Dvoretsky's Endgame Manual Page 6

  Let’s add one more white pawn at c4. Now the move 1…b6 no longer works, because of 2 cb cb 3 c5.

  Now let’s move the a-pawn to a4. In this case, Black can stop the breakthrough for good by playing 1…c6! 2 a5 a6!

  This is the sort of structure we find in the Ruy Lopez Exchange Variation. Black to move can create a passed pawn by 1…c4! 2 Kg3 c5, followed by …b5-b4, …a5-a4 and …b4-b3. (Strictly speaking, the term “breakthrough” is not really appropriate here, since no pawn sacrifice is involved; but the effect is just the same.)

  White to move can stabilize the situation on the queenside by 1 c4!, which guarantees him a decisive advantage, thanks to the outside passed pawn he will create on the opposite side of the board.

  Maslov – Glebov

  USSR 1936

  Black’s position looks difficult, since the enemy king rules the queenside. But the possibility of a pawn breakthrough changes the evaluation of the position completely.

  1…h5! 2 Ka3 (2 g4 g5!) 2…g5 3 K×a4 f5! 4 Kb5

  There is no defense: 4 hg f4!, or 4 ef g4! 5 fg e4.

  4…f4 5 gf gh, and the h-pawn queens.

  The errors committed in the following examples are quite instructive. They could have been put in the “Tragicomedies” section, except that I already had plenty of material for that section without them.

  Havasi – Peko

  Budapest 1976

  Black resigned, never suspecting that the queenside pawn structure contained the possibility of a breakthrough.

  1…c4! 2 bc

  If 2 dc, then 2…a4 3 ba b3 4 cb d3–+, while 2 Kg3 is met by 2…a4! 3 ba b3 4 cb c3.

  2…a4 3 c5 a3 4 ba ba 5 c6 a2 6 c7 a1Q 7 c8Q

  The pawns queen simultaneously; but Black has an easy win by once again obtaining a pawn ending. Note Black’s working the queen up to h4 – a standard technique in these positions, ensuring that the g4-pawn is captured with check.

  7…Qf1+ 8 Kg3 Qf4+ 9 Kh3 Qf3+ 10 Kh2 Qf2+! 11 Kh3 Qh4+ 12 Kg2 Q×g4+ 13 Q×g4 K×g4 14 Kf2 K×h5–+.

  Capablanca – Ed. Lasker

  London simul 1913

  1…Ke5?? 2 h6! ( 3 g6) Black resigned.

  The draw would have been assured after 1…Ke6! 2 K×d4 f4 3 Ke4 f3 4 K×f3 Kf5 5 g6 hg 6 h6 Kf6=.

  Kharlov – Ernst

  Haninge 1992

  The position is drawn: 1…Kd6! 2 Kd4 Kc6 3 c5 g5 4 Ke4 gh 5 gh K×c5 6 Kf5 Kd6=.

  In the game, Black played 1…g5?? 2 g4!+–, and after a few more unnecessary moves (2…hg 3 h5 f5 4 h6 f4+ 5 Kf2 g3+ 6 Kg2 Ke4 7 h7), he resigned.

  Tragicomedies

  Ed. Lasker – Moll

  Berlin 1904

  Black wins easily by 1…f6 2 g5 (2 h6 gh 3 f4 Kd5) 2…h6; 1…Kd4 is also good.

  1…h6??

  A terrible blunder, allowing the breakthrough 2 f6! gf 3 f4 Kd5 4 g5 fg 5 fg Ke6 6 gh Kf6 7 Kc2+–.

  But White failed to exploit his unexpected opportunity and lost after 2 f4?? f6! 3 g5 Kd4.

  Svacina – Müller

  Vienna 1941

  Black cannot capitalize on the active position of his king: 1…g4 2 Ke1 Kc2 3 Ke2=; or 1…f4 2 gf gf 3 ef=. He thought up an amusing psychological trap: retreating his king instead.

  1…Kc4 2 Kc2 Kb5 3 Kb3 Kc6 4 Kb4 Kd6 5 Kb5 Kd7 6 Kc5 Ke6 7 Kc6?

  And it worked! White, having no idea what his opponent was up to, naively marched his king deep into enemy territory – no doubt, he was already expecting to win. But now, Black plays the pawn breakthrough.

  7…g4! 8 Kc5 f4! 9 ef h4! 10 gh g3 11 fg e3 White resigned.

  Nakagawa – Day

  Buenos Aires ol 1978

  This position is drawn: the potential threat of a kingside breakthrough is counterbalanced by Black’s outside passed pawn – but no more than that. Here is how matters should normally develop: 1…Kb6 2 h5 Kc6 3 h6! Kb6 4 e5 (4 f5 Kc6 5 e5 Kd5! 6 K×b5! K×e5 7 Kc6=) 4…fe 5 fe (5 f5 e4 6 Kc3! Kc5 7 g5 is also possible.) 5…Kc6 6 e6! Kd6 (6…fe? 7 g5+–) 7 ef Ke7 8 K×b5 K×f7 9 Kc4=.

  But the game continued 1…h6??, and now White could have won easily by 2 h5! ( 3 g5). Instead, he chose 2 g5??, which lost after 2…fg 3 fg h5! 4 e5 Kd5, etc.

  Süss – Haakert

  BRD ch, Kiel 1967

  1 g4!? or 1 Bb6!? would have retained excellent winning chances for White. However, he forgot about the possibility of a pawn breakthrough, and obtained the opposite result instead.

  1 g3?? g4! 2 gh gh 3 Be5 B×h4 4 f3 Bf6! 5 Bh2 B×c3+ 6 Ka4 Ke3 7 f4 Ke4 White resigned.

  Averbakh – Bebchuk

  Moscow 1964

  Black has an inferior, but defensible endgame. Bebchuk, however, misjudged the pawn ending.

  1…b5? 2 R×b5! R×b5 3 ab+ K×b5 4 e4 Kc6 5 e5! fe

  If 5…Kd5 6 e6 Kd6, White brings his king to b6 (or, if Black plays …b7-b6, to b5), and after Black replies …Kc8, uses his reserve tempo h4-h5 to win.

  6 g5 hg

  Black finds no relief in 6…Kd6 7 f6 Ke6 8 fg Kf7 9 gh b5 (the floating square for Black’s pawns does not reach the last rank, and the distance between the pawns is the unfavorable two files) 10 Ke4 b4 11 Kd3 and 12 Kc4+–.

  7 f6!

  Not, however, 7 h5? Kd6 8 f6 Ke6 9 fg Kf7–+. Black resigned, in view of 7…gf 8 h5.

  Black had a better defense in 1…h5!?, when White could have tried 2 Rb5!. Here, too, the exchange of rooks leads to a loss: 2…R×b5? 3 ab+ K×b5 4 e4! Kc6 (4…hg 5 e5) 5 e5 Kd7 6 e6+ Kd6 7 gh b5 8 Kc3 b6 (8…Kc7 9 Kb4 Kc6 10 Ka5) 9 Kb4 Kc6 10 h6! gh 11 h5 Kd6 12 K×b5 Kc7 13 e7+–. Black would have to play 2…Re8! 3 gh Rh8, with good drawing chances.

  Gazik – Pétursson

  Groningen ech jr 1978/79

  After 1…Kh8! the draw is obvious. In the game, however, Black allowed the trade of queens.

  1…Kf8?? 2 Qf5+! Q×f5 3 gf Kg7 4 c4

  A simpler way is 4 Kg4 Kh6 (4…Kf6 5 h6 +–) 5 c4 f3 (5…Kg7 6 Kf3) 6 K×f3 K×h5 7 f6 (7 Kg3 is good, too) 7…Kg6 (7…ef 8 c5 +–) 8 fe Kf7 9 Kg4 K×e7 10 K×g5 +–.

  4…f3 5 h6+??

  White returns the favor by being in too much of a hurry for the breakthrough. The win was 5 Kg3 g4 6 Kf2! Kh6 (6…Kf6 7 h6) 7 c5 (or 7 f6 ef 8 c5) 7…dc 8 f6 ef 9 d6.

  5…K×h6 6 c5 dc 7 f6 Kg6! White resigned.

  Wade – Korchnoi

  Buenos Aires 1960

  Korchnoi ignored the possible pawn breakthrough on the queenside.

  1…Kg5?? 2 b5! Kh5 3 a5!

  Black resigned, in view of 3…ba 4 b6 cb 5 d6.

  He would also have lost after 1…g5? 2 b5 g4+ 3 Ke2 (3 Ke3 Kg5 4 f3 g3 5 hg hg 6 Ke2! is also possible) 3…Kg5 (3…h3 4 a5!) 4 f3 or 4 h3. Black’s king would be tied to the kingside, while White could break through on the queenside at the right moment.

  The only way to ward off White’s threat was by 1…b5! 2 ab b6, and we have a draw: 3 Ke3 (3 h3 g5 4 Kg2 g4?! 5 f4! g3!= is also good) 3…g5! (3…Kg5? 4 f4+! ef+ 5 Kf3+–; or 3…Kg4? 4 f3+ Kh3 5 f4 ef+ 6 K×f4 K×h2 7 e5+–) 4 h3 (4 f3 is weaker due to 4…g4 5 f4 h3!) 4…g4 5 f4! gh 6 Kf3 ef 7 e5, etc.

  Exercises

  The Outside Passed Pawn

  An outside passed pawn usually means a positional advantage sufficient to win. This pawn will draw off the enemy king, allowing the other king to be the first to attack the enemy pawns.

  For example, in Diagram 1-103, after 1 c4!, stopping the threatened enemy breakthrough, we broke off our analysis, since the further exploitation of the outside passed pawn is elementary here.

  Of course, that is not always the case. In the endgames Kharlov-Ernst (Diagram 1-107) or Nakagawa-Day (Diagram 1-110), the proper outcome would have been a draw, despite the presence of an outside passed pawn. And in the game Hansen-Nimzovitch (Diagram 1-83), Black met the threat of an outside passed pawn with the activation of his king, which even won for him.

  Lombardy – Fischer

  USA ch, New York 1960-61

  The game hangs in the balance after 1 Ra1, despite Black’s material advantage – it is not so easy to find a way to break through the ene
my defenses. However, Lombardy committed “harakiri”: he allowed Fischer to obtain an easily won pawn ending, based on an outside passed pawn.

  1 Re1?? R×c3+! 2 bc R×e5+ 3 Kd2 R×e1 4 K×e1 Kd5 5 Kd2 Kc4 6 h5 b6

  Black gets an outside passed a-pawn by force, which draws the white king to the edge of the board.

  7 Kc2 g5 8 h6 f4 9 g4 a5 10 ba ba 11 Kb2 a4 12 Ka3 K×c3 13 K×a4 Kd4 14 Kb4 Ke3 White resigned.

  Martynov – Ulybin

  Daugavpils 1986

  White had to play 1 Rh1!, intending 2 Rh5. After 1…Kc6! (neither 1…Rc7 2 b3, nor 1…Re7 2 Rh5 Re2 3 R×d5+ Kc6 4 Ra5= is dangerous) 2 Rh5 Rd6 3 f4!?t, White’s more active king and (even more importantly) rook position assure him good compensation for the pawn minus.

  In the game, he played 1 Re1?, misjudging the force of the reply 1…Re7!

  Ulybin allowed his opponent to reestablish material equality, because he knew that he would have a decisive positional advantage in the pawn endgame, thanks to his unstoppable threat to create an outside passed pawn.

  2 R×e7 K×e7 3 K×d5 g6! 4 c4 h5 5 gh gh 6 Ke5 h4 7 Kf4 f5 8 b4 Kd6 9 Ke3

  9…a5! 10 a3 ab 11 ab h3! 12 Kf2 Ke5 13 Kg3 Kd4

  Here we see why Black exchanged a pair of pawns with 9…a5! In this way, he wins the queenside pawns quicker, and can queen his b-pawn before White gets anything going on the kingside.

  14 K×h3 K×c4 15 Kg3 K×b4 16 Kf4 Kc4 17 K×f5 b5 18 f4 b4 19 Ke6 b3 White resigned.

  Tragicomedies

  Nimzovitch – Tarrasch

  San Sebastian 1911

  The actual move made, 1 Kh5?, lost: 1…Rb5! 2 Kg4 R×f5 3 K×f5 a5 4 Ke4

  The outside passed pawn draws the white king to the queenside – but Nimzovitch probably hoped that his pawns would be able to defend themselves, as in the variation 4…a4? 5 Kd3 f5 6 g3! However, Tarrasch does not allow his opponent to connect his pawns.

  4…f5+!

  White resigned, since after 5 Kd4 (5 K×f5 a4) 5…f4! 6 Kc4 Kg6, his pawns are lost.

  White had a draw with 1 Kh7! Rb5 2 R×b5! (I suggest the reader establish for himself that White loses after 2 g4? R×f5 3 gf a5 4 h5 a4 5 h6 a3 6 Kh8 a2 7 h7 Ke7!) 2…ab 3 g4 b4 4 g5=.

  Brüggemann – Darius

  Botzov 1969

  A draw was agreed here, but White can win.

  1 Kf5 Kh7 2 Ke4 Kg6 3 Kd3 K×f6

  If Black does not take the pawn at once, White’s king will capture Black’s queenside, and still have enough time to get back to the kingside: 3…Kf5 4 Kc4 K×f6 5 Kb5 Kg6 (or 5…Kf5 6 K×a5 Kg4 7 K×b4 f5 8 a4 f4 9 a5, with a winning queen endgame) 6 K×a5 f5 7 K×b4 f4 (7…Kh5 8 a4 f4 9 a5+–) 8 Kc3 Kh5 9 Kd3 K×h4 10 Ke2 Kg3 11 Kf1+–.

  4 Ke4!

  There is not time to go after the queenside anymore; however, the situation on the kingside is now a simple win because of the outside passed pawn.

  4…Kg6 5 Kf4+– (after 5…Kh5, both 6 Kg3 and 6 Kf5 are good).

  Exercises

  Two Rook’s Pawns with an Extra Pawn on the Opposite Wing

  Positions in which two rook’s pawns are facing each other, with one side having a distant passed pawn, are fairly common in practice, so it is useful to have a quick and accurate way of evaluating them. The plan to play for a win is obvious: the king will go after the rook’s pawn. His opponent, meanwhile, must eliminate the pawn on the other wing, and then rush the king over to the corner where it can stop the rook’s pawn. Under what circumstances can he succeed?

  White to move wins: 1 a5! Kg7 2 Kf4 Kf6 3 Ke4 Kf7 4 Kd5 Kf6 6 Kc6 K×f5 6 Kb6 Ke6 7 K×a6 Kd7 8 Kb7

  If it is Black to move, after 1…a5!, the position is drawn, as you may easily determine: Black’s king has enough time to get to c8.

  But let’s say that we move the kings and the f-pawn one rank down, or one file to the left; then, once again, Black loses. But what if we also move the queenside pawns one rank down?

  Of course, with the position right in front of us, any question is easily answered. But in practice, such situations often occur at the end of long calculations, and extending such calculations a few moves further still could be most difficult. It would be good to have a definite evaluation of this position immediately, as soon as we lay eyes on it.

  Bähr demonstrated such a means of quick appraisal in 1936. I did not find his rule very convenient for us; in addition, it was not designed to work when the king would be, not to one side, but ahead of the pawn. So therefore I offer a somewhat different method of quickly evaluating this sort of position.

  (1) The first rule is similar to Bähr’s rule: If the rook’s pawn of the stronger side has crossed the middle of the board, it is always a win.

  (2) We shall designate a “normal” position, in which:

  (a) the rook’s pawns, which block one another, are separated by the middle of the board; and

  (b) Black’s king, aiming for the c8-square, can reach it without loss of time. This is because the passed pawn has either traversed the key diagonal h3-c8, or stands upon it.

  The “normal” position is drawn.

  (3) For the kingside passed pawn, every square behind the h3-c8 diagonal is an extra tempo for White. For example: the pawn at f4 means one extra tempo; the pawn at e4 – two. And if the king is not beside the passed pawn, but in front of it, that is another extra tempo.

  And every square the queenside pawns are behind the “normal” position is an extra tempo for the defending side. With pawns at a3/a4, Black has an extra tempo in his favor; with pawns at a2/a3 – two.

  White wins only if the relative number of tempi calculated by the means shown above is in his favor.

  The formulation may seem a bit ungainly, but once memorized, it is quite easy to apply. For example:

  White of course is on move (if it were Black to move, the f-pawn would queen). White wins, because the count is 3:2 in his favor. Black has two tempi, because the queenside pawns are two squares behind the “normal” position. White’s f3-pawn is two squares lower than the h3-c8 diagonal (the f5-square), and his king, being in front of the pawn, gives him three tempi.

  1 Ke4! Ke6 2 Kd4(d3)+–

  1 Ke3? Ke5(f5)= would be a terrible blunder, because then we would have a position where the tempi are 2:2 (White’s king is no longer in front of the f-pawn, but next to it) – which makes it a draw.

  One more useful addendum to the rule. Let’s suppose that White’s passed pawn is a rook’s pawn, with the king in front of it, but the enemy king is boxing his opposite number in on the rook file. This situation is the same as the one in which the king is next to his pawn.

  According to the rule formulated above, this is a draw. And in fact, after 1…Kf5 2 Kh5 (2 Kg3 Kg5 is the “normal” position), Black does not play 2…Kf6? 3 Kg4, when White has an extra tempo, because his king is in front of his pawn, but 2…Kf4! 3 h4 (3 Kg6 Kg3) 3…Kf5 4 Kh6 Kf6 5 Kh7 Kf7(f5) 6 h5 Kf6!, etc.

  It must be noted here that this last rule is inoperative with the pawn on its starting square.

  Black has one tempo (since the queenside pawns are one rank back), but if it is his move, he still loses. The problem lies in the fact that the standard 1…Kf3 is impossible in view of 2 Kg5 Kg2 3 h4, while after 1…Kf5 2 Kg3 Kg5, White has not one, but two tempi (the pawn is below the c8-h3 diagonal, and the king is in front of the pawn).

  With White to move, the position is drawn, even if the queenside pawns are placed as in the “normal” position, because the h-pawn will have to go to h3: 1 Kh5 Kf5 2 h3 (2 Kh6 Kg4=) 2…Kf4!=.

  Let’s look at some more complex examples, in which understanding my proposed rule considerably simplifies the calculation of variations.

  Privorotsky – Petersons

  Riga 1967

  Black has an obvious positional plus. His plan is clear: …Kg6-f5-e4, and then attack the queenside pawns with either bishop or king. This plan can be forestalled by offering a trade of bishops, but this requires accurate calculation.

  1 Bd4! B×d4+ (1…Bh6 2 Kf2 Bc1 3 K×f3 B×b2 4 a4=) 2 cd Kf5 3 Kf2 Ke4 4 d5
! (otherwise 4…K×d4 5 K×f3 Kd3) 4…K×d5 5 K×f3 Kd4 6 Ke2 c3

  On 6…h4 7 Kd2 a5, White plays either 8 Kc2 or 8 a4.

  7 bc+ K×c3 8 h4!!

  Right! Otherwise, Black would play 8…h4! himself, and after picking up the a3-pawn, he wins, because his pawn on the opposite wing has crossed the center of the board. After the text, we have the “normal” – that is, the drawn – position.

  8…Kb3 9 Kd3 K×a3 10 Kc3 a5 11 Kc4! Ka4 12 Kc3 Kb5 13 Kb3 Draw.

  The calculation of this endgame resulted in positions we have examined in one form or another. If White had not been able to appraise them “mechanically,” using the rule shown above, but had had to extend the variations to the end, he would have had to take each of his calculations a dozen moves further – certainly not a simple process.

  The following endgame created even more complex problems for both sides.

  Matanovic – Botvinnik

  Belgrade 1969

  In his notes, Botvinnik analyzed two approaches for White: 1 Rd5, and 1 Rd6+ Ke7 2 Ra6. In fact, he had a third try: 1 Kf2! For example, 1…ed 2 Ke3 Ra1 (2…Rg1 3 Kf2) 3 R×d3 R×a4 4 Rd6+ followed by 5 Ra6, when White must draw.

  But suppose we forget about this possibility, and try to choose the more exact of the two possible rook moves.

  First, we must try some short variations, in order to establish the differences between them, to compare their advantages and their shortcomings.

  On 1 Rd6+ Ke7 2 Ra6, a clear draw follows 2…R×d3 3 R×a5, or 2…Rd2+ 3 Nf2 e3 4 Kf3! (4 R×a5 R×f2+ 5 Kg1 is also possible) 4…e2 (4…ef 5 Kg2) 5 Ra7+. However, the pawn capture on d3 is unpleasant: 2…ed! 3 R×a5 Kd6. Now, 4 Kf2? is bad, in view of 4…Rg1!; so White must continue 4 Ra8, allowing the black king to get closer to his passed d-pawn. Is this rook ending lost or drawn? It is hard to say – which means that it is time to break off the calculation here, and look at the alternative.