But it also rids yourself of your more problematic prelate.

This retreat is understandable, but it doesn't really change anything. White might as well have saved a tempo and played 36.Qxe3 Rxe3 37.b5, obtaining some useful counter-play. E.g. 37...cxb5!? 38.Bxd5ch Kf8 39.Bc6 Rxc3 40.Bxb5 Rxa3? 41.Rd8ch ... and suddenly Black has a problem. That sample line does to show, I think, how Black would have had to take care to retain winning chances.

Observe that with the extra tempo, Black shut down White's chance for counterplay by herself playing ...b5. That it leaves the c-pawn backward is of little consequence here, as that pawn also protects the passed pawn at d5. The thing about 'opposite coloured bishops' is that, in this instance, Black is a piece ahead on the dark squares, White a piece ahead on the light squares. The bad part for White is that the enemy pawns are keeping his bishop penned in, whilst Black's Bishop is free to roam, without let or hindrance, the length and breadth of England.

All but one black pawn is in white squares. This means that white can make it hard for black to move any pawn, since after any exchange black pawns will be exposed (for instance: Rc2, d4?, Bxc6). Black has plenty of mobility with his bishop, and white's queenside pawns are stucked on black squares. Black has the e column, with joined threats from the bishop and the rook.

For white: 41. Rxc6 Rxc6 42. Bxd5+

Well spotted. Black's 40th was a terrible mistake that ought to have thrown away the win. No doubt this is noted in the GK analysis. But even on general grounds I dislike this passive move. Surely Black could have done better? 40...Re3! comes to mind at once, with the ideas [A] 41.Be2 Rxg3 threatening mate; [B] 41.Bd1 Rc3 (or ...Rxg3) 42.Re2 Be3 43...d4; and [C] 41.Kg2 Rc3 etc.

Despite her two extra pawns, Black is leaving herself open to effective counter-attacks. I was beginning to think the win has already gone - especially if White can contrive without cost to exchange rooks. But Black comes up with as fine an endgame combination as I've seen in this type of game.

It's best to reinforce strength than weakness. Black chooses right in my view.

What this does is give Black the option, when it becomes necessary, to sacrifice the bishop to stop the h-pawn. Black's four connected passed pawns will have no difficulty barging past the White bishop to promote. Black is now winning.

I think 52...c5 offered better chances, with this sort of thing in mind: 53.h5 c4 54.h6 Bd2 55.Be4 Bxh6 56.Bc6 a6 57.Bb7 a5 58.Ba6 c3 and wins.

In my view, White should draw from here, beginning 59.Kc5! Bxe5 60.Kxc6 b4 61.Ba2 Kf6 62.Kb5 Bc3 63.Kc4 then 64.Kb3 ..., 65.Bb1... and then play the bishop back and forth along the b1-h7 diagonal.

It seems that 60.Bb3 ought to have been given the preference: a 'waiting' move.

An interesting ending - very interesting - and it need not have gone all Black's way, neither. But 2 extra pawns even in this type of end game is not easy to pull back. (One thing I am pleased about: I made all my comments before looking at the GK analysis, and by and large they concur.)