is the position I think - before 27 c4 below.

[Event "4NCL"]
[Site "Nottingham ENG"]
[Date "2005.04.10"]
[Round "8"]
[White "Howell, D"]
[Black "Parker, J"]
[Result "1-0"]
[WhiteElo "2348"]
[BlackElo "2537"]
[ECO "C67x"]
[EventDate "2005.01.01"]

1.e4 e5 2.Nf3 Nc6 3.Bb5 Nf6 4.O-O Nxe4 5.d4 Nd6 6.Bxc6 dxc6 7.dxe5 Nf5 8.Qxd8+ Kxd8 9.Nc3 a5 10.h3 Ke8 11.Bd2 Ne7 12.Rfe1 Nd5 13.a3 Nxc3 14.Bxc3 Be6 15.Nd4 c5 16.Nxe6 fxe6 17.a4 Be7 18.Rad1 b6 19.Rd3 Rf8 20.g3 Rd8 21.Red1 h5 22.Kg2 g6 23.Rxd8+ Bxd8 24.Rd3 Rf7 25.Bd2 Rd7 26.Rxd7 Kxd7 27.c4 Ke8 28.f4 Kf7 29.Kf3 c6 30.g4 Bh4 31.Ke4 Ke7 32.gxh5 gxh5 33.f5 Kf7 34.b3 Ke7 35.Bh6 Kf7 36.Kf4 Be1 37.Ke4 Bh4 38.Be3 Ke7 39.Bg1 Bg3 40.Be3 Bh4 41.Bc1 Kf7 42.Bh6 Ke7 43.Kf4 Kd7 44.Bf8 Be1 45.Kg5 Bg3 46.Kf6 exf5 47.Kxf5 h4 48.Kf6 Ke8 49.Bd6 Kd7 50.Kf7 Kd8 51.Bb8 Kc8 52.e6 Kxb8 53.e7 Kc7 54.e8=Q 1-0

I'm writing a book on the Berlin, and it includes a section on typical endings.

One of these is W: a4 b2 c4 e5 f2 g3 h3 Bd2 Kg2; B: a5 b5 c7 c5 e6 g6 h5 Be7 Kd7. This is Howell-Parker, 4NCL 2003, but similar endings have occurred in other games, eg Deep Fritz-Kramnik 2002.

I have been analysing this position on and off for four years and it's very close. I still don't know whether White is winning or not and why. I think I know, but then a year ago I thought I knew something different. I know we have some pretty cute endgame analysts on the forum, so if anyone cares to tell me where I'm wrong I'll be very grateful.

The game went 1 c4 (fine, Black was threatening ...c4). Ke8 (? according to me; Black had to play 1...g5. White can then go 2 g4 hg 3 hg Ke8 4 f4 gf 5 Bxf4 Kf7 6 Kh3 Kg6 7 Be1 Bg5 (7...Bf8 8 Bh4 Bg7 9 Bf6) 8 Bh4 Bf4 9 Bf6 Kf7, but I don't see how he can win from here. He can win c7 and b6 in return for b2, or alternatively win b6 while playing b3, but either way a position arises with say Bb6 against Bb4 and Ke4 against Kg6, and I don't see how White wins.

2 f4 (? according to me. White had to play 2 h4 after which his plan is Kh3/g4/Kg2/Kf3/Bg5/f4/gxh5/Ke4/f5/Bf6. Black has some moves, but I don't see that they're useful. Black has to be Kd7 and Bf8 at this point, and now b3! Bh6; fxe6 Kxe6; Bd8 Kd7; Bg5! wins f5 for the king and wins) 2...Kf7 3 Kf3 c6 (? according to me; Black had to avoid this and sit tight with say 3...Bd8. then 4 Be1 Ke8 5 g4 Kd7 (4...gh? loses because White recaptures, plays Kh3 and organises Bh4, when the pawn endgame is winning) 6 gh gh; now White takes his king to e4 and plays f5, but Black replies either ...Bg5 or ...Bh4 with this king on d7, and he can stop the king coming in and thus holds.)
4 g4 Bh4, and now White won by playing gxh5 and then organising Bh6 against Bh4 and Ke7, when ...Kd7 loses to Bf8 and ..Kf7 to Bg5 (and ...Be1 to Bg5+ and Bd8). It seems to me that he would have won just the same if the h-pawns were exchanged: he organises the same position after f5 gf; gf with Kf4, Bh6 against ...Ke7, ...Bh4, and wins with ...Be1; Bg5+ Kd7; fe6 Kxe6; Bd8 Bd2+; Ke4 Bc1 (...Be1; Bxb6 Bb4; b3 wins); Bxb6 Bxb2; Bxa5 Bxe5; Bb6 Bd4 (say); a5 Kd7; a6 Kc8; Kf5 Bg7; Ke6 Bf8 and now a tempo trick with; Kf6 Bd6; Kf7 Bg3; Ke6 Bf4; Bxc5 Kc7?; Bd6+, and otherwise White wins with B-d4-e5, Kd6xc6 and so on.

Anyone care to tell me where I'm going wrong or introduce any more completely new plans?