The Kibitzer
by Tim Harding

The Cold Heaven of Rook Endings

Why have rook endgames suddenly become sexy? 

No sooner had I sent off issue 11/99 of my magazine Chess Mail to
the printer, including a glowing review of GM Victor Korchnoi's
book on the subject, than two more books on this class of endgame
arrived in the post. 

One was John Nunn's new edition of his Secrets of Rook
Endgames and the other was The Survival Guide to Rook Endings
by another English grandmaster, John Emms.

Each of the three books is worthwhile and has its own area of
interest. Korchnoi's book concentrates on practical endings he has
played himself, Nunn's uses the Ken Thompson database to
examine endgames with only five pieces on the board while the
Emms book is more wide-ranging and is aimed at the average
player rather than the master or expert.

Instead of reviewing these books, which I have done elsewhere, I
want to take another look at a few rook endings I have played, or
misplayed, down the years which readers may find amusing or
even instructive. By all means inform me of errors that you
discover! I certainly don't claim to know the truth about them and
have not re-analysed them specially for this column; I am relying
on my contemporary notes.

I shall also be looking at some other examples of rook endings
which I think will mostly or entirely be new to readers as I haven't
seen them in books. During the 1970s, long before chess databases
on computers, I collected many endings that interested me in a card
index classified by material on the board. I recently rediscovered
this and will show you some of them. I won't include five-man
endgames which fall within the scope of Nunn's book, because
computers accessing the database can invariably provide the right
result and best move.

Let's start with a very simple example, which occurred in the 8th
game of the Women's World Championship match in Moscow,
1958. Study the diagram carefully before reading on; White has
just played 68 Ke5-f4 and it's Black to move. Elizabeta Bikova-
Olga Rubtsova (See Diagram)

White: Kf4, Rg6; pawn -  g7
Black: Kh2, Rg8; pawn - h4

The game ended 68...h3?? 69 Rg3 (zugzwang) and 1-0, for if
69...Rxg7 70 Rxg7 Kh1 71 Kg3 and 72 Ra7 mating. Does that look
like the natural finish to you? Rubtsova overlooked the saving idea
68...Kh1! which draws by a stalemate resource: 69 Kg4 (69 Rg4 h3
is no different.) 69...h3 70 Kxh3 Rxg7!.

When preparing the endgames for this article, I was reminded of
some lines by the Nobel prizewinning Irish poet W.B. Yeats that
always puzzled me when I was studying for final school exams

Suddenly I saw that cold and rook-delighting heaven That seemed
as though ice burned and was but the more ice...

Whatever the literary critics and Yeats biographers make of those
lines, for a chess player the concept of a 64-square chequered
heaven (or hell) of icy purity populated only by rooks (apart from
kings and pawns of course) has a certain appealing logic. Rooks
may seem the most obvious of chess pieces, "just" moving in
straight lines back and forth, but what awesome power they wield
in a land with no queens, knights and bishops   and yet how
strangely vulnerable they can be against two or three well-
advanced enemy pawns!

My next example also comes from a women's master event. Black
was Mrs Rowena Bruce from Devon, England, who died recently
at the age of 80. She was a major figure in British chess for many
decades and here she came close to outwitting a Russian opponent
who was soon to become one of the top female players in the
world.

Zatulovskaya - Bruce Women's olympiad, Oberhausen, 1961 (See
Diagram)

White: Kc1, Re6, Rh1; pawns - a2, b2, c2, f3, g4, g5 
Black: Kg7, Rc7, Rh8; pawns - a6, b7, d6, g6, h5

This position is quite interesting in itself but is only the prelude to
a rather weird finish. I have never had to play the ending of rook
versus three connected passed pawns with either side and you will
probably be lucky to avoid it, too, but that is what will arise here.

24...Rf7 25 Rxd6 Rxf3 26 gxh5 gxh5 27 g6 h4 28 Rd7+ Kxg6 29
Rxb7 h3 30 Rb6+ Kg5 31 Rxa6 Rhf8 

Black is two pawns down but her passed pawn is very dangerous
while White's have yet to move. 

32 Rd6 Rf1+ 33 Rd1! 

Not 33 Rxf1 Rxf1+ and Black gets a new queen at once. 

33...Rxh1 34 Rxh1 Kg4 (See Diagram)

As the white king is cut off, evidently White will sooner or later
have to give up her rook. If her pawns and king were already part
of the way down the board, Zatulovskaya would probably be
winning but in the actual situation she has to be careful.

35 Rg1+ 

If White had just played 35 a4 Kg3 36 b4 Kg2 37 Rxh3 the black
king would nhave bee on h3 instead of h2, but maybe it doesn't
matter in this case.

35...Kh4 36 a4 h2 37 Rh1 Kh3 38 b4 Kg2 39 Rxh2+ Kxh2 40 a5
(See Diagram)

White Kc1, Pawns a5, b4, c2; Black Kh2, Rf8 It is important for
White to get at least one pawn across the "demarcation line". 

40...Ra8 41 Kb2 

41 Kd2 also came into consideration, as shouldering off the enemy
king is a major technique in such endings.

41...Kg3 42 Kb3 Kf4 43 c4 Ke5 44 Ka4 Kd6 45 Kb5 Rb8+ 46
Ka4 Kc6 47 Kb3 Rc8 48 Kc3 Ra8 49 Kb3 -.

I think a draw is probably the right result? It's a shame this was not
played on, but it's understandable that the players were happy to
agree a draw. The most important work on this ending was done by
Maizelis and later by Kopayev (1958), and is summarised in
Averbakh's standard work on rook endings.

Generally White is recommended to keep his king back to prevent
the rook attacking the pawns from behind (i.e. the first rank). If
White can be forced to place her pawns on a6, b5 and c4, with the
white king on b3 or c3, and the black king on c5 then the pawns
may be vulnerable but the compact position White achieved in the
game may be impregnable?

The nearest I ever got to playing something like that occurred in a
British weekend tournament before the days of blitz finishes. After
a comedy of errors, White reached this position with two rooks and
no pawns against one rook and five pawns.

Tim Harding-Graham P.Burton Leicester open (5), 1971 (See
Diagram)

White: Kf7, Rf6, Rh7 
Black: Kh5, Rg5; pawns - a2, c6, d5, h4, h6.

With only a short time to eat lunch before the final round, this
position (White to move) was left to an adjudicator to sort out. We
returned to find round 6 just begun on the basis that Black had won
and White had lost! A hasty appeal had to be processed, and it was
soon discovered that British midlands stalwart P.N. Wallis had
mistakenly thought the pawns were moving the other way! 

However, with the black pawn on a2 not h7 and White to move,
the win is quite simple: 49 Rfxh6+ Kg4 50 Rxh4+ Kf3 51 Ra4
because with two rooks it is much easier to mop up pawns.

Positions that reduce to even one rook against one pawn can be
very hard to judge and I think the average player or even expert is
quite likely to make serious mistakes. I have certainly lost at least
one game where I was a pawn up in a rook ending but soon ended
up in a lost rook versus pawns position. Any time the weaker side
can obtain a passed pawn is a potential danger situation. The
following ending, between two of Lancashire's strongest players,
was published in B.H. Wood's Chess magazine.

John J. Carleton - John E. Littlewood Southport, 1969 (See
Diagram)

White: Kg1, Ra2; pawns - e4, f3, g3, h2
Black: Kg8; Rb8; d6, e7, f7, g6, h7

How can Black lose this game? Let us see what happens... 

35...Kg7 36 Kf2 Rb3 37 g4 g5 38 Ra5 Kf6 39 h4 gxh4?! 

Now Black is two pawns ahead! Of course it is not for long and he
has weakened his structure. I think we must assume he was short of
time and counted too much on getting a passed pawn. 

40 Rh5 Rb2+ 41 Ke3 Rh2 42 Rxh7 e6 43 Rh5 Ke7 44 Rg5 h3 45
Rh5 Rh1 46 Kf2 Kd7 47 Rh7 (See Diagram)

White: Kf2, Rh7; pawns - e4, f3, g4 
Black: Kd7, Rh1; pawns - d6, e6, f7, h3

Here Black took a silly risk without calculating; his idea is to get
split passed pawns at any price. 

47...d5?! 48 exd5 exd5 49 Rxf7+ Ke6 50 Rf8 d4 51 Rd8 Ke5 52
Kg3 

Now Black's pawns are doomed and White has connected passed
pawns! 

52...Rd1 53 Kxh3 Kf4 54 Kg2 Ke3 55 Re8+ Kd2 56 g5 Ra1 57
f4 d3 58 g6 Ra7 59 f5 Ra4 60 Kg3 Ra1 61 Kf4 Rg1 62 Rd8 Ke2
63 Ke5 d2 64 Kf6 Rg3 

Threatening...Rd3 and promotion but it's too slow. 

65 g7 1-0. Black resigned for if 65...Rd3 66 Rxd3 Kxd3 67 g8Q
d1Q 68 Qd8+ and the f-pawn will decide. 

A few years later I remembered this when I reached an ending of
rook and three pawns versus rook and four in a weekend
tournament game and managed to win. This time there was a blitz
finish. I haven't been able to reconstruct the position but basically
White's king wandered down to c7 to force me to give up my rook
for the pawn. I was able to set up a kingside passed pawn and
White, who was still at least drawing, panicked and managed to
lose.

In correspondence play it is easier to keep your nerve and check the
correct technical procedure than in an over-the-board finish against
a ticking clock.

The following case was a happier experience for me. Tim
Harding-Al Khateeb Ahmed Correspondence Olympiad XII
Preliminaries, 1992-96 (See Diagram)

White: Kh2, Qe4, Rd2; pawns - a2, c4, d3, g2, g3
Black: Kg8, Qe6, Rf8; pawns - a6, b4, d7, g7, h7

Having been outplayed by the IM from Qatar in both the opening
and middle game, White seeks salvation in the rook ending.
Although White still has level material, the position looks lost with
an inferior pawn structure, king well away from the action and
potential outside passed pawn for Black. My active rook is the only
hope. I only give some of the main lines of this ending, which was
analysed in enormous detail at the time.

31 Qxe6+ dxe6 32 Re2 Kf7! 33 Re5 Rd8 34 Ra5 Rxd3 35 Rxa6
Ra3 

Black offers transition to a pawn ending. I decided eventually that
this was lost and I had to make the best of it with rooks on the
board.

36 Rb6 Ra4 37 c5

I think this position should be lost for White (bad King position
and all pawns vulnerable). 

37...Kf6

I thought 37...e5 38 Rb7+ Ke6! was his most dangerous try,
bringing his King to support the b-pawn. Now the position
simplifies so he only has pawns on one wing and my king position
is therefore not quite so serious.

38 c6 Rxa2 39 Rxb4 Rc2 (See Diagram)

40 Rf4+!

It is obviously hopeless to defend the c-pawn as his king comes
across; the king must be driven back. If Black brings his king to the
middle I can attack his kingside pawns.

40...Kg6 41 Rg4+ Kf7 42 Rf4+ Kg8 43 Ra4 

The idea is to keep the rook active at a distance before improving
my king position.

43...Rxc6 44 Ra7 Kf8 45 Kh3 Rc2 46 g4 Rc3+ 47 Kh4 Rc8 48 g5
Re8 

Black has got his rook behind his passed pawn but White's
position doesn't look so bad now. The doubled g-pawn has little
defensive value so I gave up a second pawn temporarily to shatter
his own pawns.

49 g6! hxg6 50 Kg5 Re7 51 Ra2 Kf7 52 Rf2+ Ke8 53 Kxg6 Kd7 

The king begins a long journey. 53...e5? 54 Re2 Kd7 55 Kf5
drawing, as Kd6 is met by Rd2+ and then Re2 again. I was
beginning to have some hope of saving the game now.

54 Rd2+ Kc6 55 Kg5!? 

With the idea K-e5-f4. Theory books say defender should keep his
Pawn off the same rank as the King to avoid his Rook being tied to
passive defence, so I wanted to keep the pawn back. 

55...Rd7

I rather expected 55...Rf7 but after 56 Kg6 Rd7 57 Rc2+ (57 Re2
Kd6 transposes) 57...Kd6 58 Kg5 (idea K-f4-e4 and to force him
back with a check if he advances King) I didn't find a clear win for
him. So first he wants to see where I put my rook.

56 Ra2!?

This It doesn't directly aid my King to get in front of his passed
pawn but gets the Rook more active. 56 Re2 was my original idea
and perhaps I should have stuck to it. What do readers think?

56...Rf7 57 Ra6+ Kd5 58 Ra5+ Kd4 

I expected 58...Ke4 but maybe it makes no difference? His move
avoids blocking his e-pawn but this may not be the crucial factor.
What turns out to matter is that, to avoid the checks, his king lands
up outside the winning zone in the eventual ending of rook versus
g-pawn. At last the pawns get moving now. (See Diagram)

59 g4 e5 60 Ra4+ Ke3 61 Ra3+ Kf2 62 Ra2+ Kg3 63 Ra3+ Rf3 

The only way to escape the checks, but now I can force him to
block his passed pawn.

64 Ra4 Re3 65 Ra7 Rf3 -.

To my surprise, my opponent now faxed a draw offer (66 Ra4 Re3
etc. repeating). Over-the-board, Black would certainly play on for
several more moves. This was the crucial half-point for me to
secure my ICCF International Master title.

I had analysed 65...e4 (instead of 65...Rf3) 66 Re7 to an apparent
draw, although the variations are long, and presumably Black had
done the same. Play could go 66...Re2 67 Kh5 e3 68 Kg5 Re1 69
Kh5 Rh1+ 70 Kg6 Kf2 71 g5 Rg1 72 Rf7+ Ke1 73 Rxg7 e2 74
Re7 Kd2 75 Rxe2+ (Entering the realm of the endgame databases)
75...Kxe2 76 Kf6 Rf1+ 77 Ke6 Ra1 78 g6 Rg1 79 Kf7 Rf1+ 80
Ke6 Rd1 81 g7=. 

Somehow the technical endings which fill the pages of endgame
books rarely arise in practice, while one sometimes get positions
which look like they should be in the books but aren't. I had to
defend the following ending for a whole adjournment session
(remember adjournments?) at the British Chess Federation Major
open, Ayr, 1977. The endgame of king, rook and g-pawn versus
king and rook with the defending king secure on the queening
square is well known to be a dead draw. Black has two 100% easy
drawing plans. But does it make a difference if White has two g-
pawns instead of one? 

The standard Averbakh text says that when the defender's king
stands in the way of the doubled pawns, the result is normally a
draw but my opponent thought it was worth a try.

The normal defensive method is to keep the black rook on the sixth
rank (a6-b6-c6) until White plays g5-g6 and then switch it to the
first rank (a1/b1/c1) and give checks from behind. I wasn't happy
that this works when White has a second g-pawn as the king
shelter, so I adopted the alternative defence of keeping the rook on
the back rank. (See Diagram)

White: Kf5, Rb7; pawns - g2, g5
Black: Kg8, Ra8

Johannes - Tim Harding Drawn with either player to move. After
about 20 moves of manoeuvring, White gave away his front g-
pawn and tried for another hour with only one pawn but I saw no
point in changing my defensive plan and eventually White offered
a draw. I spent the whole morning walking around looking at all
the other adjournments while White sat at the board trying to
concoct futile traps. 

(Health Warning: the back rank defence only works with b-pawns
and g-pawns! Does this mean there are winning chances with
central pawns by keeping the rear pawn well back to block
checks?)

Here is another ending from my card index. The 1968 Monaco
open was the first international open I ever played in; this finish
arose between two of the leaders and I noted it down. Kupreyanov
- Masic (See Diagram)

White: Kf3, Ra7; pawns - c6, f7 
Black: Kh2, Rf8, Bb8

The question now is what rook ending will arise? White continued
53 Rb7? Kh3 54 Re7 Bh2 55 c7 Kh4 56 Ke4 Bxc7 57 Rxc7 Kg5
58 Ke5 Kg6 59 Ke6 Kg7 60 Re7 Ra8 61 Rd7 Ra1 62 Ke7 Re1+
63 Kd8 Kf8  . I made notes from the post-mortem where it
seemed that White had missed a win by 53 Rd7 Bg3 (53...Kh3 54
c7 is a decisive tempo ahead of the game Black's king on h3
instead of h4.) 54 Kg4! (54 c7 Kh3!=) 54...Kg2 (or 54...Be5 55
Kf5; or 54...Rc8? 55 Rd2+; or 54...Bb8 55 c7 Ba7 56 c8Q) 55 c7
(threatening Rd8), and White wins.

I will conclude by showing a few more practical endgames that I
have been involved in, for better or worse. Mostly worse.

Tim Harding- John Elburg Evans Gambit Thematic, corr 1989
(See Diagram)

White: Kh1, Rf1, Nf3; pawns - b3, c3, e5, f4, g2, h2 
Black: Ke7, Ra8, Bc6; pawns - b7, b6, c7, f7, g7, h7

Black has secured his positional objectives in Lasker's Defence
and now mops up

23...Bxf3 24 gxf3 Ra2 25 Rd1 Rb2 26 b4 Rc2 27 Rd3 Ke6 28
Kg1 h5 29 h4 b5 30 Kh1 b6 31 Kg1 f6 32 Re3 Rd2 33 exf6+
Kxf6 34 Re5 Rd3 35 Rxb5 Rxc3 36 Rxh5 Rxf3 0 1.

Positional advantage was decisive in the next case too. Harding -
Peter Sowray North Atlantic Team Tournament-2, corr 1985. (See
Diagram) 

White: Ke2, Rd1, Be3; pawns - a5, b4, c3, c2, f2, g4, h3 
Black: Ke6, Rf8, Bb6; pawns - a6, c4, c7, d5, e5, g7, h7

28...Bxe3 29 fxe3 c6 30 Rh1 h6 31 Rg1 g6 32 Rh1 e4 33 h4 Rf3!
34 h5 g5 35 Rg1 Kd6 36 Rg2 Ke7 37 Rg1 Rh3 38 Kf2 Kd6 39
Rf1 Rh2+ 40 Kg3 Re2 41 Rf6+ Ke5! -+

Black has found the only way to win the ending. I could have
resigned here but played on for a few moves as we had another
game in a different event.

The start of another rook ending is shown in the next diagram.
Probably Black can hold but I didn't put up very effective
resistance to GM McNab. Colin McNab - Tim Harding
NATT3 02 9094 corr, 1990 (See Diagram)

White: Kf4, Rb6; pawns - d4, e4, f2, g3, h4 
Black: Ke6, Rc7; pawns - c6, d5, f7, g6, h5

50 e5 f6 51 exf6 Kxf6 52 f3 Rc8 

This is passive. Maybe Black can hold by 52...Kg7 53 Kg5 (or 53
Ke5 Re7+ 54 Kd6 Re3) 53...Rf7 54 g4 Rxf3 55 Rxc6 (55 Rb7+
Kf8) 55...hxg4 56 Rxg6+ Kf7 57 Kxg4 Rd3.

53 g4 Rc7 54 Ra6 hxg4 55 fxg4 Rc8 56 Ra7 g5+ 57 hxg5+ Kg6
58 Ra6 Rf8+ 59 Ke5Kxg5 60 Rxc6 Kxg4 61 Kxd5 Kf5 62 Kc5
Re8 63 d5 Re7 64 Rc8 1 0.

In the next example, I drew when I probably should have won. M.
Krecmer - Tim Harding 4th European Team Ch, Preliminaries,
corr 1988 (See Diagram) 

White: Kh5, Rf2, Rf3; pawns - a4, b3, h2 
Black: Ke6, Rb4, Ne3; pawns - a5, b7, c6, e5, h6

Black has fought back from a difficult early middle-game where he
had to sacrifice the exchange. Now his mating threats force the
return of the material and a favourable rook ending. 

51...Ng4 52 Re2 Nf6+ 53 Kxh6 Rh4+ 54 Kg5 Rg4+ 55 Kh6 Kf7!
56 Rxf6+ 

Not 56 Rxe5?? Rg6 mate, nor 56 Rg3? Rh4+ 57 Kg5 Rh5 mate. 

56...Kxf6 57 Kh5 Rf4 58 Rc2 Kf5 59 h4 Rf3 60 Rc5 Rxb3 61
Rxa5 (See Diagram)

What a change from the first diagram. White should surely be lost.
Unfortunately here I misjudged White's counterplay and let him
off by 61...Kf4?! 62 Kg6 Rh3 63 h5 e4 64 Rf5+! Kg4 65 Rg5+
Kf3 66 Rf5+ Ke2 67 h6 Kd2 68 h7 Rxh7 69 Kxh7 e3 70 Rf7
White avoids the last trap 70 Re5?? e2 71 Kg6 c5! 72 Kf5 c4 73
Kf4 c3 74 Kf3 e1Q +. 70...e2 71 Rd7+ Kc2  .

Returning to the last diagram, I later showed the ending to Irish IM
Mark Quinn. He asked me why I hadn't simply played 61...Rd3 62
Ra8 (if 62 Ra7 Rd7 with the idea ...e4 etc.) 62...Rd7 63 Kh6 Re7.
It seems that this consolidating plan would have won quite
comfortably.

The last example from my own games came in an over-the-board
tournament and it is chiefly interesting for a trap that I managed to
spot in time. Harding,T - Mabbs,D Islington open, London, 1971
(See Diagram)

White: Kf2, Rb7; pawns - d3, e4, f5, g2, h3 
Black: Kg8, Rd8; pawns - c5, e5, f6, g7, h7

33 Ke3 

Now the weak c5-pawn is a big factor.

33...Rc8 

If 33...Ra8 34 Rb2! Ra3 35 Rc2 Ra5 36 Kd2.

34 Rb2 g5?! 

Evidently Black worried that he cannot reach the standard 3 pawn
v 4 pawn drawing formation because of White's pawn on f5, but
this weakening advance seems to have no other point than to hope
White will be tempted to show he knows the en passant rule?
However, there was a subtle idea. If 34...Kf7 35 Kd2 Ke7 36 Rb7+
while if 34...c4 35 Rc2 c3 White can either play d4 to win the c-
pawn or try marching his king around via the a-file.

35 Kd2 Kg7

Objectively 35...Rc7 is better.

36 Rb7+ Kh6 37 Rf7 Rc6 38 Kc3 Ra6 39 Kc4 Ra2 (See
Diagram)

40 Kd5! 1-0.

Black resigned because 40...Rxg2 41 Ke6 followed by Rxf6+, Rf7,
f5-f6 is an easy win for White. On the other hand, 40 Rxf6+ might
be messy. However, what Black was really hoping for was this40
g4? Ra4+ 41 Kxc5 Rc4+!! and White's king can never escape the
checks, because capturing the rook stalemates Black!