The ancient art of Selfmate December 28, 2005
Click "Stop" on your browser to stop the music and "Refresh" to start it.

Nigel Short confessed that he found them, "well, a little bit kinky". But selfmates have been around for almost as long as chess itself, at least in Europe. The first know selfmate puzzle appeared in the late thirteenth century and were originally known as sui-mates.

In a selfmate the task is for White, who moves first, to force Black to deliver checkmate, against his will, in a specific number of moves. The black defence is to do anything to avoid mating the opponent, even to the extent of getting mated himself.

15th century woodcut showing a monk solving a chess problem

Selfmates did not appear in the well-known Arabic collection of mansubat (or chess problems), and were very likely an innovation that appeared soon after chess reached Europe. The earliest example is to be found in the famous Bonus Socius manuscript, a compendium of chess problems that dates back to the 13th century and was written by a scribe named Nicholaus, probably hailing from the town of Nicolai in today's Belgium.

57a Bonus Socius, 13th Century

Selfmate in 15 moves exactly

This first selfmate is problem number 57a in the Bonus Socius manuscript. In its original form it had a fers on h2. This was a piece from the original game of Shatranj, which moved just one square diagonally. It was a very weak piece and later replaced by today's queen. We have replaced the fers with a pawn in the above diagram. That does not change the character of the problem.

If you have never solved – or even seen – a selfmate before you may be baffled as to how White, with such a massive material advantage, can force Black to deliver mate (with the one lonely pawn). Remember that Black can repeat moves, or break out with his king and move aimlessly around to thwart the white intentions. A very precise manoeuvre is required to bring the black king to the square f1. On the very next move White must move a piece to g2 and, having made sure that the black king cannot move, force Black to take the piece on g2 and deliver mate.

The solution given in Bonus Socius is as follows: 1.Nc6 Kb6 2.Na5 Kb5 3.Nc4 Kb4 4.Na3 Kb3 5.Nc2 Kb2 6.Na1 Kb1. Beautiful how the knight has forced the black king, which has always had just one move, to the bottom rank. 7.Rcb8+ Kc1 8.Ra2. Now the black king will be constrained to the first rank. 8...Kd1 9.Rc8 Ke1 10.Rd8 Kf1. It is being forced towards the white king. 11.Rd3 Ke1 12.Nb3 Kf1 13.Nd4 Ke1 14.Nf3+ Kf1. Now the black king is where we want it, and it is completely immobilised. So: 15.Rg2 hxg2 mate. [Click to replay]

We must note that problemists in the 19th century discovered a faster selfmate in the above position, one that could be forced in just 13 moves, without the use of the knight in the second part: 9.Nb3 Ke1 10.Rd2 Kf1 11.Rc8 Ke1 12.Re8+ Kf1 13.Rg2 hxg2 mate.

Here is a simple selfmate for you to practice with.

Noam D. Elkies, Harvard 2002

Selfmate in two

The author composed this simple selfmate as an illustrative position for a seminar on Chess and Mathematics for first-year students at Harvard. Remember: White has to force an uncooperative Black to deliver mate in two moves at most. There is an astonishing little subtlety contained in this problem.

And here is one provided by our problem expert (and problem solving world champion) John Nunn:

Edgar Holladay, British Chess Magazine 1965

Selfmate in 10

John writes: "It’s not easy to see how White can force Black to deliver mate with his rook, and you may need a leap of the imagination to spot the first move. After that the rest is (relatively) easy." For problem enthusiasts we leave it at that. For those who still feel uneasy about this kinky type of problem here are some pointers.

Frederic Friedel