (with the
threat of 14 . . . .l:!.e2) 14 'i!i>b2(a2) b4 the black
pawns reach their goal sooner than the g
pawn . No better is 1 3 'i!i>c3 b4+ 14 'it>d3 a3!
1 5 'it>c2 .l:!.e2+ etc.
Let us examine 1 2 'i!i>c3. Here the win for
Black is not obvious.
I n lessons devoted to the technique of
converting an advantage, an important
principle has been mentioned : to make use
of any opportun ity to improve even slightly
your own position and weaken the oppo
nent's. Here Black can move his king
forward , but in this case the wh ite pawn
advances and there is no longer a win. The
only way to the goal is to interpose the check
1 2 . . . .l:!.c5+ ! . If 1 3 Wb2, then 1 3 . . . .l:!.g5, and
the b-pawn advances to the 4th rank. I n the
event of 1 3 'it>d4 Black can either advance
h is pawn immed iately, or fi rst play 1 3 . . . .l:!.g5 .
There only remains 13 Wb4 l:tc4+ 1 4 'it>a3 .
The d i rect 1 4 . . .a5? does not ach ieve
anyth ing : 1 5 'i!i>a2 b4 1 6 g5 b3+ 1 7 'it>a 1 ! a3
1 8 g6 b2+ (for 1 8 . . . a2 and . . . 'i!i>b4-a3 Black
is just one tempo short) 1 9 'i!i>b 1 Wb4 20 g7
'it>b3 2 1 .l:!.g3+ 'i!i>b4 22 .l:!.g 1 .
The correct move is 1 4 . . . .l:!.c2! (with the
threat of 1 5 . . . 'it>a5) 1 5 'it>b4 .l:!.b2+! (nothing
is g iven by 1 5 . . . l:tf2 1 6 'it>c3 ; fi rst the position
of the wh ite king must be clarified ) , and 1 6
Wa3 l:tf2 1 7 Wb4 .l:!.f3 is bad for White , while
after 1 6 Wc3 there fol lows 1 6 . . . a3 1 7 g5 b4+
1 8 Wc4 a2 1 9 lla1 ( 1 9 g6 .l:!.b1 ) 1 9 . . . b3 20
g6 l::tb1 21 g7.
Typical Endings with Connected Passed Pawns lZJ 91
21 . . . .l:i.xa1 22 g8'ii' .l:tc1 + 23 'it>xb3 a 1 "i'i.
[Nowadays, for the analysis of both opening
and endgame positions, increasing use is
made of computers. John Nunn and Graham
Burgess checked the concluding position of
this variation on a computer, and it tran
spired that after 24 "YIVb8+ Black cannot
avoid perpetual check. For example, with
the king on h7 there follows 1 "Y/Ve4+ 'lt;g7 2
'le7+, and with the king on c8 - 1 'ii'g4+
rttb8(b7) 2 'ii'b4+.
Nevertheless, Black has a way to win - but
instead of 21 . . . I:.xa1 ? he should play 21 . . . b2!.
Here is the analysis by Nunn and Burgess:
22 gB'fi (22 l:lxa2 .l:i.c 1 + 23 'lt;d4 b1"YIV 24
ga"fk 'ikb4+ with a quick mate or win of the
white queen) 22 .. .1J.c1+! 23 Wd5 (23 J:!xc 1
bxc1"fk+ 24 'lt;b3 'ilb 1 + 25 Wc4 'fib5+)
23 . . . .l:!.c5+ 24 'it;d6 .l:!.c6+!
24 . . . bxa1"ik? is premature: 25 "fibB+ 'lt;a6 26
'laB+ 'it;b5 27 "YIVb7+ 'lt;c4 28 "ilif7+ 'it;b4 29
'lb7+ 1:.b5 30 "ike4+ 'it;b3 31 "ilie6+! with
perpetual check.
25 cJ,;d7
After 25 cj;d5 bxa 1"YIV 26 "YIVb8+ 'lt;a5 27
'fa7+ 'it;b5 28 "ilib8+ I:.b6 29 'ii'e8+ 'it>b4 30
'ffB+ 'it;b3 the checks come to an end.
25 . . . bxa 1"YIV 26 "YIVb3+
No better is 26 "ikb8+ 'it;a5 27 "ik a8+ 'it>b4 28
'lb7+ 'it;a3 29 "ilia7+ 'it;b3 30 "ilib7+ 'it;c2! 31
"ilixc6+ "Y/Vc3 32 "Y/Va4+ 'it;b233 ikb5+ 'it;c 1 34
"YIVf1 + 'lt;c2.
26 . . . 'lt;c5 27 "ilia3+ (27 'flc2+ 'it;b4 28 "iie4+
'lt;a3 29 'ii'e7+ �b3 30 'ilif7+ 1Ic4) 27 .. .'ild4!
28 "iib4+ 'it>e3 29 ii'e7+ 'it;f3, and Black
blocks the next check with his queen or rook
- Dvoretsky.]
As you see , the win is very compl icated .
Besides, Wh ite's defence can be improved
at the very start of the variation . I nstead of 9
Wb3?! we can play more cunn ingly - 9
Wa3! , so that the pawn should move to a4
without check.
9 . . . a4 (no better is 9 . . .b6 1 1 Wb4 leads to a
position of mutual zugzwang , and with Black
to move. 1 1 . . . .l:te5 is pointless, since 1 2 . . . l:Ie3
is not a threat and White can s imply
advance h is pawn. After 1 1 . . . Wa6(c6) 1 2
.l:!.g 1 .l:te5 the move 1 3 . . . l:te3 i s n o longer
deadly and again 1 3 g5 can be played .
Let us try 1 0 . . . Wc6!? 1 1 Wb4 Wb6.
Now it is Wh ite who is in zugzwang. We
a l ready know that he loses after 1 2 .l:!.g 1
.l:!.e5; let us see whether 1 2 .l:!.g2 l:Ie5 1 3
Wa3! helps h im . The d ifference compared
with the position of the rook on g 1 is
immediately apparent: after 1 3 . . . Wa5 14 g5
the black rook cannot invade at e2 . B lack is
forced to advance h is pawn : 14 . . . b4+ .
92 � Typical Endings with Connected Passed Pawns
To where should the king move? The
outcome depends on this. Of course, 1 5
'it>a2! , so that the a-pawn should advance
without check. All the same Black plays
15 . . . a3 (but now without gain of tempo ! ) 1 6
g6 'it>a4.
Threatening 1 7 . . . b3+ with mate . White loses
after 1 7 .l:.g 1 b3+ 1 8 'it>a 1 a2, but he finds the
defence 17 Zig4! l:te2+ 18 'it>a1 ! . And now
18 . . . a2 is dangerous only for Black: 1 9 g7
'it>a3 20 .l:!.g3+ b3 21 .l:!.xb3+! 'it>xb3 22 g8'it'+.
Thus we have establ ished that in the event
of 8 . . . '1t>c5 White gains a draw.
In the game I moved my king the other way.
8 . . . 'it>e5
You see , the king has nevertheless reached
e5, but instead of going there immed iately it
has wasted time, by wandering about on the
e6- and d6-squares.
9 'it>b3
White intends by Ua 1 to force the advance
of one of the pawns, and then to establ ish
his k ing between them.
9 . . . 'it>f4
1 0 Ita1
Here I did not bother to ponder over which
pawn to advance , and this was a mistake -
one pawn move leads to a win , and the other
to a draw. Let us first see what happened in
the game.
1 0 . . .
1 1 'it>b4
1 2 .l:!a3?!
a4+?
'it>xg4
My opponent embarks on a ru inous course.
He probably thought that he would be able
to restrict my king along the 3rd rank, but in
fact his rook is badly placed here .
1 2 . . . 'it>f4
1 3 .Uc3??
Any move along the 3rd rank loses - the
rook should have moved off it.
1 3 . . . .Ug3
My rook goes to b3, after which the pawns
queen of the i r own accord . B lack won easi ly.
But how should my opponent have de
fended? Let us assume that we do not know
the theory of endings with such a pawn
arrangement - let us try acting simply by
using common sense .
Let us ask the question : 'What does Black
want?' Undoubtedly, to take his king to b2,
after which it wi l l be possib le to g ive up the
b5-pawn and queen the a-pawn . Let's try
and h inder the movement of the king, by
cutting it off along a fi le .
12 .l:!.f1 .l:!.f5 1 3 .l::!.e 1 'it>f4 1 4 1:te2 .l:!.e5 1 5 1:td2
�e3 1 6 ltd 1 .l:t.g5 1 7 .l:!.d8 �e4 1 8 .:td 1 1:tf5. It
is important not to place the rook on e5 -
then a check on e 1 wi l l d rive the king away.
1 8 . . . l:td5 is premature in view of 1 9 Ite1 + .
Black must play for zugzwang. The white
king on b4 is ideal ly placed , so the position
of rook must be improved .
1 9 �d2 Itd5 20 .Uc2 (now 20 11e2+ 'it>d3 is
hopeless for White) 20 . . . 'it>d3 2 1 .l:tc8 (21
l1h2 is also not bad ) 2 1 . . . 'it>d2 22 l:tc7 'it>d 1 .
(see next diagram)
Up to this point White has not been in any
particu lar danger, and he could have de
fended in various ways . But here he must
make an accurate move (23 .:tc3 or 23 l:f.h8),
s ince Black has created the concrete threat
Typical Endings with Connected Passed Pawns ttJ 93
23 . . . l:.d2 . For example, 23 'R.c7? l:td2 24
l:.c3 (24 �xb5 a3 25 �b4 a2 26 l:ta7 �c1 )
24 . . . .U.b2+ 25 �a3 .l:.b1 . After 26 l:th3 there
follows 26 . . . �c2 with the threat of 27 . . . .l:.b3+,
while if 26 l:tc8 , then 26 . . . �d2 , and the king
approaches the pawns. I t has acqu i red an
excellent shelter from the s ide checks at a5 .
This is one of the important winn ing posi
tions.
And now - the main d rawn position , which it
is also essential to know.
H. Kasparian
Could we have reached this position? Quite
possibly. Wh ite could a lways have placed
his rook on the 3rd rank. The only plan to
play for a win is 1 . . . �c2 2 l:th2+ l:td2 3 l:th3!
( it is important to control the a3-square)
3 . . . �b2 . At fi rst sight White is in trouble
Black intends 4 . . . .l:1d4+ and 5 . . . a3 . But let us
verify: 4 .l:.g3 l:td4+ 5 �c5! (5 �xb5? a3 6
.l:.g2+ �c3 7 .l:.g3+ .l:ld3) . If the rook goes to
e4 , it is now possible to capture the b5-pawn
and after 6 . . . a3 to begin side checks. The
king has to step onto the d-fi le , but then the
rook attacks the a-pawn and this leads to a
draw. This was Kasparian 's conclusion.
I n this position I d iscovered another curious
subtlety: B lack can try 5 . . . :1d 1 ! . Again it is
not possible to capture on b5, and 6 �b4 is
necessary, but then there fol lows 6 . . . l:.b1 .
Now it is essentia l to take the pawn: 7 �xb5!
a3 8 �a4 a2 9 l:tg2+ , and the king is
deprived of the important b 1 -square - draw!
[There is another way to draw: 6 .l:tg2+ �b 1
7 1:tg3 or 6 . . . �c3 7 �xb5 a3 8 �a4 -
Dvoretsky.)
Let us return to the position after White's
1 Oth move. We have seen that 1 O . . . a4+?
leads to a d raw. Let us now analyse 1 O . . . b4!
1 1 �a4 ( 1 1 'R.g 1 .l:.x:g4 is hopeless) 1 1 . . .
�xg4.
This pawn configuration is obviously stronger
than a4-b5, s ince after the sacrifice of the
a5-pawn the remain ing b-pawn is more
dangerous than the a-pawn , and affords
more winn ing possib i l it ies. For the moment
the black king is free to approach the
queenside ( 1 2 .l:.f1 1:tf5 etc. ) . Let us see what
methods of defence White has against
pawns on b4 and a5.
The first: playing for stalemate.
(see diagram)
If the black king goes to c3 , there fol lows
llc2+ ! . But this mechanism is easi ly de
stroyed - playing the black rook to the 2nd
rank proves decisive.
94 � Typical Endings with Connected Passed Pawns
The second: the attempt by White to
place his own king in the way.
The drawback to the position of the king on
b3 is that it comes under check along the 3rd
rank.
1 . . . 'it>d4 2 1:!.a2 l:te3+ 3 'it>a4 (forced) 3 . . .'�c3,
or 2 l:!.a4 l:!.e3+ 3 'it>b2 'it>c4! 4 .l:.xa5 l:!.e2+,
and a won position, well known in theory, is
reached .
Black wins in roughly the same way with the
enemy rook on a8 ( instead of a 1 ) : 1 . . . 'it>d4
with the idea of . . . .Ue3+ .
The third : to cut off the black k ing along
What is the s implest way to win here? Let's
improve the position of the rook: 1 . . . l:!.e5.
The threat is 2 . . . .l:.e3+ , l ift ing the blockade of
the pawns . Both 2 l:.d 1 + 'it>c5 3 'it>a4 .l:!.e3
and 2 'it>a4 'it>d3 (not immediately 2 . . J:te3? 3
'it>xa5 b3 4 'it>b4) 3 .l:tc8 (or 3 'it>b3) 3 . . . 1:!.e3 4
l:.c7 'it>d2 are hopeless for White . You see
that the key square for the king in th is type of
ending is d4 - it is very important to occupy
it! And after this - accord ing to ci rcum·
stances : if the wh ite king is at a4, the route
. . . 'it>d4-c3-b2(c2) becomes possible. With
the king on b3 it comes under checkand the
black king can then go to c5 and b5.
I t a l l seems to be very s imple, but look at the
fol lowing position :
the fi le. What is the evaluation? Draw! The king
Typical Endings with Connected Passed Pawns ltJ 95
cannot break through anywhere.
We arrive at a general ru le for this type of
end ing :
- If the black king is cut off in i ts own half of
the board , the position is d rawn ;
- But if it breaks through into the opponent's
ha lf of the board , the position becomes won .
A general conclusion for this type o f ending
also suggests itself:
Black's plan of moving his king towards the
white pawn and capturing it leads to a win ,
whereas the plan o f playing the king to the
help of the queenside pawns only d raws.
However, later analysis sometimes intro
duces serious corrections into seemingly
establ ished conclusions. On one occasion I
looked more carefu l ly at the position after
8 . . .�e5.
Why did White play 9 �b3 ? The immediate
9 J:ia1 ! was far more log ica l . Now the reply
. . . b5-b4 is no longer possible, and after
9 . . . a4 1 0 'lt>a3 (not 1 0 �c3 .l:!.xg4 1 1 .l:!.b 1
l:ic4+) White need not fear 1 O . . . l:!.xg4 1 1 .l:!.b1
with an immediate d raw. Black repl ies
1 0 . . .'it>d5 , but 11 'it>b4 �c6 1 2 l:Lg1 leads to
an already fami l iar d rawn position .
The number of mistakes that the two players
made in this endgame! The reason was an
inadequate knowledge of the theory of rook
endings. They had no 'beacons' by which
they could be gu ided .
Thus in the game the moving of the king to
the kingside (alas, rather belatedly) should
have led to a d raw. Then I again returned to
the position after 8 . . . 'it>c5 9 'lt>a3.
We have seriously stud ied only 9 . . . a4 , but
there is a lso another idea : 9 . . . b4+! 10 �a4
l':!.d5! . The th reat is 1 1 . . .�d3. There is no
point in return ing with the king : 1 1 �b3 �b5
(with the th reat of 1 2 .. J:td2) , and White
loses. In the event of 1 1 .U.c1 + �b6 1 2 l:tf1
l:td3 1 3 l:tf6+ �c5 1 4 l:tf5+ �c4 1 5 �xa5 ( 1 5
l:tf4+ l:td4 and 1 6 . . . l:txg4) 1 5 . . . b3 Black
wins, s ince h is passed pawn advances more
qu ickly than the opponent's, and also it is
supported by the king .
Let us verify 1 1 �xa5 l':!.d3 1 2 �a6
(otherwise mate ; 1 2 l:!.c1 + l::tc3 is bad for
White) 1 2 . . . b3 1 3 g5 .
(see diagram)
96 � Typical Endings with Connected Passed Pawns
The direct 1 3 . . . b2? (with the idea of 1 4 . . . .l:!.a3+
and 1 5 . . . .l:!.a 1 ) leads only to a draw in view of
1 4 .l:!.b1 .l:!.a3+ ( 1 4 . . . .l:!.d2 1 5 g6) 1 5 �b7 .l:!.b3+
16 cJi;c? 'it>b4 1 7 'it'd?! 'it>a3 1 8 'it>e6(e7) , and
the white king, paradoxical ly, succeeds in
uniting with its pawn .
1 3 . . . 'it>b4? is hopeless: 1 4 g6 b2 1 5 g7 lld8
1 6 g8'ii' l:txg8 1 7 l:.xg8, and if 1 7 . . . b 1 �? 1 8
ltb8+. After 1 3 . . . 'it>c4? both 1 4 g6 b2 1 5 g7
l:td8 16 l:tg4+ and 14 'it>a5 b2 1 5 g6 'it>c3 1 6
'it>a4 are possible [this last move is a mistake
in view of 16 .. .'i:J.d8 17 g7 :aB+ 1 8 cJi;b5 l:tg8;
a draw is given by 16 g7 .l:!.d8 17 .l:!.g3+ 'it>c4
18 .l:!.g4+ 'it>c5 19 .l:!.g5+ 'it>c6 20 l:tg6+ 'it>c 7
21 .l:!.g1 - Dvoretsky.]
However, Black finds a subtle solution :
13 . . . �d7 ! ! 1 4 g6 �g7 . I n th is way the
mobil ity of the wh ite king is restricted - now
it can neither approach the b-pawn , nor
move to the kingside. After 1 5 .l:!.g5+ 'it>b4 1 6
'it>b6 ( 1 6 .l:!.g4+ 'it>a3 1 7 'it>b5 b2) 1 6 . . . b2 1 7
.l:!.b5+ 'it>c3 1 8 .l:!.c5+ ( in the hope of d riving
the king to b1 and return ing to g5) B lack
repl ies 18 . . . 'it>d4! 19 l:tb5 .l:!.xg6+ and wins.
[In fact this way to win from the last diagram
is not the only one: in all the alternative
variations Black's play can be improved.
A) 13 . . . 'it>b4? 14 g6 :dB! (instead of 14 . . . b2)
15 g7
(see diagram)
In the event of 15 . . . .l:!.g8 White saves himself
by continuing 1 6 .l:!.g4+ (or 16 .l:!.g2) 16 . . /ila3
17 .l:!.g3 1;;a2 18 .l:!.g5! b2 19 .l:!.a5+ �b3
( 1 9 . . . cJi;b 1 20 .l:!.g5) 20 .l:!.b5+ cbc3 21 .l:!.c5+
'it>d4 22 1lb5. If instead 15 . . . .l:!.a8+, then 1 6
'it>b7 (or 1 6 'it>b6 b2 1 7 'it>c6) 1 6 . . . 'J:.g8 1 7
'it>c6 b2 1 8 'it>d5 (a very important tempo;
Black cannot reply 1 8 . . . 'IDc2 20
.l:.g2+ cJi;b3 21 1:!g1 ! 1;;a2 22 1;;e6, and the
king succeeds in joining up with the gl
pawn.
B) 13 . . . �c4 14 g6 (in the event of 14 'it>a5 b2
15 g6 the simplest win is by 15 . . . .l:!.a3+!, but
15 . . . .l:!.g3 16 .l:!.xg3 b 1"ik is also possible - in
view of the unfortunate position of the white
king) 14 . .. 1:l..d8! 15 g7 l:!a8+! (we already
know how a draw is gained after 15 . . . l:!gB?)
16 'it>b7 ( 1 6 'it>b6 b2 is no better) 16 . . . 11g8
17 'it>c6 b2 18 'it>d6 'fJ.xg7! (with the king on
b4 this move would not be possible), and
Black wins .
C) 13 . . . b2 14 .l:!.b 1 .l:!.d2! 15 g6 1:!g2 16 gl
r!g6+!, and on the next move Black will
capture on g7 either with check, or with a
threat of mate - Dvoretsky.]
Thus our in it ia l conclusion has been re
versed : the plan of moving to the help of our
pawns proves to be stronger than the march
of the king to the g4-pawn .
tb 97
Mark Dvoretsky
Adventu res on Adjourn ment Day
I t is said that winner of the first prize is
always lucky. From the examples demon
strated below, you wi l l see that i n the USSR
Cup (the club team championship of the
country) in 1 976 in Tbil isi our Burevestn ik
team was indeed lucky. But when you have
by no means the most impressive l i ne-up ,
competitive good fortune alone is not enough
to win by an enormous marg in (before the
last round we were a l ready T'h points ahead
of our nearest riva ls) . Our success was
largely secured by the friendly atmosphere
reigning in our team , the benevolence and
mutual help. An important role was a lso
played by our superiority i n the analysis of
adjourned positions (although from the
exam ples g iven below you would probably
not say this) - when they were resumed the
resu lts of a good dozen games came as a
pleasant surprise to us .
On the even ing before the adjournment day
our leader Vasi ly Smyslov adjourned h is
game against Mikhai l Tal in what was a
dangerous position for h im . At a team
meeting he said that the fol lowing day he
would need help with the analysis .
'Of course, of course, let's look at the
position together, ' grandmaster Ta imanov
offered his services.
'Thank you , Mark Evgenevich , but I would
l i ke to work with Mark lzrai levich , ' Smyslov
repl ied .
Of course , it is flattering to have the
reputation of being a good ana lyst, a lthough
at t imes it is sl ightly onerous - after al l , my
game was a lso adjourned . But the fol lowing
morning Smyslov and I sat down to look at
his position . After three hours of exception
a l ly i ntensive work my head was l itera l ly
spl itting , but on the other hand it appeared
that we had found a way to save the game.
Tal - Smyslov
Tbi l is i 1 976
Of course, it is unfavourable to play 42 g5?
.l::i.xh5 43 tt::lf6+ �g6 , and therefore Tal 's
sealed move was obvious .
42 tt::lxd6 cxd6
Now the capture of the b6-pawn a l lows
Black to activate h is king : 43 .l::i.xb6 �h6! 44
.l:!.xd6+ (44 f4 !? exf4 45 .l:!.xd6+ also does not
win ) 44 . . . �g5 45 f3 e4 ! with counterplay
sufficient for a d raw, for example: 46 .l:!.g6+
'it>f4 47 fxe4 a4.
Wh ite must play more sharply.
43 .l::i.xd6!
We have a choice between 43 . . . b5 44 .l::!.a6
a4 and 43 . . . .l::i.b 1 . In both cases the opponent
sends his k ing forward . The queenside
98 � Adventures on Adjournment Day
pawns do not advance very qu ickly - during
th is t ime danger impends over theblack
king. The fol lowing variation is an instructive
one, i l lustrating the typical ideas in the
position and the d ifficulties facing Black.
43 . . J�tb1 44 'it>f3! (44 g5? .l:tg 1 is prematu re)
44 . . . a4 (44 . . . .l:tb4? 45 'it>g3 and 46 g5) 45
'it>e4 a3 46 .l:td7+ 'it>h6 (46 . . . 'it>g8 47 .l:ta7
.l:tb3 48 'it>f5 or 48 g5 is completely bad for
Black) 47 \t>f5! (threatening 48 g5+ \t>xh5 49
.l:!.h7 mate) 47 . . . .l:!.g 1 48 .l:!.a7 .l:tg2 49 f4 ! exf4
50 .l:!.xa3 .l:!.g3 5 1 .l:!.a 1 ! f3 (not 5 1 . . .llc3 52
g5+) 52 'lt>f4 �g2 (52 . . . �h3 53 �b1 f2 54
IDb6+ 'it>h7 55 �b1 �b3 56 �f1 'it>h6 57
.Uxf2 also does not help) 53 'it>xf3 �c2 54
.Ub 1 , and Black is short of the one tempo
which wou ld enable him to block the pawns
securely and set up a fami l iar d rawn position
with king on g5 and rook on c5.
Even in such sharp endings, where every
thing can depend on a single tempo,
sometimes it is not worth immed iately
delving into the mass of variations . You
should fi rst logica l ly weigh up the situation
and look for a p lan , a general idea , which
should be carried out. What, do you th ink, is
this idea?
It turns out that Black should switch his rook
to the 8th rank. Fi rstly, from here it covers
the king - to checkmate it the opponent wi l l
have to br ing forward his king and both
pawns, and this demands time. Secondly, it
may be possible to place the rook behind its
own pawn and advance it, while g iving up
the other.
So, the general p lan has been found . It is
merely necessary to firid the most accurate
way of implementing it.
43 . . . b5
As Tal commented after the game, 43 . . . .Ub1
was nevertheless possible, but only in
connection with the plan indicated above :
after 44 'it>f3 a4 45 'it>e4 Black should play
45 . . . .l:tb4+ ! 46 'it>f5 .l:!.f4+ 4 7 'it>g5 .Uf8 .
44 �a6
45 'it>e3
a4
We considered 45 'it>f3 to be more accurate.
Tal was concerned about the reply 45 . . . a3,
but in th is case Black loses: 46 c.t>g2 .Ua1 47
g5 b4 48 g6+! (not 48 �a7+? Wg8 49 g6 Iic1
50 h6 l:!.c8) 48 . . . Wh6 49 .l:!.a7 with the
decisive th reats of 50 l:th7+ or 50 g7 'it>h7 51
h6 and 52 l'la8 . Therefore Smyslov would
have played as in the main variation of the
analysis - 45 . . Jlc1 ! . But after the move in
the game Black acqu i res an add itional
possib i l ity.
45 . . . lle1 +
4 6 'it>f3
46 . . . �c1
46 . . . e4+ 4 7 'it>f4 �e2 was also qu ite possi
ble. We analysed sharp variations such as
48 'it>g5 �xf2 49 �a7+ 'it>g8 50 h6 e3 51 'it>g6
�f8 and d id not see how Wh ite could win.
But on the other hand the main plan of
defence a lso seemed sufficient for a draw,
so that it was not easy for Smyslov to make
a choice . He real ised perfectly wel l that, i n
view of the lack of t ime for ana lysis, in any
branch a mistake cou ld creep in . The on ly
question was, where was this more prob
able?
47 c.t>e4 �c4+?!
Adventures on Adjournment Day C2J 99
This was how we intended to switch the rook
to the 8th rank. True, in this way the position
of the wh ite king is improved , although the
e5-pawn remains invul nerable. We rejected
47 . . . J:!.c8 ! , because we considered that
Black was lost in the position arising after 48
'it>xe5 .l:!.b8 49 g5 b4 50 .U.a7+ 'it>g8 51 �a4
b3 52 l1a 1 b2 53 l:!.b1 .
Not long before the resumption Vasi ly
Vasil ievich came up to me.
'You know, ' he said , ' i t would appear that
White's three pawns do not win . '
'That can't be so ! ' I said in su rprise , and I
tried to refute h is conclusion , but without
success. Here is the key position .
1 . . . l:tb3+ 2 'it>d4 ( 2 'it>d2 .U.b4 3 f5 .i:i.b5)
2 . . . J:.b4+ 3 Wc3 l:txf4 4 l:txb2 .U.h4! with a
draw.
The discovery made by Smyslov is instruc
tive and pretty, and I th ink that it is of
considerable importance for the theory of
rook endings. But we no longer had t ime to
verify it thoroughly.
{Many years later grandmaster Carsten
Muller nevertheless found a winning plan for
White. He suggested 2 '>t>e4 l:!.b4+ 3 �f5
l:!.b5+ 4 '>t>e6 (4 �g4 l:i.b4 is hopeless)
4 . . . 1:!b6+ 5 �d5 l:!.b5+ 6 �c6 l:!.b4 7 f5 tlg4 8
h6+ rtlh7 9 l:!.xb2 l:i.xg5 10 l:!.f2 etc. If Black
waits: 7 . . . .U.b8 8 h6+ �h 7, the most accurate
is 9 Wd5! (but not 9 �c5? .U.g8!) 9 ... l:!.b4 10
�e5 (because of zugzwang Black is forced
to allow the king into the lower half of the
board) 10 ... l:!.b5+ 1 1 �f4 l:!.b4+ 12 �g3 f!.b5
13 '>t>g4 .i:i.b3 14 �h4 with a decisive
zugzwang.]
To the grandmaster's question, which plan
of defence it wou ld be better to choose, in
reply I merely shrugged my shoulders .
Without waiting for advice , he sa id that he
would th ink about i t once more at the board .
And he made h i s choice in favour of the
main variation , which we had planned from
the very start. Alas, it was here that a
mistake had crept i n .
48 'it>f5
49 'lt>g5
50 h6!
.l:!f4+
l:i.f8
Here it a l l became clear to Smyslov. I n our
analysis we had somewhere g iven a check
on a?, after which there is no win . In sharp
endings such as this , every tempo is
precious - White leaves the king on h7, in
order to advance h is pawn to g6 with check.
50 . . . b4
51 l:txa4?
An unexpected amnesty at the very last
moment. Wh ite could have won by 51 �h5!
b3 52 g5 l:tb8 53 g6+ 'it>h8 54 h7 'it>g7
(54 . . . b2 55 'it>h6) 55 l:ta7+ Wf6 56 g7 .
1 00 � Adventures on Adjournment Day
51 ...
52 .l:.a7+
.l::tb8
It is now pointless to play 52 �h5 b3 53 g5
b2 54 g6+ 'it>h8!.
Draw.
52 . . . 'it>h8
53l:ta2
54.l:.b2
55 'it>f4
b3
e4
'it>h7
That same day I too resumed by game (also
a sharp endgame with passed pawns for
both sides). It was adjourned before Smys
lov's game, and so I had managed to look at
it, although , obviously, I no longer had time
to check the variations.
V. Kozlov - D voretsky
Tbi l is i 1 976
Analysis showed that, amazingly enoug h ,
the position was a forced draw.
41l:txd1!
The sealed move .
41 ...
42 f6
tt:lxd1
.l:.a8
After 42 . . . tt:le3+ 43 .ixe3 dxe3 44 'it>f3 lla8
45 tt:lb3 .Ua3 46 tt:lc1 b3 47 tt:lxb3 llxb3 48
g6 .if8 49 tt:lxe5 Black is unable to convert
h is extra rook.
43 tt:lb3
If 43 g6 I was intending 43 . . . hxg5 44 hxg6
.l::txa 1 45 .ih6 .l::ta7! 46 f7 .l:.xf7 47 gxf7 b3. 1t
later transpired that after 48 tt:lxe5 b2 49
tt:lc4! b 1 1i' 50 tt:lxd6 White does not lose, for
example, 50 . . . 1i'a2+ 5 1 'it>g3 (51 'it>h3?
'iie6+ and 52 . . .'iVxh6) 5 1 . . .'iVf2+ 52 'it>h3 (52
'it>g4? tt:le3+) 52 . . .'ii'f3+ 53 'it>h4 'ii'f6+ 54
'it>h5. Apart from 45 .ih6, a lso possible is 45
tt:lh6 l:ta7 46 g7+ .Uxg7 47 fxg7+ 'it>xg7 48
tt:lf5+ 'it>g6 (48 . . . 'it>f6 49 tt:lxd6 b3 50 tt:le4+
'it>f5 5 1 .ib4 b2 52 tt:ld2) 49 tt:lxd6 b3 50 lt:Jc4
b2 5 1 tt:lxb2 tt:lxb2 52 'it>f3 tt:lxd3 53 'it>e4 or
52 . . . 'it>f5 53 'it>e2 .
43 ...
44 g6
45 hx g6
46 tt:lxe5!
l:ta3
hx g6
.itS
The s implest way to d raw. 46 tt:lc1 b3 47
tt:lxb3 :txb3 48 .ih6 tt:le3+ 49 .ixe3 dxe3 50
'it>f3 was also possible. I merely wanted to
check whether my opponent might mix up
the move order by choosing 46 .ih6?. In this
case after 46 . . . tt:le3+ 47 .ixe3 (47 tt:lxe3
.ixh6 48 tt:lf5 .if8 49 g7+ .ixg7 50 fxg7+
�h7 5 1 tt:lc5 b3) 47 . . . dxe3 48 tt:lc1 Black
does not play 48 . . . b3? , but 48 . . . .U.c3!.
46... .l:tx b3
47 tt:lf7+
Adventures on Adjournment Day ltJ 1 0 1
I was expecting 4 7 i..h6 �b2+ 4 8 'it h 3 (after
48 'it>g3 there is the u npleasant reply
48 ... �d6!, and if 49 lt::lg4, then 49 . . . �g2+! 50
�xg2 i.xf4) 48 . . . I:i.f2 49 i..xf8 .l:.xf6 50 .iL.xb4
with a drawn endgame. I n my analysis the
move made by White in the game was not
even made on the board , s ince I thought that
after 47 . . .Black's extra pawn. And when my
opponent nevertheless went in for th is
variation , I instantly (a typical mistake!)
made the moves that I had p lanned before
hand.
You a lways have to reckon with the possibi l
ity of 'holes' i n your preparatory analysis -
after all , not a l l its detai ls wi l l have been
worked out with identical thoroughness.
Perhaps there was no point i n again
checking al l the previously prepared varia
tions, but at least I should have taken a fresh
look at the position , to avoid any bad
oversight.
47 ... 'it>g8
48 lt::lh6+ i..xh6??
48 . . . h i losopher was probably taken on
trust. I , on my own experience, have been
fortunate enough to satisfy myself of the
unusual capabi l ities of a knight.
Gheorghiu - Yusupov
Luzern 1 985
The Romanian player went in for this
position , erroneously assuming that he
would be able to construct an impregnable
fortress.
45 ...
46 'it>e2
f4!
If 46 gxf4 , then 46 . . .f6 49 .te1 'it>g7) 4 7 . . . gxf3 48
'it>e3 lLlf5+ 49 '>t>xf3 lbxd4+ 50 'it>g4 lbxb3,
and Black wins.
M ichel Montaigne
46 ... lbd6!
As Florian Gheorgh iu informed me after the
game, he overlooked th is knight move in his
adjournment analysis. 46 . . . 'it>f5 would not
have g iven anyth ing because of 4 7 'it>d3,
whi le if 46 . . . lbf6 there would have fol lowed
47 f3!.
47 'it>d3
Black's task would have been more compli
cated after 47 f3 . I t would appear that
47 . . . gxf3+ 48 '>t>xf3 lLlf5 49 'it>xf4 lbxd4
th rows away the win , s ince White activates
h is b ishop: 50 .tf2 lbxb3 51 .tb6 lbd2 52
.txa5 b3 53 i.c3 lbc4 54 g4 b2 55 i.xb2
lbxb2 56 a5, or 51 . . . lbc1 52 'it>e3! lba2 53
.txa5 b3 54 Wd2 h5 55 .tc3!. I nteresting
play results if instead of 54 . . . h5 Black plays
54 . . . d4!? 55 g4 'it>d5. In reply 56 'it>d3 !?
lbc1 + 57 'it>d2 comes i nto consideration, but
Wh ite can also go in for a sharp variation
suggested later by Mark Dvoretsky: 56 g5
'it>e4 57 h5 d3 58 i.c3! ttJxc3 59 'it>xc3 'it>e3
60 g6 d2 6 1 g7 d 1 'if 62 g8iV 'it'c2+ 63 'it>b4
b2 64 'ifg3+, and the king wi l l hardly be able
to avoid perpetual check.
The correct continuation is 47 . . . lbf5! 48 fxg4
ttJxd4+ 49 'it>d3 lLlf3 50 .tf2 lbe5+ 5 1 'it>e2 (if
51 'it>d2 there fol lows 5 1 . . . ttJxg4 52 .tb6
fxg3) 51 . . . f3+ 52 'it>f1 (s imi lar variations
occur after 52 'it>d2 lbxg4 53 .tb6 'it>e5 54
.txa5 'it>e4 55 .txb4 f2 56 'it>e2 d4)
52 . . . lbxg4 53 .tb6 'it>e5 54 .txa5 'it>e4 55
i.b6 (or 55 .txb4 '>t>e3 56 i.e1 f2 57 .txf2+
Solo for a Knight ctJ 1 03
I'Llxf2 58 a5 d4 59 a6 d3 60 a? d2 6 1 a8'ii
d1'i'+ 62 �g2 'i!Vh 1 mate) 55 . . . d4 56 a5 f2!
57 'it>g2 (57 a6 'it>f3 58 ii.xd4 lDh2 mate)
57 . . . d3 58 a6 d2 59 a? f 1'iV+ , and White
loses.
47 ... lbts
Now White is in zugzwang, and he h imself is
forced to break up h is fortress.
48 hS
48 �d2 would a lso not have saved White in
view of 48 . . .fxg3 49 fxg3ltJxg3 50 �f4ltJf5!
51 !Ji.c7 g3 .
48 ...
49 fx g3
50 h6
fx g3
�f6
White's last hope is the vulnerable placing of
the black pawns on the queenside. Thus the
careless 50 . . . ltJxh6?? is answered by 5 1
ixb4!.
50... g6!
Not so convincing is 50 . . . �e6 5 1 �f2 lDxh6
52 �e3 with chances of a d raw. When he
made this move, Black had to calculate the
variation which occurred in the game.
51 �d2 ltJx g3
52 .tx b4 ax b4
52 . . . ltJe4? 53 �xa5 g3 54 �c7 .
53 aS
head start: it requ i res just three more
moves, whereas the black kn ight can reach
the aS-square only i n fou r moves. D isap
pointment awaits Black if he tries to queen
his own pawn: 53 . . . ltJf5? 54 a6 g3 55 a? g2
56 a8'ilt' g 1 'ii 57 'ilt'g8+. However, as we
know, a wel l -tra ined horse is capable of
unusual feats . . .
53 ...
54�e3
lDhS!!
54 a6 ltJf4+ 55 �e3 lbe6 56 a? CDc? and
wins.
54 ... lbf6
55�4 �x h6
56 a6 ltJd7
57 a7 lbb6
58�x g4 �g6
59�4 �
60�g4 ltJa8
White resigned .
Yusupov - Li Zunian
Luzern 1 985
In th is position the game was adjourned for
the second t ime. Although during the fi rst
adjournment session I managed to win a
pawn thanks to the enthusiasm of the wh ite
knight , which accompl ished an heroic ra id in
I n the race to queen, the white a-pawn has a the enemy rear - ltJg4-f6-g8xh6-g8-e7-
1 04 � Solo for a Knight
c6-d4-e2--g3 - a draw sti l l seemed to me to
be the most probable outcome. After a brief
ana lysis it transpired that apart from the
exchange of the g-pawn there was no other
real istic plan of playing for a win . After this
White is left with a single target - the e4-
pawn . The impression was that Black could
fa i rly easi ly solve the problem of its defence.
However, serious work on the position
inspired hope, and I began to real ise that the
last wh ite piece , standing modestly at g3 ,
was tru ly a 'Montaign ian ' knight.
59. . . 'it>e5
60 'it>h5 f6
61 g5 fx g5
62 'it>x g5
Black is at the crossroads , s ince the bishop
can defend the pawn from various sides. For
a long time the plan chosen by the Chinese
player also seemed the strongest to me.
62... �c6
Black keeps h is bishop on the b7-a8
squares, and when h is king is evicted from
e5 it aims for d3 .
63 tt:Jts �as
Of course, not 63 . . . 'it>d5 because of 64
tt::le7+ .
64 tt::le7!
White must prevent the passage of the black
king to d3. For example, 64 tt::lh6? leads to a
d raw after 64 . . . 'it>d5 65 'it>f4 Wc4 66 "2lf5
'it>d3.
64 ... 'it>d6
The more accurate 64 . . . �b7 wi l l be ana
lysed later.
65 tt::lg6 'it>d5
66 'it>f4
White's p lan takes shape. The winn ing idea
is to occupy the key e5-square with the
knight . From there it not only covers the d3-
and c4-squares, but also a ims for d7 or f7.
66 . . . 'it>c5
In the event of the natural 66 . . . Wc4 White
would have won by 67 tt::le5+! 'it>c3 68 "2ld7!.
The threat is 69 tt::lc5, after 68 . . . �c4 there is
the fork 69 tt::lb6+ , whi le if 68 . . . 'it>b4, then
simply 69 tt::lf6 .
67 tt:Je5! �b7
Or 67 . . . 'it>b4 68 tt::ld7 , and Black loses.
68 tt::lf7!
Since there is no satisfactory defence
against the threat of 69 tt::lg5 ( if 68 . . . 'it>c4 69
tt::ld6+) , Black resigned .
Let us return to the position after 64 ttJe7.
I nstead of 64 . . . 'it>d6 Black had the more
cunning 64 . . . �b7. If now 65 tt::lg6+ 'it>d5 66
�f4, then 66 . . . �c4 67 tt::le5+ 'it>c3 .
This is a position of mutual zugzwang. With
Black to move he would be lost: 68 . . . �d2 69
Solo for a Knight lZJ 1 05
ti:ld7, 68 . . . i..a8 69 lZ'ld7 '01tc4 70 lZ'lb6+, or
68 .. . '01tb4 69 tt:Jf7 '01tc3 70 lZ'ld6. But it is
White to move and he is unable to win : 68
ti:lf7 'it>d3 , or 68 lZ'ld7 '01tc4 69 'iti>e5 i.c6.
The natural 66 '01tf4? was a mistake; Wh ite
can win by 66 '01tf5!! '01tc4 (66 . . . i..c8+ 67 '01tf4
ib 7 68 lZ'le5 i.a8 69 lZ'ld7, and Black has no
defence) 67 lZ'le5+ '01tc3 68 '01tf4 , and the
situation analysed by us is reached , but with
Black to move.
It remains to clarify what would have
happened if Black had carried out h is p lan
more accurately, i .e . obta ined the position in
the last but one diagram with h is bishop on
aB (with the bishop on b7 White wins by 1
ti:lf5,needs to remember about the g iven type of
end ing . As you see , not real ly so much and
not rea l ly so d ifficult!
Let us see how another, rather more ex
tensive section in our system of endgame
knowledge is constructed - the theory of
endings, in which a rook fights against
pawns. A basis for study can be provided by
any endgame manual , for example, l lya
Maizel is's monograph Ladya protiv peshek
(Rook against pawns) , publ ished in 1 956 ( in
contrast to opening books, those on the
endgame hardly date at a l l ) . Here some 400
How to Study the Endgame ltJ 1 1
positions are examined . Clearly we are not
able to study and remember all this informa
tion . We need to select the most important
key endings for practical purposes.
But how to choose the most important
material? This is the main problem . What
tel ls here is the player's i ntel lect , and h is
abil ity to work with books , to general ise, and
to draw conclusions. He is a lso helped by
the knowledge (even if incomplete) that he
already has, and by h is own practica l
experience in the g iven field .
The play i n endings with rook against pawns
is dynamic in character, and every tempo
has a decisive influence on the outcome. It
fol lows that here there is no large-scale
strategy, battle of plans, or deep regu larities
(as , say, in endings with opposite-colour
b ishops). There are also hardly any exact
positions, by relying on which we could
avoid the need for concrete calculation . The
main role is played by a knowledge of typical
techniques, which help the correct move to
be found more qu ickly and variations to be
calculated more certa in ly.
The procedures are best mastered with the
help of elementary positions, in which they
are employed and where their action is not
obscured by extraneous analytical deta i ls .
Subsequently the exact pattern of the
position my be forgotten , but the impression
of the techn ique wi l l remain . Sometimes
such a posit ion - the conveyor of the
technique - is s imu ltaneously an exact
position which is important for us ; in this
case, of course, we should memorise it .
So, using some very simple schemes, let us
examine the main ideas which apply in
endings with rook against pawns.
Cutting off of the king
White wins, by playing 1 .l:tg5! . When the
pawn reaches a3 , it can be e l iminated by
llg3 (or with the pawn on a2 - by l:.g 1 and
l:r.a 1 ) . If it is B lack to move , then 1 . . .'it>b5(c5) !
leads to a draw - as it is easy to see, cutting
off the king along the 4th rank by 2 l:tg4 does
not ach ieve anyth ing .
Promotion of the pawn to a knight
1 l:.h2+ 'it>c1 2 'it>c3 b1lD+! 3 'it>d3 lb a3 4
l::ta2 lbb1 1 with a draw, but not 4 . . . lbb5? ( in
end ings with kn ight against rook, the knight
should not be separated from the k ing) .
A draw also results from 1 . . . 'it>b1 2 'it>b3
'it>a 1 ! 3 l:txb2 - sta lemate. But with a
12 � How to Study the Endgame
bishop's pawn or a central pawn only the
promotion to a knight saves Black.
However, with a rook's pawn this idea no
longer helps.
1 'it>b4(c4) a2 2 'it>b3 a 1 4J+ 3 'it>c3, and
Black is in zugzwang.
I t is usefu l to note that if B lack also had a
pawn on b5, a l l the same this would not save
h im: 3 . . . b4+ 4 'it>xb4 4Jc2+ 5 'it>c3 tt:Je3 6
'it>d3 4Jd5 7 .Uh4 'it>b2 8 .Ud4, and the knight,
which is separated from the king , is soon
lost. 6 l:.h4! ( instead of 6 �d3) wins even
more quickly: 6 . . . �a2 (6 . . . 4Jd 1 + 7 'it>d2 4Jb2
8 .l:i.b4 'it>a2 9 �c2 'it>a 1 1 0 l:.b8 ; 6 . . . 4Jd5+ 7
�b3 'it>c 1 8 .l:!.c4+ �b 1 9 .Ud4) 7 .l:.a4+ �b 1 8
.Ue4 4Jf5 9 .Ue5 4Jd6 1 0 b3 c1 1 1 .i.:i.c5+
'it>b 1 1 2 .Ud5.
Stalemate
We have already examined one very impor
tant practical instance of stalemate . Here is
another example (wh ich , incidental ly, consti
tutes one of the few 'exact' positions that it is
usefu l to memorise) .
I t is hopeless to play 1 . . . a2? 2 .l:.b8+ 'it>a3 3
'it>c2! a 1 4J+ 4 'itc3 'it>a2 5 .Ub7 with
zugzwang. The only way to save the game
is 1 . . . 'it>b2! 2 .l:!. b8+ (2 .l:.h2+ 'it>b3 ! , but not
2 . . . 'it>b1 ? 3 'it>c3) 2 .. . 'it>c1 ! 3 .l:ta8 'it>b2 4 'it>d2
a2 5 ktb8+ �a1 ! .
As you see, when learn ing new ideas one
can sometimes repeat material that has
been covered earl ier ( in this case - promo
tion of the pawn to a knight) .
An intermediate check to gain a tempo
The diagram position arose in the game
Korchnoi-Kengis (Bern 1 996) . B lack re
signed , after calculating the fol lowing forced
variation .
How to Study the Endgame tZJ 13
1 . . .'it12 2 .l:tf8+!
2 'it>d3? g3 3 l::tf8+ 'it>e1 ! leads to a d raw.
2 . . . 'it>e2 3 l::tg8! 'it>f3
Thanks to the intermediate check, the king
has been driven one square back - from f2
to f3.
4 Wd3 g3 5 l:!.f8+ 'lt>g2 6 'lt>e2 etc.
'Shoulder-charge'
1 l:!.h2+ 'lt>a3!
By not al lowing the enemy king to approach
the pawn , Black ga ins a draw. It is i ncorrect
to play 1 . . .'1t>b 1 ? 2 'lt>b3 a 1 lD+ 3 'lt>c3.
Now let us examine a sl ightly more compl i
cated example.
1 . . . a5? does not work i n view of 2 .l:!.h5! - we
a l ready know this idea. But 1 . . . 'it>b5? is also
bad : 2 W a5 3 'lt>e6 'it>c4 (3 . . . a4 4 'it>d5) 4
.l:ta8 ! 'lt>b4 5 'it>d5 a4 6 'it>d4 'lt>b3 7 'it>d3 a3 8
.l:tb8+. The only saving move is 1 . . . '1t>c5! ,
preventing the approach of the enemy king
to the pawn.
Outflanking
The ideas of 'shoulder-charge' and 'out
flanking' are vividly expressed in a famous
study by Richard Reti ( 1 928) .
1 l:td2(d3) 1 ! d4 2 .l:i.d1 ! 'it>d5 3 'it>d7! , and
Black is i n zugzwang: if 3 . . . '1t>c4 4 'lt>e6, or
3 . . . '1t>e4 4 'lt>c6.
1 .l:td 1 ? is a mistake : 1 . . . d4 2 'it>d7 (2 'lt>f7
'lt>e4 3 'it>e6 d3) 2 . . . '1t>d5! (Black prevents the
outflank ing) 3 �c7 'it>c5! (3 . . . '1t>c4? 4 'it>d6!
d3 5 'it>e5) , and it is Wh ite who ends up in
zugzwang .
Let us now turn to positions in which a rook
fights against two connected passed pawns.
Mate threats to the opponent's king
I f the pawns are far advanced (two black
pawns on the 3rd rank , or one on the 2nd
rank and the other on the 4th ) , then the rook
is unable to stop them. However, sometimes
14 � How to Study the Endgame
it is possible to save the game, by pursu ing
the opponent's king when it is p inned to the
edge of the board .
B. Horwitz, J. Kling
1 851
1 �f5 �h4 2 'it>f4 �h3 3 �3 'it>h2 4 'it>e3 !
'it>g2
After 4 . . . �g3 5 l:tg 1 + �h4 6 Wf4 Wh3 7 'it>f3
bad is 7 . . . Wh2?? 8 l:.b1 , when Black loses
because of zugzwang.
5 Wd3 Wf3 6 Wc3 a2 7 'it>xb2 (or 7 .S.f1 + )
with a draw.
Intermediate check before the capture
of a pawn
In this position Herman Fridstein resigned
against Anatoly Lutikov (Riga 1 954 ) . He
considered the variation 1 �xb3 c2 2 l:.b4+
'it>d5 3 .l:tb5+ 'it>d6 4 .U.b6+ �c7 , but d id not
notice the saving i ntermediate check 1
.U.b4+ ! .
I should mention , i ncidental ly, that d ifferent
players can single out d ifferent ideas and
ru les, depending on their experience and
knowledge. I n the above example attention
should be paid to the manoeuvre with which
the black king escapes from the checks
(after 1 l:.xb3?) . But you can a lso d isregard
it , if this idea is a l ready wel l known to you .
The best position for the rook i s beh ind
the more advanced pawn
1 .S.g6! Wd7 2 .S.g4! g2! 3 .S.xg2 'lte6 4 l:!.g5! ,
winning thanks to the fact that the black king
is cut off from the pawn along the 5th rank.
I n Maizel is's book he g ives a position by
Sozin , wh ich d iffers only in that the wh ite
king is on a7. In this case after 1 l:.g6! Wd7
there is a second solution: 2 'it>b6 'it>e7 3
'it>c5 'it>f7 4 .S.g4 �f6 5 �d4! (5 .S.xf4+? 'it>g5 6
l:!.f8 'it>g4 7 �d4 g2) 5 . . . Wf5 6 .S.g8 and wins.
But with the king on a8 theand if 1 . . . '01td5, then 2 '01tf4 with the
irresistible threat of 3 lZ'lg3) . In this case 1
ti:lf1! leads to a win . After 1 . . . 'it>d5 there
follows 2 lZ'ld2 'it>e5 (3 '01tf4 was threatened )
3lLlc4+ '01te6 (3 . . . '01td5 4 lZ'lb6+) 4 'it>f4 and 5
ti:ld2, whi le if 1 . . . i..b7 - 2 tZ'lh2 '01td5 3 lZ'lg4
'it>c4 4 lZ'le5+ 'iti>c3 5 '01tf4 , and aga in a
familiar position is reached (cf. the last
d iagram).
Thus the system of defence with the bishop
on b7-a8 runs in to a far from obvious
refutation .
Black could have stuck to another l ine of
defence with h is bishop on g2-h 1 . We wi l l
examine the fol lowing important positions.
Here White wins i rrespective of the turn to
move.
1 lZ'lg6+ '01td5 2 'it>f4 i.f1
If 2 . . . i..e2, then 3 lZ'lf8! i..f3 4 lZ'lh7 '01tc4
(lZ'lf6+ was threatened ) 5 lZ'lg5 .
3 tt:Je 7 + '01te6
3 . . . 'it>d6 loses immediately in view of 4 lZ'lg8!
i..g2 5 lZ'lf6 .
4lZ'lc8!!
This at fi rst s ight r idiculous move becomes
understandable, if the goal of the knight's
unusual route is noticed - the c3-square .
4 ... i.d3
Other continuations also fa i l to save Black:
a ) 4 . . . i.g2 5 lZ'lb6 '01td6 6 lZ'la4 '01td5 7 lZ'lc3+;
b ) 4 . . . '01td7 5 lZ'lb6+ '01tc6 6 lZ'la4 and 7 lZ'lc3 ;
c) 4 . . . i..b5 5 '01txe4 '01td7 6 lZ'lb6+ 'it>c6 7 lZ'ld5;
d ) 4 . . . 'it>d5 5 lZ'lb6+ '01tc5 6 lZ'le7+ 'it>d6 7 lZ'lf6 .
5lZ'lb6 i..c2 (otherwise lZ'lb6-a4-c3) 6lZ'lc4
Although White has not i n fact managed to
transfer h is knight to c3, h is achievements
are very considerable : the black bishop has
been forced onto the b 1 -h7 diagona l , where
it is less wel l placed .
6 ... 'it>d5 7lZ'ld2 i.d3
8 'it>f5
Black is in zugzwang and is forced to al low
the knight to go to f1 (8 . . . i.e2 9 lZ'lxe4 i.d3
1 0 f3) .
1 06 � Solo for a Knight
8 ... �c2 9 tL'lf1 �d1
Or 9 . . . �c4 1 0 tL'lg3 �d5 1 1 tL'lh5 and wins .
10 tL'lh2 �c2 11 t2Jg4 'it>c4 12 tL'lf6 �d3 13
t2Jxe4 �e2 14 �f4, and White wins.
Here I should make a sl ight d igression and
refer the reader to the start of th is interest ing
endgame, where the author praises the
'enthusiasm of the white knight' . In order to
el iminate the last bu lwark of Black's defence
- the e4-pawn , the wh ite knight had to
complete a veritable round-the-world jour
ney (f4--g6--e 7 -c8-b!H;4-d2-f1 -h2--g4-f6-
e4) .
I n the position from the last but one d iagram
it could have been Black to move.
1 ... Sl.f1 2 tL'lg6+ �d5 3 �4 Sl.g2 4 tL'lh4!
Sl.f1
If 4 . . . .th 1 , then 5 �g3! �c4 6 �h2 , forcing
the exchange of bishop for knight .
5 tL'lf5 Sl.g2 6 t2Jg3
A very important position of mutual zug
zwang . If it is Black to move he is forced to
occupy the f3-square with h is bishop,
al lowing the knight to go to f1 .
6 ... Sl.f3 7 tL'lf1 Sl.d1
7 . . . Sl.g2 is bad because of 8 tL'ld2 ! Sl.h 1 9
�g3.
8 tL'lh2 Sl.c2 9 �f5
I f 9 t2Jg4, then 9 . . . �e6 .
9 •.• �c4 10 tL'lf1! �d5 ( 1 0 . . . 'it>d3 1 1 tL'lg3 ) 1 1
tL'lg3 fol lowed by tL'lh5, transposing into a
winn ing position which is a l ready famil iar to
us .
Although the b ishop is at g2 , a l l the same
Wh ite is able to win .
1 tL'lf5!
Noth ing is g iven by 1 tL'lh5 .tf3 ! (but not
1 . . . .t h 1 2 t2Jf4 .tf3 because of 3 tL'lg6+ '.t>d5
4 �f4 .th 1 5 tL'lh4 �c4 6 �g3 �d3 7 '.t>h2
�e2 8 �xh 1 �xf2 9 t2Jf5) 2 t2Jf4 .t h 1 ! .
1 ... �d5
Or 1 . . . .tf3 2 tL'lh4 .td 1 3 tL'lg6+ �d5 4 'ii'f5
Sl.f3 5 �f4 , s imi lar to the main variation.
1 . . . Sl.h 1 2 tL'lh4! is bad for Black.
2 tL'lh4! .tf1 3 �f5! �e2 4 tL'lg6 Sl.f3 5 �f4
.tg2
Black a lso loses after 5 . . . .te2 6 tL'le7+ �e6
7 t2Jf5 �d5 8 tL'lg3 .tf3 (the position of
mutual zugzwang from the last but one
d iagram has been reached ) 9 tL'lf1 etc.
6 tL'lh4 Sl.f1 7 t2Jf5 .tg2 8 tL'lg3
Again a fami l iar mutual zugzwang position .
Wh ite wins .
I t is more d ifficult to win when Black moves
fi rst:
1 ... Sl.h3! (preventing the important manoeu
vre t2Jg3-f5-h4) 2 �h4! (White tries to give
Solo for a Knight C2J 1 07
his opponent the move) 2 ... .i.c8
Or 2 . . . .i.g2 3 'it>g4 .i.f3+ 4 'it>g5 .i.g2,
reaching the position from the last d iagram
with White to move.
3 'it>h5! .i.d7
If 3 . . . .i.h3, then 4 'it>g5 , and White has
succeeded in giv ing his opponent the move .
4 . . . i.c8 5 lt:Jh5 .i.h3 6 lt:Jf4 fol lowed by lt:Jg6+
and r.t>f4 transposes i nto variations analysed
earl ier.
4 'it>g6! .i.g4
Or 4 . . . �c8 5 lt:Jh5 Wd5 6 tt:Jf6+.
5 lt:Jh5 .i.f3
If 5 . . . �h3 , then 6 lt:Jf4 is possib le .
6 lt:Jf4 .i.h1 7 'it>g5 ..tf3
This position has a l ready occurred i n the
notes. I wil l remind you of the winn ing
method : 8 lt:Jg6+ 'it>d5 9 'it>f4 .i.h 1 1 0 lt:Jh4
'it>c4 1 1 'it>g3 'it>d3 1 2 'it>h2 'it>e2 1 3 'it>xh 1
'it>xf2 1 4 lt:Jf5 .
With th is the author concludes h is analysis
of th is interesting endgame. I wil l be very
indebted to readers for any corrections ,
refinements or refutations.
1 08 �
Mark Dvoretsky
More about the 'Montaign ian' Kn ight
W
hen I saw the analyses by Artur
Yusupov in the previous chapter I
was reminded of several stud ies on the
same theme, which in their t ime made a
strong impression on me. I hope that they
will also appeal to you . Their beauty l ies in
the unusual amount of work carried out by
the wh ite knight, the paradoxical manoeu
vres of the wh ite pieces, and above al l the
precision and depth of logic behind these
manoeuvres.
Before enjoying the solutions , try to find the
answers yourself. I should warn you before
hand: the problems are very d ifficult , and
you wi l l most probably have to move the
pieces on the board (perhaps only the
second example might be solved in you r
head ) . But even so, don't b e in a hurry to
make moves - first reason to yourself about
the final and intermed iate aims of both
sides, the plans they wi l l carry out, any
important intermediate positions, and so on .
N. Grigoriev
1 932
A knight can stop a rook's pawn, if it
succeeds in 'touching' any square in its
path (apart from the corner square h1). In
the g iven instance it is clear that the knight
wi l l aim for the h2-square . How to reach it
seeing as the black k ing stands in its way?
Here are some logical considerations, which
wi l l make it easier to find the solution . The
knight can reach h2 via g4 or f1 . Each of
these routes can be control led separately by
the black king . I t is necessary to create a
'double attack' - by threaten ing to go to h2
by both ways . The kn ight can reach f1 via
d2, and g4 via e5. Do you see the
intersection point of these two routes?
1 tt'lb4! h5
2 tt'lc6!
Not 2 tt'ld5+? 'itof3 ! , and the pawn cannot be
stopped . Note that a king restricts a
knight most effectively when there is one
square between them along a diagonal,
or two squares between them along a
rank or a file.
2 ... 'it>e4!
Of course , not 2 . . . h4? 3 tt'le5 , when the g4-
square can be covered only by 3 . . . Wf4 , but
then there fol lows 4 tt'lg6+.
3 tt'la5!!
Only in th is way can the knight reach the key
c4-square , from where it can a im for both f1,
and g4 . 3 tt'ld8? would have lost after 3 . . . h4
4 tt'le6 'it>f5 ! 5 tt'ld4+ 'ltog4 .
3... h4
4 tt'lc4!
4 tt'lb3? �e3 .
4 ... 'itof3! ?
More about the 'Montaignian' Knight lLJ 1 09
The last trap . 4 . . . h3 5 lZ'ld2+ and 6 lZ'lf1 leads
to an immed iate d raw.
5 lDe5+!
5 liJd2+? (hoping for 5 . . . f2 , and B lack wins . I n th is
variation he is able to deflect the knight from
its route to h2 , and , as you can see ,
'touch ing ' the h 1 -square with the knight
does not help White .
5 . . . 'it>g3
Forced: the g4-square must be guarded , but
if 5 . . . 'it>f4, then 6 ltJg6+.
6lZ'lc4! h3
7lZ'le3!
White has achieved h is a im : If 7 . . . h2 8 lZ'lf1 + ,
while after 7 . . . 'itf3 there fol lows 8 lZ'lf1 g2 1 0 ltJg4 e4? 3 lZ'la3 ! 'itd3 (3 . . . h4 4 ltJc4) 4 'it>b5!
and lbc4, but 2 . . .'�f2 ! leads to a win .
D . Gurgenidze
1 970
This study is a n a rtistic adaptation of a
position by N ikola i Dmitrievich Grigoriev.
The threat of win n i ng the kn ight is easi ly
parried by approach ing the b5-pawn with the
k ing : 1 ..t>b3(a3) 'it>f7 2 'it>b4 'it>g7 3 'it>xb5
'it>xh7 4 'it>c4, and the king enters the square
of the h-pawn. What, then , is the problem?
It turns out that B lack can save a very
important tempo by avoid ing the attack on
the knight and satisfying h imself with merely
restrict ing its mobi l ity : 1 . . .b4 'itf5 3
'it>xb5 h5 , and the pawn queens. This means
that i n reply to 1 . . . 'it>e6 White must immedi
ately tackle the h-pawn with his knight.
In a practical game without much hesitat ion
many players would play 1 'it>b3 'it>e6 2lZ'!f8+
'it>f5 3 ltJd7 h5 , and only now th ink about
where next to d i rect the kn ight . The para
doxical feature of the position is that here
such a genera l ly-accepted way of acting
does not work - it is necessary to th i nk
earl ier!
1 'it>a3!!
2lZ'lf8+!
3lZ'ld7
4 ltJc5
5lZ'lb3!!
f5
h5
h4
This is why it is important to calculate a l l the
variations as early as the fi rst move - the b3-
square must be left free for the wh ite knight .
5 lZ'ld3? would have lost after 5 . . . h3 6 lZ'lf2
h2 7 f4 8 'it>xb5 f4 7 ltJf1 leads to a fami l iar drawn
position .
7 lZ'lf1 h1�
8lZ'lg3+
(see diagram)
1 ltJg1
I n contrast to the previous position , here the
fi rst move can be made without th i nki ng . But
what next - how to drive away the e nemy
king? For a start we at least need some idea .
Let's see where the kn ight should a im for, i n
order to create d ifficu lties for the opponent.
1 1 0 � More about the 'Montaignian' Knight
R. Reti, A. Mandler
1 924
We find the square c2 . From here the knight
takes away the important e3-square , and it
is itself invulnerable in view of the reply 'i!te2 .
The black king has to be at d2 or d 1 . But with
the king on d1 Wh ite has the decisive tt'lb4!
fol lowed by tt'ld5. I t is clear that here we
beg in to have mutual zugzwang positions.
Black has to p lay accurately from the very
start. 1 . . . 'it>d3? is bad in view of 2 tt'lf3 'i!te3 3
tt'le 1 ! Wd 2 4 tt'lc2! Wd 1 ( 4 . . . 'i!td 3 5 'it>e 1 !
'>t>xc2 6 �e2) 5 tt'lb4! Wd2 6 tt'ld5.
1... 'iit>d2!
2 tt'lf3+ �d3!
Now it is pointless playing 3 tt'le1 + 'it>e3 ! 4
tt'lc2+ '1t>d2 5 tt'lb4 'it>e3 6 tt'ld5+ 'it>e4 7 tt'lf6+
'ite3. To win , the opponent must be given
the move. But how can this be ach ieved?
3 'iit>e 1 ! 'it>e3
4 tt'le5 'it>e4
In the event of 4 . . . 'iit>d4 White wins by 5 tt'lg4!
'it>d3 6 'it>d 1 , breaking forward with the k ing,
since 6 . . . f3 fai ls to 7 tt'le5+.
5 tt'lc4! '.t>d3
5 . . .f3 6 tt'ld2+, or 5 . . . 'it>d4 6 'it>e2 ! .
6 tt'ld2 'it>e3
7 tt'lf3 'it>d3
8 'it>f1!
The kn ight's circu lar journey has enabled
White to achieve h is a im - he has given the
opponent the move . The rest is al ready
fami l iar to us .
8 ... 'it>e3
9 tt'le1 'it>d2
10 tt'lc2! 'it>d1
11 tt'lb4! 'it>d2
12 tt'ld5
R. Reti, A. Mandler
1 924
Here White's task is even more d ifficult than
in the previous study. The winn ing plan
suggested there (wh ich in itself was not at all
obvious) does not work here: the left edge of
the board interferes. After a l l , the knight has
no square equ ivalent to the important b4-
square in the previous example .
True, a new possib i l ity has appeared - the
activation of the king along the route d 1 -e1-
f1 -g2-f3 . I t is obvious that Wh ite has no
other winn ing pla n . Black wi l l try to prevent it
by attach ing h is k ing to the e2-pawn from d2
or d 1 . I t is easy to imagine that here too
mutual zugzwang positions cannot be
avoided . Let's try to work th i ngs out!
Let's suppose that Wh ite's knight is on d3,
his king on f1 , and the black king on c2 .
More about the 'Montaignian' Knight ttJ 1 1 1
Then Black loses after 1 . . . 'it>d2? 2 lt:Jf4 (2
tt'le5 �d 1 3 lt:Jf3 is a lso good ) 2 . . . 'it>d 1 3
Wg2 . He must play 1 . . . 'it>d 1 ! 2 lt:Jf4 (2 lt:Je5
lt>d2 3 lt:Jf3+ �c3) 2 . . . 'it>d2 ! , and if 3 'it>g2 ,
then 3 . . . d 3 ! with a draw.
From this the fol lowing conclusions can be
drawn : if Wh ite plays We 1 , then with the
knight on d3 Black must reply . . . 'it>c2 ! , whi le
with the knight on f4 , e5 or c5 the correct
rep ly is . . . 'it>c3 ! . The position with the knight
on c5 , b lack k ing on c3 and wh ite k ing on e 1
is one of mutual zugzwang .
The fol lowing step i n our logical analysis of
the position is to clarify the importance of the
e4-square for the kn ight . Let us suppose
that the kn ight stands on e4 , the wh ite k ing
on d 1 , and the black king on b3. Then if it is
Wh ite to move 1 �c1 ! is decisive. But things
are no easier for Black if it is h im to move : if
1...Wb2 there fol lows 2 tt:Jc5 ! 'it>c3 (2 . . . �b1 3
tt'le6) 3 �e1 ! , and the afore-mentioned
position of mutual zugzwang is reached with
Black to move .
Thus the knight must be brought to the e4-
square. This is not at a l l easy to ach ieve ,
seeing as White constantly has to watch out
for . . . d4-d3 .
1lLle1 �b2
2lt:Jd3+ �c3
Noth ing is changed by 2 . . . �b1 3 tt:Jc1 �b2 4
tt'la2 , whi le 2 . . . �b3 shortens the solution : 3
tt'lf4 Wb2 (3 . . . 'lt>c3 4 �e 1 ! 'it>c2 5 lt:Jd3 , and
Black is i n zugzwang) 4 tt:Jd5 �b3 5 lt:Jc7!
etc.- cf. the main variation .
3 lt:Jc1!
But not immed iately 3 lt:Jf4 because of
3 . . . Wb3 ! , and if 4 lt:Jd5 there is the reply
4 ... '.t'c4 ! .
3... �b2
4 lt:Ja2!
A manoeuvre , found in the solving of the
previous study, also comes in usefu l here .
As you remember, 4 . . . �b3? 5 �c1 ! �xa2 6
'it>c2 is bad for Black. He is forced to move
his k ing along the 1 st rank , away from the
c3- and c4-squares.
4... �b1
5 lt:Jb4 'it>b2
6lt:Jd5 'it>b3
6 . . . 'it>b1 ? loses immediately to 7 lt:Jc7 �b2 8
lt:Jb5.
7lt:Jc7!
The shortest route to e4 is via f6 . However,
the d i rect 7 tt:Jf6? is refuted by 7 . . . �c4 ! ! 8
'it>c2 d3+! 9 exd3+ �d4 and 1 O . . . e2. There
fore the knight chooses a more i ntricate
route : d5-c7-b5-d6-e4.
7... �c3
Here 7 . . . 'it>c4 8 'it>c2 is now pointless.
8lLlb5+ �c4
9lt:Jd6+
9 . . . �d5 1 0 lt:Jf7 .
10 lt:Je4+
1 0 . . . c1 .
11lt:Jc5!
12 'it>e1!
'it>c3
�b2
'it>c3
White has ach ieved h is a im - he has set up
the requ i red position of mutual zugzwang
with h is opponent to move .
12 . . . 'it>c2( c4)
13 lt:Jd3
Aga in zugzwang!
13 ...
14 �f1
15 lt:Jf4
�c3
'it>d2
The fi na l , decis ive zugzwang . 1 5 lt:Je5 �d 1
1 6 lt:Jf3 is equal ly good .
15 .. . 'it>d1
16 �g2
Thanks to the lengthy knight manoeuvre ,
the wh ite k ing has fina l ly gained the opportu
n ity to break free.
1 1 2 �
PART II
Technique
Mark Dvoretsky
Converti ng an Advantage
C
hess players suffer from many ai l
ments . One of the most common and
serious is poor techn ique in the conversion
of an advantage. Even champions some
times suffer from th is i l l ness - it is sufficient
to remember the 1990 match for the world
championship between Garry Kasparov and
Anatoly Karpov.
How frequently after an unsuccessfu l game
do we state with vexation to our opponent,
trainer, ora casual spectator: ' I had a
completely won position ! ' But it is pointless
complain ing about fate - it is better to th ink
about the causes of the mistakes made, and
endeavour to understand what playing or
personal deficiencies are behind your er
rors . I wi l l now dwel l briefly on the main
factors which h inder the normal conversion
of an existing advantage.
I . Tiredness to wards the end of a game
It is clear that after several hours of
intensive struggle a player becomes tired .
But some become more t ired than others . I t
is in the last few minutes that the fate of a
game is often decided , and therefore many
additional points can be gained by a p layer
who reta ins a sufficient reserve of energy
towards the end of a round .
The play of grandmaster Yusupov is a lways
time and energy on solving the problems
wh ich confront h im in the fi rst half of a game.
But for its later part he often lacks strength,
and here he makes serious errors . It was
only for th is reason , for example , that he did
not win h is Candidates semi-fina l match in
1989 against Karpov. Yusupov constantly
outplayed h is formidable opponent, but was
u nable to convert th is into wins on account
of t iredness towards the end of a game.
Here is one of the most annoying examples.
Karpov- Yusupov
Candidates Match , 6th Game,
London 1989
deep and interesting , and he spends much Yusupov saw that in the variation 38 . . . a3?
Converting an Advantage ttJ 1 1 3
39 .l:la6 .Uf2+ 40 'it>e1 a2 41 f5 he would not
have time to win the rook: 4 1 . . . .Uxh2 42 f6
l:ih 1 +? 43 'it>e2 a 1 'i!V 44 .l::!.xa 1 .l::!.xa 1 45 f7.
Therefore he played 41 ... 'it>d7 , and after 42
f6 'it>e6 43 .l:ta8 ! 'i.t>xd6 44 f7 .l::!.xf7 45 .U.xa2
lt>e5 46 lla6 the players agreed a d raw.
Yet Black could s imply have captured the
pawn:
38 . . . l:txh2!
39 .l:.a6
39 f5 .Uf2+ and 40 . . . .l:txf5 .
39 . . . l:.f2+
40 'it>e1 .l:txf4
41 'i.t>e2 .U.e4
With an easy win , for example : 42 .l:ta5
(otherwise . . . h6-h5-h4) 42 . . . '1t>d7 43 l:.d5
h5! 44 IIxh5 'it>xd6.
Why d idn 't Artur play this? By h is own
admission , at that moment his mind had
simply switched off, and he d id not see any
other possib i l it ies apart from 38 . . . a3?.
I f you are let down by ti redness, perhaps i t
means that a l l is not wel l with your physical
preparation? The prescription in such cases
is clear - you must do more physica l
exercise and devote more t ime to sport , i n
particu lar exercises for stamina (for exam
ple, slow but long-distance runn ing) . Con
sider devising a rational dai ly reg ime during
a competition , enabl ing you to relax properly
and regain your energy before a new game.
F inal ly, a lso during a game you can husband
your strength , by using for relaxation those
brief minutes of respite when it is your
opponent's turn to move . However, a l l these
are fa i rly serious questions , demanding a
special d iscussion , and not just a brief
mention .
II. Insufficiently stable nervous system
Throughout a game it is very important to
mainta in fu l l concentration and u nbroken
attention to everything that is happening on
the board . But it is not everyone's nervous
system that is ready for such prolonged
tension . Often a player composes h imself
only at especia l ly important moments of the
strugg le , but when the main problems seem
to h im to be resolved , he loses his vig i lance
and beg ins act ing carelessly. It is here that
mistakes usual ly occur.
Mestel - L. Popov
Olympiad , Malta 1 980
White is a sound pawn to the good , but he
now has to resolve a d ifficult problem: what
posit ion to go in for, so that the opponent wi l l
have the least in the way of counter
chances. The fol lowing possib i l ities suggest
themselves:
a ) 25 .l:te 1 .l:I.d2 (25 . . .'it'c5 is less accurate in
view of 26 'iWa6! .Ud2 27 .l:.e2) ;
b ) 25 'i�Nxc6 .l:.xd 1 + 26 .i.xd 1 'iVxa2 (stronger
than 26 . . . 'i�Nxe5 27 'ifa6! ) 27 'ifb5 'iVa 1 28
�e2 a5 , and it is not easy for Wh ite to
strengthen h is position ;
c ) 25 .Uxd8+ 'i!Vxd8 26 'ii'xc6 'i�Nd2 ;
d ) 25 .l:I.xd8+ 'i!Vxd8 2 6 �xa7 'iVd2 .
I n every case Black reta ins cou nterplay, and
the outcome remains unclear.
Jonathan Mestel found an excel lent solu
t ion .
25 .l:.xd8+! 'i�Nxd8
1 1 4 � Converting an Advantage
26 i.c4! !
The bishop wi l l securely defend the king
from f1 . For the moment the queen remains
on b?, from where it defends the b2-pawn.
The a?- and c6-pawns are weak, and soon
White is sure to create a passed pawn on
the queenside.
26 . . . �d2
27 i.f1 'lt>g7
27 .. .'ti'c2 was more tenacious, preventing
the fol lowing strong move by the opponent.
28 a4! aS
29 'ir'b6 h5
Black's last fa int hope is to break up the
wh ite king's defences by the advance of h is
g- and h-pawns. He no longer has any other
possibi l it ies.
30 'ii'xa5 i¥xb2
31 'ir'b4 li'a1
32 a5 g5
33 a6 g4
It is clear that Black's position is completely
hopeless. But it is very dangerous, in
bel ieving th is , to weaken your attention and
stop checking variations. For example, if
Wh ite plays 34 'ii'b8 'ii'c1 35 a?? (35 'ii'b6! is
correct) , then after 35 . . .'!"i'le3! 36 fxe3 'ii'xe3+
37 'it>h 1 "it'c1 (e1 ) 38 'it'h8+ (38 'ii'b6 �xf1 +
39 �g 1 li'a6 with equal ity) 38 . . . 'it>xh8 39
a8'6'+ 'it>g? 40 'ii'a6 "ii'xc3 Black ga ins qu ite
good saving chances - at any event, a
lengthy struggle sti l l l ies ahead.
34 'ii'b?! was strong , ensuring the advance
of the pawn whi le reta in ing the option, in
case of necessity, of defending the bishop
from a6. 34 g3 ! also came i nto considera
tion , after which Black does not have a
s ingle sensible move .
34 'it'b6 h4
35 a7?!
Was it worth a l lowing the open ing up of
White's own king position? 35 g3 ! would
have g iven an elementary win .
35 . . . g3 !
36 ii'a6?
Here it is - relaxation when only one step
away from victory! 36 fxg3 hxg3 37 h3 would
have won , for example , 37 .. .'i!Va2 38 'lt>h1
tt'lh4 39 ii'a6 'ii'f2 40 a8'ii' tt'lf3 41 'i!V8a7 c5
42 'i!Vxf?+ ! 'i£txf7 43 'ii'b?+ and 44 'i!Vxf3.
However, such a variation is too compl i
cated for a player in time-trouble . In any
case , with more accurate preceding play the
game could have been concluded far more
s imply.
36 . . . gxf2+
37 'i£txf2
I n the event of 37 Wh 1 'ii'c1 38 a8'it'? White,
paradoxical ly, even loses - after 38 . . . tt'lg3+!
39 hxg3 hxg3 there is no defence against
mate by the queen on h6 .
37 . . . i¥xc3
Aga in mate is threatened .
38 i.d3
Draw.
39 i.e2
40 We1
41 Wf2
'i!Vd2+
'iid4+
i¥c3+
How can the nervous system be tra ined to
endure prolonged tension? Here too it is
probably not possible to get by without
Converting an Advantage ttJ 1 1 5
physical preparation ( 'healthy in body, healthy
in mind ! ' ) ; self-tra in ing exercises or even
yoga are probably usefu l . Specific chess
training is a lso possib le . You can practice
playing specia l ly selected exercises, i n
wh ich you have to find a long series of the
only correct moves. Try conducting ind i
vidual games or even entire tournaments
with the aim of concentrating to the maxi
mum throughout the entire game.
Ill. Time-trouble
Nearly every player can remember depress
ing instances of t ime-trouble adventures, in
which the fru its of all the preced ing work
were ru ined. But even so I wi l l show you one
more example, together with an instructive
assessment of his own actions , which was
given in his notes by an ex-world champion .
Alekh ine - Tylor
Nottingham 1 936
Black's posit ion is, of course, completely
hopeless. If 46 . . . '>i'xd5 there fol lows 47
::l.d1 + '>i'e6 (47 . . . '.i?e5 48 i..c6) 48 '>i'c3 . Tylor
tries h is last time-trouble chance.46 . . . tbxa4
47 '>i'd3??
Regarding this Aiekhine writes in the tou rna
ment book:
An awful move, the fact that White was very
short of time is, to my mind, as little to be
considered as an excuse, as for instance the
statement of the law-breaker that he was
drunk at the moment that he committed the
crime. The inability of an experienced
master to deal with the clock should be
considered as grave a fault as a miscalcula
tion.
White would have won by 47 i..xa4 .U.Xb 1 48
'.i?xb 1 '>i'xd5 , and now, if there is noth ing
better, 49 �e8!? We5 50 h5 Wf4 (50 . . . gxh5
5 1 �xh5) 5 1 hxg6 hxg6 52 i..d7 (Aiekhine) .
47 . . . 'it>xd5
48 i..c4+ '>i'd6
49 l:!.xb7 tbc5+
50 We3 tbxb7
Soon the players agreed a d raw.
Aga in I wi l l not speak in deta i l about how to
combat t ime-trouble . I wi l l merely mention
two main methods: 1) 'anti-time-trouble'
games; 2) time-study of games, with a
subsequent analysis of the reasons for
getting into time-trouble.
Points are lost not on ly in your own time
trouble , but also in the opponent's. This
happens because a player often d isregards
a wel l-known principle of how to act in such
s ituations . If you have the better position,
never play on the opponent's time
trouble. Act exactly as you normally
would, not even remembering about
your opponent's lack of time. Why? By
playing qu ickly, so as not to a l low the
opponent to th ink over his moves, you
essential ly d rive yourself i nto the same time
trouble as h im . But in a d ifficult situation the
opponent is fu l ly composed and mobi l ised ,
whereas you , by contrast, l u l led by your
advantage i n t ime and position , are awa iting
the fa l l of his flag and are unable to play at
fu l l i ntensity.
116 � Converting an Advantage
I n the hope of exploiti ng this psycholog ical
effect, in d ifficult situations some players
del iberately get themselves into time-trou
ble, and there they often change the
unfavourable course of the play.
Mark Tseitlin - Makarychev
Krasnoyarsk 1 98 1
Only 1 4 moves have been made, but Black's
position is d ifficult , and in addit ion he had
already spent nearly a l l the t ime on his clock
- he had just 6 ( ! ) m inutes left for 26 moves.
Obviously it is important for Wh ite to
develop h is rook at c1 as soon as possib le ,
in order to create pressure on the c-fi le .
However, after the immediate 1 5 l:tac1 he
has to reckon with 15 . . . h6 .
15 h3 !? looks strong , and after the retreat of
the knight - 1 6 l:tac1 . Black would probably
have repl ied 1 5 . . . h6, i ntending to create
compl ications after 1 6 ..tf4 cxd4 ! . But Wh ite
can very wel l waste a tempo: 1 6 ..ltc1 ( 1 6
hxg4 hxg5 1 7 l:lac1 i s also not bad ) 1 6 . . . tt::lf6
1 7 ..lte3 cxd4 1 8 tt::lxd4 fol lowed by .l:!.ac1 . An
unhurried method of playing, when you
simply strengthen your position without
allowing any counter-chances, is the
most unpleasant for an opponent who is
in time-trouble.
1 5 dxc5 tt::lxc5
1 6 h3
Before placing his rook on c1 , Tseitl in wants
to drive away the knight . I n the event of the
immed iate 1 6 l:.ac1 he was concerned
about the reply 1 6 . . .'ilt'f5 , when Black ac
qu i res tactica l ideas associated with the
weakness of the f2-point. The fol lowing
combinative variation is i nteresting : 1 7 e4!
tt::lxe4 1 8 tt::lxe4 'ifxe4 1 9 tt::ld4 (the bishop at
c4 is u nder attack) 1 9 . . .'iVe5 20 ..tf4 'ifh5 21
h 3 tt::lf6 (21 . . . tt::lxf2 is worse: 22 �xf2 e5 23
g4! 'iih4+ 24 ..tg3 �f6+ 25 �g 1 ) 22 g4!
tt::lxg4 23 hxg4 'ifxg4 24 'it'g3 'ii'xg3 25 .bg3
..txa2 (25 . . . ..ta6 26 b4 .l:.ad8 27 tt::lb3) 26
..txb7 .l:1ad8 27 tt::lc6 . The concluding posi
t ion is undoubted ly in Wh ite's favour, but the
outcome is sti l l not clear - too many pawns
h ave been exchanged .
1 6 . . . tt::lxf2 !?
Th is p iece sacrifice is B lack's best practical
chance . If 1 6 . . . i.xc3 the s imple 1 7 hxg4
..tg7 1 8 l:tac1 would h ave fol lowed .
1 7 �xf2 i.xc3
1 8 bxc3 tt::le4+
1 9 �g1 ?
Th is is what Sergey Makarychev had to say:
Such a disdainful attitude to one's own
material can be explained only by the
opponent's time-trouble. In the event of 19
Converting an Advantage lZJ 1 1 7
d6 35 lixc8
ti:'Jxf6 36 Uxc5 �xc5 37 g4 with a serious
advantage. B lack's play can be improved on
the 28th move , by placing his rook not on c5,
but more actively on e3. And instead of
26 . . . g5 it probably makes sense for him to
choose 26 . . . L'De2!? .
24 . . .
25 Jl..e3
26 il..xd4
27 Ua1
White resigned.
Uad8!
Uxd4
'ifc2
l:l.xe5!
I t is curious that on a l l the moves that we
have seen , Makarychev spent just three
minutes - one half of h is reserve of t ime.
1 1 8 Converting an Advantage
IV. Inadequate knowledge of
endgame theory
In the majority of cases the conversion of an
advantage has to be carried out i n the
endgame. It is clear that, if you are not
fami l iar with theory, you are much more
l ikely to make a mistake.
Wolff - Browne
USA Championsh ip , Durango 1992
It is qu ite possible (although not i nevitable)
that Black wi l l lose h is e-pawn , and therefore
it is usefu l to have some i nformation about
endings with two pawns against two on the
same wing. The most important conclusion
is this: by placing h is pawns on h5 and g6,
Black sets up an impregnable fortress - the
opponent's k ing is unable to approach h is
pawns.
(see diagram)
It obviously makes sense for Wh ite to
prevent the erection of th is defensive sys
tem, by playing g3-g4! But ifit is B lack to
move, he should play . . . h7-h5 ! . But, a las ,
neither player knew this position .
50 'it>g2?! �d4?!
51 �f3?! g6?
A strange move, wh ich sign ificantly worsens
Black's position . H is king is now forced to
retreat to the edge of the board (and yet he
cou ld have kept it at f6) , and the h-pawn for
ever remains backward .
52 'it>e4 �f6
53 l:ta7+
54 g4!
At last!
54 . . .
'it>g8
�c3
Why g ive up the e6-pawn without a fight?
54 . . . h6 55 .l:tb7 �f8 suggests itself.
55 l:!.e7 i..f6
56 l:lxe6 'it>f7
57 Ua6
57 . . . �c3?
Black should have placed his bishop on h4
Converting an Advantage lZJ 1 1 9
and played . . . h7-h6 . If Wh ite's pawn were
at h3 , there would be a ltogether noth ing that
he could do ( I once found this fortress i n the
course of a joint analysis with Boris Gu lko of
one of h is adjourned positions) . With the
pawn on h2 it is possible to march the king to
h3 fol lowed by 'it>g3 and h2-h4 . However,
this plan is not easy to carry out, and a lso in
reply to h2-h4 White has to reckon with the
strong reply . . . h6-h5!
58 .l:i.a7+ 'it>g8
59 l:Id7?
If during the game Patrick Wolff had known
about the system of defence with the bishop
on h4 , he would u ndoubted ly have played
59 h4! fol lowed by 60 h5 .
59 . . . .if6
60 'it>f4 .ib2?
60 . . . h6! .
61 �c7?!
61 h4! .
6 1 . . .
62 g5
.if6
.id4
63 h4 .ib2
64 �g4 .ie5?!
64 . . . .i.a3 ! 65 h5 gxh5+ 66 �xh5 .ib4 was
more tenacious. For many years this posi
t ion was considered drawn , but in 1 993 the
chess composer Noam Elkies nevertheless
found a winn ing p lan .
65 l:tc6! .ii.b2
66 l:!.a6 .ii.c3
67 .t!.a4!
68 h5!
.ie5
.ic3
No better is 68 . . . gxh5+ 69 �xh5 i.d6 (the
threat was 70 .Ua8+ 'it>g7 71 l:!.a7+ 'it>g8 72
�h6) 70 .Ua8+ �g7 (70 . . . i.f8 71 g6) 7 1
.!:!.a?+ Wg8 7 2 g 6 hxg6+ 7 3 'it>xg6.
(see diagram)
Another important theoretica l position ! B lack
loses, if his king will be shut in the corner
(with a l ight-square bishop, by contrast, it
would be a draw). As is not d ifficult to see,
fleeing from the dangerous corner does not
work in view of the u nfortunate position of
the bishop: 73 . . . �f8 74 �f6 'it>g8 (74 . . . �e8
75 'it>e6) 75 l:!.g7+ �h8 (75 . . . �f8 76 l:!.d7) 76
Wg6 and wins .
69 h6 �
70 .Uc4 i.e5
71 �3 i.d6
72 .l::!.c8 'it>e6
73 .t!.h8! �5
74 l:!.xh7 �xg5
75 lid7
Black resigned .
I n the second issue of the American Chess
Journal grandmaster Wolff gave an excel
lent commentary on this endgame. By
studying his analyses, you , for example, wi l l
learn how White wins if he remains with a
pawn on h5 or g5 against a black pawn on
h7 . Al l th is is very i nterest ing and usefu l , but
nevertheless not essentia l . Whereas the
fortress with black pawns on g6 and h5
should defin itely enter you r store of end
game knowledge.
Why in particu lar th is position? F i rstly, here
it is sufficient to learn the assessment of the
1 20 � Converting an Advantage
position and the basic idea of the defence
(not to al low the king to approach the
pawns) - you don't have to remember any
compl icated variations. Second ly, th is as
sessment (draw! ) is automatica l ly trans
ferred to positions with a wh ite h- or g-pawn
against a pawn on g6 (after a l l , Wh ite can
play g3-g4 and capture on g4 with the pawn
or a piece) . And above a l l , th is position is the
most un iversal and informative . Very often
the pawns of both sides have not advanced
further than the 2nd or 3rd rank , and then it
is clear that Black should aim to play
. . . h7(h6)-h5 ! , and White - g2(g3)-g4 ! .
One of the methods of converting an
advantage is to transpose into an endgame
position that is known to be theoretica l ly
won .
Larsen - Torre
I nterzonal Tournament, Len ingrad 1 973
The simplest way to win is to sacrifice the
c5-pawn. After 78 'i.t>g5! .Uxc5 79 'it>g6
fol lowed by 80 l:!.h8+ we reach an elemen
tary theoretica l ending, which is completely
hopeless for Black, s ince h is king is on the
' long' side of the pawn .
78 .U.c7?!
Bent Larsen decided to keep both of h is
extra pawns, which , of course , is a lso good
enough to w in . Why then should h is decision
be criticised? Wel l , because after 78 'it>g5!
the game would have concluded - theory,
wel l known to Larsen , would have begun.
Here it would no longer be possib le to go
wrong. But after the move chosen by h im ,
p lay contin ues in an u nfami l iar position ,
which means that the probabi l ity of a
mistake remains .
78 . . .
79 .l:!.c6
80 .l:!.d6+
81 f6+?
�d8
c.t>d7
'it>e7
And here is the decisive mistake , which
leads to a draw. Wh ite should have played
either 81 lle6+ Wf7 82 c6 , or 81 .Ud5 .
81 . . . �f7
82 c6
83 'i.t>f3
'l.t>g6
l:.e1 !
This is the whole point - the king cannot
break through to either of its pawns.
84 �f4 .Ue2 85 .l:td5 .l:!.c2 86 .l:!.d6 .l:!.e2 87 f7
'i.t>xf7 88 c.t>f5 c.t>e7 89 .l::!.d7+ 'it>e8 90 'iiif6
.Ue1 91 .l:!.d5 .l:!.c1 92 .l:!.d6 .l:!.f1 + 93 �e6
l!e1 + 94 �d5 .l:!.d1 + 95 c.t>c5 .l:!.xd6 96 �xd6
'i.t>d8 Draw.
V. Poor technique in the conversion
of an advantage
We wi l l consider this problem in more detai l .
Grandmaster Igor Zaitsev once suggested a
deep and val id idea : Technique is the art of
the past' . If th is is so, then a rel iable way of
improving your technique is to study classic
models , and , above a l l , examples from the
games of p layers who were famed for their
mastery in this field . P layers , such as Akiba
Rubinste in , Jose Raul Capablanca , Alexan
der Alekh ine , Tig ran Petrosian , U lf Anders
son . . . When playing through their games,
th ink about why they were able so easi ly and
Converting an Advantage t2J 1 2 1
natural ly to exploit even a seemingly imper
ceptible advantage, and you wi l l g radual ly
beg in to notice their approach to such
situat ions, those principles of converting an
advantage which they consciously or sub
consciously fol lowed , and the techniques
wh ich they employed . We wi l l now examine
the most genera l of these principles and
methods .
Maximum restriction of the opponent's
counterchances
It is very important for every player to
possess 'prophylactic thinking' - the
ability to constantly ask yourself: 'What
does my opponent want, and what would
he do now if it were his move?' But the
role of prophylactic th inking increases par
ticularly when convert ing an advantage,
when the maximum restrict ion of the oppo
nent's possib i l it ies, the e l im ination of the
sl ightest counterplay or any usefu l opera
tions to improve h is own position , becomes
probably the main princip le .
I wil l show two examples from my own
games .
Dvoretsky - Butnoris
Kiev 1 976
nent has a bad dark-square bishop, and as a
consequence - weaknesses on the l ight
squares. How can I strengthen my position?
The p lan of playing the knight from d2 to d5
suggests itself: f2-f3 , i.e3-f2 and tt:'ld2-f1 -
e3. It would seem that it can be begun with
either 22 f3 , or 22 tt:'lf1 . I t a lso makes sense
to play 22 g3 , depriving the enemy knight of
the f4-square . Wh ich of these three continu-
ations is the most accurate?
White must carefu l ly look to see what active
resources the opponent has, and how he is
intending to play. The move 22 . . . tt:'lf4 should
not concern us too much - after 23 g3 tt:'le6
the knight does not create any threats from
e6 and does not control the weak d5-square,
for which Wh ite is a im ing .
The attempt to create cou nterplay on the
kingside with 22 . . . 'ife6 ! fol lowed by 23 . . .f5
looks more serious . For example, 22 g3
'ii'e6 23 'it'b3 f5 ! , and there is no time for 24
'ilt'xb7? in v iew of 24 . . .f4 . Or 22 tt:'lf1 'it'e6
(with ga in of tempo! ) and 23 . . . f5 .
I n the second variation White has an interes
t ing tactical resou rce : 23 'il:Vb3 f5 24 exf5
gxf5 25 i.xh6! jLxh6 26 .l:!.d6 �f7 27 .l:!.xh6,
although after 27 . . . tt:Jdf6 h is rook is stuck in
enemy territory, and 28 tt:'lfe3! f4 29 tt:'lf5
'ike6 30 tt:'lh4 e4 leads to a rather tense
situation . And in genera l , when you have
such a sol id positional advantage, why
calculate such compl icated variations? After
22 f3 'YWe6 23 �b3 the capture on b7 is now
seriously threatened , and Black is forced to
d ivert either his rook, or his queen, to the
defence of the pawn . I t is th is move order,
therefore , that enables White to be fu l ly
prepared for the opponent's counterplay.
22 f3 ! tt:'lf4
23 g3 tt:Je6
24 tt:'lf1 f6?!
25 'i!tg2!
Another accurate move . 25 i.f2? is prema-
Of course, Wh ite stands better. The oppo- ture in view of 25 . . . tt:Jg5 fol lowed by . . . 'iie6,
122 � Converting an Advantage
when 26 h4 does not work because of
26 . . . tt:Jxf3+ 27 'it>g2 tbd4 28 cxd4 exd4.
25 . . . 'it>h7
26 i.f2 tt:Jc7
27 tt:Jfe3
White has carried out h is p lan and increased
his advantage.
Usual ly, when we look at a game, we do not
pay any attention to modest moves such as
22 f3 ! and 25 'it>g2! . And yet it is thanks to
them that events have developed in the
desirable, qu iet way for Wh ite , and the
opponent has not gained the sl ightest
opportun ity to activate h is game or compl i
cate the play. But few are happy to defend
passively without any counterchances, and
in such cases new mistakes or inaccuracies
are l ikely, making it easier for the stronger
side to convert h is advantage.
27 . . . tDb8
28 tt:Jb6 'iie6
29 ii'a4! tt:Jb5
30 .l:!.d5 l::txd5
31 tt:Jexd5
Threatening 32 c4 tt:Jd4 33 'i*'xe8 ! .
31 . . . .l:!.d8
32 'ii'c4 �c6?
32 . . . 'ifd6 was more tenacious. Now White
lands a decisive blow.
33 tt:Je7 ! i.xe7
34 ii'f7+ 'it>h8
35 'ifxe7 'i!Vd6
36 'i!Vf7 g5
37 i.xc5! 'iid2+
38 'it>h3 tt:Jd7
39 il..e7 .l:!.g8
40 tt:Jd5! .l:!.g7
40 . . . g4+ 41 fxg4 'iig5 42 tt:Jxf6 .
41 'iie8+
Black resigned .
Zakharov - Dvoretsky
Ordzhon ikidze 1 978
Black clearly has a great positional advan
tage. When I ask, how he should continue in
th is position , usual ly either 29 . . . .l:!.c2 or
29 . . .f6 and 30 . . . .l:!.c4 is suggested . And
indeed , why not - seeing as Wh ite has no
cou nterplay?
But nevertheless, th ink how you would play
if it were White to move. And then you wi l l
fi nd an idea which promises qu ite good
chances of a successfu l defence - the
manoeuvre tt:Je5-g4-e3 with the a im of
exchanging the mighty kn ight on d5 . This
knight dominates the position , cementing
together Black's queenside and making the
rook at b5 a passive, inoperative piece . But
Converting an Advantage ttJ 1 23
in the event of the kn ight exchange, the rook
wi l l immediately be transformed - after a l l , i t
is attacking the b6- and g5-pawns.
It becomes clear what the best move is .
29 . . . h5 !
Black reta ins a l l the advantages of h is
posit ion and prevents the opponent's on ly
promising idea .
The game did not last long .
30 �d2 f6
31 tt:'lf3
32 b3
l::tc4
l:lc6
The wh ite rook has ended up in a trap .
33 h4 g4
34 tt:'le1 tt:'lc7
35 l:!.xh5 'it>g6
Wh ite resigned .
'Do not hurry! '
The inherent a im wh ich a player should
fol low when trying to convert an advantage
is not to win as qu ickly as possib le . As yet no
one has offered prizes for the smal lest
number of moves. You should endeavour to
play with the utmost safety, exploit ing a l l the
resources of your posit ion and completely
restrict ing the active possib i l it ies for your
opponent. I t is not a misfortune i f you h ave
to make an extra dozen moves on the board ,
if thanks to th is you make your task easier
and reach your goal more su rely. In a sharp
midd legame you may be tempted by the
image of t iger, swiftly h u nt ing down its
quarry and tearing it apart , but when
converting an advantage in the endgame
you should rather imitate a python , slowly
suffocating its victim .
The principle ' do not h u rry! ' was fi rst
formulated (but not expla ined to a sufficient
degree) i n instruct ional material on the
endgame prepared by the Soviet master
Sergey Belavenets . I n fact, beh ind th is brief
formula are concealed various aspects of
endgame techn ique, some of which we wi l l
encounter in the fol lowing examples.
'Do not hurry!' does not imply that you
can carelessly squander tempi. On the
contrary, every opportunity to gain a
tempo should definitely to taken into
account and exploited.
Leonhardt - Spielmann
San Sebastian 19 12
One must possess great presence of mind,
to not seize the booty immediately, but do
this after several strong preparatory moves.
The anticipation of victory often hinders the
objection evaluation of a position.
It is to this factor that I prescribe the loss of a
half point in what was a very important game
for me - the most annoying instance in my
chess career. This was at the finish of the
San Sebastian tournament of 1912, at a
moment when I had excellent chances of
winning first prize. For me it was sufficient to
win against Leonhardt . . . (Rudolf Spielmann) .
The game went:
46 . . . c;t>xd4?
47 g6
It transpires that Black is i n zugzwang . I
should mention that here the zugzwang is
1 24 � Converting an Advantage
mutual - any move by White worsens h is
position .
47 . . .
48 l:td7
49 g7
'it>d3
d4
After advancing h is passed pawn to the 7th
rank, Wh ite easily parries all his opponent's
attempts . For example: 49 . . . '1t>c3 50 .l:!.c7+
'1t>b3 51 ki.d7 , or 49 . . . .l:!.g 1 + 50 Wb2 'it>e3 5 1
'it>b3 ( 5 1 �e7+ 'it>d2 5 2 ki.d7 d3 5 3 �e7 'it'd 1
54 �d7 d2 55 :I.e? is also possible) 5 1 . . . d3
52 'it'xb4 d2 53 'iit>c3 .
Draw.
49 . . . �g6
50 'it>b2 Itg1
51 'it>b3
Of course, Black should have played
46 ... 'it>c3 ! . If 4 7 'it>d 1 , then 4 7 . . . l::.g 1 + 48
'it>e2 b3 is decisive . After 47 .l:!.c7+ 'it'xd4
Black captures the pawn with ga in of tempo,
thanks to the attack on g5. S ince 48 .l:!.b7?
.l::!.xg5 49 .l:txb4+ 'it'c3 is bad for Wh ite , he
has to reply 48 .l:i.g7 . Now another tempo
can be won by 48 . . .'it>c3 ! 49 ki.c7+ 'it>d3.
Look at the posit ion after 50 ki.g7 : as yet
Wh ite has not done anyth ing usefu l , whereas
Black has e l iminated the pawn and placed
his king on d3 .
And yet, contrary to Spielmann's op in ion ,
even here he apparently did not have a win .
For example: 50 . . . d4 5 1 g 6 .l:tc2+! ( 5 1 . . . b3
52 'it>b 1 ! , but not 52 �b7? b2+ 53 'it'b 1 .l:txg6
54 'it'xb2 'it'd2) 52 'it'b1 ! (52 'it'd 1 ? .l::!.c6 53
l:tg8 'it>c3 54 g7 Itc7) 52 . . . .Uc6 53 'it>b2! 'it>c4
54 l:!.g8 (54 'it>c2) 54 . . . d3 55 llg7 ! (but not
55 g7? �c7 with zugzwang) , and Black is
not able to strengthen h is position .
If the opponent is deprived of counterplay,
before changing the pattern of the play
and initiating decisive action you should
first strengthen your position to the
maximum, by making all moves that are
even slightly useful.
Reti - Romanovsky
Moscow 1 925
28 .l:!.c4!
White is p lann ing 'it'f3 and e2-e3. It is very
important that, after the black bishop moves,
the c5-pawn remains en prise. It wi l l have to
be defended by the rook, but then the white
bishop wi l l occupy an active position on d5 ,
the k ing wi l l obta in the excel lent square e4,
and the rook may be switched along the 4th
rank to the kingside.
28 . . . 'iit>f8
29 'it>f3
30 e3
lieS
.i.c3
It would have been better to play the bishop
to b2.
Converting an Advantage ctJ 1 25
31 a4!
Method ical ly p layed ! This move does not
come d i rectly into White'sp la n , but in itself it
is usefu l - it is clear that the pawn stands
better at a4 than at a2 . The opponent wi l l no
longer have counterplay involving the switch
ing of his rook to a6, and if Wh ite should
somehow be able to capture the a 7 -pawn he
wi l l obta in a passed a-pawn . I t is not clear
whether these considerations wi l l p lay any
role, but th is is not important . Any opportu
nity to strengthen the posit ion even s l ightly
should be exploited .
31 . . . �e7
32 �d5 l1c7
33 l:i.h4!
The black king is ready to go to d6,
defending the c5-pawn , and therefore there
is no longer any point in keeping the rook at
c4 . It is switched to the kingside to support
the pawn offensive on that part of the board .
33 . . . h6
34 'it>e4 �f6
35 l1h5
Now Black must do someth ing , s ince g3-g4,
h2-h4 and g4-g5+ is threatened . He should
probably have played 35 . . . g6! 36 fxg6 (36
l:lxh6? ;t>g5) 36 . . . 'it>xg6 37 l1f5 a5! (prevent
ing a4-a5-a6 fol lowed by l1f5-f2-a2-a5-
b5). The outcome would have remained
unclear, although Wh ite could have devel
oped his i n itiative by h2-h4-h5+ and .Uf5-
f1 -h 1 -h4-g4.
I checked a pretty attempt to prevent the
clos ing-up of the queen side by 37 a5? ! . The
capture of the rook leads to an immediate
mate: 37 . . . �xh5?? 38 Wf5 . Black has a
d ifficu lt position after 37 . . . .l:td7 38 g4 .Ud6 39
.tc4 . U nfortunately, there is a spectacu lar
refutation : 37 . . . c4 ! ! , and 38 dxc4? is not
possible in view of 38 . . . 'it>xh5 39 �f5 e4 40
.txe4 .Uc5+.
35 . . . l1d7?!
Pyotr Romanovsky tr ies to solve the prob
lem by tactical means, but the combination
he has prepared meets with a spectacu lar
refutation .
36 g4
Of course, not 36 h4? g6 and the wh ite rook
is trapped .
36 . . .
37 .Uxh6!
38 l1h7
g6
Wg5
�xg4
Now it is clear what Romanovsky had i n
m ind . I n t he event o f 39 fxg6?? f5 or 39 �e6
fxe6 40 .Uxd7?? gxf5 White is unexpectedly
mated . I f 39 f6 Black was intending 39 . . . �g5 .
However, after 40 �xf7! �xf6 4 1 �xg6
.l:lxh7 42 �xh7 he most probably is unable to
save the ending with opposite-colour bish
ops. Wh ite attacks and captu res the a7-
pawn (perhaps after fi rst playing h is bishop
to c4) , after which one of the two passed
pawns , a- or h-, is bound to decide the
outcome. But with the pawn on a2 such an
end ing would certain ly be drawn .
Richard Reti found a prett ier and more
convincing solution .
39 �e6! fxe6
39 . . . .l:te7 40 .l:lxf7 .l:lxf7 4 1 fxg6+ .
40 fxg6! l1d8
41 .l:lxa7 'itg5
1 26 Converting an Advantage
42 g7
43 a5
'it>h6
The passed a-pawn comes decisively into
play. Now we can assess the true worth of
the move 3 1 a4 ! . With h is pawn on a2 White
would have been unable to win .
43 . . . 'it>h7
44 a6 .l:td6
The threat was 45 llb7 and 46 a? .
45 h4
White's last reserve joins the batt le.
45 . . . ..te1
46 h5 ..th4
47 h6
Black resigned .
In cases when one side's advantage is
insufficient for a direct win, it makes
sense to manoeuvre, without changing
the pattern of the position, in order to set
the opponent various problems, even if they
are not too compl icated . And on ly when ,
after fai l ing to withstand the prolonged
pressure , he blunders or makes some error,
can you then turn to decisive action .
Such tactics of sett ing the opponent an
endurance test can sometimes a lso make
sense in positions with a big advantage. By
provoking h im into making a mistake , you
can make it sign ificantly easier to convert
your advantage.
Mikhai l Botvinn ik remembers :
In 1936 in Moscow during the 3rd Interna
tional Tournament I witnessed the resump
tion of the Capablanca-Ragozin game. The
ex-world champion had an extra pawn and
hence a won endgame. To my surprise,
however, Capablanca did not undertake any
positive action, but stuck to waiting tactics.
Finally his opponent made an inaccuracy,
and the Cuban won a second pawn and
soon the game.
'Why didn't you immediately try to convert
your material advantage?' I ventured to ask
the great chess virtuoso. My companion
condescendingly smiled: 'It was more practi
cal to wait. '
Dvoretsky - Cooper
Ph i ladelph ia 1 990
Wh ite u ndoubtedly has an appreciable
positional advantage. He finds a convincing
plan to exploit it , i nvolving the creation of
threats on the dark squares on the kingside.
1 8 ..txg7
1 8 ..tg5 was also not bad . However, 1 8
lt:ih2? would have been a serious inaccu
racy in view of the strong reply 1 8 . . . f5 ! .
1 8 . . . 'it>xg7
1 9 tt:ih2 h5
I a lso had to reckon with cou nterplay on the
c-fi le . If 1 9 . . . .Uc2 there was the strong reply
20 lt:ig4 �h4 21 .Uac1 ! .UacB (2 1 . . . .l:txb2 22
lie? ..te8 23 llec1 is equal ly cheerless) 22
.l:.xc2 .Uxc2 23 g3 ! (23 l:l.c1 .Uxc1 + 24 �xc1
is a lso not bad , when the wh ite queen
breaks into the opponent's position along
the c-fi le) 23 . . ."it'xh3 24 "ii'f6+ 'it>gB 25 "ii'd8+
'it>g7 26 tt:if6 .
20 .Uac1
21 .Uxc1
llxc1
llc8
Converting an Advantage ttJ 1 27
22 l::!.xc8
23 g4!
..txc8
This is the whole point! Now 23 . . . hxg4 24
tt:lxg4 , and if 24 .. .'it'h4 - 25 'ii'f6+ (of course ,
it is a lso possible to delay th is check)
25 . . . 'it'xf6 26 exf6+ is completely bad for
Black. He does not want to a l low the
exchange of pawns on h5, whi le if 23 . . . 'i!Vh4
there fol lows simply 24 '.t>g2 and 25 lLlf3 .
23 . . . g5
24 '*'e3 h4
After 25 f4 !? the h4-pawn is , of course,
doomed . But in th is case the wh ite k ing
becomes somewhat exposed , which may
g ive the opponent some counter-chances .
For the moment I preferred not to change
the pattern of the posit ion and I tried to
ach ieve success in positional manoeuvring ,
by tying the black pieces to the defence of
the weak g5-pawn . Especia l ly, s ince the
possibi l ity of f2-f4 wi l l never run away.
25 ltJf3 'it>g6
26 'ikd3+ 'it>h6
Here I noticed that I could win a pawn by 27
'id2 (with the threat of 28 ltJxh4) 27 . . . 'it>g6
28 'i'c2+ Wh6 29 'ir'c1 �g6 30 ltJxg5 'i!Vxg5
31 'ir'xc8 . The queen endgame is a lmost
certa in ly won , but again I did not want to
force matters and I tried to obta in benefit
from the fact that my opponent's p ieces
were tied down .
27 'ii'a3
28 1i'c5
29 '.t>g2
aS
'it>g6
A usefu l prophylactic move, which in some
variations prevents the black queen from
giv ing a check on c1 .
29 . . . b6
30 �c2+
30 'it'd6? does not work in view of 30 . . . ifxd6
31 exd6 f6 ! . But if Black should play h is
b ishop to a6 , then by p lac ing h is queen on
d6 White wi l l immed iately decide the out
come. It makes sense to check whether the
opponent wil l go wrong.
30 . . . '.t>h6
31 �c6 ..ta6?
He does! Of course , 31 . . . 'it>g6 was correct.
Then I would probably have nevertheless
agreed to win a pawn by 32 l\Vc2+ 'it>h6 33
'i'c1 �g6 34 tt:Jxg5 "ikxg5 35 'i!kxc8 , although
fi rst I would certa in ly have pondered over
whether I had extracted everyth ing possible
from the type of position now on the board .
3 2 'it'd6! 'ifcB
33 "ilie7 'ii'g8
34 'ilff6+
Black resigned .
See how more easi ly (not more qu ickly, but
more easily) I was able to win , thanks to the
fact that I did not hurry to force matters .
(see diagram)
White has the advantage. But what is it
better to play: 38 ..lkxd4 or 38 f6 ?
When you have a choice between advan
tageous positions with different material
balances, all other things being equal
you should choose the one in which the
material balance is the most usual, the
most standard. Here you will have more
1 28 � Converting an Advantage
Dvoretsky - Baikov
Moscow Championsh ip 1 972
experience and hence there is less
chance of a mistake in the evaluation of
the position or the subsequent play.
I n the event of 38 f6?�e3 39 f7 'iVxe6 40
fB'iV l::txf8 41 'il*'xf8+ tLleB or 41 �xf8 ttle4
Black, with a pawn for the exchange,
compl icates the play. Moreover, on a more
carefu l examination of the result ing position
it becomes clear that the advantage has
now passed to the opponent.
The simple captu re on d4 is much safer.
38 �xd4 iVa4?
38 . . . ttlb5! was much stronger. In reply 39
�c8? does not work in view of 39 . . . 'iVc6 ,
whi le after 39 �e5 ttld4 40 �xd4 .l:Ixd4
Black has sufficient compensation for the
pawn deficit. Possibly Wh ite should reply 39
�d5!? , when in the event of 39 . . . �d7 40
�e3! �xe3 41 l::txe3 tt:ld4 42 iVe4 ttlxc2 43
l::te2 or 43 l::!.d3 he reta ins the advantage.
However, after 39 . . . tt:lxd4 40 l::txe8 ttlxf3 41
l::txd8+ �xd8 the ending with opposite
colour bishops is probably drawn .
39 �e5 ttlc4
Now the simple move 40 'iVe4! would have
forced the transition into an absolutely won
endgame (40 . . . ttlb6 41 'ii'xa4) - Black's
position would have been resignable. I saw
it, of course, but I wondered whether it
wouldn ' t be possib le to ach ieve even more .
After noticing that 40 b3? is refuted by
40 . . . ttlxe5 , I for some reason completely
forgot about the same possib i l ity in reply to
40 .l:Ie4 and I considered only 40 . . .'�xc2.
After d iscovering 41 �xc7+! and ca lculating
its consequences, this is what I p layed .
40 �e4?? ttlxe5
White's incorrect move could a lso have
been refuted i n a d ifferent way: 40 . . . ttld2 !?
41 .l:Ixa4 ttlxf3 , and after the bishop moves
B lack has the decisive 42 . . . h3 .
41 l::txe5 �xc2
42 nds �c1 +
43 Wh2
44 Wh3
45 .ltxd5
�xb2+
l::txd5
c6
Here the game was adjourned . The situation
has completely changed - Wh ite's position
is absolutely hopeless, not on ly because of
the opponent's extra pawn , but also in view
of the dangerous position of the wh ite king.
However, on the resumption I managed to
confuse matters and save the d raw.
I regarded my b lunder on move 40 as
merely an inexpl icable 'eccentricity' . But
when I showed the game to ex-world
champion Tig ran Petrosian , he took a qu ite
d ifferent view of th ings.
'How do you expla in why you avoided
transposing into the endgame? You were in
no doubt that it was won . But if you see an
elementary solution , why then calculate
variations and get involved in an exchange
of blows?'
The moral of this sad episode is obvious .
Always give preference to the simplest
way of converting your advantage, in
which case the probability of a making a
mistake will be minimal. Avoid unneces
sary complications, and never play 'for
brilliancy'.
Converting an Advantage '2J 1 29
Any 'trifles', capable of facilitating the
conversion of an advantage, should
without fail be taken into consideration.
If, for example , you do not have much t ime
left to the t ime control , make use of every
opportun ity to repeat moves. And after the
time control has been reached , defin itely
adjourn the game, if you have a winn ing
position . I f you don't do th is , because of
ti redness you may make a m istake and spoi l
your position .
This last p iece of advice has ceased to be
topical i n view of the change in the rules of
chess competitions - games are no longer
adjourned . The fol lowing example is never
theless sti l l instructive : it shows that over
confidence in success, combined with a
certain haste and inaccuracy, is capable of
having a negative effect on the play of even
such a player, deserved ly famed for his fine
techn ique, as Anatoly Karpov.
Karpov - Korchnoi
World Championship Match ,
22nd Game, Baguio 1 978
White's position is absolutely won . Karpov
should have sealed h is next move , after
which his opponent would probably not have
bothered to resume the game. But for some
reason the world champion made a few
more moves at the board , and i n the end he
squandered his entire advantage.
41 .l:l.xd6 lt:lxd6
42 .Jtc7?!
Wh ite avoids the obvious 42 .l:txa4 because
of the reply 42 . . . h5 , which of course,
however, does not change the evaluation of
the position . In pri nciple, the desire to find
the most accurate way of exploiti ng your
advantage is commendable, but in so doing
you must accurately check the variations,
which Karpov d id not do. I ncidenta l ly, a
s imi lar mistake was made in h is commen
tary by M ikha i l Tal , who recommended 42
.l:ld4 lt:lc8 43 .Jtc5 . I nstead of 42 . . . lt:lc8?
Black plays 42 . . . .l:l.e 1 + 43 �c2 .U.e2+ 44 �c1
(44 �d3 llxb2 45 l:!.xd6 .l:lxa2) 44 . . . a3! 45
l:txd6 .l:lxb2 , cast ing doubts on whether
White can win . So that the s imple capture of
the a4-pawn is the most rel iable way to win .
42 . . . lle1 +
43 �c2 lt:le8
Karpov simply missed this straightforward
reply. Now, to avoid further mistakes , it was
essential for h im to ask the arbiter for an
envelope and to seal his move. But the
world champion contin ued in the same vein .
4 4 .ta5 a3
45 llb8 Ite7
45 . . . lle2+ 46 'it>d3 l:!.xb2 d id not work in v iew
of 47 llxe8+ 'it>h7 48 lle2 .
1 30 � Converting an Advantage
46 i.b4??
Through inertia Karpov decided that now too
the check on e2 was not dangerous for h im .
Of course, 46 bxa3 (or 46 b4) would have
given an elementary win .
46 . . . .l:i.e2+
47 'it>d3?
Showing the same inert ia . 4 7 i.d2! axb2 48
a4 was essentia l , sti l l reta in ing excel lent
chances of success. It is hard even to
understand what exactly Karpov overlooked ,
since now both captures on b2 enable Black
to save the game. Sensing th is , Victor
Korchnoi decided to adjourn the game at
this precise moment, so that the opponent
would not know which choice he had made.
47 . . . axb2
I n the variation 47 . . . .l::!.xb2 48 .l:i.xe8+ 'it>h7 49
.ltxa3 (49 �c3 .l:i.xa2 50 l:!.f8 f6 or 50 'JJ.e7
'it>g8) 49 . . . .l:i.xa2 Black then plays . . . f7-f6
and . . . h6-h5 , obta in ing a drawn position
there are too few pawns left on the board .
48 i.d2
If 48 .ltc3 , then 48 . . . b 1 'ii'+ ! 49 l:I.xb1 l:I.xa2 50
l:I.b8 l:I.g2 51 l:I.xe8+ 'it>h7, and roughly the
same drawn situation arises as in the
47 . . . l:I.xb2 variation . There can fol low 48
l:I.e4 h5 49 gxh5 l:I.h2 50 l:I.g4 f6 5 1 'it>e4
l:I.xh5, and to avoid . . . g7-g6 the wh ite rook
has to stay on the g-fi le .
48 . . . l:I.e7
49 a4 l:I.d7+ 50 'it>c2 'it>h7 51 l:I.xb2 h5! 52
gxh5 tt:ld6 53 l:I.a2 tt:lxf5 54 a5 tt:ld4+ 55
�c3 (55 'it>b1 tt:lb3, then giving up the knight
for the a-pawn ) 55 . . . tt:lc6 56 a6 l:!.d5 57 ii.f4
l:I.f5 (57 . . . .l:i.xh5? 58 l:I.h2 ! ) 58 i.d6 l:I.d5 59
ii..g3 l:I.g5 60 i.f2 l:I.xh5 61 'it>c4 tt:la5+ 62
�c3 tt:lc6 63 l:I.a4 'it>g8 64 'it>c4 tt:la5+ Draw.
The princip le of two weaknesses
This princip le is essentia l ly one of the
consequences of the more genera l ru le of
converting an advantage, which we have
just been d iscussing - 'do not hu rry! ' . If the
opponent is condemned to passivity,
don't try to achieve success at one point
alone - to hold it the defensive resources
may prove quite adequate. Play more
widely, and try to exploit weaknesses
(and if possible - create new ones) on
different parts of the board - then it will
be much more difficult to defend.
Alekhine - Samisch
Baden-Baden 1 925
How to convert the extra pawn? Advance it
to the queening square? But Black wi l l set
up a blockade on the b6-square , the wh ite
k ing wi l l be exposed and there wi l l be a
danger of perpetual check. I should remind
you that queen and knight form a rather
dangerous duo, if they are in the vicin ity of
the enemy king . Only in the event of the
queens being exchanged wi l l the wh ite king
be able to advance fearlessly to the help of
its passed pawn .
34 l\Yd4!
With this move and the next one White finds
the correct winning plan, which is toad
vance his kingside pawns. The passed b
pawn must not advance until later, once the
danger of perpetual check has been re
moved by the exchange of queens. A
Converting an Advantage t2J 1 31
concrete and clear evaluation of the posi
tion , typical of Alekh ine - in his commentar
ies one can find n umerous instructive
features such as th is .
34 . . .
35 il.d31
"ike7
Perhaps the most difficult move in the game.
Its purpose is to prepare an attack on the
point h 7. The winning method which follows
leaves Black powerless to resist. (Aiekhine)
35 . . . 'ii'c7
36 g4! �f7
37 h4 tt:'!b6
38 h5
39 gxh5
gxh5
On the kingside Black has been saddled
with a second weakness, and a very serious
one ( I should remind you : the fi rst 'weak
ness' is the opponent's passed pawn , and
Black constantly has to reckon with the
threat of its advance) . I f now 39 . . . �g7 , then
40 h6+ ! .
39 . . .
39 . . . h6 was more tenacious.
40 Jl.e4!
Of course, not 40 ..txh7? "ikxf3 4 1 'ilt'xb6?
'ld 1 + with perpetual check. I f Wh ite de
s i red , he could now have exchanged the
queens and after 40 'ii'e4 'i!i'xe4 41 ii.xe4 h6
42 'it>c2 gradual ly won the minor piece
ending. However, the move in the game is
far stronger, s ince it enables h im to fix the
weakness on h7 .
40 . . .
41 h6
42 ..tc2 !
"ikb5
'it'b3
Now that the pawn on h 7 has been
blockaded, the next step is to force the
exchange of queens. (Aiekh ine)
42 . . . 'itb5
42 . . . 'ii'e6 43 'i!Ve4 .
43 'iVd3
44 ..txd3
45 il.xh7
Black resigned .
'ii'xd3
tt:'!c8
Kotov - Pachman
Ven ice 1 950
White's positional advantage is determined
by his better pawn structu re , by the weak
ness of the c6-pawn . But th is factor alone
would have been insufficient for a win , if
B lack had now played 42 . . . h5 ! .
42 . . . �f6?
43 g4!
A typical move . Wh ite fixes a second
weakness in the opponent's position - the
h7-pawn. This was why it should have been
advanced to h5. I should mention that
42 . . .f5?! was much weaker in view of 43 h3
fol lowed by g3-g4 , and if Black repl ies
43 . . . h5 , then he again acqu i res a second
weakness - this time on g6.
43 . . . �e6
43 . . . �g5 44 h3 h5 45 f4+ �h4 46 'it>g2 .
44 'it>g2
In the endgame you should never forget
about improving the position of your
king.
1 32 Converting an Advantage
44 . . .
45 .l:!.e8+
l:i.b7
Before attacking the h-pawn it is useful to
lure the black rook to a more passive
position .
45 . . .
46 .Uh8
47 h4
48 'it>f3
49 .Ue8+
50 l:!.d8!
.Ue7
f6
l:tb7
llf7
.l:i.e7
White wants to place h is knight on c5. It is
important that after the exchange of minor
pieces the black rook should be t ied to the
defence of the c6-pawn . Passivity of the
rook is a very serious drawback in rook
endings.
50 . . .
51 lt:Jc5+
52 .Uc8 !
.Ua7
rJiie7
A method ical move . 52 .Uh8 was incorrect i n
view of 52 . . . �xc5 53 dxc5 .Ua5! 54 .Uxh7+
Wf8 . Now the c6-pawn is under attack and
Black does not manage to activate his rook.
Thus if 52 . . . .Uc7 there now fol lows 53 .Uh8.
Do you sense how uncomfortable it is to
simu ltaneously defend two weaknesses -
c6 and h7 , and how much easier it would be
to defend the weak c6-pawn alone?
52 . . .
53 dxc5
54 .Uh8
�xc5
rJiid7
'it>e6
Now 54 . . . l:i.a5 55 l:txh7+ �e6 (the king
can not go to f8) 56 l:lg7 is bad for Black.
Such 'trifles' play a very important role in the
conversion of an advantage.
55 .:td8
Alexander Kotov has successfu l ly carried
out h is p lan of transposing into a favourable
rook end ing . In h is commentary he judged
the result ing endgame to be won for White.
However, in Jonathan Speelman's book
Endgame Preparation· th is evaluation was
cal led into question .
The Engl ish grandmaster's conclusion ap
pears not be to compatible with the log ic of
the preced ing play - after a l l , White has
consistently outplayed h is opponent and, it
would appear, has the right to count on
success. But what can be done - the
defensive resources in chess are g reat,
especial ly in rook end ings, wh ich according
to Tarrasch are not usual ly won . The work
done by Kotov was not in va in - almost out
of noth ing he has developed a dangerous
i n itiative and posed serious problems for the
opponent, which at the board the latter was
unable to solve.
55 . . . We7?!
Converting an Advantage lZJ 1 33
Consideration should have been g iven to
55 . . J:tc7! 56 .Ud6+ 'it>e5 57 'it>e2 g5 58 hxg5
fxg5 , and if 59 'it>d3 , then not 59 . . . .Uc8? 60
l:td7 h6 61 l:td6 (Kotov) , but 59 . . . .U.f7! 60
l:txc6 .Uxf2 61 .l:tc8 'it>e6! (Speelman) . I n
stead of 56 .Ud6+ White can try 56 'it>f4 !? , but
after 56 . . . .l:!.d7 ! 57 .Uc8 d4 58 .l:i.xc6+ 'it>e7
(58 . . . 'it>d5? 59 .l:!.d6+) 59 exd4 (59 .Ud6? d3)
59 . . . .l:!.xd4+ Black again reta ins real d rawing
chances.
56 .l:td6
57 g5!
.l::i.a6
Wh ite clears a way into the enemy posit ion
for his k ing .
57 . . .
58 hxg5
59 'it>g3
fxg5
Wf7
Not immediately 59 Wf4 .Ua4+ 60 'it>e5??
l:te4 mate.
59 . . . 'it>e7
60 f3
61 'it>f4
62 'it>e5
.l:!.a3
.Ua4+
It would have been a mistake to play 62 e4?
dxe4 63 fxe4 .l:!.c4 64 .l:txc6 Wd7 65 .Ud6+
We? 66 .l::i.d5 .Uc3 ! 67 'it>e5 .l:tc4 . The rook on
d5 is too passive and therefore it is not
possible to convert the pawn advantage.
62 . . . .Ua3!
63 J::!.xc6 !?
Serious consideration should a lso have
been g iven to 63 .Ue6+ !? 'it>d7 (63 . . . 'it>f7 64
Wd6) 64 'it>f6 , for example, 64 . . . d4 65 .Ud6+
We? 66 .Uxd4 .Uxe3 67 f4 , and B lack's
position is very dangerous.
63 . . .
64 �xd5
.Uxe3+
.Ud3+
64 . . . .Uxf3 65 .l:!.c7+ and 66 .l::i.xh7 is hopeless
for Black.
65 'it>e4 .l::i.c3
66 f4 .l:!.c1
67 .l::i.c7+ �dB?
In Speelman's opm1on , even now, two
moves before resignation , it was sti l l possi
ble for Black to save the game, and ,
moreover, very prett i ly : 67 . . . 'it>e6! 68 Uxh7
iic4+ 69 Wf3 l:txc5 70 l:tg7 .Uc6 ! ! . Now 70
.UXg6+ 'it>f5 71 .Uxc6 leads to stalemate , and
70 'it>g4 'it>d5 to a stra ightforward d raw (71
.Uf7 .Ua6 72 .l::i.f6 llxf6 73 gxf6 'it>e6 7 4 Wg5
'it>f7) .
And yet Wh ite's position would appear to be
won . Having seen through the opponent's
stalemate trap , he should 'take a move back'
- 68 .l::i.c6+ ! 'it>e7 , then play his k ing to the
queenside: 69 'it>d5 Ud 1 + 70 'it>c4 .Uc1 +
(70 . . Jif1 7 1 .Uf6) 7 1 'it>b5 lib1 + 72 'it>a6 (with
the th reats of 73 .Uc7+ or 73 .Ub6) , and after
72 . . . Wd7 bring it back, exploit ing the fact
that the important f6-square is now accessi
ble for i nvasion : 73 .Ub6 .Uc1 74 'it>b5 .Ub 1 +
75 �c4 .Uf1 (75 . . . .Uxb6 76 cxb6+ �c6 77
�d4) 76 Wd5 etc. (suggested by Sergey
Dolmatov) .
68 .Uxh7 .Uxc5
69 .Uf7
Black resigned .
134 w Converting an Advantage
Spassky - Korchnoi
Candidates Match , 5th Game, Kiev 1 968
This is what grandmaster Korchnoi had to
say:
Despite the occasional inaccuracies com
mitted, I consider my play in the middle
stage of this game to be my best achieve
ment in the match. But I wasn't quite able to
complete the strategic picture - at the
decisive moment I failed to display the
necessary know-how. What was the prob
lem facing Black? I will allow myself to quote
Bondarevsky: 'White's pieces are tied to the
weakness at c2, but a single weakness he is
able to defend. Korchnoi was faced with the
problem of starting play on the kingside, so
as to create a new weakness in the enemy
position. '
I realised that the move of the h-pawn
appeared too routine to be the best. And 1
rejected 29 . . . g5 on account of the concrete
variation 30 "illd2 f6 31 "ille1!, when White
neutralises his opponent's advantage. But
the best move- 29 . . . f5 (suggested by Flohr)
completely escapedmy attention! The point
of the move is not only that after the
exchange on g4 White's f- and g-pawns will
be further weakened; a/so of considerable
importance is the fact that, after the ex
change of queens, Black can create an
outside passed pawn by . . . g7-g6 and . . . h6-
h5.
29 . . .
30 'it>h2
31 hxg4
32 g5!
h5?
hxg4
g6?!
Now a draw becomes the most probable
result: the pawn position is fixed on both the
queenside, and the kingside. (Korchnoi) .
The game ended in a draw on the 51 st
move.
Exchanging
Grandmaster Kotov remembered for a long
t ime the advice g iven to him by the
experienced master Vlad im i r Makogonov at
the i nternational tournament in Venice in
1 950.
Don't sharpen the play - what for? Ex
change the queens, and arrive at a position
where each side has a rook and two or three
minor pieces left. Which piece should you
exchange, and which should you keep?
There are few modern players who can
solve this question correctly. They under
stand tactics, but in this you are superior to
them.
When try ing to convert an advantage you
constantly have to th ink about the advisabil
ity of this or that exchange. One of the most
general gu ides is g iven by the fol lowing rule:
Having a material advantage, the stronger
side should aim to exchange pieces,
whereas the weaker side should aim to
exchange pawns.
(see diagram)
Converting an Advantage ltJ 1 35
Vidmar - Thomas
Nottingham 1 936
Wh ite has a decisive advantage. He should
now move h is kn ight from c5 and then play
J:l.c5 , aiming to exchange the active black
rook . 32 tbe4 �ad8 33 .l:!.c5 (the prophylactic
move 33 g3!? is a lso strong) 33 . . . �d3+ 34
'iite2 �3d4 35 l::t 1 c4 is possib le . 32 tbd7
(with the threat of 33 lbb6) and 33 .l:!c5 is
even simpler.
Mi lan Vidmar tried to carry out the same
idea , but he d id it i n a very inaccurate way,
overlooking the opponent's counterplay in
volving an exchange of pawns.
32 lbb7? g5!
33 g3 gxf4+
34 gxf4 l::tg8
The fi rst unpleasant consequence of White's
mistake - the h itherto passive rook at a8 has
come into play.
35 .l:!4c2 f6!
Another pawn exchange, and moreoever
the strong wh ite e-pawn is forced off the
board .
36 exf6+ �xf6
It is obvious that the last exchanges have
considerably increased Black's drawing
chances. (Aiekh ine)
37 tbc5 l:::.g4
Now it made sense to switch the knight to
e5: 38 tbd7+! �e7 (38 . . . �f5 39 lbb6! is bad
for Black) 39 tbe5 . I nstead , Wh ite for some
reason returns his knight to the rear.
38 tbe4+ We 7
39 lDf2 llg8!
40 �f3
If 40 tbd3 , then 40 . . . l:i.f5 ! , preventing 4 1
tbe5.
40 . . .
41 �c5
lbb5
.l::!.c8
After the exchange of the e5-pawn , the
passed c6-pawn has been sign ificantly
weakened , s ince B lack has acqu i red the
opportun ity to attack it with his king from d6.
42 .l:txd5?! exd5
43 .l:!.c5 tbd4+
44 'it>e3 tbf5+!
Much worse was 44 . . . t2Jxc6? 45 J:.xd5 with a
s ign ificant advantage for Wh ite .
45 'it>d3 �d6
46 ti.xa5
47 l:i.a7
48 �xh7
.l:.xc6
.l::!.c4
.l:!xf4
Black has managed to exchange a further
two pai rs of pawns, and al l h is remain ing
p ieces and pawns are excel lently placed . A
draw is now the most probable outcome.
49 'it>e2 .l:!.c4
50 'it>d2 l:i.d4+
51 'it>e2
52 �d1
.l:!.c4
d4
52 . . . l:td4+ 53 Wc2 .i:lc4+ was simpler, seeing
as 54 �b3? wi l l not do in view of 54 . . . tbd4+
55 �a4 b3+ 56 'it>a3 l::ta4+! 57 �xa4 bxa2.
53 �d2 b3!
George Thomas forces the exchange of
another pa i r of pawns .
54 axb3 .l:!.b4
55 tbd3 .l:!.xb3
56 l:td7+?!
1 36 � Converting an Advantage
56 h4 was stronger, but even then Black
would have successfu l ly defended by acti
vating his rook: 56 . . . .l:!.b8 fol lowed by . . . .l::i.g8 .
56 . . . �xd7
57 lLlc5+ �d6
58 lLlxb3 lL:le3!
There are too few pawns left on the board for
Wh ite to hope for success in the knight
ending.
59 h4 lL:lc4+ 60 'it>c2 'it>e5 61 lL:lxd4 'it>xd4 62
b4 'it>e4 63 'it>c3 lL:lb6 64 b5 'it>f5 65 Wd4
'it>g4 66 'it>c5 lL:la4+ Draw.
It should be remembered , however, that the
ru le we have just formulated is too genera l
to be trusted uncond itional ly - in chess such
universal laws do not exist. This is merely
one of the gu ides; the concrete features of
the position often d ictate a completely
d ifferent course of action .
Eh lvest - Andria nov
Tal l i nn 1 98 1
The passed a-pawn promises Black defin ite
counter-chances, but even so Wh ite's mate
rial advantage should be sufficient for a win .
However, not with the plan chosen by Jan
Ehlvest.
36 �g6? 'it>c7!
37 �e4?
38 �xc6
�c6
'it>xc6
The exchange of bishops was bad , since
now the wh ite rook is forced to take up a
passive position i n front of the enemy pawn.
39 'it>f1 a4
40 'it>e2
41 'it>d3
42 .l:ta1
a3
a2
'it>xc5
The d raw has become obvious ( if 43 'it>c3,
then 43 . . . l:tg8 44 g3 .l::!.f8 ) .
I nstead of the incorrect exchange, White
could have activated h is k ingside pawns: 36
g4!? (threaten ing g5-g6-g7) . But it was
safer fi rst to centra l ise the k ing : 36 f3 ! a4 37
'it>f2, and only then play g2-g4 . Such
strategy would have been in accordance
with a principle of endgame play, formulated
by Aaron N imzowitsch : 'The advance must
be a collective one/ '.
V I . Lack of concrete action at
the decisive moment
Let us suppose that your opponent has no
real cou nterplay and that you , i n accordance
with the principle 'do not hurry! ' , are accumu
lating advantages l ittle-by-l ittle . But against
tenacious resistance by the opponent you
wil l probably be unable to win the game by
technique alone - at some point you wil l
certa in ly have to switch from positional
manoeuvring to the precise calculation of
variations, and seek a concrete way to the
goa l . Many players stumble at this point, with
various factors playing their part. There is
carelessness, ar is ing in anticipation of a
qu ick win , about which we have a l ready
spoken . There is the fu l ly understandable
aim to act 'with every comfort' , not exces
sively exerting yourself, and not subjecting
yourself to the r isk of making a mistake in
forcing play. There is the d ifficulty of deter
m in ing that turn ing point , when you have
Converting an Advantage ctJ 1 37
already extracted the maximum from play
ing accord ing to the pr inciple 'do not h urry! ' ,
which means that i t i s t ime to find a concrete
variation , one which exploits the advantage
gained and advantageously changes the
character of the play.
I have noticed that bri l l iant positional players
such as, for example, Salo F lohr or Anatoly
Karpov, would successfu l ly convert an ad
vantage against opponents i nferior to them
in class. They manoeuvred , suppressed al l
active possib i l it ies by their opponents , and
when the latter fa i led to withstand the
pressure , they made mistakes and them
selves broke up their positions . But against
opponents of equal class they often d id not
manage to convert even a big advantage .
For the reason that, when faced with
tenacious resistance, you cannot afford to
miss an appropriate moment for concrete
and precise action , and this is by no means
the strongest aspect of such positional
p layers .
F lohr - Keres
1 8th USSR Championsh ip , Moscow 1 950
Queen's Indian Defence
1 lt'lf3 c5
2 c4 lt'lf6
3 g3 b6
4 i.g2 i.b7
5 0-0
6 lt'lc3
7 d4
e6
i.e7
lt'Je4?!
A dubious move , which could have been
ca l led into question by the energetic 8 d5 !
tt:lxc3 9 bxc3 , and if 9 . . . i.f6 , then 1 0 e4!
.bc3 11 i.g5 (Udovcic-Kovacevic, Zagreb
1 969). The usual continuation is 7 . . . cxd4.
8 'i!Vc2 lt'Jxc3
9 'iWxc3 i.f6
1 0 i.e3 tt:Jc6
1 0 . . . i.xf3 !? 1 1 i.xf3 lt'lc6 came i nto consid
eration .
1 1 .l:tad1 .l:i.c8?analogous
variation no longer works: 1 l:.g6 ! Wd7 2
�b 7? rtie 7 3 rtic6 'it>f7 4 l:.g4 'it>f6 5 Wd5 'it>f5
6 .l:l.g8 f3 ! 7 Wd4 (7 l:.xg3 'it>f4 8 .l:tg8 f2 ; 7
How to Study the Endgame ctJ 15
.l:!.f8+ Wg4 8 'it>e4 f2 9 'it>e3 �h3 with a draw)
7 . . .f2 8 'it>e3 f1lt:l+ ! .
Which pawn to advance?
Mar6czy - Tarrasch
San Sebastian 1 9 1 1
There was a n easy win by 1 l:txh2 'it;xh2 2
'it>a6! (the immed iate 1 'it;a6! is also possi
ble) 2 ... �g3 3 b5 �f4 4 b6 'it>e5 5 b7 .l:tb1 6
Wa7 �d6 7 b8�+. Note the move 2 'it>a6 ! .
Fi rstly, Wh ite advances the pawn beh ind
wh ich the rook is not stand ing . Secondly, h is
remain ing pawn is further away from the
enemy k ing, which does not manage to
attack it.
2 a6? is a mistake in view of 2 . . . 'it;g3 3 'it;b6
Wf4 4 a7 'it;e5 5 �b7 'it>d5 6 b5 'it>c5 , when
the black king succeeds in ' locking on' to the
b-pawn . Or 4 b5 'it>e5 5 'it>a7 'it>d6 6 b6 .l:!.b1 !
7 'itb7 (7 b7 'i;;c?) 7 . . . 'i;;c5.
The game went 1 'i;;c6? �c1 + 2 'it>b6 �c4!
(threatening the interference 3 . . . .l:!.h4) 3
.l:!.xh2 l:txb4+ 4 'i;;c5 .l:!.a4 5 �b5 �xa5+ with a
draw.
It wou ld be possible to expand further the
store of typical ideas, but for a start it is
sufficient to l im it ourselves to these, the
ones most used .
Some of the ideas mentioned operate not
only in endings of the g iven type. Thus, for
example, the rook should be placed to the
rear of the more advanced pawn in nearly
every case, when it is fight ing against two
connected passed pawns.
Alekhine - Tartakower
Vienna 1 922
Alexander Alekh ine analyses the natural
continuations 36 'it>c2 , 36 �c4. 36 g5 and 36
l:th2, and shows that they are sufficient for a
d raw at best. There is only one way to win .
36 .ti.d5! !
White's fantastic move finds a precise
explanation , from the standpoint of typical
ideas for such endings.
The variations springing from this rather
unlikely move (it attacks one solidly de
fended pawn and allows the immediate
advance of the other) are quite simple when
we have descried the basic idea: The black
pawns are inoffensive:
1) When they occupy squares of the same
colour as their bishop, for in that case
White 's king can hold them back without
difficulty, by occupying the appropriate white
square, for example 36 . . .f2 37 l:!.d1 e4 38
'it;c2 j)_f4 39 �f 1 and 40 'it>d1.
16 w How to Study the Endgame
2) When the rook can be posted behind
them, but without loss of time, for example
36 ... e4 37 'af5 Ji..g3 38 g5 e3 39 'ii.xf3 e2 40
.U.e3 (Aiekh ine) .
I t makes sense to also examine endings
which are closely l inked to those being
stud ied . I n the g iven instance - sharp rook
endings, transposing into endings with rook
against pawns. In them we encounter ideas
with which we are a l ready fami l iar.
Alekhine - Bogoljubow
World Championship Match ,
1 9th Game, 1 929
In the game there fol lowed 70 .. . 'it>g4? 71 b7
f5 72 b8'ik l:Ixb8 73 l::txb8 and White won
easily by approach ing the pawn with h is
king . But Efim Bogoljubow could have saved
the draw by employing the 'shoulder-charge' .
70 . . . 'it>e4!
The black king must be placed in the path of
the opponent's k ing.
Of course , we wi l l also meet new ideas
which operate in sharp rook endings. The
most important of them is inteference. We
have a l ready encountered it in the analysis
of the Mar6czy-Tarrasch ending. Now we
wi l l examine a far more compl icated exam
ple.
This posit ion could have occurred i n the
game Lapin - Utyatsky (Bryansk 1 965) .
1 . . . 'it>c2
2 .l:i.c7+ 'it>b2 ! !
Only this paradoxical move , suggested by
Utyatsky, leads to a win. I t i nvolves the idea
of i nterference. For example, if 3 'it>g4 B lack
decides matters with 3 . . . .l::ta5! 4 .l::tc6 'it>a3! 5
.l:r.xg6 b2 6 :b6 l:ta4+ and 7 . . . l:tb4 .
3 .l::tc6 .l:i.a4!
4 .U.xg6 'it>a3
4 . . . 'it>c3 or 4 . . . 'it>a2 is a lso not bad .
5 l:tb6
5 l:tf6 b2 6 l:tf1 l:Ic4 7 l:tb1 l:tc1 .
5 . . . b2
Threatening the interference 6 . . . .l::tb4 .
6 l:txb2 'it>xb2
7 g4 'it>c3
8 'it>h4 'it>d4
9 'it>g5
1 0 h4
And B lack wins easi ly.
'it>e5
'it>e6
Thanks to the threat of i nterference, B lack
forced his opponent into hurrying to g ive up
his rook for the pawn. In the event of the rou
tine 2 . . . 'it>b1 ? , interference no longer occurs
and White can wait unti l the pawn reaches
b 1 . From there, incidental ly, it takes longer
for the king to reach the opposite wing.
How to Study the Endgame
3 'it>g4 b2 (3 . . . �a5 4 �c6 b2 5 .l::txg6 Wa2
6 J:(b6 with a d raw) 4 �g5 �a 1 (4 . . . .l::i.b3
5 'it>xg6 or 5 g4) 5 �b7 b 1 'iV 6 �xb 1 + 'iti>xb 1 .
Wh ite can now ach ieve a draw i n various
ways. It is usefu l to examine the result ing
variations, s ince in this way we wi l l repeat
and consol idate our knowledge of endings
with rook against pawns. I n the analysis
extreme care has to be taken - despite the
apparent s impl icity, here one can easily go
wrong .
1 ) 7 'it>xg6 'it>c2 8 g4 'iti>d3 9 h4 'iti>e4 1 0 h5
'it>f4 1 1 h6 �a6+ 12 'iti>h5! with a draw
(shoulder-charge). It is amusing that Utyatsky
suggests 1 2 'iti>g7? 'it>g5 1 3 h 7 .l:f.a 7 + 1 4 'it>g8
'.t>g6 1 5 h8tt:J+ 'it>f6 1 6 g5+ 'it>xg5 1 7 tt:Jf7 + ,
but we already know that, accord ing to
theory, after 1 7 . . . 'it>f6 1 8 tt:'Jd6 .l::ta5 (or
18 . . . We6) Black wins.
2) 7 Wxg6 'it>c2 8 h4 ( in Utyatsky's opin ion ,
this move loses) 8 . . . .l::txg3+ 9 'it>f6 .l::i.h3 1 0
'it>g5 Wd3 1 1 h5 'it>e4 1 2 h6 �e5 1 3 'iti>g6
'.t>e6 14 'it>g7! (but not 1 4 h 7? l:tg3+ 1 5 'it>h6
'it>f7 1 6 h8tt:J+ 'iti>f6) 14 . . . 'iti>e7 ( 1 4 . . . l:tg3+ 1 5
'it>f8!) 1 5 h7 l:tg3+ 1 6 'it>h8 ! , saving the game
thanks to stalemate.
3) 7 g4 Wc2 8 h4 .Ug3 9 'it>f4! .l::th3 1 0 'it>g5
'.t>d3 11 h5 gxh5 1 2 gxh5 'it>e4 1 3 h6 'it>e5 1 4
'lt>g6 'iti>e6 1 5 'it>g7! with a draw, a s i n the
previous variation .
Thus we should bu i ld up our theory of the
endgame in the most economical way, by
s ing l ing out the most general ly used tech
n iques and the most important exact posi
tions. How best to ass imi late and consol i
date this material is another matter. Here
one cannot get by without a fami l iarity with
addit ional examples, i nclud ing compl icated
practical endings (such as the one we have
just been ana lysing) . I t is usefu l to try and
solve a series of tra in ing exercises on the
g iven topic. And above al l , I recommend that
you analyse i ndependently those endings
which you happen to encounter.
What does an i ndependent analysis of
endgame positions g ive us?
1 ) We learn new ideas and methods,
expandir;�g our system of knowledge, and we
refine the information we a l ready have.
2 ) After analysing a large amount of mate
ria l , we have a better u nderstanding of what
featu res are typical and important and
should therefore be i ncluded in the 'system' ,
and which are accidental i n character. As a
result we form our endgame impressions
most clearly, economical ly, at the same t ime
without omitt ing anyth ing important.
3) I t improves our analytica l mastery.
4) At times some players ga in the impres
sion that they largely u nderstand the secrets
of chess and that to find the best move in the
majority of cases is no problem . They only
need not to b lunder, and to obta in the
open ings they want. Analysis helps to rid
themselves of such i l l us ions, and shows
what an enormous wealth of ideas is
sometimes concealed in the seemingly
most modest position . I t guards against
superficial ity, and a ids the development of
such important tra its as precision , accuracy,
industriousness, and so on .
5) An analysis of you r own games enables
deficiencies i n you r play to be objectively
diagnosed .
18 � How to Study the Endgame
6) AnalysisA serious mistake , after which Black fa l ls
s ign ificantly beh ind in development and
ends up in a d ifficult position . He should
have castled .
1 2 'iVa3 ! lt'la5
1 2 . . . cxd4 real ly was better.
1 3 b3 ii.e7
1 4 dxc5 ffi
It is a bad sign , if moves such as this have to
be made. But if 1 4 . . . bxc5 there fol lows 1 5
tt:Je5 i.xg2 1 6 �xg2 d6 1 7 'if a4+ 'it>f8 1 8
lt'ld7+ �g8 1 9 lt'lxc5.
1 5 ii.h3 'i;;f7
1 6 i.xe6 was th reatened , and if 1 5 . . .'iVc7 ,
then 1 6 cxb6.
Wh ite has an und isputed advantage . He is a
pawn up , the black king is stuck in the
centre , and the d7- and e6-points are
obviously weak. But note that a l l these
factors are not constant, but temporary.
I magine that Black plays . . . bxc5 and . . . d7-
d6 - then he wi l l consol idate h is position .
This means that Wh ite must act swiftly and
decisively.
1 6 li.d2?
An instructive commentary on the move
1 38 � Converting an Advantage
made by Flohr was g iven by grandmaster
Isaak Boleslavsky:
In this position could White really not find
anything better than the strictly positional
doubling of rooks? If White really wanted to
play positionally, he should have continued
1 6 li:Jd4 ii..xc5 17 'ika4 ( 1 7 'ir'c1 !? -
Dvoretsky), and to avoid the worst Black
must exchange on d4. But the position
demanded other measures, and after the
energetic stroke 1 6 b4! White would have
gained an irresistible attack. Here are some
sample variations:
1) 1 6 . . .Chc6 17 cxb6 ( 1 7 l:td2 is a lso not bad
- Dvoretsky) 17 .. . axb6 18 'ikb3 li:Jxb4 (if
1 8 . . . ii..xb4 both 1 9 c5 and 1 9 a3 are strong ,
and even 19 ii.xe6+ ! 'lti>xe6 20 ii.xb6! 'ii'e8
21 c5+ cj;;e7 22 a3 ii..a5 23 ii.xa5 li:Jxa5 24
'ii'b4 - Dvoretsky) 19 ii.xe6+! cj;;xe6 20
ii.xb6 'i!lixb6 (20 . . . 'fle8 21 c5+ li:Jd5 22 e4)
21 c5+ ii.d5 22 1J.xd5 (22 'ir'e3+ - Dvoretsky)
22 . . . ii.xc5 23 1J.fd1! (of course, 23 1J.xc5+ is
also good enough to win ; general ly speak
ing , you should not continue calculating
variations, if the evaluation of the continua
t ion being analysed has become obvious -
Dvoretsky) 23 . . . ii.xf2+ 24 �g2 li:Jxd5 25
'iVxd5+ 'lto>e7 26 'ii'xd7+ cj;;f8 27 't!Vxc8+, and
wins.
2) {6. Jhxc4 17 'iVxa7 liJxe3? 1 8 fxe3 ii.xf3
19 'fud7.
3) 16. Jhxc4 17 fixa7 ii..c6 1 8 ii.xe6+! (or 1 8
cxb6 l:ta8 1 9 ii.xe6+! - Dvoretsky) 1 8 . . . 'lto>xe6
19 li:Jd4+ rtJfl 20 li:Jxc6 .:txc6 21 '!J.xd7 'ike8
22 cxb6 li:Jxe3 23 fxe3, and White, with four
pawns for the piece and an overwhelming
position, wins without difficulty.
4) 1 6 . . . li:Jxc4 17 'flixa7 ii.d5 1 8 �d5 exd5
19 'ikb7 cj;;e8 20 'iixd5 li:Jxe3 21 fxe3 'Wic7 22
1J.d1 'i:J.d8 23 cxb6 'iic6 (23 . . . 'ii'xb6 24
ii.xd7 + ..tis 25 li:Jd4 ii.xb4 26 li:Je6+ cj;;e 7 27
li'e4! - Dvoretsky) 24 b 7 'iixd5 25 1:.xd5 rtJfl
26 b5, and White's powerful pawns decide
the game.
After the move made by White, the picture
changes amazingly rapidly.
I should also add that after 1 6 b4 ! li:Jxc4 the
move 1 7 'ir'xa 7 is the strongest - 1 7 'i'b3
(hoping for 1 7 . . . b5? 1 8 ii.xe6+! 'it>xe6 19
li:Jd4+ and 20 li:Jxb5) is much worse in view
of 1 7 . . . li:Jxe3 1 8 fxe3 ii.xf3 . However, also
after 1 7 . . . i.d5 1 8 �xd5 exd5 1 9 'ii'd3 lbxe3
2 1 fxe3 'lti>e8 compared with the analogous
variation with 1 7 'ikxa7 the a7-pawn would
have remained a l ive.
1 6 . . .
1 7 I!fd1
1 8 lLle1
bxc5
d6
Another passive move . 1 8 ii.f4 suggests it
self, forcing the uncomfortable reply 1 8 . . Jk6
( if 1 8 . . . .i.xf3 1 9 exf3 li:Jc6 , then either 20
.i.xd6 li:Jd4 2 1 .i.xe6+ ! , or 20 �xd6 �xd6
2 1 �xd6 'file 7 22 .i.xe6+ ) .
1 8 . . . 'i!lib6
The d6-point is easi ly defended , and there is
noth ing more with wh ich to attack it - the f4-
square wi l l be taken away from the white
bishop by . . . g7-g5.
19 'ir'c1 h5 !
Black has a l ready seized the in itiative. If 20
li:Jd3, then 20 . . . g5 2 1 b4 ii'c6. Possibly
Wh ite should have tried 20 ii.g2 h4 2 1 b4!?
- after 21 . . . 'ii'xb4 22 .Ub2 'it'a4 23 �xb7 (23
ii.xb7? �b8) 23 . . . li:Jxb7 24 ii.xb7 l::tbB
fol lowed by 25 . . . hxg3 26 hxg3 'it'xa2 the
result ing position is d ifficult to evaluate .
20 f3? ! h4
21 g4 li:Jc6
22 li:Jg2?
22 li:Jc2 was better.
22 . . . li:Jd4
23 1J.xd4
The time for combinations was earl ier. In the
subsequent play Black converted h is ex
change advantage , although the opponent
d id not exploit a l l h is chances.
Converting an Advantage CtJ 1 39
23 . . . cxd4 24 i.xd4 'iii'a6 25 g5 fxg5 26 f4
g4! 27 i.xg4 h3 28 'ii'e3 l:Ih6 29 tt:Je1 l:tg6
30 'ii'xh3 'ifc6 31 ti:Jf3 'ii'e4 32 'iVg3? (32
f2 llh6 33 'ii'g3) 32 .. . 'it>g8 33 .l:r.d3 l':tf8 34
.ie3 e5! 35 'ii'g2 exf4 36 i.d2 i.d8 37 h3
l:te8 38 'iti>f1 d5 39 l:td4 'ii'b1 + 40 i..e1 dxc4
41 l:.xc4, and White resigned in view of
41 . . . i.h4! (but not 4 1 . . . i.a6? in view of 42
tt'ld2) .
The entire game convincingly i l lustrates a
wel l-known aspect of Ste in itz 's theory - the
player with an advantage must attack, as
otherwise he risks losing his advantage.
I n th is clear formula the word 'attack' must
be interpreted broadly - often it is necessary
to find some precise variation , forcing
combination etc. , i n short - a concrete and
energetic way to exploit your advantage.
Petrosian - Spassky
World Championship Match , 1 2th Game,
Moscow 1 969
23 l:.c1
A natura l move, reta in ing for White a serious
positional advantage. I ndeed , the c6-pawn
is weak, the knight has an excel lent square
at c5 , and the bishop at h5 is out of play.
But couldn't Wh ite have played more accu
rately - 23 i.h3 ? After a l l , after 23 . . . l:.b7 24
l:tc1 .l:.c7 compared with the game White
has ga ined a tempo - he has brought out h is
b ishop to a more active position . I f instead
23 . . J1c7, then 24 i.e5 i.d6 25 i..xd6 tt:Jxd6
26 e4 , exploiti ng the fact that the rook has
remained on d 1 .
But it can a lso be exploited by Black! By
giv ing up two minor p ieces for a rook:
26 . . . tt:Jxe4 ! 27 fxe4 (27 g4 tt:Jg5) 27 . . . i.xd 1
28 l:.xd 1 dxe4 , he reta ins excel lent chances
of saving the game. There is no point in
Wh ite going in for such an exchange, and
the move made by Petrosian must be
deemed the strongest.
23 . . .
24 i..e5
25 i..xd6
26 l:tfd1
l:.c7
i.d6
tt:Jxd6
Threaten ing both 27 l:.xd5 , and 27 e4 tt:Jxe4
28 g4 .
26 . . . ti:Jb5
White's advantage has crystal l ised . He now
has numerous tempting continuations, but it
is not so easy to choose the strongest. If 27
tt:Jc5 (with the threat of 28 tt:Ja6) , then 27 . . . a5
28 ti:Jd3 (threaten ing 29 ti:Jf4 , then 30 a4 and
3 1 tt:Jxd5) 28 . . . a4 29 ti:Jf4 i..g6, and no d i rect
win is apparent. To 27 l:.c5 Black repl ies
27 . . .f5 (defending against 28 e4) 28 l:tdc1
l:te7 !? (28 . . . ti:Jd4 29 'iii>f2 l:tac8) 29 'iii>f2 i.e8
1 40 � Converting an Advantage
or 29 .l:!.xc6 tt:'ld4 .
The strongest was 27 g4! .tg6 2a f4 , relying
on a tactical subtlety: 2a . . . f6 (or 2a . . .f5) is
not possible because of 29 l:!.xc6 ! . I n the
event of 2a . . . .te4 29 .txe4 dxe4 30 �f2
Black's position is hopeless in view of the
weakness of h is c6- and e4-pawns. But
2a . . . .th7 is also no better: 29 f5 (29 tt:'lc3 !? )
29 . . . g6 30 e4 dxe4 3 1 .txe4 .Uea 32 tt:'lc5
with an overwhelming advantage than ks to
the trag i-comic position of the black bishop
and the terrible threat of 33 a4 .
Wh ite also had another promising possibi l
ity: 27 tt:'lc3!? , emphasising the vulnerabi l ity
of the opponent's centra l pawns . The idea of
g3-g4 and f3-f4 could have been put i nto
effect sl ightly later.
27 '.t>f2 f6!?
28 e3?!
Petrosian continues strengthening h is posi
tion , but now his advantage is somewhat
reduced , s ince the black bishop is included
inthe defence of the queen side pawns . But
meanwh i le White sti l l had a concrete way of
achieving a won position : 2a tt:'lc5 ! .l:!.e7 (or
2a . . . a5 29 tt:'le6 l:!.cca 30 .th3 with the
unavoidable 31 tt:'ld4) 29 tt:'la6 .tea 30 a4!
tt:'ld6 31 e4, and Black loses a pawn .
28 . . . .tf7
29 .tf1
30 l:!.c3
tt:'ld6
30 .ta6!? came into consideration .
30 . . . �f8?
A far from obvious mistake . Black should
have taken control beforehand of the impor
tant f4-square , by playing 30 . . . g5 ! . After 3 1
tt:'lc5 a 5 3 2 l:!.dc1 .l:te7 fol lowed b y . . . .tea it
would not be easy for Wh ite to strengthen
his position .
31 tt:'lc5 a5
32 l:!.dc1 .l:!.e7
33 .th3
Wh ite has prevented the important defen-
sive move 33 . . . .tea (34 tt:'le6+ and 35 tt:'ld4)
and created the threat of 34 tt:'ld7+ .
33 . . . .Uaa7
B lack is only just hold ing on . One senses
that it is time to find a concrete way to break
through the opponent's defences. And there
is such a way. After 34 tt:'ld3 ! .tea 35 .!Llf4
White is threaten ing both 36 tt:'le6+ fol lowed
by 37 tt:'ld4 or 37 tt:'lda , and 36 l:!.xc6 .txc6 37
tt:'lg6+. I n the event of 35 . . . �f7 he decides
matters with 36 l:!.xc6! .txc6 37 .l:!.xc6 tt:'lb5
3a .te6+ l:!.xe6 (forced) 39 tt:'lxe6 , and the
conversion of the extra pawn is not too
d ifficult ( if 39 . . . a4 there fol lows 40 b4).
However, we nevertheless do not have the
right to say that Wh ite's position is defin itely
won . Even in a seemingly d ifficult situation
one can usual ly find resources , enabl ing
defeat to be avoided or at least the
opponent's task to be s ign ificantly compl i
cated . That is a lso the case here . For
example, there is a clever exchange sacri
fice : 35 . . . l:!.a6 !? 36 tt:'le6+ l:!.xe6 37 .txe6 f5
3a g4 g6 39 gxf5 gxf5 40 .Ug 1 �e7, and the
unfortunate position of White's bishop means
that t he conversion of h is material advan
tage is problematic. I nstead of 36 tt:'le6+, i t is
probable that 36 a4! is stronger - subse
quently the knight may be switched to d4 not
only via e6, but also via e2.
Converting an Advantage lZJ 1 4 1
Often the best defence is active defence. I
recommend checking 35 . . . a4 ! . Here is an
approximate variation : 36 .:.xc6 .i.xc6 37
t:Llg6+ (37 l:txc6 axb3 3a axb3 .l:.a2+ 39 'i!tg 1
:xe3) 37 . . . 'i!tea 3a CiJxe7 'i!txe7 39 l:!.xc6
axb3 40 axb3 .l:ta2+ 4 1 'i!tg 1 d4!? 42 exd4
t:Llb5. For the moment the outcome of the
game remains unclear - B lack's counter
attacking resources should not be under
estimated .
Even so, the manoeuvre of the knight to f4
was the correct p lan . Petrosian played a
weaker move .
34 a4?!
The position can be unhurriedly embel l ished ,
if during this time the opponent is not able to
do anything to strengthen h is defences. This
is not the case here. I t is dangerous to abuse
the principle 'do not hurry! ' .
Apparently Petrosian was i ntending 35 tiJd3
.iea 36 CiJf4 'i!tf7 37 tiJe2 fol lowed by CiJd4
and he wanted to prevent the opponent from
replying 37 . . . tiJb5. Genera l ly speaking , it is
usefu l to fix the black pawn on a5 and to
deprive the knight of the b5-square . But if
th is is played , it should be after the switching
of the kn ight to f4 , and therefore now Boris
Spassky forestal ls the main danger.
34 . . . g5 1
The f4-square is taken u nder contro l .
35 l:.d1
An exchange sacrifice came i nto considera
tion - 35 tiJd3 .tea 36 .llxc6 .i.xc6 37 llxc6
tOea 3a tiJc5 . Petrosian wants to play h is
knight to d4 v ia a lengthy route - d3-c 1 -e2 ,
but during this t ime Spassky is able to
activate h is forces.
35 . . .
36 tiJd3
37 CiJc1
38 tiJe2?!
'i!tg7
.tea
f5
3a .i.g2 g4 39 f4 was better, with a probable
draw.
38 . . .
39 .i.g2
40 .txf3
41 .i.xe4
42 CiJd4
43 'i!tg2
g41
gxf3
ttJe4+
fxe4
.l:r.f7+
lif6
Black now has some in itiative in connection
with the strateg ic th reat, after the exchange
of a pa i r of rooks, of playing h is k ing to d6.
On the resumption the game ended in a
draw.
Transformation of an advantage
The best way of exploit ing an advantage
sometimes i nvolves a favourable change in
the character of the position , giving up some
advantages that you a l ready have in favour
of others . Such a method is ca l led 'transfor
mation of an advantage' .
Taimanov - Stein
34th USSR Championsh ip , Tbi l is i 1 966/67
Black has a great positional advantage. He
has securely blocked the opponent's central
pawns, the bishop at g2 is 'bad ' , and the
wh ite knight a lso lacks mobi l ity. The most
natu ra l p lan , which Leonid Stein undoubt
edly had in m ind , i nvolves the advance of
the queenside pawns.
1 42 � Converting an Advantage
26 'it>f1
White intends to reinforce h is centra l pawns
by taking h is king to e3 and if necessary
placing h is bishop on f1 . H is rooks would
then be freed to take action against Black's
queenside pawn offensive. At this moment it
probably seemed to Stein that the conver
sion of h is advantage by normal methods
would not be so easy.
If the opponent makes an unexpected
move, hindering the implementation of
your plans, it is useful to ask yourself:
'What may be the drawback to the
opponent's move?' But even after asking
yourself such a question , it is not easy to
come to the decision found by Stein - it is
very much not in keeping with the unhurried
character of the preced ing play, and with
Black's intended plan .
26 . . . f5 !?
By opening the f-fi le , on which the wh ite king
stands for a moment, Black creates a th reat
to the d3-pawn . Of course , such a move ,
freeing White's bishop and knight , could
only be made by a h igh ly dynamic, non
routine player.
27 exf5 �xf5
28 'lt>e2
I would have preferred to part with a pawn
immediately, by returning with the king to g 1 .
28 . . . lbg4!
Threatening 29 . . . lbxf2 30 ;t>xf2 �g4+ 3 1
;t>e3 i.xd 1 .
29 l:i.b2 lbxh2
The conversion of the extra pawn
difficult.
30 'it>e3 l:i.a4
31 �e4 �xe4
32 lbxe4 lbg4+
33 'it>d2 lLlf2!
is not
Remember: with a materia l advantage it is
advisable to exchange pieces.
34 lLlxf2 lbf2+
35 '>t>c3 I:!.a3+
36 .J::f.b3 l:i.axa2
37 .l:tb5 �g6
38 l::td5 l:i.f5
39 I:!.d6+ l:i.f6
40 l:i.d7 l:i.g2
41 d4 l:i.xg3+
White resigned.
Ste in 's energetic actions were crowned by
complete success. But it seems to me that
what main ly told here was the psycholog ical
effect of Black's unexpected operation -
there are nevertheless some doubts about
its objective strength .
Serious consideration should have been
g iven to the reply 27 �h3!? suggested by
Grigory Kaidanov. After 27 . . . g4 28 i.g2
Black cannot play 28 . . . lbxd3? 29 �xd3
l:i.xd3 30 lbxd3 fxe4+ 3 1 lLlf4 , and nothing
particu lar is promised by 28 . . . fxe4 29 i.xe4
or 28 . . . f4 29 gxf4 l:i.xf4 30 '>t>e2 .
I nstead of 27 . . . g4 combinations involvi ng a
sacrifice on d3 look tempting . However, if
27 . . . l::txd3 there fol lows 28 lbxd3 fxe4+ 29
lbf2 �xh3+ 30 '>t>e2 . 27 . . . lbxd3 is stronger,
hoping for 28 �xf5?! lbb4 ! ! 29 l:i.xd4 tt:lxc2
30 �xe6 .Uxf2+ 31 ;t>xf2 lbxd4 with a won
minor p iece ending (32 i.f5 is bad in view of
32 . . . lbxf5 33 exf5 g4! 34 We3 ;t>f6 35 Wf4
h5) . Wh ite defends by 28 exf5 ! tt:lxf2
(28 . . . i.xf5 29 i.xf5 l:i.xf5 30 'it>g2 or 28 . . . tt:lb4
30 I:!.xd4 lbxc2 3 1 fxe6 l:i.xf2+ 32 'it>xf2 tt:lxd4
33 'lt>e3) 29 I:!.xd4 lbxh3 30 g4 with an
u nclear end ing .
I th i nk that in reply to 27 �h3 Black should
move h is bishop: 27 . . . �d7 ! . However, here
too White reta ins some saving chances in a
position where he is the exchange down : 28
�xf5 �a4 29 I:!.dd2 �xc2 30 l:i.xc2 , or a
pawn down after 30 . . . lbxd3 3 1 l:i.d2 lLlxf2 32
.J::f.xd4 lbxe4 33 g4.
The strongest response to 26 .. . f5 wou ld
Converting an Advantage ltJ 1 43
seem to be the cool-headed 27 'it>e2 ! . I n the
event of 27 . . .f4 28 gxf4 gxf4 there is the
satisfactory reply 29 ..th3 ! , whi le after
27 . . .fxe4 28 ..txe4 ltJg4, as shown by Ph i l ipp
Schlosser, Wh ite has the s imp le move 29
l:tf1 ! (29 . . . ltJxh2 30 l:th 1 ) .
The transformation of an advantage - giv ing
up some benefits that you a l ready have for
the sake of achieving other benefits - is a
rather compl icated techn ique, accessible
only to players with a subtle u nderstanding
of the game. After a l l , you have to assess
the situation correctly, and precisely weigh
up the p luses and minuses of the decision
being taken , i n order not to 'buy a pig i n a
poke' . And psychologica l ly it is not easy in a
favourable position to take sharp decis ions,
depriving yourself of some advantages
gained earl ier.
Petrosian - Bann ik
25th USSR Championsh ip , R iga 1 958
Wh ite unexpectedly offered an exchange of
bishops.
1 8 it.c5!
Why? Here is Petrosian's explanation :
Before deciding on this move, i t was
essential to thoroughly weigh up everything
'for ' and 'against'. It looks illogical, since
White voluntarily exchanges his 'good' bishop
for the opponent 's 'bad ' bishop, instead of
exchanging it for the knight (1 8 ..txb6+) and
consolidating his advantage. But on a
deeper investigation of the position it be
comes clear that after the possible ex
change of rooks on the d-file and the
transfer of his king to e6, Black covers his
vulnerable points and sets up an impregna
ble position. In this case his 'bad' bishop
would play an important role.
For my part I should comment that after 1 8
g4 .l:xd 1 + 1 9 l:txd 1 l:td8 20 l:txd8 'it>xd8 2 1
it.xb6+ axb6 22 Wc2 White a lso reta ins
excel lent chances of success. He plays h is
k ing to e4 and h is kn ight to d3 , with the idea
of a pawn offensive on the queenside, and in
some cases even e2-e3 and f2-f4 .
1 8 . . . l:txd 1 +
Petrosian recommended the pawn sacrifice
1 8 . . . ..txc5 1 9 ltJxc5 .l:Ihe8 20 I!.xd8 'it>xd8 2 1
ltJxb7+ 'it>c7 2 2 ltJc5 e 4 (with the threat of
23 . . . a5 and 24 . . J::te5) , but it is i ncorrect in
v iew of 23 ltJa6+ 'it>b7 24 liJb4 fol lowed by
ltJc2 .
1 9 l:txd1
20 ltJxc5
21 ltJe4
..txc5
l:te8
.l:.e6
2 1 . . . Itf8 was no better: 22 g4 .l:.f7 (22 . . . ltJc8
23 liJc5 .l:.f7 24 liJe6+) 23 .l:l.d6 .
22 g4 a5
23 l:td3 liJd7
24 'it>c2
24 'it>d2 !? .
24 . . . b6
Anatoly Bann ik hopes to ease h is defence
by exchanging knights with 25 . . . ltJc5 . Wh ite
prevents th is .
25 l:tf3 ! 'it>d8
(see diagram)
1 44 � Converting an Advantage
26 a3!
Again Petros ian takes measures against the
threat of an exchange - if 26 . . . ..t>e7 ( intend
ing 27 . . . l2Jc5) he had prepared 27 b4 axb4
28 axb4 . Then there fol lows c4-c5, when
the opponent is a ltogether unable to breath .
26 . . . c5
27 �c3 �e7
28 Ild3
After provoking . . . c6-c5, which has weak
ened the d5-point , Wh ite returns his rook to
the d-fi le .
28 . . . .Uc6
29 Ild5 tt:'lf8
30 l2Jg3
31 lDf5+
32 e3
l2Je6
..t>e8
l2Jc7
32 . . . lDd8 and 33 . . . l2Jf7 was more tenacious.
33 l::!.d1 l2Je6
34 �d3!
The time has come to activate the king .
34 . . . .Uc7
35 �e4
36 l2Jd6+
37 tt:'lf5+
38 l2Jd6+
39 l2Jf5+
l::!.c6
rt;e7
'it>e8
rt;e7
'it>e8
When converting an advantage, experi-
enced players often resort to repeating
moves, not only to gain time on the
clock, but also in the hope that the
opponent will try to change the unfa·
vourable course of the play and, by
avoiding the repetition, worsen his own
position. However, th is should be done
carefu l ly, avoid ing the three-fold repetition
of the position which occurred in the present
game. It is strange that neither Petrosian,
nor his opponent, noticed that after 39 . . . 'it>e8
the position wou ld be repeated for the third
t ime and Black had the right to cla im a draw.
Most probably th is a l l happened in a severe
t ime scramble.
40 a4 lDd8
41 lDh6!
Not a l lowing 41 . . . l2Jf7 .
41 . . . l2Je6
42 l2Jg8 lDf8
42 . . . 'it>f7 43 �d7+! 'it>xg8 44 'it>d5 is an
elegant variation .
Now 4 3 'it>d5 rt;d? i s pointless, whi le i f 4 3
�f5 there fol lows 43 . . . 'it>f7 44 tt:'lh6+ rt;g? 45
�d8 l2Je6 46 l::!.e8 tt:'lc7 , and White loses h i s
knight .
How then can he break through the enemy
defences? When the opponent is con
demned to passivity, one is very often
Converting an Advantage ctJ 1 45
aided by a very important endgame
device - zugzwang.
43 l:td2 ! �f7
In the event of 43 . . .'�Jd7 White wins by 44
lttf5 Wd8 45 e4 '.t>e8 46 f3 �d8 47 l:.xd7+!
lttxd7 48 tt:Jxf6+. Note that, before sacrific
ing the exchange, it makes sense, i n
accordance with the principle ' do not hu rry ! ' ,
to make two preparatory pawn moves,
strengthening the posit ion to the maximum.
I f 43 . . . ne6 there a lso fol lows 44 Wf5 �f7 45
�d8 �c6 46 tt:Jh6+ �g7 4 7 �e4! tt:Je6 48
�d7+! '.t>xh6 49 '.i?d5 .
44 tt:Jh6+
45 tt:Jf5
46 �d6!
�e8
tt:Je6
The exchange of rooks, strengthening the
threat of an i nvasion by the wh ite king, leads
to a won knight end ing .
46 . . .
47 tt:Jxd6+
48 tt:Jb5
�xd6
�d7
tt:Jg7
This leads two moves later to zugzwang , but
that is a lso how things conclude in the
variation 48 . . . tt:Jf8 49 'it>f5 'it>e7 50 tt:Jc3 tt:Jd7
51 tt:Jd5+ '.t>f7 52 e4 h6 53 f3 .
49 h6
50 'it>d5
51 '.t>xe5
52 tt:Jc3
53 tt:Je4
54 �f5
55 fxg3
56 tt:Jg5+
57 �e6
58 '.i?d7
59 e4
60 e5
61 e6
B lack resigned .
tt:Je8
f5
fxg4
�e7
'.t>f7
g3
g4
'lt>g8
tt:Jc7+
tt:Ja6
tt:Jb4
tt:Jd3
An excel lent ending - in it Wh ite used many
of the principles for converting an advantage
that we have been d iscuss ing .
I n conclusion I offer a few exercises, i n each
of which you have to choose the most
methodical way of proceed ing .
1 46 � Converting an Advantage
Exercises
1 . Black to move 2. Wh ite to move
3. White to move 4. Black to move
Converting an Advantage tD 1 47
5. Black to move 6. Wh ite to move
7. White to move 8. Wh ite to move
1 48 � Converting an Advantage
Sol utions
1 . Koberl - Szabo (Budapest 1 951 )
23 . . . a5!
24 tt:'lc1 a4!
By advancing his a-pawn , Black has pre
vented the equal is ing manoeuvre tt:'le2-c1 -
b3, prepared the development of h is rook by
. . . l:ta8-a6-b6 or . . . a4-a3 fol lowed by . . . l:!.a4,
and, final ly, created the precond itions for an
attack on the opponent's queenside. If now
25 tt:'ld3, then 25 . . . l:!.d8 26 �f1 l:td4 ! .
The game concluded a s fol lows: 25 'it>f2 a3
26 'it>e2 �b2! 27 .Uc2 (27 tt:'ld3 tt:'la4 ! )
27 . . . l:r.d8 28 �f1 tt:'la4 29 tt:'ld3 tt:'lc3+ 30
�e3 tt:'lxa2! 31 tt:'lxb2 tt:'lb4 32 �c1 axb2 33
l:!.b1 tt:'lc2+! 34 'it>f4 (34 'it>e2 tt:'la3 35 �xb2
tt:'lxc4) 34 . . . g5+ 35 'it>e5 l:td6! 36 c5 l:!.e6+ 37
'it>f5 tt:'le3 mate.
2 . Bastrikov - Kiselyov (Sverd lovsk 1 946)
Noth ing is g iven by 22 l:tg 1 + 'it>h 7 23 l:tg7 +
'it>h6 or 23 l:!.g5 f6 (23 . . . Wh6? 24 ..te3) 24
Ir.xh5+ Wg6. Black's important defensive
move . . .f7-f6 must be prevented .
22 �e1 ! l:tfe8
If 22 . . . e6 or 22 . . . l:tae8 , then 23 tt:'lc5 is
strong .
23 l:!.g1 +!
23 . . . 'it>h7 24 l:!.g7+ 'it>h6 25 l::i.xf7 is now bad
for Black.
There fol lowed : 23 .. . Wf8 24 tt:'lc5 1:1ed8. As
was pointed out by grandmaster Matthew
Sadler, 24 . . . Ir.ec8!? was more tenacious,
when White should continue 25 tt:'ld7+! ( less
good is 25 .l::tg5 b6 26 .Uxh5 f6 or 25 tt:'lxb7
.l:l.ab8 26 ..tg7+ ! WeB 27 l:!.b1 f6 28 il.h6
'it'd?) 25 . . . �e8 26 tt:'le5 'it>f8 (26 . . . 'it>d8 27
tt:'lxf7+; 26 . . . e6 27 l:tg8+ �e7 28 l:tg7) 27
..te3 e6 28 �c5+! .Uxc5 29 tt:'ld7+ 'it>e7 30
tt:'lxc5.
25 l:tg5! b6 26 .Uxh5 e5 27 �xeS! bxc5 28
i..f6 �e8 29 .Uh8+ 'it>d7 30 l:txd8+ l:txd8 31
�xd8 'it>xd8 32 �3 (Wh ite's outside passed
pawn ensures h im an elementary win)
32 . . . 'it>e7 33 �e4 �e6 34 'it>f4 f5 35 h4 'ittf6
36 h5 We6 37 Wg5 Black resigned .
3 . M i les - Nikolac (Wijk aan Zee 1 979)
Noth ing is g iven by 48 l:!.f5 l:tg5.
48 a4!
With th is unhurried move Wh ite forestal ls
the opponent's only sensible plan of . . . b6-
b5-b4 and puts h im in zugzwang. After any
move by the knight from e4 , 49 l:tf6+ is
decisive. 48 . . . l:!.g5 49 .Uh7 is bad for Black,
whi le if 48 . . . l:!.h4, then 49 tt:'lg6! and 50 tbe5,
but not 49 .Uf5? because of the pretty reply
49 . . . l:!.h 1 + ! .
48 . . . �c6
49 l:tf5!
It transpires that 49 . . . .Ug5 no longer defends
the pawn in view of 50 tt:'lxd5 ! l::i.xf5 51 tbe7+
and 52 tt:'lxf5.
49 . . . tt:'ld6
50 l:!.f6
51 g3
52 �g2
.Uh4
Itg4
Black's position is now completely hopeless.
Tony M i les qu ickly converted his advantage.
52 . . . h5 53 tt:'lxh5 �d7 54 'it>f3 .Ug8 55 tLlf4!
.Uxg3+ (55 . . . tt:'le4 56 tt:'lxd5 ! ) 56 'it>xg3 tt:Je4+
57 �g4 tt:'lxf6+ 58 'it>f5 tt:'le4 59 tt:'lxd5 tt:Jd6+
60 �e5 tt:'lf7+ 61 'it>f6 B lack resigned .
4. Skembris - Torre
(Olympiad , Luzern 1 982)
White's p ieces have hardly any active
possib i l it ies. However, he nevertheless has
Converting an Advantage tZJ 1 49
one chance to become active : 'iie2 ! , i ntend
ing 'ii'b5! . For example, 30 . . . h6? (general ly
speaking, th is pseudo-prophylactic move is
usefu l , but it does not parry the opponent's
concrete threat) 31 'iie2 ! 'ikxb3 (in the
endgame White gains a d raw without d iffi
culty) 32 'ikb5 b6 33 lt:Jf3, and the weakness
of the f7-point ensures White sufficient
counterplay.
30 . . . a6!
The opponent's only active idea is parried ,
and Black wi l l soon create threats on the
queenside by moving h is kn ight across to
there .
31 g4 lt:Je7 32 lt:Je2 .id2 33 lt:Jg1 lt:Jc6
(threaten ing 34 . . . lt:Ja5) 34 i...c7 lt:Jb4 35 i...a5
lLJc2 36 i...xd2 'ikxd2 37 �g3 lt:Jxe3 ! 38 'ii'a3
lLJd1 39 lt:Jf3 'ii'xf2+ 40 �4 g5+ White
resigned .
5. Gragger - Barcza (Olympiad , Varna
1 962 , variation from the game)
I f a passed pawn is blocked by a bishop, the
winn ing p lan usual ly i nvolves breaking
through with the king towards the passed
pawn . But doing this immed iately does not
work: 1 . . . �e4? 2 �e2 .ih5+ 3 �f2 �d3 4
i.e? ! a4 5 i...d6 �c2 6 i...a3 with a d raw.
Black must first tie the wh ite king to the
defence of the queenside pawns, and only
then break through with h is king on the
opposite wing .
1 . . . �c4!
2 i...c7
3 i...e5
4 �c1
4 . . . .ih5 also wins .
5 i...d6
Or 5 c4 b6.
a4
�b3
i...c2!
5 . . . a3
6 bxa3 �xc3
Then . . . .ia4, . . . b7-b5 and . . . �d3-e2-f3-g2.
6. Smir in - Vogt (Saltsjobaden 1 988/89)
In the event of 33 .l:.xa5? .l:i.f3 34 l:t.h5 l:txg3
35 .l:lxh7+ �g8 the passed g-pawn ensures
Black sufficient counter-chances. The attack
on the g3-pawn must be foresta l led , and the
move 33 i...e4!? , made by l lya Smiri n , looks
a sensible solut ion to the problem . There
fol lowed 33 . . . i...xa4? 34 l:txa5 i...e8 (34 . . . i...c6
35 i...xc6 bxc6 36 l:tg5) 35 i...xb7 ltf1 + 36
�d2 i...g6 37 c4 l:tf2+ 38 �c3 �g7 39 l:tg5! ,
a n d Black, find ing no way out, lost on t ime.
Lothar Vogt cou ld have exchanged either
the rooks , or the bishops. Try to estimate ( I
mean estimate - to ca lculate everything is
not possible and you have to trust your
i ntu ition ) whether one of the exchanges (or
both) offers real istic chances of saving the
game. If you r answer is positive, th is g ives
grounds for seeking an a lternative move to
the one chosen by White in the game.
F i rst let us examine the bishop ending:
33 . . J:te8 34 ltxe8+ i...xe8.
In the magazine 64 - Shakhmatnoe oboz
renie ( 1 996 No. 1 2) , grandmaster Igor Zaitsev
suggested a clever breakthrough plan : 35
b4! ? b6 (Black loses immediately after
35 . . . axb4? 36 aS or 35 . . . i...xa4? 36 bxa5
fol lowed by 37 i...xb7) 36 b5 i...f71 ( it is
important to forestal l Wh ite's main threat
c4-c5) 37 �d2 �g7 38 �d3 , restricting
1 50 � Converting an Advantage
himself to the variation 38 . . . h6 39 c4 'it>f6 40
c5 'iite5 41 cxb6 'iitd6 42 'iite3 �b3 43 �c6
�xa4 44 b7 'it>c7 45 b6+ 'iitb8 46 �xa4 .
Here I do not agree with h im - Black's
resources are not yet exhausted . To say
noth ing of the attempt, by sacrificing the h7-
pawn with 38 . . . 'it>f6 , to bring the king to the
queenside as qu ickly as possible in order to
h inder c4-c5, he can also play more
accurately in the course of the plan exam
ined by Zaitsev.
38 . . . h5 39 c4 'iitf6 40 c5 'it>e7! (but on no
account 40 . . . 'it>e5? - the place for the king is
not in the centre, but in front of the pawns)
41 cxb6 'iitd8 ! . How can White win here? I t
is not possible to queen a pawn : 42 �f5
�b3 43 'iitd4 �xa4 44 'it>c5 �d 1 45 �d6
�f3 - the bishop has arrived just in t ime.
And the position arising after 42 'it>e3 �b3
(42 . . . 'it>c8) 43 �c6 'it>c8 44 b7+ 'it>b8 45 b6
.ii.d 1 is drawn .
The main continuation is 35 �xb7 �xa4 36
c4. Black avoids an immed iate loss by
playing 36 . . . i.b3 37 c5 �g7 .
If 38 c6 , then 38 . . . i.d5! 39 �a8 �e6 ! . I n the
event of 38 'iitd2 'iitf6 39 'iitc3 Black finds the
excel lent manoeuvre 39 . . . �d 1 ! with the
idea of . . . i.f3 . I t is not apparent how White
can win . For example, after 40 c6 �f3 4 1
'iitc4 h 5 4 2 �a8 �e7 4 3 'it>b5 i t i s bad to play
43 . . . 'it>d6(d8)? 44 «t>b6 h4 45 c7 or 43 . . . h4
44 gxh4 g3? 45 c7! (now it is clear why the
king avoided the d4- and c5-squares - so
that after 45 . . . g2 the pawn should not queen
with check), but Black continues 43 . . . h4! 44
gxh4 «t>d6! , l u ring the king to b6 where it wil l
be checked , and then 45 . . . g3 (analysis by
Zaitsev) .
I n the event of 38 �c8 ! Black has two
possib i l it ies:
a ) 38 . . . h5 39 c6 h4 40 gxh4 g3 41 �h3 �e6
42 �g2 'it>g6 43 �d2 �h5 44 'it>e3 'it>xh4 45
'it>d4 'it>g5 46 Wc5 �f4 4 7 'it>b5 'it>e5 (after
47 . . . 'it>e3 48 'it>xa5 Wf2 49 �h 1 the king
does not manage to return to the queenside
in time) 48 'it>xa5 'iii>d6 49 b4 , and White
should apparently win .
b ) 38 . . .f6 39 i.xg4 'it>e5 ( in th is way, i n
Zaitsev's op in ion , B lack ga ins a draw) 40
'it>d2 Wd5 4 1 'ito>c3 i.a2 42 b4 axb4+ (or
42 . . . «t>c6 43 i.f3+ 'ito>b5 44 �e2+ 'ito>c6 45
bxa5 'ito>xc5 46 'it>d2 ) 43 'it>xb4 'ito>c6 44 .ii.f3+
'ito>c7 45 'it>c3 , and there is noth ing to be done
against the decisive breakthrough of the
king to the king s ide. And th is means that the
exchange of rooks most probably would not
have saved Black.
How can the defence be improved? The
best saving chances i n such situations are
usual ly promised by transposing into a rook
ending (everyone knows the saying : ' rook
endings are never won' ) . And so, 33 . . . �xe4!?
34 .l::txe4 . However, after 34 . . . .l:!.g8? (34 . . . h5?
35 l::te5) 35 Wd2 !? with the idea of 'it>e3-f4
Black is condemned to complete passivity
and should certain ly lose.
In rook endings you should aim to
activate the rook. After 34 . . . .l::tf1 + 35 'it>d2
.l:!.f2+ 36 Wd1 (36 �d3 .l::tf3+ 37 .l::te3 .l::tf2 38
b3 'it>g7) 36 . . . h5 ! 37 .l::te5 .l::th2 38 .l::txa5 'it>g7
fol lowed by . . . 'ito>f6(h6) and . . . h5-h4 Black
gains cou nterplay, but is it sufficient to save
the game?
There is a lso another way of transposing
Converting an Advantage LtJ 1 51
into a rook end ing : 33 . . .l:tf1 + 34 �d2 l:tg1
35 i..xc6 bxc6 36 l:txa5 l:tg2+ 37 'it>d3
l:'l.xg3+ 38 �e4, and now either 38 . . . l:tg 1 39
l:'l.c5 .Uf1 ! (38 . . . g3? 40 �f3 g2 41 b3 ! ) 40
l:'l.xc6 �g7 41 a5 h5, or 38 .. . I:tg2 !? 39 �c5
l:'l.f2! 40 a5 g3 41 'it>e3! (4 1 a6 g2 42 a7?
g 1 'i' 43 a8'ii'+ �g7 44 'ii'b7+ �f7 or 42 .Ug5
l:'l.xc2 43 a7 .Uc4+) 41 .. . 'it>g7 !? (4 1 . . . �f1
would appear to be worse : 42 I:tg5 c5 43
l:'l.xg3 .Ua 1 44 '.te4 I:txa5 45 'it>d5 , and Wh ite
is threatening 46 c4 fol lowed by 'it>c6-b6) 42
a6 l:f.f1 , and the position is most probably
drawn .
I n the rook endgame Black would have
retained good drawing chances. This factor
casts doubts on the p lan beg inn ing with 33
..ie4 , and forces us to seek other ideas.
Here is a suggestion by grandmaster Viorel
Bologan .
3 3 .Ug5!
I n the fi rst instance , as we know, 33 . . J:U3
must be prevented .
33 . . . �xa4
34 I:txa5!
An unexpected change of d i rection ! I n h is
commentary Smirin considered only 34
l:'l.xg4 �c6 with a probable draw.
34 . . . �c6
34 . . . �e8!? would seem to be more tena
cious , aiming at the fi rst conven ient opportu
n ity to play . . . h7-h5 .
35 �g5!
The rook resembles an annoying fly.
35 . . . �f3
35 .. ..!:!.g8 36 l:f.h5 is no better.
36 l:f.h5
The immediate 37 'it>d2 !? �f6 38 b4 is also
good .
36 . . .
37 �d2
I:tf7
After ski lfu l ly tying down the enemy pieces ,
White now wants s imply to strengthen h is
position by advancing h is queenside pawns.
I n reply to 37 . . . �c6 noth ing is g iven by 38 b4
.i::tf2+ (but not 38 . . . .Uf3? 39 b5) 39 �e 1 �g2
40 b5 �f3 41 .Uxh7+ 'it>g8, but 38 �h4! is not
bad .
Such a p lan for convert ing an advantage
(domination and the absence of cou nterplay
for the opponent) is fu l ly i n the spirit of
Anatoly Karpov. From the viewpoint of the
practical player, it is very important that here
practical ly nothing needs to be ca lculated ( in
contrast to the 33 �e4 variation , where one
has to delve both into the b ishop, and the
rook endgame), and this means that the
probabi l ity of mistakes is reduced .
I ncidenta l ly, the fina l conclusion about there
being on ly one solution to the in it ia l end
game position (and a lso about there being
only one winn ing method in the bishop
ending) is fu l ly i n accordance with Zaitsev's
view: My many years' experience of analysis
have convinced me that in tense, balanced
positions there cannot be two ways to win.
The same thought was a lso expressed by
another experienced analyst, i nternational
master Gavri i l Veresov: In positions on the
border between a draw and a loss, we
normally find there is only one solution.
7 . Smyslov - Botvinnik (World Champion
ship Match , 3rd Game, Moscow 1 954)
Although Black has three pawns for a piece,
his position is d ifficult. Vasi ly Smyslov could
have decided the outcome in the middlegame,
by breaking up the opponent's pawn cha in
and open ing l ines for h is p ieces by h2-h3.
27 Wkg2!
Threatening 28 .Ue5 .
27 . . . .Ufe8
28 h3!
I n the game there fol lowed 27 �e6+? �xe6
28 .Uxe6.
Usual ly p iece exchanges are the easiest
1 52 � Converting an Advantage
way of converting a material advantage. But
here, fi rstly, material is nominal ly balanced ,
and secondly (and this is more important) ,
the fewer the pieces remain ing on the board ,
the greater the role played by the pawns.
28 . . . 'it>f7 29 Itfe1 .
If 29 l:te5, then 29 . . . Itfe8, when 30 Itfe 1
i..c7 1eads to roughly the same position as in
the game. And if 30 Itxd5 there fol lows
30 . . . Ite3 31 i..b 1 (31 �d 1 'it>e6) 31 . . . .l:!.e2
with sufficient counterplay for Black.
29 . . . �fe8 30 .l:!.xe8 .l:!.xe8 31 l:txe8 (31 l:td 1
l:!.e3 32 'it>f2 l:th3) 31 .. .'lt>xe8.
White is not able to convert h is extra piece ,
since his k ing has nowhere to break through
- the black pawns prevent th is . But what a
wonderfu l target they presented in the
midd legame!
32 i..c3 'it'd? 33 a5 i..d8 34 i..b4 b6 35 a6
i..f6 36 i..c3 'it>e6 37 'it>g2 g5 38 �e2 g6 39
i..d1 i..e7 40 ii..d2 i..d8 41 ii..e3 .
Here the game was adjourned , and the
players agreed a draw without resuming .
8. Dvoretsky - Zilberstein
(Ordzhonikidze 1 978)
To where should the rook move , e2 or d 1 ?
But isn't it a l l the same - after a l l , i n both
cases White remains a sound pawn to the
good? But you should not approach the
conversion of an advantage so frivolously -
otherwise very often disappointment wi l l
await you . You should try to d iscover the
d ifference between moves and choose the
one which is in some way better, more
accurate than the other.
If 26 l:!.e2 there fol lows 26 . . . �c8 , when 27
.l::!.e6 .l::!.c2 is pointless . Having an obvious
advantage, you don't want to compl icate the
play and weaken your queenside pawns by
27 a3 i..f8 . The normal continuation is 27 g3
Wf7. We note that the black king prevents
our rook from becoming active on the e-fi le,
whereas the black rook on the c-fi le , a long
way from the wh ite k ing , is very active and it
restricts the wh ite pieces .
I myself wanted to seize the c-fi le , Therefore
I began checking 26 l:td 1 .
26 �d1 ! 'it>f7
Black hardly has the right to sacrifice a
second pawn by 26 . . . I:tc8 27 ..ll.xf6 M.c2 28
a4 bxa4 29 bxa4 (29 . . . i..c5 30 i..d4; 29 . . . .l::ia2
30 .l:i.a 1 ) . In reserve I a lso had the transition
into a bishop end ing : 27 l:!.c1 M.xc1 + 28
i..xc1 f5 (28 . . . 'it>f7 29 'it>e2 'it>e6 30 Wd3 '.iid5
31 g4) 29 'it>e2 'it>f7 30 'it>d3 'it>e6 31 '.iid4,
and in a l l probabi l ity Wh ite should gradual ly
win .
27 l:!.c1 !
Now 27 . . . i..d2 is poi ntless in view of 28
.l:!.c7+ and 29 g3. White wants to calmly
strengthen h is position by g2-g3, l:!.c2 , and
Wg2-f3 ; his rook is constantly th reatening to
break into the opponent's position along the
c-fi le . The result ing situation is more com
fortable for Wh ite than after 26 l:i.e2 .
The further course of the game confi rmed
that my evaluation was correct - the
conversion of the advantage proved to be an
a ltogether easy matter.
27 . . . .l:!.d8 28 .l:!.c2 .Ud1 + 29 'it>e2 l:!.e1 + 30
Wf3 llb1 31 i..d4! .l:i.d1 32 e4 a5 33 g4
i..d6 34 l:!.c6 �e5 35 i..xe5 (35 �e3
fol lowed by f2-f4 is a lso strong) 35 . . . .Me1 +
36 'it>d3 .l:!.xe5 3 7 f4 l:i.d5+ 3 8 'it>e4 l:i.d2 39
h4 Itxa2 (39 . . . h5 40 g5 fxg5 41 hxg5) 40
Wf5 Itf2 41 Itxf6+ 'it>g8 42 .l:!.a6 Black
resigned .
ltJ 1 53
Artur Yusupov
Tech n ical Proced u res i n a
G randmaster Battle
T
he game which I would l i ke to show you
was played in a grandmaster tourna
ment in the Span ish town of L inares. In its
i n it ial stage the two players engaged in a
d ifficult manoeuvring battle in a roughly
equal position . Then an end ing , s l ightly
better for Black, was reached . I t is instruc
tive to fol low those typical endgame proce
dures, thanks to wh ich I was able fi rst to
increase, and then successfu l ly convert my
advantage.
Salov - Yusupov
L inares 1 99 1
Reti Opening
1 l2lf3 l2lf6
2 g3
3 i..g2
4 0-0
5 c4
d5
c6
ii.g4
A normal position for the Reti Opening has
arisen . In my view, 5 l2le5!? i..f5 6 c4 is
interesting , s ince in the game after B lack's
reply the active knight advance is no longer
possible.
5 . . . l2lbd7
6 d3
7 b3
8 tt'la3
e6
i..d6
An unusual p lan . Now in the event of . . . e6-
e5 the manoeuvre l2la3-c2-e3 wil l h igh l ight
a certa in weakness in Black's centra l pawns;
however, if he avoids occupying the centre ,
Wh ite's idea does not present any danger.
8 . . . 0-0
9 l2lc2 l:te8
Black is not in a hu rry to d isclose h is p lans .
1 0 i..b2 a5
Real is ing that for the moment the advance
. . . e6-e5 is unfavourable, I carry out another
idea that is typica l i n such positions, trying to
' latch on' to the opponent's queenside. If
now 1 1 a3 �b6, and Wh ite has problems
with the defence of h is b3-pawn .
1 1 l:tb1 !
A deep prophylactic move . I n reply to
1 1 .. . a4 , apart from 1 2 b4 Black a lso has to
reckon with 1 2 bxa4 !? J::txa4 1 3 i..xf6 and 1 4
l:txb7.
11 . . . i..h5
11 . . . e5 is premature in view of 1 2 cxd5 cxd5
1 3 tt'le3 (attacking the bishop) 1 3 . . . i..h5 1 4
l2lh4 . So why not retreat the bishop i n good
t ime?
1 2 l2le3
Aga in Wh ite prevents . . . e6-e5. For both
sides it is d ifficult now to do anyth ing active .
As is usual in such situations, manoeuvring
beg ins without any clearly defined plan . The
two players merely operate with 'short'
positional or tactical ideas.
1 2 . . . i..c5
1 3 �d2
After 1 3 d4 i..f8 the e4-point is weakened .
1 3 . . . �b6
1 4 a3!? �a7
Black intends in some cases to play . . . a5-
a4. For example, if 1 5 l:tfd 1 there can fol low
1 5 . . . a4 1 6 b4 i..xe3 1 7 fxe3 dxc4 .
1 54 � Technical Procedures in a Grandmaster Battle
1 5 lLlc2 ii.f8
The bishop moves away from the tempo
gaining b3-b4 .
1 6 ii.d4
1 6 lLle5 with equal ity came into considera
tion .
16 . . .
1 7 i.a1
�b8
White thought that he had sl ightly improved
the position of h is bishop and worsened the
position of the opponent's queen .
1 7 . . . e5!?
After a l l these clever manoeuvres I decided
it was time for activity in the centre , s ince the
move lLlc2-e3 does not have to be feared -
the a3-pawn demands constant concern .
Even so, this advance a lso has defin ite
minuses - it weakens the d5-pawn and the
f5-square.
1 8 lLlh4 1i'd8!?
The opposition of the queens is advanta
geous to Black - in some cases the
undefended state of the wh ite queen may
tel l .
1 9 lLlf5
The position is roughly equa l . The s l ight
pressure of the wh ite pieces is neutra l ised
by Black's superior pawn formation .
1 9 . . . ..tg6
20 tL:lh4
21 lLlf5
22 i.h3?!
..th5
ii.g6
The 'grandmaster draw' after 22 lLlh4 would
have been the logical outcome. In h is desire
to play on , Valery Salov commits a sign ifi
cant inaccuracy - he loses control of the e4-
point . I was able to exploit the 'hanging'
position of the wh ite pieces .
22 . . . a4!
The long-awaited advance!
23 cxd5
I had reckoned with th is possib i l ity and I had
prepared a s imple intermediate operation .
But White had no choice: it was bad to play
23 bxa4? dxc4 24 dxc4 tt:Je4 (with the threat
of 25 . . . lLlg5) 25 'ir'xd7 'ir'xd7 26 lLlh6+ gxh6
27 i.xd7 .:.e7 28 .:.xb7 tL:lc5, or 23 b4? dxc4
24 dxc4 lLle4 25 'iYxd7 (25 'ii'e3 tL:lg5)
25 . . .'ifxd7 26 lLlh6+ gxh6 27 ii.xd7 .:.e7 28
i.h3 (28 .:.bd 1 lLlf6) 28 . . . lLld2 (28 . . . tL:lg5).
23 . . .
24 �xb3
axb3 !
24 dxc6 bxc2 25 �xb 7 did not work in view of
25 . . . tt:Jc5 26 c7 itd5 .
24 . . . tL:lc5
25 .U.bb1 tL:lxd5
Now Black's position is preferable - the
opponent has a weak pawn on a3 .
26 lLlfe3
If 26 J:ifd 1 Salov was apparently concerned
about 26 . . . e4! 27 d4 e3 ! .
2 6 . . . 'it'g5
Black contin ues the idea of exploit ing the
s l ight superiority of h is pawn structu re .
26 . . . tL:lc7 !? 27 tL:lc4 f6 , mainta in ing the
tension , was a lso possib le .
27 ii.g2 tL:lxe3
28 �xe3
28 h4!? came i nto consideration . In the
event of 28 . . . 'ir'h6 29 tL:lxe3! .l:i.xa3 White's
pawn deficit would have been compensated
Technical Procedures in a Grandmaster Battle a 1 55
by the bad position of the black queen , whi le
after a d ifferent retreat he would have
captured on e3 with h is queen , avoid i ng a
further spoi l ing of h is pawn structure .
28 . . . 'i!Vxe3
29 fxe3
Of course, the endgame is more pleasant
for Black, but it is not easy for him to
increase his advantage.
29 . . . f6
Black reinforces h is e5-pawn and prepares
. . . i.f7.
30 ..lli.c3!
31 �b4
32 �f2
Ua7
..lli.f7!
Salov defends in accordance with a l l the
rules of the endgame - he covers h is
weaknesses and br ings h is k ing towards the
centre .
32 . . .
33 .i.xf8
34 1:i.b2
tt::'lb3
�xf8
Ue7
When short of time it is usefu l to over
protect the important b7-pawn .
35 �fb1 tt::'ld2!
I n time-trouble Black gains t ime on the clock
by repeating moves.
36 Ud1 tt::'lb3
37 �e1 tt::'lc5
Al l the same the knight cannot be main
ta ined at b3, and Black switches it to a4,
intending to advance h is queenside pawns.
38 .Ub4 tt::'la4
39 �d2?!
A natu ra l move , but not the best, s ince it
does not prevent Black from carrying out h is
p la n . 39 lldb 1 ! ii.a2 40 .Ua 1 ii.e6 4 1 .Uab1
was stronger, when the pressure on b7 does
not a l low . . . c6-c5 to be played .
39 . . . c5
40 Ubb1 ?
A t ime-trouble error. After 40 .Ub5! .i.e8 4 1
l:tbb 1 b 5 Wh ite could have i ncluded h is
b ishop i n the defence: 42 ..lli.d5 ! .
40 . . . l:i.d7!
41 l:tdc1 c4
Here Salov spent a lot of t ime, apparently
weigh ing up wh ich was the lesser evi l - the
loss of a pawn or passivity, - and he chose
the latter. In the variation 42 Ub5 cxd3 43
tt::'lb4 (or 43 exd3 ..lli.c4 44 .Uxb7 .Uxd3+)
43 . . . dxe2+ 44 �xe2 White does not have
fu l l compensation for the pawn .
42 tt::'lb4 tt::'lc5
43 .Uc3
Now 43 . . . tt::'lb3+ 44 �e 1 Uxa3 is unfavour
able because of 45 dxc4 .
1 56 � Technical Procedures in a Grandmaster Battle
43 . . . e4!
Formally the move made by me is against
the ru les (pawns are supposed to be kept on
squares of the opposite colour to your own
bishop), but in fact it is very strong , s ince it
shuts in the bishop on g2.
44 d4
Forced .
44 . . . h5 !
44 . . .'�Jb3+ 45 'it>e1 �xa3 46 .lli.xe4 l:!.a4 also
looks good , driving back the knight and then
advancing the pawns. But in th is case the
wh ite pieces would have become active . I
preferred to play for a second weakness in
the opponent's position - shutting the
bishop out of p lay (the fi rst weakness is the
a3-pawn) .
45 'it>e1
45 h3 was better, in order after 45 . . .f5 to
have the reply 46 g4 (or 46 .:f1 g6 47 g4) .
45 . . . lZJa4
46 .l:i.c2 f5
The bishop on g2 has ended up on a square
from which it is unable to escape. After the
game Salov heated ly exclaimed that it
wou ld have been better if it hadn't existed at
all - then he could at least have tried to
obtain some counterplay on the kingside.
47 h3
Trying to activate the bishop.
47 . . . g6!
The last black pawn occupies a square of
the same colour as its bishop. Rules are
ru les, but concrete considerations come
first! I t is important to be able to answer g3-
g4 with . . . h5-h4!
48 lZJa2 l::.a5
49 g4 h4
The trap has snapped shut!
50 'it>d2
Probably the only chance. Wh ite intends
play against the h4-pawn with 5 1 gxf5 gxf5
52 l:tf1 fol lowed by l::tf4 . I n the event of 50
lZJc3 lZJxc3 5 1 l::txc3 b5 (and then . . . l::tda?)
Black is effectively a p iece up .
50 . . . �g7
51 .l'::f.f1 .i.e6
I decided not to defend the h4-pawn with the
king , fearing that after 5 1 . . . �h6 52 .l:If4 �g5
53 gxf5 gxf5 54 I::f.c1 it would be attacked by
the rook from g 1 . I nstead of this Black
exploits the departure of the white rook from
the queenside and beg ins decisive action
there . All fu l ly in accordance with the
pr inciple of two weaknesses.
52 l:tf4 ctJb6
53 gxf5
54 .l:!.xh4
gxf5
54 . . . c3+!
The qu ickest way of converting the advan
tage. The pawn moves into a three-fold
attack, but noth ing can capture it. For
example, if 55 ttJxc3 , then 55 . . . lZJc4+ 56
�c1 lZJxe3 is decisive .
55 '.t>c1 .lli.b3
56 lZJxc3
57 '.t>xc2
.lli.xc2
.l:!.xa3
I was expecting my opponent to capitulate,
but unexpectedly White sacrificed h is knight.
58 ttJxe4 fxe4
Technical Procedures in a Grandmaster Battle lZJ 1 57
59 .i.xe4
60 .i.d3
61extra rook I could have
continued playing , but it is my conviction
that in a won position one should not do th is .
Ti redness after six hours' play sometimes
leads to mistakes such as 61 . . . lt:'Jd5?? 62
:Ih7+ .
61 . . .plan : 39 h4 with the
idea of 40 g4, 41 .l::!.h2 and so on - the wh ite
rook wi l l occupy an ideal position to the rear
of the passed h-pawn .
As you see, Wh ite has numerous tempting
possib i l it ies. In order to make the correct
choice, it wi l l be essential to take into
account the opponent's cou nterplay.
1 60 � Lessons from One Particular Endgame
Let's take them in order and beg in with 39
c5. If 39 . . :lt>e5?! there fol lows 40 l:Id7. I n the
event of 39 . . . l:.b4?! noth ing is given by 40
l:.d6+ 'it>e5 41 l:.xc6 l:.xe4+ and 42 . . . l:.a4,
but 40 �f4 ! fol lowed by 4 1 l:.d6+ is far more
dangerous. The best defence was sug
gested by Alexander Alekh ine: 39 . . . l:.b5! 40
.l:!.d6+ 'it>e5 41 l:!.xc6 (4 1 l:!.d7 l:!.a5 or
41 . . . l:!.xc5) 4 1 . . . l:!.a5 . With such an increase
in the activity of his pieces, Black should not
lose.
Now let's examine 39 Wd4. Obviously, the
king cannot be a l lowed to go to c5 .
39 . . . 'it>d6? 40 e5+ does not help , and
therefore the reply 39 . . . l:.d8+ is forced . After
41 'it>c3 the threat of c4-c5 has become
more serious, s ince now the c5-pawn may
be defended by the king . However, it is not
hard to foresta l l Wh ite's p lan : 41 . . . .l:!.h8! 42
h3 .l:!.h5 (42 . . . .l:!.h4 also comes into consid
eration) . The rook is wel l placed on the 5th
rank - it controls the c5-square (if 43 'it>b4
a5+) and is able to attack any of the enemy
pawns. I t is evident that Wh ite has not
ach ieved much .
It remains to verify 39 h4. The reply
39 . . . .l:!.h8! suggests itself (39 . . . f5? 40 exf5+
is bad for Black). Wh ite plays 40 g3,
preparing 41 l:!.h2 and 42 g4 . How can th is
p lan be countered? Black is saved by the
same rook manoeuvre: 40 . . . llh5! 41 llh2
l1a5 ! . Now 42 g4? is unfavourable because
of 42 . . . 'it>e5 43 h5? l:.a3+ and 44 . . . l:Ixa2+,
and if 42 �f4 there fol lows 42 . . .f6 ! , prepar
ing in the event of g3-g4 to exchange the
opponent's most dangerous pawn by . . . g6-
g5+ ! .
We have establ ished that White does not
ach ieve anyth ing with the d i rect implemen
tation of any of our intended plans. How can
he nevertheless continue playing for a win?
Note that everywhere Black was saved by
the switch ing of h is rook onto the 5th rank .
Let's remember about prophylaxis and try to
find a way to prevent the opponent's main
defensive idea.
Alekh ine suggests the surprising move 39
h3! ! . Now if 39 . . . l:Ih8 the h-pawn is not
hanging and Wh ite repl ies. 40 c5. After
40 . . . I:th4 the reply 4 1 .l:!.d6+ 'it>e5 42 l:Ixc6
.l:!.xe4+ and 43 . . . .l:!.a4 is unconvincing, but 41
l:Id8! is very strong . At the same time Black
must now seriously reckon with 40 'it>d4 , for
example: 39 . . . .l:!.b 1 (b4) 40 '>t>d4 '>t>d6 41 e5+,
or 39 . . .f6 40 'it>d4 .lld8+ (40 . . . '>t>d6 41 c5+
'it>e6 42 Wc4) 41 �c3 .Ub8 42 c5 'it>e5 43
l:Id6 with an obvious advantage. 39 . . . 'it>e5 is
dangerous because of 40 l:Id7 . There only
remains 39 . . . c5 40 l:Id5 ( if 40 h4, then
40 . . . .llb4 ! , but not 40 . . . l:Ih8 41 g3 l:Ih5 42
.l:!.h2 , and the 5th rank is too short) 40 . . . .l:!.b2
41 g4 (4 1 .l:!.xc5 .l:!.xg2 42 l:Ia5 is also good )
4 1 . . . .l:i.xa2 42 l:!.xc5 .l:!.a3+ 43 'it>d4 .l:!.xh3 44
.l:!.a5 with excel lent winn ing chances for
Wh ite .
It is s ign ificant that Jose Raul Capablanca
a player with bri l l iant i ntuit ion - was unable
to come to the correct decision , suggested
by Alekh ine - a player with a tota l ly d ifferent
way of th ink ing . A move such as 39 h3 ! !
cannot be cal led intu itive , based on 'general
considerations' - it cou ld be found only after
a deep and very concrete penetration i nto
the secrets of the position .
Many years ago I was help ing Botvinn ik by
giv ing some lessons at h is schoo l . On one
occasion , at the request of Mikha i l Moisee
vich , I prepared an extensive endgame
lesson for the young Garry Kasparov, which
included in particular an independent analy
s is of the Capablanca-Aiekhine ending.
Garry found another way of foresta l l ing the
switch ing of the black rook to the 5th rank
the move 39 g3! ! . It appeals to me perhaps
even more than Alekh ine's recommenda
tion , s ince it conta ins an add itional active
idea : 40 h4 ! . And there do not appear to be
any drawbacks : for example, if 39 . . . g5 there
Lessons from One Particular Endgame ctJ 1 6 1
i s a pleasant choice between 4 0 h 4 a n d 40
l:.f2 with the threats of 41 .l:.f5 or 4 1 d4? c5+ .
42 . . . f6!
The main danger has been e l iminated - if
43 g4 there is the reply 43 . . . g5+ ! . The play
takes on a manoeuvring character. Capa
blanca ski lfu l ly sets his opponent one
problem after another, so that Alekh ine is
required to defend with exceptional care .
43 l:tc2 l:f.e5
Otherwise after 44 c5 the rook would have
been cut off from the kings ide and would no
longer have been preventing White from
playing g3-g4 and h4-h5 .
44 c5
A double-edged move , but otherwise the
position cannot be strengthened . Wh ite
restricts the mobi l ity of the enemy rook , but
his own rook wi l l be tied to the defence of the
c5-pawn.
44 . . . l:th5
45 l:.c3
Threaten ing an exchange of pawns advan
tageous to White : 46 lta3 l::txc5 47 l:.xa7.
45 . . . aS!
46 .l:.c2 l:te5
47 l:lc3 l:1h5
48 g7!
Wh ite's subtle manoeuvres have forced the
black king (which is obl iged to control the
g6-square) to move away from the centre .
Capablanca sees that the most appropriate
moment for transforming h is advantage has
arrived . He g ives up h is extra pawn , but
activates h is rook to the maximum and
drives the opponent's k ing onto the back
rank .
51 .l:td4! I!xc5
162 Lessons from One Particular Endgame
52 l:td7+ 'it>f8
52 . . . Wh6? would have been risky: 53 I:!.f7
.ti.c4 54 'it>f3 �c1 55 'lt>e3 .l:i.f1 56 We2 l:ta 1 57
�f6 .l:!.xa2+ 58 �3.
53 'lt>f4
If 53 l:ta7 there is the reply 53 . . . l:tc4 !? , but
after the move in the game Black could have
played 53 . . . .l:i.c2 !? .
53 . . .
54 l:ta7
55 a4!
'it>g8
'it>f8
'it>g8
White has done everyth ing possible to
strengthen his position and now is the time
for decisive action . The logical consequence
of h is preced ing strategy would have been
the variation 56 'it>e3 ! .ti.c3+ 57 'it>d4 .l:!.xg3 58
.ti.xa5 cj;;f7 (in the event of 58 .. J:tg4 noth ing
is g iven by 59 .ti.a7 .l:!.xh4 60 a5 .ti.h5! , but the
reply 59 h5 is unp leasant) 59 .ti.a8 or 59 h5.
I n Alekhine's opin ion , Black is able to hold
the position , but at any event he would have
had to defend accurately.
Unfortunately, Capablanca d id not want to
sharpen the play and he chose a d ifferent
p lan , which leads by force to a draw.
56 g4 g5+!
57 hxg5 l:txg5!
Of course, not 57 . . . fxg5+? 58 We3 - there is
no reason to g ive the opponent a passed
pawn .
58 .i::ta6
59 We3
60 'lt>d4
61 ltxc6
62 .l:!.c5
.l:!.c5
'it>f7
.ti.g5
.l:!.xg4
.ti.g5!
In th is posit ion the players agreed a draw in
view of the variation 63 l:txg5 fxg5 64 'it>e5
e5 (66 e5?! We8 ; 66 Wd7 'iii>f6)
66 . . .'it>g6! .
With what topic should the ending of the
Capablanca-Aiekhinegame be l inked? Af
ter a l ittle thought you wi l l see that there is
no clear answer - in the process of studying
the ending various aspects have come to
l ight, identica l ly important for the practical
player. Let's remember what we have seen:
1 ) Excellent example of a practical rook
endgame. Among the numerous evalua
tions and methods typical of this type of
ending , with which the two players operated ,
I wi l l s ingle out a comparatively less trivial
idea , which is very clearly expressed here.
An open l ine , for wh ich a rook should a im ,
may be not only a fi le , as usua l , but
sometimes also a rank .
2) Model example of accurate defence. I t
is i nstructive to fol low how Alekh ine, without
losing his presence of mind in a d ifficu lt
situation , move by move patiently resolved
the problems facing h im .
3) Various aspects of the problem of
converting an advantage. Here we can
mention : the importance of looking for and
foresta l l ing the opponent's cou nter-chances
(at the very start of the ending) ; the
maximum strengthening of Wh ite's position
before changing the pattern of the play; the
timely transformation of an advantage (the
Lessons from One Particular Endgame lZJ 1 63
5 1 st move) ; fi nal ly, the need at some point
(the 56th move) to abandon positional
manoeuvring and choose a concrete course,
involving precise calcu lation .
4) Demonstration of the importance of
prophylactic th inking. Without it, of course,
it is not possible to find the bri l l iant solution
to the position on the 39th move . And
subsequently too Alekh ine's defence was
based on taking account of al l the oppo
nent's active plans and forcefu l ly opposing
them.
5) Grounds for reflection about chess
players with an intu itive way of th inking.
We have seen which decisions are d ifficult
for them or a ltogether inaccessib le . The
conclusion suggests itself, that even if you
possess splendid intuition , you should de
velop in yourself the abi l ity to constantly
delve into the concrete deta i ls of the position
and if necessary to accurately ca lculate
variations .
For a chess player it is very important to
evaluate objectively the strengths and weak
nesses of a forthcoming opponent. A source
of such evaluations is provided by an
analysis of games played by h im. Some of
them wi l l prove especia l ly i nformative.
I n the 1 920s Alekh ine was preparing for h is
due l for the world crown against Capablanca .
This is what he recorded for h imself after the
New York tournament of 1 924:
I took home with me from this tournament
one valuable moral victory, and that was the
lesson I learned from my first game with
Capablanca, which had the effect of a
revelation on me. Having outplayed me in
the opening, having reached a won position
in the middlegame and having carried over a
large part of his advantage into a rook
ending, the Cuban then allowed me to
neutralise his superiority in that ending and
finally had to make do with a draw. That
made me think, for Capablanca had cer-
tainly been trying very hard in this game, so
as to draw nearer to Dr. Lasker, who was in
the lead, and who had won against me the
previous day. I was convinced that if I had
been in Capablanca 's position I should
certainly have won that game. I had finally
detected a slight weakness in my future
opponent: increasing uncertainty when con
fronted with stubborn resistance! Of course I
had already noticed Capablanca committing
occasional slight inaccuracies, but I should
not have thought that he would be unable to
rid himself of this failing even when he tried
his utmost. This was an exceedingly impor
tant lesson for the future!
Later, in a famous a rticle 'The 1 927 New
York tou rnament as a prologue to the battle
in Buenos Aires for the world championsh ip ' ,
Alekh ine once again emphasised the ro le
that the game with Capablanca had played
for h im :
This game, incidentally, was the starting
point for my understanding of Capablanca 's
chess individuality.
I wi l l a lso g ive some other assessments by
Alekh ine of the style of h is h istoric oppo
nent, wh ich are confi rmed by the ending we
have examined . They may seem exces
sively sharp, but to some extent th is is
expla ined by the very tense personal rela
t ions which developed between the two
champions. But objectively these assess
ments seem to me to be just (of course, only
'on a grand scale' - taking i nto account the
very h igh standard of play in question) .
. . . Capablanca is by no means an excep
tional master of the endgame; his skill in this
stage of the game is mainly of a technical
character and other masters in certain fields
of the endgame surpass or used to surpass
him (for example, Rubinstein in rook end
ings).
. . . In Capablanca's games with the years
one observes increasingly less delving into
164 � Lessons from One Particular Endgame
the details of the position, and the reason for
this is his unshakeable (I am talking all the
time about the period before Buenos Aires)
confidence in the infallibility of his intuition.
The saddest thing for Capablanca is that this
system of his of operating with 'good' moves
almost without exception proved sufficient,
since to a great extent he was opposed in
the positional sense by a more or less
helpless weapon. On account of this 'lack of
punishment' in employing not the best
moves, he, on the one hand, got out of the
habit of that concentration of thought during
a game, which alone can give a guarantee
against possible elementary oversights, and
on the other hand- his self-confidence grew
immeasurably and turned almost into self
worship . . .
ltJ 1 65
Mark Dvoretsky
G randmaster Tech n i q ue
I
n August 2005 I gave a lectu re in the
London Chess Centre and showed the
fol lowing position , taken from the magazine
64 Shakhmatnoe obozrenie - the ending of
the game was publ ished there with notes by
the winner Evgeny Najer.
Yandemirov - Najer
Russian Club Championsh ip ,
Dagomys 2004
Grandmaster Jonathan Rowson , who was
present at the lectu re , surprised me by
remarking that this position is reached more
or less by force in one of the modern
open ing variations, and that he h imself had
once played it .
1 e4 c5 2 tt:lf3 d6 3 .i.b5+ .i.d7 4 .1Lxd7+
'i'xd7 5 c4 tt:lf6 6 tt:lc3 g6 7 0-0 .i.g7 8 d4
cxd4 9 tt:lxd4 0-0 1 0 f3 IreS 1 1 b3
11 ... d5 !
A clever blow i n the centre , fi rst employed , I
th ink , by Vasi ly lvanchuk against Alexander
Delchev in the 2003 European Champion
sh ip . In th is way Black solves his opening
problems - in fact, it is now h is opponent
who has to act carefu l ly, to avoid getting into
d ifficu lt ies.
12 exd5 ( 1 2 e5 is dubious in view of
1 2 . . . tt:leB 1 3 tt:lxd5 e6 1 4 tt:lc3 .1Lxe5)
1 2 . . . tt:lxd5 1 3 tt:lxd5 e6 14 .i.h6 exd5 1 5
.1Lxg7 'it>xg7 1 6 cxd5 ( 1 6 tt:lb5!?) 1 6 . . . 'iVxd5
1 7 tt:le2
Here peace was concluded in the game
Gdanski-Kempinski from the 2004 Pol ish
Championship , and a move (and a year)
earl ier i n a game Yandemirov-Biryukov. And
in genera l , looking in a computer database, I
saw that in most of the games where th is
variation occurred there was effectively no
play - the contestants agreed a draw
somewhere between the 1 1 th and 23rd
moves.
1 66 � Grandmaster Technique
In the summer of 2003 I publ ished an article
in the Russian newspaper Shakhmatnaya
nedelya and on the Chesscafe site (it was
also publ ished somewhere else on the
I nternet) with a suggestion that a ru le should
be introduced forbidding conversations be
tween players during play and , hence,
premature draw agreements . Two years
later my suggestion was successful ly adopted
at the super-tournament in Sofia . Its un iver
sal adoption would , I am sure, not only
lengthen games, butalso lower the percent
age of draws. As we wi l l now see , even in
such a qu iet and seemingly l ifeless situa
tion , where a draw is indeed the most
probable outcome, it is possible to seek
resources and pose problems for the oppo
nent.
17 .. .'it'e5 1 8 'it'd4 tt:Jc6 1 9 'if xeS+ tt:Jxe5 20
�ac1
We have reached the position in the fi rst
d iagram. The at fi rst sight mysterious move
found by grandmaster Najer is the strongest
- it was thanks to it that he won the game.
20 . . . b6! !
Let us try and reproduce Black's logic. F i rst
he probably checked 20 . . . tt:Jd3 and real ised
that after 21 .l:r.xc8 .l:r.xc8 22 .l:r.d 1 the position
was equa l . Then h is attention was drawn to
the possibi l ity of 20 . . . .l::.xc1 21 �xc1 �d8
with the idea of 22 . . . �d2 . Najer is an
experienced player and he knows that in the
fi rst instance you must check active repl ies
by the opponent - in the given case 22 �c7 .
There appears to be noth ing better than
22 . . . .l:r.d2 23 �f2 .l:r.xa2 24 .l:r.xb7, but here
Black's advantage is of a purely academic
natu re, and he has practica l ly no chances of
success. As was confi rmed by the game Li
Ruofan-Rowson , 2004 .
But after . . . b7-b6 in this variation Black
would remain a pawn up, s ince from a2 his
rook defends the a7-pawn .
21 .l::tfd1 ? !
After the game Valery Yandemirov sug
gested that he should have played 21 f4. Of
course, advancing the f-pawn is not some
th ing that one wants to do - one can decide
on this move only after employing prophy
lactic th inking and clearly appreciating the
danger threaten ing White.
21 . . .
22 .l:r.xc1
.l:r.xc1
.l:r.d8
Now after 23 �c7 .l:r.d2 the only way of
avoid ing the loss of a pawn is 24 tt:Jc3 (bad is
24 tt:Jc1 ?? .l:r.d 1 + 25 �f2 �xc1 ) , wh ich al lows
the unpleasant pin 24 . . . .U.c2 . The situation
after 25 tt:Jb5 .t!.xa2 26 �xa 7 �b2 looks
dangerous for Wh ite: the b3-pawn is under
attack, and he has to reckon with the
manoeuvre . . . tt:Jd3-f4(e 1 ) .
He shou ld probably have restricted h imself
to the accurate move 23 .l:r.c2 . But one does
not want to place the rook passively,
especia l ly s ince after 23 . . . .l:r.d 1 + 24 'it>f2
tt:Jd3+ the king has to be moved to the side -
25 'iitg3 , s ince in the event of 25 'it>e3?! tt::le1
26 .l:r.c7 tt:Jxg2+ 27 'it>f2 .U.d2 28 .t!.xa7 ctJf4 29
'it>e3 .l:r.xe2+ 30 'it>xf4 l:!.xh2 White comes out
a pawn down .
It is qu ite probable that after 23 �c2 , and
perhaps also after 23 l:!.c7 , the position
would objectively have remained drawn . But
it is one th ing to calmly analyse at home,
Grandmaster Technique ttJ 1 67
and with the aid of a computer, and qu ite
different to take a decision at the board . A
forced draw is not apparent, defending is
unpleasant - in such a situation it is easy to
lose your bearings.
23 tt::lc3?
A serious mistake! Wh ite wants to exchange
the rooks, but fa i ls to take account of the fact
that the black k ing wi l l be the fi rst to reach
the centre . In a knight ending a more active
king is a very important factor.
23 . . . �d2
24 .Ue1 ?!
Obviously, when he made h is previous
move , Wh ite was intending the manoeuvre
.l::!.e 1-e2 . Of course , the enemy rook on the
2nd rank cannot be tolerated , but he should
have exchanged the rooks in a sl ightly more
favourable way: by 24 �f1 (with the same
idea : �e 1 -e2) 24 . . . tt::ld3 (24 . . . f6? 25 tt::le4)
25 .l:td 1 .l:txd 1 + 26 tt::lxd 1 �f6 27 tt::lc3 �e5 . I n
the game this same position was reached
with Black to move .
White can make use of h is extra tempo by
playing 28 g3 (28 'it>e2 is evidently weaker:
28 . . . tt::lf4+ 29 �f2 a6 30 g3 'it>d4! 31 tt::le4
l2Jd5) . But here too Black has a pleasant
choice between 28 . . . a6 29 'it>e2 'it>d4 30
Wd2 tt::le5 31 tt::le2+ �c5 32 �e3 �b4 and
28 . . . �d4!? 29 tt::lb5+ �c5 30 tt::lxa7 tt::lc1 31
�e1 tt::lxa2 32 �d2 tt::lb4 (and 33 . . . tt::ld5) with
an appreciable advantage.
24 . . . tt::ld3
25 .Ue2
26 tt::lxe2
27 tt::lc3
28 �f1
l:txe2
�6
'it>e5
a6
From the previous note it fol lows that
28 . . . �d4 was also possib le, but in th is
situation the move in the game is , of cou rse,
more accurate .
29 �e2
30 �d2
tt::lf4+
If 30 �2. then 30 . . . �d4 is strong, as is the
suggestion of Carsten Mu l ler: 30 . . . f5 !? , and
if 31 g3? �d4 ! .
3 0 . . .
3 1 tt::la4
32 'it>e2
33 tt::lc5
33 . . .
tt::lxg2
tt::lh4
b5
a5!
Najer was not satisfied with the variation
33 . . . �f4 34 tt::lxa6 tt::lxf3 35 tt::lc7 tt::lxh2 36
tt::lxb5, i n which the play becomes sharper.
I n advancing h is a-pawn he undoubtedly
foresaw the tactical subtlety on the fol lowing
move , which sign ificantly faci l itates the
conversion of his advantage .
168 Grandmaster Technique
34 a4 'it>d5!
I n this way Black ensures the safety of h is
important a5-pawn .
35 tt:Jb7 'it>c6
36 tt:Jd8+
36 tt:Jxa5+ Wb6 37 b4 bxa4 was completely
hopeless for Wh ite .
36 . . . 'it>c5
37 tt:Jxf7
38 bxa4
39 tt:Je5
bxa4
'it>b4
'.txa4
Black is a sound pawn to the good . The
outcome is decided .
40 'it>d3 'it>b3 41 tt:Jc4 a4 42 tt:Ja5+ Wb4 43
tt:Jc6+ 'it>c5 44 tt:Je5 'it>d5 45 tt:Jc4 tt:Jxf3 46
tt:Jb6+ We5 47 'it>e3 tt:Jxh2 48 tt:Jxa4 'it>f5,
and Black won .
I t was pleasant for m e to read the fol lowing
comment by Najer about the move 20 . . . b6 ! ! :
A useful move, from which one can identify a
pupil of Mark Dvoretsky or at least a careful
reader of his books.
Evgeny Najer twice participated in my
tra in ing sessions and has stud ied my books
- obviously, his work has not been in va i n . I
th ink that the grandmaster's comments
were evoked by the close s imi larity of this
ending with one of the examples g iven in my
book School of Chess Excellence 3 -
Strategic Play, which I should now l i ke to
show you .
(see diagram)
Black's positional advantage is determined
by the active placing of his pieces. But it is
rather frag i le : the sl ightest inaccuracy, and it
wi l l evaporate . Which is what happened in
the game: 28 . . . h5? 29 '.tc1 ! l::tf5 30 l::txf5
tt:Jxf5 31 tt:Je4 g4 32 tt:Jc5+ We5 33 tt:Jxb7
tt:Je3 Draw.
A very important principle in the conversion
of an advantage is the maximum restriction
Stean - Hort
Biel 1 98 1
of the opponent's possib i l it ies, and the
suppression of any cou nterplay, any useful
operations a imed at improving his own
position . To put th is pr inciple successfu l ly
i nto effect, use must be made of 'prophylac
tic th ink ing ' .
Let us ask ourselves what White wants to
play. H is choice is l im ited . There is no point
in attacking the kn ight : 29 �d2 lle1 + 30 'it>b2
�e5. Obviously the only sensible operation
is to bring the k ing to the centre : �b1 -c1-
d2-d3 . I t is th is that shou ld be prevented .
28 . . . l::te 1+ !
29 'it>b2 h5
Grandmaster Technique ltJ 1 69
Having foresta l led the opponent's i ntention ,
Black can now calmly strengthen h is posi
tion . I t is not easy for Wh ite to defend. For
example, if 30 �f8 there fol lows 30 . . . .l:!.g 1 3 1
.!:!.f2 �e5 32 ll'le2 � e 1 33 ll'lxd4 'i£txd4, and
the dominating position of B lack's king
guarantees h im a great advantage in the
rook ending .
But a completely d ifferent interpretation of
the position is a lso admissible. The black
king is far more active than its wh ite
opponent, which can be especial ly percepti
ble in a pawn or knight ending (Mikha i l
Botvinn ik once remarked : 'A knight end
game is the same as a pawn endgame' ) .
Artur Yusupov suggested 28 . . . .l:i.f5 !? . Wh ite
repl ies 29 �xf5 (29 �d2 .l:i.f1 + 30 �b2 '.te5
is unfavourable for h im) .
Now Black would l i ke to captu re with the
knight, but after 29 . . . ll'lxf5 30 lbe4! g4 3 1
lt:Jc5+! (weaker i s 3 1 lbg5+ �e5 3sometimes leads to interesting
results, afford ing creative pleasure .
Once when I was looking through Chess
lnformator, an ending annotated by one of
my pupi ls , Alexey Dreev, caught my eye.
Dreev - Moskalenko
USSR Young Masters Championsh ip ,
Lvov 1 985
A draw resu lts from 1 l:!.e 1 ? a2 2 lla 1 '1td7 ,
or 1 .l:l.e3? .l:.b4+ 2 '1tf5 l:.a4 3 .l:.e 1 a2 4 h5 (4
.l:ta 1 'it>d7) 4 . . . a 1 'ii' 5 .l:txa 1 l:txa 1 .
1 l:te6+ !
The black king faces a choice . I n the game it
moved towards the kingside, but by placing
his rook beh ind the passed pawn White won
easi ly: 1 . . . '1td7 2 lla6 a2 3 g5 '1te7 4 '1tg4
'1tf7 5 '1th5 llh2 6 .Ua7+ '1te6 7 '1tg6 l:tb2 8
h5 .l:tb8 9 h6 l:tg8+ 1 0 'it>h5 '1tf5 1 1 l:.a5+
B lack resigned .
1 . . . '1tb5
In lnformatorthe fol lowing analysis is given :
2lie5+ 1 ! '1tb6 (2 . . . 'it>b4 3 l:te8 '1tb5 4 J:tb8+ !
and 5 .U.a8) 3 .l:te3! (a l l the exclamation
marks are by the annotator) 3 ... .l:!.b4+ 4 c;tf5
l:.a4 5 h5 a2 6 l:Ie1 a1'ili' 7 .l:txa1 l:.xa1 8 h6
'1tc7 9 g51 '1td7 (9 . . . lih 1 1 0 g6! l:txh6 1 1 g7
l:lh5+ 12 c;tf4 lih4+ 1 3 'it>f3 l:.h3+ 14 'it>g2)
1 0 h7l::!.h1 1 1 g6 and wins.
U nfortunately, th is entire variation is a
comedy of errors , at the basis of wh ich is
Dreev's natura l , but in the g iven instance
incorrect striving to defin itely place his rook
behind the opponent's passed pawn .
After 9 g5 B lack saves the game by 9 . . . l:th 1 !
(the place for the black rook is to the rear of
the more advanced pawn) 1 0 g6 llh5+!
(remember the Fridstein-Lutikov ending) , or
1 0 '1tg6 '1td7 1 1 '1th7 '1te6 1 2 g6 llg 1 ! (and
this we have a l ready seen in one of the
variations of the Mar6czy-Tarrasch ending) .
Wh ite should not g ive up h is rook. I nstead of
5 h5? he wins by 5 .l:te1 ! a2 6 l:r.a1 '1tc7 7 h5
'1td7 8 h6.
But before this B lack went wrong : he could
have drawn by 4 . . . a2! ( instead of 4 . . . l:.a4 ? ) 5
l:ta3 l:.b5+ 6 '1tg6 l::l.a5 7 :Xa2 .l:!.xa2 8 h5
'1tc6 .
Even so, the ending is won . On ly, the rook
should not be p laced on e3.
2 l:te1 !
3 .l:.a1
4 h5
5 h6
a2
'1tc6
'1td6
lih2
5 . . .'.ti>e6 6 h7 .l:.b8 7 llxa2 '1tf6 8 l:.h2 llh8 9
l:Ih6+ '1tg7 1 0 '1tg5 .
6 '1tg5(f5)
Also possible is 6 h7 l:txh7 7 :Xa2 .l:th8!? 8
l:.a4! or 8 l:ta6! fol lowed by 9 'i.t>g5, but not 8
g5? '1te6 and not 8 l:te2? lif8+ 9 '1tg3 l:tg8
with a d raw.
6 . . . '1te7
7 '1tg6
7 h7 is a lso strong .
7 . . . '1tf8
8 h7
And White wins.
I t remains to analyse the rook check at e5,
which i n fact deserves not two exclamation
marks, but more probably one question.
mark. Let us verify 2 :e5+?! '1tb4 !? .
How to Study the Endgame l2:J 19
If 3 J:le8? B lack plays not 3 . . . 'it>b5? , but
3 . . . a2 ! . Then 4 l:ta8 .l:!.c2 ! with the threat of
5 . . . l:lc4+, 6 .. Jk5(c3)+ and 7 . . J:ta5(a3) leads
to an immed iate draw. And if 4 l:tb8+, then
4 ... Wc4 (or 4 . . . 'it>c3) 5 l:ta8 l:tb4! (threaten
ing interference: . . . '.tb3+ and . . . .l:ta4) 6
�a2 '.tb3+ 7 '.t>f5 'it>xa2 8 h5 .l:!.b5+! (the
king must be driven to the unfortunate h4-
square - this is not d ifficult to ach ieve, using
the rook's long range) 9 �g6 l:!.b6+ 1 0 �g5
l:lb5+ 11 Wh4 l:i.b 1 ! (and now the rook
moves to the rear of the pawns) 12 h6 ( 1 2 g5
'it>b3 13 g6 l:tg 1 ! ) 12 . . . .l:!.h 1 + 1 3 'it>g5 'it>b3 1 4
'it>g6 'lt>c4 1 5 g5 �d5 1 6 'lt>h 7 'it>e6 1 7 g6
J:lg 1 ! with a draw.
After 2 J:le5+?! 'it>b4!? a l l the same the rook
has to be returned to the fi rst rank . But then
it is clear that the check was pointless -
White is forced to ca lculate the lengthy
variation 3 .l:te1 a2 4 .l:ta 1 'it>b3 5 h5 l:!.b1 6
l::!.xa2 '.txa2. Now it is incorrect to play 7 h6?
J:lh 1 8 g5 '.tb3 with a draw. B ut as yet the win
has not been missed : 7 g5! .l:th 1 8 g6! or
7 . . J:tf1 + 8 'lt>g4! (8 'it>e5? l:th1 ! ) 8 . . . 'it>b3 9 g6.
In 1 976 the USSR Championship was held
in Moscow. I n the very fi rst round my friend
Boris Gulko adjourned h is game against
grandmaster Taimanov in a compl icated
rook end ing. Before the resumption h e
asked m e to join in the analysis.
In order to figure out precisely some very
intricate variations, we had to turn to the
theory of the rook endgame with f- and h
pawns. The elementary information about
these endings, wh ich was g iven above, was
not sufficient for us. However, the necessary
positions cou ld not even be found i n books
on the endgame, so that we had to
supp lement 'officia l ' theory with our own
analyses. Here is a very important key
position that we found (see diagram).
White's king is cut off on the back rank. Does
th is mean that he is bound to lose? It turns
out that the answer is no. After a l l , the black
king too is not wel l placed - it is cut off on the
h-fi le.
1 'lt>g1 ! h4 2 l:tg8 f3 3 llf8 'lt>g3 (or 3 . . . l:i.g2+
4 'it>f1 'lt>g3 5 l:tg8+ 'lt>h2 6 l:If8) 4 l:tg8+ 'it>f4
5 l:tf8+ 'lt>e3 6 l:te8+ 'it>d3 7 l:td8+ 'it>e2 8
l:te8+ 'it>d1 9 l:tf8(e3) with a draw.
However, if it is Black to move he wins, by
depriving the enemy king of the important
g 1 -square .
1 . . . Wh2! 2 l:!.g8 ( in the event of 2 l:!.f7 or 2
J:th7 Black wins by 2 . . . �g3) 2 . . . h4 3 l:!.g7 (3
.l:tg4 h3 4 l:txf4 'lt>g3 5 .Uf8 .l:tb1 + 6 �e2 h2)
3 . . . h3 4 l:tg8 f3 (or 4 . . . l:tg2) and wins.
Taimanov - Gulko
44th USSR Championsh ip , Moscow 1 976
20 � How to Study the Endgame
42 . . .
The sealed move.
43 l:ie2+
44 gxhS
45 l:ie5
'it>e3
'it>xf3
gxhS
�g4
Weaker is 45 . . . .l:txb2 46 ltxh5! 'it>e4 47 �h8,
and White should ga in a draw.
46 l:txd5 �xb2
47 'it>e1
After 4 7 .Ud4+ 'lt>g3 48 'lt>e 1 f5 49 l:id5 f4 50
�g5+ (50 l:ixh5 .Ub 1 + 5 1 'it>d2 f3) Black does
not play 50 . . . 'it>f3? 51 l:ixh5, but s imply
50 . . . Wxh4! 5 1 �g8 'lt>h3 52 'it>f1 �h2 ! ,
achieving a winn ing position , s ince h i s king
succeeds in reach ing h2.
But now we have reached the cu lmination of
the entire endgame.
The natural move 47 . . .f5? is a m istake. After
48 Wf1 .Uh2 49 �g 1 .Uxh4 50 �g2 f4 51 l:id3
a 'normal' ( i .e . drawn) position with f- and h
pawns is reached , and with the black rook
bad ly placed . If 48 . . .f4 there fol lows 49
�g5+ 'it>xh4 50 l:ig8 'it>h3 51 'it>g 1 with a
draw, since the wh ite king has reached g 1 .
And if 49 . . . �3 ( instead of 49 . . . 'it>xh4) , then
50 �g 1 l:ib1 + 51 'it>h2 'it>f2 52 �xh5 f3 53
lta5 'it>f1 54 �g3 f2 55 .Ua2 �b3+ 56 �g4,
and White gives up h is rook for the f-pawn .
47 . . . 'itoxh4!
48 .Ud7?
We assumed that 48 .Uf5 also d id not help in
view of 48 . . . �b7 49 'it>f1 (49 'it>f2 'it>g4 50 l:tf6
h4) 49 . . . 'itog4 50 ltf2 .l:!.b 1 + ! 5 1 'it>g2 f5 . To
prevent the king from being pushed onto the
back rank , the wh ite rook must guard the
2nd rank , where it is too passively placed .
Black wins easi ly, by advancing h is pawns.
Alas, a mistake crept i nto our analysis. By
continu ing 49 .Ua5! ( instead of the losing
king moves) White exploits the long-range
power of h is rook and draws by d riving the
opponent's king to a less good position :
49 . . . 'it>g4 50 �a4+! 'it>g3 5 1 lta3+ 'it>g2 52
�a2+ 'itg 1 53 'ite2 ! (now in a number of
variations it becomes possible to shut the
king on the edge of the board after playing
the rook to the g-fi le) 53 . . . '1t>g2 54 'ite1 + ! .
48 . . . f6!
We thought that this subtle move was the
only correct one, s ince in the event of 48 . . . f5
49 .Ug7 'itoh3 50 'it>f1 the wh ite k ing succeeds
in reach ing g1 (50 . . . 'ith2 does not work in
view of 5 1 .Ug5) . But here too we were
wrong! After 50 . . . h4! 51 �g 1 Black has the
winn ing resource 51 . . . .Ub4 ! , which he did not
have in our basic position - there the pawn
was a l ready stand ing at f4 . 51 .Ug8 'it>h2!2 lbxh7
lt:Je3 33 �c1 lbxg2 34 'it>d2 �f4 35 �e2
tt:lh4) 31 . . . '.te5 32 ll'lxb7 he can hardly hope
to win .
This means that he must play 29 . . . �xf5 ! ,
intending 3 0 . . . 'it>e5 a n d 3 1 . . . ll'lf5. Events
can then develop roughly as fol lows : 30 '.tc1
We5! 31 'it>d2 ll'lf5 32 Wd3 (if 32 �e2 or 32
g3, then 32 . . . �d4 is strong ) 32 . . . lbh4 33 g3
tt:lf3 34 h3 ll'lg1 35 h4 gxh4 36 gxh4.
And now 36 . . . �f4 , 36 . . . ll'lf3 or 36 . . . h5 . The
conclud ing position of the variation is very
promis ing for B lack. But is it won? Couldn't
Wh ite have played more accurately some
where earl ier? I t is clear that here everyth ing
hangs by a thread : the sl ightest add itional
improvement to the defence, and the game
wi l l end i n a d raw.
Black's play can be improved . One is struck
by the fact that he delayed sl ightly - h is
kn ight d id not immediately reach the neces
sary square f5 . Of course, it is desirable to
capture on f5 with the knight . It is th is that
expla ins the recommendation by Vadim
Zviagintsev: 28 . . . h6 !? (a mysterious move
at fi rst sight, wou ldn 't you agree! ) 29 �c1
.l:i.f5.
After 30 llxf5 l2Jxf5 3 1 'it>d2 (31 l2Je4 is now
pointless - the g5-pawn is defended)
31 . . . ll'lh4 Black must win .
However, the exchange on f5 i s not essen
tial - 30 lbd 1 ! (but not 30 lbe4 We5) is far
more accurate for White . For example:
30 . . . .l:i.xf2 3 1 ll'lxf2 ll'lf5 32 ll'ld3, intending 33
lbc5+ or 33 �d2 ll'lh4 34 ll'le1 . Little is
changed by 30 . . . We5 31 �d2 �xf2+ 32
lbxf2 lbf5 33 �d3, s ince if 33 . . . lbh4 there is
34 lbg4+ or 34 g3 fol lowed by lbg4+. We
see that the position of the pawn at h6 is far
from idea l .
1 70 � Grandmaster Technique
And yet Zviagintsev's idea is logical - simply
it must be put into effect s l ightly d ifferently. I
suggest another mysterious move - inciden
ta l ly, the same one that Najer made.
28 . . . b6 ! !
29 Wc1 llf5!
White's position is d ifficult : 30 llxf5 tt'lxf5 or
30 tt'ld 1 llxf2 31 lbxf2 tt'lf5 32 tt'le4 h6,
threatening either to attack the g-pawn
(33 . . . tt'lh4 or 33 . . . tt'le3), or to penetrate with
the king onto the 4th rank .
As you see, the s imi larity with the Yande
mirov-Najer endgame is not restricted to the
fact that in both cases the key to the position
was an imperceptible pawn move. Here
there was also the identical nature of the
material and the pawn structure , and also
the highly important role played by the
activity of the king in a knight ending - a
recurrent theme of Black's play in both
examples.
However in the Stean-Hort game funda
mental ly d ifferent approaches to exploit ing
the advantage were possible. Which one do
you l ike more? The fi rst is purely techn ical
(restriction of the opponent's plan ; unhurried
improvement of the position ) , but it does not
demand deep calculation and therefore it
enables time and energy to be saved . The
second approach - the transit ion into the
knight ending - is much more concrete and
requ i res carefu l checking . The qu iet pawn
move on the queenside, preparing the
exchange of rooks in the most favourable
version , can be found only as a result of
delving thoroughly into the secrets of the
position .
There are two aspects to the process of
converting an advantage. On the one hand ,
it demands accuract and methodical play,
and on the other hand - an abi l ity some
where to cut short the manoeuvring , and
find and ca lculate a concrete way to the
goal . I t is not easy to sense which is more
correct at a particu lar moment. I n the given
example both approaches seem to me to be
equal ly good , but th is does not often occur.
I I
I t can happen that a position looks (and
i ndeed is) completely won , and yet a player
does not manage to convert it i nto a win .
Very often (if of course, th ings do not occur
in severe time-trouble) the cause is a loss of
concentration and insufficient attention to
those few resources wh ich are sti l l avai lable
to the opponent or which suddenly occur. A
very important ski l l , enabl ing th is type of
mistake to be avoided , is prophylactic
th ink ing , about which I write in l iteral ly every
book of mine . The essence of it is that you
need to put yourself in your opponent's
place , constantly asking yourself the ques
tion : what can he undertake , and what would
you do in his place if it were him to move?
I n some of the examples g iven below the
employment of prophylactic th inking ena
bled a player to find the optimal ways of
exploiti ng an advantage, which at fi rst sight
were far from obvious. I n others , by con
trast, such possib i l it ies were not exploited
and the advantage evaporated .
Grandmaster Technique ttJ 1 7 1
Xie Jun - Larsen
Monaco 1 994
Pirc Defence
1 e4 g6 2 d4 �g7 3 tt:Jc3 c6 4 tt:Jt3 d6 5 h3
tt:'lf6 6 a4 0-0 7 ..ie3 tt:'lbd7 8 1i.e2 e5 9
dxe5!?
In the event of 9 0-0 White has to reckon
with the reply 9 . . . d5 !? .
9 . . . dxe5
10 0-0
Instead of th is s imple-minded move, 1 0
tt:ld2 ! was stronger, and if 1 0 . . . it'e7 1 1 tt:Jc4
.!::i.d8, then 1 2 'ii'd6! with advantage to White
(suggested by grandmaster Vlad imir Potkin ) .
10 . . . 'ife7 11 'ii'd3 a5 !?
Black foresta l ls the b ind on h is queen s ide by
a4-a5. He cou ld a lso have considered
1 1 . . . tt:'lh5!? , a typical manoeuvre in such
positions, with the idea of i nvad ing with the
knight on f4 , or even find ing a conven ient
opportun ity to play . . . f7-f5 .
12 'ii'c4 .!::i.e8 1 3 .l:.fd1 h6
Black's last two moves are log ical ly l i nked:
first the rook vacates the f8-square for the
knight (which has just been prevented from
going to c5) , or perhaps a lso for the bishop,
and then control is taken of the g5-point , i n
order to safeguard the f7 -pawn from an
attack by the wh ite knight . However, th is
method of playing is too slow. He should
probably have preferred 1 3 . . . �b4, i ntending
to answer 14 tt:'lg5 with 14 . . . .!::i.f8 , and 14 tt:'ld2
with 1 4 . . . 1i.f8 .
1 4 tt:'ld2
White i ntends to retreat the queen to a2 and
occupy the c4-square with her knight .
14 . . .'iib4 !? was now essentia l , but Bent
Larsen contin ues manoeuvring on the
kingside, underestimating the strateg ic dan
ger facing h im .
1 4 . . . tt:Jh7?! 15 'i¥b3 tt:Jg5 16 tt:Jc4 tt:Jc5?
The last chance was probably 1 6 . . . tt:Je6 ,
with the idea of sacrific ing the exchange i n
t he variation 17 tt:'ld6 tt:'ld4 ! 1 8 1i.xd4 exd4 1 9
tt:Jxe8 iixe8 .
17 'ii'a3 tt:Jce6 18 'ii'xe7 �xe7 19 tt:'lb6 .:b8
20 ..ig4! .U.e8 (not 20 . . . h5? 2 1 ..ixg5)
The knight on b6 is cramping the opponent's
position , and the pin on the h3-c8 d iagonal
is extremely u npleasant. There is no doubt
about Wh ite's enormous positional advan
tage; the only question is how to method i
ca l ly exploit it .
The attempt to force matters , by invad ing
with the rook on d7, is premature : 21 tt:Jxc8?!
l:tbxc8 22 .i:f.d7? (22 1i.xg5 hxg5 23 1i.xe6
.l:i.xe6 24 .l:i.d7 is stronger, sti l l reta in ing a
s l ight advantage) 22 . . . tt:Jd4! 23 .l:i.xb7 .i:f.b8.
1 72 � Grandmaster Technique
Black has activated his pieces, and h is
position is not worse .
Let us ask ourselves what the opponent
would do if it were h im to move . Obviously
he would l i ke to get rid of the p in by . . . h6-h5.
2 1 h4 suggests itself, and after 2 1 . . . lt:lh7 22
lt:lxc8 !? Itbxc8 23 ii.b6 Black's position
remains d ifficult. But Wh ite has to reckon
with 2 1 . . . h5 !? . For example: 22 hxg5 hxg4 ,
or 22 ii.e2 lt:lh7 fol lowed by 23 . . . lt:ld4 - in
both cases with good chances of equal is ing .
Wh ite nevertheless reta ins a sol id advan
tage, by continu ing 22 ii.xh5! lt:lh3+ (22 . . .
gxh5 2 3 hxg5 with ais
hopeless, as is 51 �g5 f4 52 'it>g 1 f3 52 l:if5
'it>g4 (the rook is placed too close to the king
and is unable to g ive checks) . A good
i l lustration of how carefu l ly and cautiously
one should use theoretical knowledge: a
sl ight change in the position , and wel l
known procedures and evaluations may
prove inval id .
49 �1
49 .Ug7 does not help in view of 49 . . . �b5! 50
�f2.l':lf5+ 5 1 'it>e3 (51 'it>g2 l:ig5+) 51 . . . 'it>h3
with an easy win. Black simply advances h is
k ing and h is h-pawn , and then blocks the g
fi le with h is rook, and the wh ite k ing proves
to be too far away from the rook's pawn.
49 . . . 'it>g4
How to Study the Endgame fLJ 21
White's position is hopeless, s ince his king
is cut off on the fi rst rank, and he has been
unable to shut in the opponent's king on the
h-file.
50 .Ug7+ g6
52 .l:!.h8 f5
53 .l:!.g8+ �f6
54 �g1 f4
55 'it'f1 �f5
56 'lt>g1 h4
57 .Ug7 'it'e4
58 .Ua7 'it>f3
59 .Ua3+ �g4
We have a lready met this position , when we
were discussing the basic ideas of endings
with f- and h-pawns.
60 .Ua8
61 .Ug8+
62 .l:!.h8
63 'it>h2
64 .Uxh4
65 .Ua4
White resigned .
'lt>g3
�f3
.Ub1 +
�f2
f3
�f1
After 66 Wg3 f2 67 l::!.a2 .l:!.b3+ 68 �h2 Gu lko
wanted to win in the qu ickest way -
68 . . . .Uf3 ! . The 'scientific' 68 . . . .U.e3 69 .Ub2
.Ue8 70 .l:!.b1 + �e2 71 .l:!.b2+ �f3 72 .l:!.b3+
.l:!.e3 73 .l:!.b1 .U.e1 is a lso good , only not
69 . . . �e1 ?? ( instead of 69 . . . lle8 ! ) , as Jose
Raul Capablanca once played in a s imi lar
position . After 70 llb1 + �e2 h is opponent
Vera Menchik could have d rawn with the
obvious 71 'it>g2! . But there fol lowed 7 1
.Ub2+?? xh4 48 .l:!.f5 .l:!.b7
(48 . . . �g4 49 .Uxf7 h4 comes into considera
t ion) 49 .U.xd5 �g4 50 .l:i.d4+ 'lt>g5 51 �d3
(51 l:!.d8 is a lso possible) 51 . . Jbb2 52 'it>e3
h4 53 �f3 h3 54 .l:!.g4+?
The only way to d raw is 54 .l:!.d5+! f5 55 l:td8
h2 56 .l:!.g8+! �f6 57 l:!.h8, or 54 . . . 'it>g6 55
'it'g4! h2 56 .l:!.g5+ �f6 57 l:!.h5 'it>e6 58 'it>f4 .
54 . . . 'it>f5 55 llf4+ 'iiie6
55 . . . �g6! wins: 56 f5 .l:!.b5+.
56 .U.h4 h2? (B lack should go back with h is
k ing: 56 . . . �f5) 57 11h6+? (57 .U.h5! is
essentia l , a im ing to provoke . . . f7-f5 in a
situation where the black king has not yet
occupied the e5-square) 57 . . . 'it'e5 58 .Uh8
.l:!.c2 59 .Uh4 (the threat was . . . 'it'e5-d4-c3-
b2-c1 etc . ) 59 .. .f5 60 'iitg3
22 � How to Study the Endgame
The move that suggests itself, 60 .. .f4+?,
does not win : 6 1 �f3 l:c3+ 62 �g2 �e4 63
�xh2 (63 l:th8 ! ) 63 . . . l:!.c2+ 64 �h3! (64
'it>g 1 ? �e3 65 l:.h8 l:.c1 + 66 c;t>h2 f3 67 l:.e8+
�2 68 l:ta8 c;t>f1 ) 64 . . . �f3 65 l:th8 .l:.c7 66
l1h6 (66 �h2? 'it>f2) 66 . . . l:te7 67 l:.h8 'it>f2 68
l:ta8! f3 69 l:.a2+ l::te2 70 l:ta 1 (or 70 .Ua8 'it11
7 1 �g3 f2 72 �f3 ! �g 1 73 lig8+) with a
draw.
Let us imagine that it is White's turn to move .
He wi l l be forced to play 6 1 .l:!.h8 (6 1 �f3 is
not possible on account of 61 . . . .l:tc1 ! 62
l:!.xh2 l:tc3+ ) , and the black king can ad
vance, bypassing its rook, in order to
approach the h-pawn along the fi rst rank.
Note that it is the c2-square that the black
rook should occupy. I f it is on d2 or e2 , the
white rook is no longer obl iged to leave the
4th rank (there is the move 'it>f3 ! ) . With the
rook on b2 the king's route via the queenside
around its rook becomes too long .
I n other words, the position is one of mutual
zugzwang. White must be given the turn to
move .
60 . . .
61 c;t>t3
l:td2
l:ta2
61 . . . l:td 1 ? 62 l:.xh2 l:td3+ 63 �e2 .
62 c;t>g3 l:tc2 ! !
63 l:th8
63 �3 l:tc1 ! .
63 . . .
64 l:te8+
65l:ld8+
�e4
�d3
65 l:th8 l:!.e2 ! , intending 66 . . . �d2 .
65 . . . �c3
66l:lh8
66 l:tc8+ �d2 or 66 . . . 'it>b2
66 . . . .l:te2!
In the event of 66 . . . �b2? 67 �f4 �c1 68
c;t>xf5 �d 1 69 c;t>g4 the black king is too late .
67 'it>f4 'it>d2
68 'it>xf5 �e1
69 �g4
70 �g3
�f1
�g1
The king has arrived just in t ime!
Th i rty years later I d iscovered the possibi l ity
of a more tenacious defence. I nstead of 66
l:th8 it makes sense to play 66 l:td1 !? .
To win , it is sufficient for B lack to return with
his king to the f-pawn whi le the enemy rook
is tied to the 1 st rank . But how to ach ieve
this? If 66 . . . c;t>c4 the opponent repl ies 67
.U.a1 (67 l:tf1 .l:!.d2 ; 67 'it>f4 l:tf2+ 68 �g3
l:td2) , after which it is pointless to play
67 . . . 'it>d5 68 l:a5+ c;t>e6 (68 . . . �e4 69 %1a4+
�e5 70 .l:!.h4) 69 .l:.a6+ �e5 70 l:lh6
( intending 71 .l:.h4) 70 . . . �e4 71 .l:.e6+ �d3
72 .l:td6+ �c3 73 .l:td 1 !, and so on .
Before bring ing the king back, it is important
to place the rook on d2. Then the manoeu
vre of the wh ite rook to h6 (by analogy with
the variations just considered ) loses its
strength - Black again advances his king ,
and the wh ite rook can no longer reach d 1 .
The most accurate is 67 . . . l:ta21 (not immedi
ately 67 . . . l:td2 68 l:ta4+ �d5?! 69 l:th4, and
it is necessary to start al l over aga in ) 68 l:tb1
lld2 69 �3 (69 l:ta 1 �d5) 69 . . . �d5
(threatening . . . 'it>e5-f6-g5) 70 l:tb5+ �e6
71 lib6+ �e5 72 .l:r.h6 c;t>d4 73 .l:ld6+ �c3 7 4
lic6+ �b2 75 .l:!.h6 �c1 , and Black wins .
How to Study the Endgame ttJ 23
Now let us again remember the ending with
which we began : G l igoric-Smyslov. In the
note to White's 3rd move the variation 3 h6
J:la1 ! was analysed .
However, we d id not consider the attempt, by
playing 4 l:!.g7+ �f6 (the retreat to the 8th
rank is hopeless, of course) 5 l:!.c7 , to reach
the position which we have just been d is
cussing . Knowing of the impending danger,
Black can avoid it without great d ifficulty -
the defensive resources are qu ite sufficient.
5 . . . 'it'g6
5 .. J::tg 1 + 6 �f3 l::th 1 is a lso good .
6 h7 .l:i.h1 !
But here it is wrong to interpose a check:
6 . . . l::tg 1 +? 7 'it>f3 l':Lh 1 8 'it>e4 .
7 'it'f3
Noth ing is g iven by 7 f5+ 'itf6 . I n reply to the
waiting move 7 l':Lb7 Black can also wait:
7 . . .l::th2 , not fearing 8 l::tb5 'it>g7 ! 9 l:tg5+
�h8! . Also good is 7 . . . l:tg 1 + 8 'it>f3 .Uh 1 9
'it'e4 l:i.e 1 + , s ince with the rook on b 7, as we
know, the king's route to outflank the rook is
too long : 1 0 �d5 .Ud 1 + 1 1 'it>c6 l:tc1 + ! 1 2
�b6 l:i.h1 with a draw.
7 . . .�5!
The s implest way of demonstrat ing that the
position is d rawn . But Black also does not
lose after 7 . . . 'it>f6 !? 8 'it>e4 l:te1 + 9 'it>d5
.Ud 1 + 1 0 'itc6 .Uh 1 ! . With the king on f6
White does not have the important move 1 1
.Ue 7 , and i n the event of 1 1 l::td7 (or 1 1 'it>b 7)
1 1 . . .'itf5 the b lack k ing succeeds , after
e l im inating the f4-pawn , in returning to g6 in
time.
I I . IMPROVI N G YOUR TECHN IQUE
I w i l l now ta lk about how you can improveyour techn ica l mastery. For this you need to
study problems which are common to a l l (or
to many) types of endings. Problems, such
as the enhanced role of the king in the
endgame, zugzwang (and a very important
specific instance of it - mutual zugzwang,
and correspond ing squares) , the advisabi l
ity of this or that exchange, and so on. I t is
especia l ly important to sense the spirit of the
endgame, to develop the optimal mood for
playing it, and to understand the psychologi
cal chess laws which apply here.
Al l th is is best stud ied by analysing practical
endings, played by great masters of the
endgame. As an example, let us look at a
game by grandmaster Ulf Andersson .
Andersson - Franco
Buenos Aires 1 979
English Opening
1 tDf3 tDf6
2 c4 g6
3 lDc3 d5
4 cxd5 tDxd5
5 e4 lDxc3
6 dxc3
Andersson l ikes and knows how to play the
endgame, and so a l ready in the open ing he
happily exchanges the queens .
6 . . . 'i!t'xd 1 +
7 'it>xd1 f6
8 i£.e3 e5
9 tDd2
9 i£.c4 !? .
24 � How to Study the Endgame
9 . . .
1 0 i.c4
i.e6
i.xc4
1 0 . . . 'it>f7 looks more logica l , but here too
after 1 1 '>t>c2 lt:Jd7 1 2 b4 Black experiences
certa in difficu lties: 1 2 . . . lt:Jb6 1 3 i.b3 or
12 . . . h5 1 3 g3, intending 14 f4 .
1 1 lt:Jxc4 lt:Jd7
1 2 b4!
I n the endgame one should carefu l ly watch
for the opponent's ideas and if possible
frustrate h is plans. Here Black wanted to
equal ise the game completely with 1 2 . . . i.c5.
1 2 . . . lt:Jb6?!
An inaccuracy! The only defect of B lack's
position is that his bishop is more passive
than the opponent's. He should have tried to
exchange it, by playing 1 2 . . . h5 !? fol lowed by
. . . i.h6. The game Andersson-Mestel (Hast
ings 1 978/79) continued 1 3 f3 i.h6 1 4 i.f2
lt:Jb6 1 5 i.xb6! ( 1 5 lt:Ja5? 0-0-0+ ) 1 5 . . . axb6
1 6 b5 'it>e7! (in the endgame the king is best
placed in the centre of the board - therefore
Black avoids queenside castl ing) 1 7 a4
l:thd8+ 1 8 '>tc2 'it>e6 , and Jonathan Mestel
managed to reta in the balance. Wh ite acted
more d i rectly in the game Log inov-Sideif
Zade (Aktyubinsk 1 985 ): 1 3 'it>c2 i.h6 1 4
i.xh6 .Uxh6 1 5 .l:thd 1 0-0-0 1 6 lt:Ja5 l:thh8 1 7
l1d3, and retained slightly the better chances.
1 2 . . . f5 !? came into consideration .
1 3lt:Ja5!
14 'itc2
0-0-0+
i.e7
Now if 1 4 . . . h5 Wh ite would have repl ied 1 5
.l::i.hd 1 .l:ixd 1 ( 1 5 . . . i.e7) 1 6 .l:ixd 1 i.h6? 1 7
i.xh6 .l::i.xh6 1 8 c4 .l::i.h7 1 9 c5 lt:Jd7 20 c6 with
advantage. Even so, 1 4 . . . h5 was a usefu l
move - the exchange of rooks would have
eased Black's defence.
1 5 a3!
Andersson prepares an offensive on the
queenside with c3-c4-c5 . This positional
th reat provokes the opponent into dubious
activity.
1 5 . . . f5? !
16 i.xb6! !
The 'automatic' 1 7 f3 wou ld have a l lowed
the opponent to gain counter-chances , by
attacking the e4-pawn ( . . . lt:Jb6-d7-f6) . An
dersson makes a t imely correction to h is
p lan . A move earl ier the exchange of minor
pieces would not have g iven anyth ing : 1 5
i.xb6?! axb6 1 6 lt:Jc4 b5, but now the e5-
pawn comes under attack.
1 6 . . . axb6
1 7lt:Jc4 i.f6?
Black defends too passively. He should
have thought about 1 7 . . . .l::i.hf8 ! . I f 1 8 l:tae 1 ,
then 1 8 . . . b5! 1 9 lt:Jxe5 fxe4 20 .l:hf1 i.g5.
After 1 8 exf5 l:lxf5 1 9 f3 Black has a choice
How to Study the Endgame lLJ 25
between the i nterest ing , although question
able piece sacrifice 1 9 . . . e4? ! 20 .l:.he1 exf3
21 I:!.xe7 fxg2 22 I:!.g 1 .l:tf2+ 23 'itb3 b5 24
ttJe5 h5! (with the idea of 25 . . . c6 and
26 . . . .l:i.dd2) and the qu ieter continuation
19 . . . i.g5!? (threatening 20 . . . e4 or 20 . . . b5)
20 l:the 1 b5 21 tt:Je3 i.xe3 with an inferior,
but tenable double-rook ending (2 1 . . . .l:.ff8 ,
intending 22 . . . e4 , a lso comes into consid
eration ) . L ittle is changed by 1 9 .l:.hf1 i.g5
(weaker is 1 9 . . . .l:.df8 20 f3 e4 21 I:!.fe 1 .l:i.g5
22 g4! .l:i.xf3 23 .l:.xe4 .ltf6 24 ki.d 1 ! with the
threats of 25 .l:!.e8+ and 25 h4) 20 .l:!.ae 1 b5.
As was pointed out by Maxim Notkin , a
simi lar double-rook ending arises after
1 7 . . .fxe4!? 1 8 l1ae1 .l:f.hf8 1 9 .Uhf1 .1L.h4!? 20
g3 .ig5 21 llxe4 (2 1 a4 !? ) 21 . . . b5! 22 tt:Je3
.be3 23 .l:!.xe3.
18 a4!
White not only consol idates the position of
his knight at c4 , but also beg ins an offensive
on the queenside. 1 8 b5 was less accurate
on account of 1 8 . . . fxe4 and 1 9 . . . 11d5.
1 8 . . . .1L.g7
19 .l:!.he1
20 b5!
.l:.he8
20 a5? b5 was less good . The target (the b6-
pawn) should be fixed , and only then
attacked .
20 . . . f4
21 aS bxa5
22 .Uxa5 b6
23 .l:.a7
Threatening 24 tt:Jxb6+.
23 . . . .ltf6
24 .l:!.ea1 .l:!.e6
25 .l:!.1 a6!
Creating the strong threat of 26 tt:Ja5 and 27
tt'lc6 .
25 . . . .l:f.de8
(see diagram)
With h is active play on the a-fi le White has
t ied down the opponent's forces and forced
h is rooks to move off the open fi le .
26 �b3!
This unhurried manner of play is typical of
Andersson . Just in case he improves the
posit ion of h is king and awaits a conven ient
moment for the further strengthen ing of h is
position . This is the way to convert an
advantage in the endgame - by do ing
everyth ing possible to restrict the oppo
nent's possib i l ities, and then, without hu rry
ing , look for new breaches in h is defences.
To many the rule 'do not hurry' may seem
paradoxical, but in fact it is seen in practi
cally all the endings of games by the great
masters of the endgame. Look carefully at
the endings of Capablanca and Flohr, and
you will see with what slowness, sometimes
bordering on tedium, they convert an advan
tage (Sergey Belavenets) .
26 . . . .1L.d8?!
I t was this that Andersson was waiting for!
27 I:l.a8+ 'it>d7
28 .l:.a2 !
A convenient moment has arrived for a
regrouping of the forces : exploit ing the poor
position of the bishop on d8, Wh ite seizes
control of the d-fi le . But Al isa Gal l iamova's
suggestion 28 ll6a7 ! , with the idea of 29
26 w How to Study the Endgame
l:tb8 and 30 tLlxb6+, was possibly even
stronger.
28 . . .
29 .Ud2+
30 l:ta7 !
ii.f6
ci;e7
Of course , there is no point in White
exchanging h is active rook for the oppo
nent's passive rook.
30 . . . llc8
31 l:.d5
32 h3
In such positions Andersson loves to make
waiting moves.
32 . . . ci;e7
33 tLlb2!
The knight has done an excel lent job at c4 ,
and now it moves to d3 , from where it wi l l
support the advance of the c-pawn, and
from where it can itself advance fu rther via
b4. Note that White did not p lay this a move
earl ier, s ince he was afra id of the reply
32 . . . c6 - he waited unti l the opponent's king
had gone to e7.
33 . . . 'it>e8
33 . . . .l:!.d6 was more tenacious, after wh ich it
was best to reply 34 J::[xd6 ! �xd6 35 c4 ,
intending 36 tLld3, 37 c5+ and 38 �c4.
34 tLld3 ii.g7
34 . . . c6 35 l:.dd7! cxb5 36 tLlb4 .
35 c4 i.f6
36 c5
37 tLlxc5
37 . . . .l:r.b6 38 tLld7! .
38 I:r.a6!
bxc5
.l:.e7
Systematic play! With gain of tempo White
gains control of the 6th rank - he prevents
. . . c7-c6 and obtains the e6-square for h is
knight.
38 . . . ii.h8
39 ci;c4!
Again , just in case , Andersson improves the
position of his king .
39 . . .
40 f3
41 tLle6
42.l:lc6
ii.g7
.l:tb8
ii.f6
Black resigned , s ince 42 . . . l:!.b7 43 .l:td8+
leads to mate, whi le if 42 . . . l:tc8 , then 43 b6 is
decisive.
A classic example of virtuoso endgame play!
The study of such endings assists the
development of taste for the endgame and
improves techn ical mastery.
From the methodological point of view it is
usefu l to see the same problems d isplayed
in a negativeform - by examin ing examples
in wh ich typical endgame mistakes are
made. The fol lowing game was played on
the women's board in a competition for
Moscow higher education establ ishments in
1 972/73.
Sicilian Defence
1 e4 c5 2 tLlf3 tLlc6 3 d4 cxd4 4 tLlxd4 tLlf6
5 tLlc3 e6 6 tLldb5 ii.b4 7 tLld6+?! ci;e7 ! 8
ii.f4? ! e5! 9 tLlf5+ 'it>f8 1 0 ii.g5 d5! 1 1 ii.xf6
gxf6?! ( 1 1 . . . 1ixf6! suggests itself) 1 2 exd5
i.xf5 1 3 dxc6 ii.xc3+ 14 bxc3 'ii'xd1 + 1 5
.l:.xd1 bxc6 1 6 ..1d3 ( 1 6 l:.d6 !? )
16 . . . e4?
How to Study the Endgame QJ 27
After achieving the better endgame. Black
immediately makes a positional mistake -
she places a pawn on a square of the same
colour as her bishop. 1 6 . . . .ie6 1 7 .ie4 �e7
18 i.xc6 .l:tac8 1 9 .ie4 l:txc3 suggests itself.
1 7 �c4 .l:!.g8?
Again a fundamenta l mistake - the player
with Black does not pay attention to her
opponent's possibi l ities. Of course, 1 7 . . . 'ite7
fol lowed by 1 8 . . . .ie6 was correct.
1 8 l:td6! �e7
19 .l:txc6 .l:i.gc8
20 l:txc8 .l:i.xc8
21 .ib3 .l:i.xc3
22 rtld2 .Uc8
23 l:te1 .ti.g8
24 g3 l:td8+
25 'itc3
The advantage is now with White , who is
th reatening 26 f3 .
25 . . . �f8?
Moving the king away from the centre in the
endgame is nearly always a mistake .
25 . . . �g6 26 f3 f5 was preferable.
26 Ite2 .ig6
27 �d2 l:txd2?
An incorrect evaluation . The bishop end
game is lost. 27 . . . Itc8+ was more tenacious .
28 'it>xd2
29 'ite3
30 'itd4
31 c4
32 c5+
33 .id5+
34 'itc4
35 'itd4
36 'it>e3
37 h3
38 'itf4
39 g4
rtle7
f5
'itd6
f6
'itc6
�c7
h6
�h5
'itd7
�e7
.ig6
A slight inaccuracy. In accordance with the
principle 'do not hurry ' , before changing the
pattern of the play White should have
strengthened her position to the maximum.
I n the g iven position - by moving the a2-
pawn off a l ight square (a square of the
same colour as the bishop) . Perhaps after
39 a3! White was concerned about the reply
39 . . . h5. But after this Black, with nearly al l
her pawns on squares of the colour of her
bishop, would undoubted ly be lost.
39 . . . fxg4
40 hxg4 h5!
When defending an inferior endgame i t is
usefu l to exchange as many pawns as
possible.
41 gxh5
With the pawn on a3 White would have
played 41 .ixe4 .if? 42 g5! , e l im inating the
f6-pawn, wh ich h inders the wh ite k ing.
41 . . . �xh5
42 'it>xe4 �g4
43 'it>f4 .ih3
44 .ie4
45 a3
SL.e6
.id7?
After 45 . . . .ic8 it was sti l l possib le to put up a
tenacious defence . The move in the game
loses immediately.
46 c6! �e8
28 � How to Study the Endgame
Otherwise 47 �f5 .
47 c7
48 �c6+!
Black resigned .
'>t>d7
A player's endgame technique is based on a
mastery of the whole arsenal of ideas he has
accumulated - from understanding the spirit
of the endgame and its most general laws,
to minor techn iques which he encounters
when studying his own or other players'
games. To i l lustrate this , I wi l l acqua int you
with one ending of my own , broken up into
elementary components .
F irst we wi l l examine four ' i ntermed iate
products ' , which , incidenta l ly, are qu ite
instructive in themselves .
1 ) It is wel l known that in the endgame the
role of logical th inking increases . One must
be able to compi le p lans, outl ine a scheme
for arranging the pieces, and so on. A classic
example is provided by the fol lowing ending .
Capablanca - Ragozin
Moscow 1 936
This is what Jose Raul Capablanca writes
about this position :
White 's plan is to prevent the advance of the
c-pawn (after which the b-pawn could
become weak) and to control the entire
board up to the fifth rank. This is achieved by
moving the king to e3 and by placing the
rook at c3, the knight at d4, and the pawns at
b4 and f4. After he has attained such a
position, White will be able to advance his
queenside pawns.
The fol lowing moves are easy to understand
- Capablanca consistently carries out h is
plan .
33 ttJd4 .l:tb7
34 b4 �d7
35 f4 �e7
36 �f2 .l:ta7
37 .l:tc3 �d6
38 .l:td3 '>t>e7
39 '>t>e3 .l:ta4
40 .l;tc3 'it>d6
41 ti.d3 '>t>e7
42 ti.c3 �d6
The requ i red arrangement of the pieces has
been ach ieved . Now Capablanca wants to
regroup h is forces, by playing h is knight to
c3 (or c5) .
43 ttJe2 g6
44 .l:td3+
45 �d4?
�e6
In winn ing positions even h igh ly experi
enced players sometimes involunta ri ly relax
and make tactical oversights, risking losing
the fru its of their correct strategy. That is the
case here : White's last move is a serious
inaccuracy (45 f5+ ! gxf5 46 ttJf4+ or 46 . . . 'it>e7
46 ttJc3! was correct) , which cou ld have
been exploited by 45 . . . �b5 ! . I n the event of
46 .l:te3+ �d6 Black creates the th reat of
47 . . . c5+ , whi le after 46 ttJc3 �xd3 47 ttJxa4
�f1 he regains h is pawn.
45 . . .
46 .l:te3+
47 ttJc3
.l:ta6?
�d6
How to Study the Endgame .Qj 29
The queenside pawns are now ready to
advance. I n passing 48 4Je4+ is threatened .
47 . . . f5
48 b5 It aS
48 . . . .llxa3 49 4Je4+ fxe4 50 .llxa3 SLxb5 5 1
.l::i.g3 .
49 'it>c4
50 'it>b4
51 bxc6
52 4Jb5+
53 .U.d3!
i.e6+
c5+
i.g8
'itxc6
Take note : Wh ite does not advance h is
passed pawn , but switches to an attack on
the enemy kingside pawns. This is fu l ly in
the spirit of an important principle in the
conversion of an advantage - the 'princip le
of two weaknesses' . After creating a second
weakness in the opponent's position , by
playing against it and then , in case of
necessity, again switching the attack to the
first weakness, you convert you r advantage
i n the most methodical way.
I n the broad sense of the word , a weakness
i n the opponent's position can be not only a
vulnerable pawn or a badly placed piece, but
also our own passed pawn, which he is
forced to blockade, or an invasion square ,
which he has to cover.
53 . . • g5
54 .Ud6+
55 fxg5
56 .llg6
57 l::i.xg5
58 4Jd4
59 .llg7+
60 l:tg6+
61 lLlb5
62 4Jd6+
63 h4
Black resigned .
'it>b7
hxg5
l:i.f8
f4
:l.c8
�b6
'itb7
.U.f8
'it>b8
2) You will have noticed , of course , that
when playing the endgame Capablanca
twice repeated moves. Here is what Sergey
Belavenets writes about th is :
The repetition of moves in the endgame
plays an important role. Disregarding the
fact that it gains time for thinking, it can be
mentioned that, by repeating moves, the
active side acquires certain psychological
gains. The defender, whose position is
inferior, often cannot withstand it, and he
creates a further weakening which eases his
opponent 's task. In addition, repeating moves
enables the position to be clarified to the
maximum extent. We know that some
upholders of 'pure ' chess will severely
criticise us for this advice. But we cannot
refrain from advising players: sometimes
repeat moves in the endgame. In the
struggle every chance has to be exploited,
and there is nothing ugly or unethical in
repeating moves.
3) Let us examine an example from one of
my own games.
(see diagram)
White has an obvious advantage, but for the
moment the invasion points are securely
defended .
30 � How to Study the Endgame
Dvoretsky - Kikiani
Kiev 1 970
35 . . . i.d8
I d id not even beg in to examine 36 tt'lbc5
seriously, s ince I noticed an opportun ity to
gain a tempo by a simple triangulation
manoeuvre with the bishop.
36 i.a7! .l:!.a8
37 i.e3
Threatening 38 tt'lbc5 .
37 . . .
38 i.b6
i.e7
If now 38 . . . i.d8 the move 39 tt'lbc5 gains in
strength - the b7-pawn is not defended .
38 . . . .l:!.ab8
We have reached the position with which we
began, but with White to move.
39 g3
The 'do not hurry' principle in action : whi le
the opponent is unable to do anyth ing , al l
the evensl ightly usefu l moves should be
made. Why not, just in case, take away the
f4-square from the knight?
39 . . . i.d8
40 i.a7
41 i.e3
.l:!.a8
White is not averse to repeating his ma-
noeuvring , the opponent does not know
what he should fear i n the fi rst instance .
Kik iani decided to prevent f3-f4 , which in
fact was hardly a threat, s ince it wou ld have
weakened the e4-pawn .
41 . . . g5?
42 tt'lbc5!
There wil l no longer be a more convenient
moment for the p lanned i nvasion at c5: the
black rook is not defending the b7-pawn,
and the bishop is stuck at d8 .
42 . . . l:tb8
43 tt'ld7 tt'lxd7
44 .l:!.xd7+ .l:!.e7
45 tt'lc5 'itoe8
46 l:!.xe7+ cJ;;xe7
47 l:Id7+ 'it>e8
48 l:!.xh7 tt'lc7
49 h4 gxh4
50 gxh4
Black resigned .
4) Let us examine another endgame by
Capablanca.
Capablanca - Yates
New York 1 924
noeuvre. Faced with such unhurried ma- Note the pretty knight circuit, thanks to
How to Study the Endgame ctJ 31
which White won a pawn.
40 t'Llc3
41 t'Lle4
42 t'Lled6
43 t'Llb7
44 t'Llbxa5
�c5
.l::tb5
.l::tc5
.l::tc7
The rest is accurate , Capablanca-style
conversion of the advantage. White's fi rst
objective is to improve the placing of h is
pieces: first h is knights , and then h is rook.
44 . . . ..ltb5
45 tt:ld6 ..itd7
46 t'Llac4 .i:!a7
47 t'Lle4 h6
48 f4 ii.e8
49 t'Lle5 .l:!a8
50 l:!.c1 ii.f7
51 l:tc6 ..ltg8
52 t'Llc5 J::.e8
After strengthening his position to the
maximum and tying down the opponent's
forces, White beg ins to prepare the advance
of his passed pawn .
53 .l:ta6 l:te7
54 'it>a3 ii.f7
55 b4 t'Llc7
56 .l:!.c6
57 'it>b2
58 .l:ta6
59 g4!
t'Llb5+
lt:Jd4
ii.e8
Aga i n , as in the game against Ragozin ,
Capablanca operates in accordance with
the principle of two weaknesses . For a t ime
he defers the advance of h is passed pawn
and launches an attack on the kingside.
59 . . . 'it>f6
60 lt:Je4+
61 lt:Jd6
62 .l:ta5
63 .l::ta8
'lt>g7
..ltb5
..itf1
g5
The threat was 64 tt:Je8+ 'it>h7 65 lZ'lf6+ 'lt>g7
66 g5 with mate.
64 fxg5
65 hxg5
66 .l:!.e8
hxg5
..ltg2
l::.c7
66 . . . .l:txe8 67 lt:Jxe8+ Wf8 68 g6 ! .
6 7 :ds tt:Jc6
68 tt:le8+ 'it>f8
69 tt:Jxc7
70 'it>c3
Central isation of the king .
70 . . .
7 1 'it>d4
72 g6
73 tt:Je8!
74 b5
75 g5
76 g7+
77 g6
Black resigned .
lt:Jxd8
..ltb7
ii.c8
tZ'lb7
lt:Jd8
'it>g8
'it>f8
'it>g8
And now see how a l l this i nformation
enabled the fol lowing ending to be won .
Dvoretsky - Privorotsky
Kiev 1 970
29 .l:!.a5 b4
32 � How to Study the Endgame
29 . . . c4 30 i.xd4 was no good , but 29 . . . 4Je6
was worth considering . After the move in the
game Black wi l l no longer have any
counterplay.
Now, fol lowing the example of the Capa
blanca-Ragozin ending , White outl ined a
scheme for the deployment of h is pieces.
Obviously, the knight must be played to e4 ,
the king brought up to f3 , the rook placed on
a6 and the b ishop on the c1 -h6 diagona l ,
and , final ly, the pawn advanced to f5 .
30 4Jd2 i.e7
31 4Je4 4Jd7
32 .l:ta6 'i;f7
33 'i;f3 4Jb8
34 l:ta8
35 i.c1
36 .l:!.a6
37 f5
38 gxf5
4Jd7
4Jb6
4Jd5
gxf5
The plan has been successfu l ly carried out.
38 . . . �d7
Here I saw that the triangulation method for
gain ing a tempo, found two rounds earl ier i n
the game against Kikian i , m ight aga in come
in usefu l . The only d ifference is that here the
tempo is won not by the bishop, but by the
rook.
39 l:Ic6
40 .l:!.h6!
41 l:Ia6
�c7
'it>g7
�f7
I t is now White's turn to move .
42 i.h6 .l:!.c8
42 . . . l:td7 is now bad because of 43 l:Ia8 and
44 �h8, winn ing the h7-pawn .
43 �a? �c7
44 �a6
I f 44 �a8 there is the reply 44 . . . .l:i.c6 , and so
the rook returns to a6. I n the event of
44 . . . l:Ic8 White would probably have played
45 i.g5!? , but it did not prove necessary to
weigh up this move.
44 . . . 4Jc3?
Belavenets was right - Black cou ld not
withstand the pressure , and he h imself
avoids the repetition of moves.
Now White carries out a curious ci rcu lar
manoeuvre with h is kn ight , resembl ing that
which Capablanca made against Yates .
45 4Jd2 ! 4Jd5
46 ttJc4 �f6
47 4Je5+ was threatened .
47 4Jd6+ �e7
48 4Je4
After making four successive moves, the
knight has retu rned to where it began . But
B lack's defences are now completely d isor
ganised . The threat is 49 4Jxf6 4Jxf6 50
�g5. If 48 . . . �f7. then 49 �d6 is decisive .
48 . . . �h8
49 �e6+
50 i.g5+
Black resigned .
�d8
Thus by reflecting on the games of great
players and the recommendations which
they g ive in their commentaries, and by
considering your own competitive experi
ence, you can sharply improve you r techni
cal mastery.
How to Study the Endgame lb 33
In conclusion I offer several exercises, i n
which a rook fights against opposing pawns.
The process of try ing to solve them wi l l offer
you tra in ing in the practical appl ication of the
theory of th is type of endgame.
Exercises
1 . Wh ite to move 2. Wh ite to move
3 . Wh ite to move 4. White to move
34 How to Study the E ndgame
5. White to move 6. White to move
7. White to move 8. White to move
How to Study the Endgame ctJ 35
Sol utions
1 . Yu . Averbakh ( 1 980) .
1 �e6! e4
2 l:i.g5! !
The only winn ing move, the point of which is
to place the rook beh ind the passed pawn
with gain of tempo, and then , after ascerta in
ing the position of the enemy king , to send
the wh ite king i n the opposite d i rection , on
an outflanking manoeuvre .
2 . . .'lt>d2(d3) 3 l:i.d5+! �c2 4 l::te5! 'it>d3 5
Wf5!;
2 . . . 'lt>f2(f3) 3 l:i.f5+! 'it>g2 4 I:te5! 'it>f3 5 'it>d5 ! ;
2 . . . 'lt>e2 3 �e5 e 3 4 �e4 .
The hasty 1 l:i.g5? leads to a d raw in view of
1 . . . �f4! 2 Wf6 e4 . Both 1 �d6? e4 2 J:tg5
�d3{d2) ! and 1 �f6? e4 2 :g5 Wf3(f2) ! are
also incorrect, s ince the outflanking ma
noeuvre becomes unreal isable.
2. E. Kolesnikov ( 1 989) .
1 'lt>f7 suggests itself, bu t after 1 . . . ..txd4!
White cannot win: 2 ..te6 ..te3 3 l::tf5 d4 4
l:\.e5+ Wf2 5 l:i.d5 We3 (6 �d5 ! is not
possible) , or 2 J:tf6 We3 3 l::te6+ Wf3 4 :d6
We4 5 �e6 d4 (6 Wd6! is not possib le) .
1 l:!.f8 ! �xd4
2 �f7 'it>e4
3 l:!.e8+! Wf3
4 �d8! 'it>e4
5 ..te6 d4
6 ..td6!
7 'it.>c5
8 ..tc4
9 'it.>c3
d3
'it>e3
d2
The king has arrived just in t ime.
3. J . Moravec ( 1 9 1 3) .
The d irect 1 Wxg7? h4 2 �g6 h3 3 'it.>g5 h2 4
'>tg4 h 1 'if leads on ly to a draw, s ince 5
�g3?? 'ifh8 is not possib le. The g7 -pawn
must be kept on the board .
1 'it.>h7 ! ! h4
1 . . . g5 !? 2 ..tg6 g4 is another try, hoping for 3
Wxh5? g3 4 'it>g4 g2 5 ..th3 � h 1 ! with a
draw. To avoid sta lemate, the h5-pawn must
be left a l ive : 3 Wg5 ! ! .
2 'it>g6
3 'it>g5
4 ..tg4
h3
h2
h 1 'if
4 . . . g5 !? a lso fa i ls to save Black: 5 'it>g3
h 1 �+ 6 'it>f3 g4+ 7 'lt>xg4 tLlf2+ 8 'lt>f3 tLld3 9
J:ta4 (or 9 'lt>e3 tLle5 1 0 l:i.a4 ), and the knight
wi l l soon be caught.
5 Wg3
4. V. Bron ( 1 929) .
1 l:!.c8+! 'it>e7!
1 . . . �d7 2 :f8 ; 1 . . . �f7 2 :c4.
2 I:tc7+
3 .l::i.c6+
�e6
..te5
4 .Uc5+ 'it>e4!
If 4 . . . 'it>d4 , then 5 J:tf5 ..te3 6 Wg5 g3 7 'it>g4
g2 8 l:!.xf3+ .
5 l:!.c4+ 'it>e3
The checks are at an end : if 6 l:!.c3+, then
6 . Aid2 is decisive .
6 �xg4! f2
7 l:l.g3+ 'it>e4
8 l:l.g4+ �e5
9 �g5+ 'it>e6
1 0 .l:i.g6+ ..te7
1 1 I:tg7+ ..tf8
1 2 l::tg5 ! f1 'if
1 3 �f5+ 'ifxf5
Stalemate .
36 � How to Study the Endgame
5. V. Sokov ( 1 940) .
The routine 1 c:Ji;e7? throws away the win in
view of 1 . . . c:Ji;b4! 2 l:.e1 (otherwise 2 . . . c:Ji;c3)
2 . . . a5 3 c:Ji;d6 a4, and the black king
'shoulder-charges' White's. The move . . . c:Ji;b4!
must be preventedMais conteúdos dessa disciplina
Tempo restante 167:19:13 Questão 10 Ainda não respondida Vale 3,00 ponto(s). Marcar questão Texto da questão O sucesso da atividade e produção agrí...- Código da questão: 51164 Consideremos uma função de produção P que dependa da quantidade X de um fator variável. Chama- se produtividade marginal d...
- A solução da integral indefinida ∫(cosy.tangy)dy é: Questão 1Escolha uma opção: a. −cosy+c b. cosy+c c. cosy.cotangy+c d. −cosy
- d.unip.br/webapps/assesment/take/launchjsp?course_assessment_ide_184688_18coursejde_418665_18contentJd=_4671936_8stepanul { Suas respostas foram sa...
- A soma de dois números inteiros e positivo é igual a 80. Logo o produto máximo entre esses dois números será igual a: Questão 5Escolha uma opção:...
- a) Cite e justifique duas vantagens logísticas que o proprietário do mencionado restaurante obtém ao oferecer seus serviços em uma plataforma terceiri
- A fatoração é uma habilidade que se desenvolve com a prática. Se você tem a fatoração 3 x 3 x 5 x 5 x 7, qual é o número que você obteve?
- Determine o valor da expressão 3x + 5xy + 3x2y dado x = 3 e y = 2. 92. 91. 93. 94
- O Românico dos séculos X, XI e XII é o primeiro estilo do Ocidente em uma edificação mais estrutural do que formal. A concatenação dos elementos d...
- O valor numérico da expressão algébrica x−−√ + 3y, para x = 9, para y = -2 é: Escolha uma opção: a. -2 b. -1 c. -4 d. -5 e. -3
- O valor numérico da expressão algébrica x2 – 7x + y, para x = –1 e y = –5 é: Escolha uma opção: a. 3 b. –4 c. 1 d. –3 e. –13
- O valor numérico de 2x + 3y para x = 5 e y = – 5 é: Escolha uma opção: a. -2 b. -4 c. -3 d. -1 e. -5
- Valor 3,75 pontos D Objetiva Questão 1 ? Estou com dúvida Ao ensinar a ginástica para adultos, o profissional deve ter ciência de, exceto: o jovem ...
- Avaliação disciplina - Estudo Transversal VI (17604)
- TRABALHO DE HIST E GEO 1TR 24
