This might turn out to be a mistake - something goes wrong hereaboouts. Maybe 34.Kg1 was the way to go, to obviate, among other things, any fork threats on f2. The knight on e3 would have no real targets.

The game was already sharply veering south, and this is symptomatic of White's increasing awareness of it. Where did he go wrong? It is not so easy to discover. A possible alternate continuation runs 35.Rxf6 Kxf6! 36.Kg1 Ke5 37.g3 f5 38.Re1 fxe4 39.Rxe4ch Kd6 40.Rxe7 Kxe7 after which there is nothing White can do to save the d-pawn against ...Ng4-e5xd3.

35...Rxf4 would have won a whole rook.

White was hoping to stymie Black's Q-side pawns. It might have worked but for the unfavourable position of his pieces.

Probably played on in the interests of his team. The point is gone, but .... you never know!

I would have played 39...Raa2 here, but everything wins now.

Excellent Charles, thanks for the annotation.

I don't think Black was ever at any real disadvantage during the whole course of the game, but, until things started to turn, didn't have any substantial edge, either. Possibly the seeds of White's downfall were sown by his decision to close down the Q-side in the interests of a K-side attack. That Black easily held White off before the attack could be properly developed suggests that the whole strategic decision might well have been mistaken. Interesting and instructive on those grounds - something of an advanced lesson. I might well have made the same strategic decision myself.