White draws

With ...hxg1=Q and ...Rxc8 looming, White stands with their back to the wall. Only 1.Rh8! keeps the game alive, hoping for liquidation into a theoretically drawn R versus B+R endgame after, say, 1...Kxg1 2.Rxd4 Re5 3.Rxh2 Bxh2 4.Ke3, etc., when the e4-pawn falls. But Black won't let their adversary off the hook quite that readily. The main line continues: 1...Kg2!? 2.Rxh2+ Bxh2 3.Rxd4 Re5 — bringing us to a critical moment.

The g1-knight is trapped and will fall on the next move. Meanwhile, the e4-pawn can no longer be pursued: 4.Ke3? now runs into 4...Bxg1+ 5.Kf4 Bxd4 — and Black wins. So, is it over for White? Note that, even though R versus B+R is a theoretical draw, that extra pawn tips the balance heavily in Black’s favour.

Indeed, White could be forgiven for resigning here, were it not for one astonishing resource: 4.Nh3!! — if the knight must perish, let it do so on a square of its own choosing. 4...Kxh3 5.Ke3! — White walks into the skewer with eyes wide open. But why? 5...Bg1+ 6.Kf4, and now 6...Bxd4 results in stalemate, courtesy of Black’s king on h3. That, all along, was the hidden idea!

Needless to say, Black isn't obliged to allow the stalemate — but avoiding it means either conceding the e4-pawn to enter a theoretically drawn endgame, or settling for repetition with 5...Bg1+ 6.Kf4 Bh2+ 7.Ke3 Bg1+, etc. One way or another, the result is the same — a draw.

(For ease of solving, the first two moves of the original study have been omitted)