<Willber: Can somebody please explain what is wrong with b6 rather than a6?>

I assume you mean this line:

33. a4 Kd6 34. a5 Kxd5 35. a6

If you look at this link under "Rule of the Square" you will get a pretty good idea of what is going on in this endgame. White is trying to keep the Black King out of the Square, and White is trying to push a Pawn so that Black's King can't get in the Square: http://en.wikipedia.org/wiki/King_a...

If 35. b6 then 35...axb6 36. axb6 Kc6 and the King easily stops the Pawn because it's inside the square (Some people call that "inside the cordon" or "inside the box").

click for larger view

If 35. b6 axb6 36. a6 Kc6 again puts the King inside the square.

click for larger view

click for larger view

Rybka 3: <23-ply>

[-0.26] 32.Ke3 Ne7 33.Ke4 Nc8 34.Be1 Nb6 35.Bc3 g6 36.Be1 Nc4 37.a4 Kc7 38.h4 Kb6 39.Bc3 Kb7 40.Be1 Kc7 41.g4

32.Bxe5;

This is a classical advantage of a B vs a N in an endgame with pawns on both sides of the board. When most of his pawns are of colour opposite to his B, W often uses his long range piece to re-arrange the configuration of pawns to his advantage.

32...dxe5 33.Ke3!

I think this has to be done immediatly so that the BK does not have access to useful squares around his center pawns.

33...Kd6 34.Ke4.

Now B has the choice between keeping the status quo in the center or forcing things. In either case the answer will be 35.a4(!). For instance:

34...c4

Keeping the status quo would lead to the kind of losing seperated passed pawns (in B's point of view) I discussed in the introductory paragraph above

35.a4! c3 36.Kd3 Kxd5 37.Kxc3 Ke4 38.a5!!

Certainly not 38.Kd2? Kd4! when suddenly the BK heads for the back of the Q-side pawns.

38...Kf4.

Notice how useful the useless(!) W's K-side pawns are!

39.b6 e4 40.b7 e3 41.b8Q+.

Time to check.

Black threatens 32... Nc4 33.a4 Nb6 winning a pawn. This and the possibility of creating a distant passed pawn suggest 32.Bxe5 dxe5 33.a4:

A) 33... Kd6 34.a5

A.1) 34... Kxd5 35.a6

A.1.a) 35... Kd6 36.b6 Kc6 (36... axb6 37.a7 + -) 37.bxa7 + -.

A.1.b) 35... c4 26.b6 is similar to A.1.a.

A.2) 34... Kc7 35.a6

A.2.a) 35... Kb6 36.d6 c4 37.d7 Kc7 38.b6+ axb6 39.a7 + -.

A.2.b) 35... Kd6 36.b6 axb6 37.a7 + -.

A.2.c) 35... c4 36.Ke3 and the white king controls both passed pawns. When Black exhausts his pawn moves, White wins by playing d6 and b6 in the appropriate order.

A.3) 34... c4 35.Ke3 as in A.2.c.

A.4) 34... a6 35.bxa6 Kc7 36.d6+ Kb8 (36... Kc8 37.a7 Kb7 38.d7) 37.d7 Kc7 38.a7 + -.

B) 33... c4 34.a5 c3 35.Ke3 e4 36.a6 c2 37.Kd2 e3+ 38.Kxc2 e2 39.Kd2 + -.

And that will be with great pleasure.

Lets'consider both 32.Bxe5! dxe5 33.a4! Kd6 34.a5 Kxd5 and:

a) 35.a6! (1st d)
b) 35.b6?? (2nd d)

(after 35.a6)

click for larger view

(after 35.b6)

click for larger view

In the 2nd diagram 35...axb6 and either 36.a6 or 36.axb6 is refuted by 36...Kc6 followed by 37...Kc7. Note that the black pawns at c5 and e5 are "self defensible".

In the 1st diagram the situation is quite different: W has a blockade. That is: 35...Kd6;

Or else 36.b6 axb6 37.a7 and the BK is outside of the aquare of the a-pawn.

36.b6;

Anyway!

36...Kc6;

Because after 36...Kd7 37.bxa7 the BK is again too far.

37.bxa7! (d)

click for larger view

And there is no time for 37...Kb6: 38.a8Q.

Now I' m assuming that you know the stuff about "square of the pawn" and "self defensible pawns". I do not know where you are in your "improvement process" on the game. But if you don't know about this stuff and wish to, feel comfortable mentionning it. I will explain in a subsequent post.

Having looked at this position for a couple of minites, my instinct is to play 32. Bxe5 dxe5 33. a4

Now after 33..... Kd6 34. a5 Kxd5 35. b6. If black plays 35....axb6 36. a6! wins (we had a puzzle with a similar position recently). If 35.... Kc6 36. Ke3. Now 36....axb6 again loses to 37. a6. So black must play 36..... Kb7.

Now I don't see a way forward - I think black wins in this line.

So it must be 35. Ke3. But then black has 35....c4.

I have played on from here but white again loses.

I have come off the track somewhere - maybe even with the first move.

32 Bxe5 dxe5
33 a4 Kd6
34 a5 Kxd5
35 a6! Kd6
36 b6 winning

Now this is the third time CG have provided this example of pawn play and perhaps finally I am getting the point! Is there any alternative for black? Obviously not. If black goes to block the a and b pawns then the d pawn will come into play, so black has to take the d pawn...

32 Bxe5 dxe5
33 a4 Kd6
34 a5 Kc7
35 a6! ...

35 ... Kb6
36 d6 winning

35 ... any other move
36 d6+ Kxd6
37 b6 winning

32 Bxe5 de
33 a4 Kd6
34 a5 Kxd5
35 a6! Kd6
36 b6 Kc6
37 ba

the king cant access!

Taking the pawn

36 ... ab
37 a7

Pushing c4 instead of moving king at the 35th

35 ... c4
36 b6

Pushing c4 one move earlier

34 ... c4
35 b6

Earlier again is clever

33 ... c4
34 a5 c3
35 Ke2! Kd7
36 Kd3

Leave, instead of moving king, two passed black pawn is not working

35 ... e4
36 b6 ab
37 a6

But only because CG recently had another problem that featured this same trick. (a6! then b6! resulting in a Queen. As I remember, this was rather recently. There also the Black King could not approach, as the BK could not cross a key square... here it will be c6.)

Alternatives:-

Analysis by Rybka 3 1-cpu:

1. = (-0.25): 31...Kd7 32.Ke3 Ne7 33.Ke4 Nc8 34.Be1 Nb6 35.Bc3 g6 36.Be1 Kc7 37.g4 Kb7 38.h3 Nc4 39.a4 Kc7 40.h4

2. (0.33): 31...h6 32.a4 Nf8 33.a5

3. (0.35): 31...Nf8 32.a4 h6 33.a5 Nd7 34.Ke2 Nf6 35.b6 axb6 36.axb6 Kd7 37.b7 Kc7[] 38.Bxd6+ Kxb7[] 39.Bxc5 Nxd5 40.Bf8 Nf4+ 41.Kf3 Ne6 42.Bb4 Kc7 43.Ke4 Kd7 44.Kf5 g6+

4. (0.43): 31...h5 32.Ke3 Kd7 33.Ke4 Ne7 34.a4 g6 35.a5 Kc7 36.Bf2 Kb7 37.Bh4 Nf5 38.Bf6 a6 39.b6 c4 40.Bc3

5. (0.56): 31...Nh8 32.Ke3 Nf7 33.a4 Kd7 34.a5 Nh6 35.Bf4 Nf5+ 36.Kd3

White's death dive:-