If a pawn could “promote” to a pawn…

Lately I’ve been visiting the forum of Chess.com, and came across a topic that posed an interesting question. Suppose it is legal for a pawn to stay as a pawn upon reaching the eighth rank; are there situations in which such a promotion – or rather, non-promotion – would be the best move? Apparently the question was answered in the affirmative back in the 19th century, when the pawn promotion rule was in dispute. A position cited by Steinitz demonstrates how a pawn promoting to a pawn (P=P) could be the only way to salvage a draw, through stalemate. Although this problem has reasonable economy with thirteen units, that seems improvable and I constructed an eight-unit setting, using only pawns besides the kings.

After 1.e8=P!, Black cannot release the stalemate. White would lose with 1.e8=Q? g3 2.Qg6 g2+ 3.Qxg2+ fxg2.

What about a position in which only a P=P move could force a win? This is a much harder task, of course, and at first I was sceptical that it was possible to arrange. In such a problem, the only practicable reason that other choices of promoted pieces would fail is that White is induced to use these pieces to stalemate Black. And given that a newly promoted rook, bishop, and knight will each control completely different squares, at least three different stalemates need to be organised in response. Be that as it may, I couldn’t resist the challenge and eventually came up with the following position.

The black queen threatens mates on a8 and f7, severely limiting White’s options. 1.exf8=Q? or 1.exf8=R? pins the knight and stalemates Black immediately. 1.exf8=B? guards g7 and allows Black to deflect the f5-knight with 1…Se7+ 2.Sxe7= (or 2.Bxe7=). 1.exf8=S? controls h7, which means Black can eliminate the f6-knight with 1…Sxf6+ 2.gxf6=. If White attempts 1.e8=Q?, Black still draws with 1…Qxf7+ 2.Qxf7 Se7+ 3.Qxe7= or 3.Sxe7=. Only 1.exf8=P! wins, avoiding all the stalemate traps. Best for Black is 1…Sxf6+ but after 2.gxf6 Kh7, White is still a piece ahead and so wins the ending easily.

I wondered if there were precursors to these problems, or if the tasks could be arranged more economically. The best place to ask such questions is the Mat Plus forum, which attracts an active community of problem composers. The answers I received were striking, and include a three-mover by the great Sam Loyd that is solved by a P=P key! The Draw task was accomplished by a few problemists in miniature, and my Win study above was modified in an amusing way. Here’s a link to check out these compositions: A pawn remaining as a pawn upon promotion.