Underpromotion Milestones in Endgame Studies

By Harold van der Heijden

Some chess composition challenges – in jargon tasks – are so difficult that it takes decades to achieve them. In 1996 I published my booklet Pawn Promotion, in which I summarized the state of the art around underpromotions in endgame studies. The most important omission then was the Babson task in an endgame study – the last underpromotion field yet to be conquered. And 28 years later that situation had not changed much.

But very recently (September 7, to be exact), the magician composer of our art, Gady Costeff (ISR/USA) sent me an original study offering it for publication in “my” magazine EG [see EG-info.docx]. As it was well after the deadline of issue 238 I almost ignored the mail as I was busy editing. But when I did have a look, I couldn’t believe my eyes reading that Gady had accomplished the “impossible” Babson task!

Of course, Gady knows of my passion for underpromotions, but obviously I was very honoured that he granted EG this scoop. Before showing Gady’s fantastic achievement, here is an overview of other underpromotion milestones.

The first underpromotion study

Obviously, a knight moves completely different from a queen, rook or bishop and may be a stronger piece than a queen (e.g. checking the opponent’s king after the promotion). In contrast, promoting to rook or bishop means that the new piece is weaker than a queen. You will appreciate that such an underpromotion is always associated with stalemate (in a win study to prevent stalemate, and in a draw study with self-stalemate by incarceration or pinning of the promoted piece).

The most elementary position requiring a promotion to rook as the only winning move is: wKb4 (or wKa4) Pb7, bKa6. Of course, 1.b8Q? would be stalemate, while 1.b8B or 1.b8S+ leave White with insufficient material for a win. But 1.b8R! wins.

That scheme was used for the first time in an endgame study by Wouter Mees, De Schaakwereld 1939. As always one wonders on hindsight why this position was not already found by the composers of the 19th century or earlier as it is so obvious.

The very first underpromotion studies were published by Ignazio Calvi, an Italian patriot of the 19th century. He published his study in the French chess magazine Le Palamède of April 1836, which by the way was the very first issue of the world’s first periodical on chess!

The study with the bishop promotion he originally published proved to be unsound. Nine (!) years later, he published a new version. But it still was the first correct study with promotion to bishop:

1.gxh7 Bxe3 2.h8B! (2.h8Q? Bd4+ 3.Qxd4 stalemate). An interesting subline is 1...Bb4 as now 2.h8B? Bc3+ 3.Bxc3 would be stalemate, while 2.h8Q! Bc3+ 3.Qc3 is not stalemate, but mate!

His rook study appeared in the same year:

1.a8R! Sh7 2.Rg8! Sg5 (threatens mate) 3.Rxg5 wins. After 1.a8Q? Sh7 2.Qg8 Sg5 3.Qxg5 would be stalemate.

The AUW task

Here is a warning: for good endgame studies we require an economical setting (number of pieces on the board and no redundant piece exchanges during play), surprise and intricate play and similar aesthetic features. Tasks are often so difficult to accomplish that we must be less critical.

The so-called Allumwandlung (AUW) task requires four different promotions, and the first examples of chess problems date back to 1859. Niels Hoeg eventually came up with a perfect AUW in an orthodox problem in 1905.

Just over a century ago, underpromotion studies gained more interest. In the late 1920’s the composition magazine L’Échiquier even organized a thematic tourney requesting AUW studies. But not a single correct study was submitted! The famous composer Henri Rinck (who had composed a study with three promotions) then wrote: “this Himalaya of chess will remain unconquered even if a million francs were offered as a reward”.

For long it was thought that Harold Maurice Lommer (picture) was the first to accomplish this task in 1933. His prize was.... twenty francs. Soon, new AUW studies were composed by others (e.g. by the judge of the tourney André Chéron). However, in the 1980’s it became known that the Swede Sigurd Clausén had already published an AUW study in 1927, i.e. well before Lommer. Unfortunately, that study was found to be unsound later (only in 1984!) but the problem was also quickly fixed. We assume that also Clausén would have been able to correct the study at the time if he only had known of the unsoundness. Despite that, we reproduce Lommer’s study here:

1.gxh7+ and now:
A) 1...Qxh7 2.exf8Q mate.
B) 1...Kh8 2.exf8R+! (2.exf8Q+? Kxh7 3.gxh5 Re1+ 4.Kxe1 stalemate) 2...Kxh7 3.gxh5 wins.
C) 1...Kg7 2.exf7B+! (2.exf8Q+? Kxh7 3.gxh5 Re1+ 4.Kxe1 stalemate) 2...Kxh7 3.gxh5 wins.
D) 1...Kxh7 2.exf8S+! Kg8 3.gxh5 wins.

Apart from an AUW with parallel lines, in studies also an interesting task is an AUW with subsequent (sequential rather than parallel) promotions. Again, Harold Lommer was the composer to manage it for the first time:

White must be careful not to walk into a mate on the first rank or the a-file. 1.fxg8Q! hxg1Q+ 2,Qxg1 Kb3 3.c8S! Rb5 4.d8B! (4.d8Q? e.g. e1Q+ 5.Qxe1 Ra5+ 6.Qxa5 c1Q+ 7.Qxc1 stalemate) 4...Re5 5.e8R! (5.e8Q? e1Q+ 6.Qxe1 Ra5+! 7.Bxa5 c1Q+ 8.Qxc1 stalemate) 5...Rxe8 6.Be7 wins.

Here are all the studies in a replay window:

Coming soon: Part two – The Babson Task