I have long held a sneaking suspicion that some of the most brilliant games in history have occurred by accident. Is there, after all, any detectable difference between a brilliant sacrifice and an oversight that turns out well?

Take this game from the recent Reykjavik tournament. When White played the artificial-looking 16.Rc3 (see diagram), had he seen Black's reply? After 16...e4 White can hardly allow 17.Nxe4 Ne5, when too many of his pieces suddenly come under attack. If White's 16.Rc3, 17.Qg3 and 18.Rcc1 was all planned in advance, it was a truly magnificent and brave conception. But even if it was a piece of desperate improvisation after an oversight, it still deserves applause.

Black's problems stemmed from his poor opening play. After 12.d5! and 13.Nd4! White was clearly on top: 13...exd4 loses to 14.Bxf6 gxf6 15.Qh5. After White's sacrifices, Black was two pieces ahead, but still had no defence. 28...Kd7 would have led to mate after 29.Qe8+ Kd6 30.Qe5+ Kd7 31.Qe6.

White: R Djurhuus

Black: K Edvardsson

1 Nf3 Nf6 20 Nxg7 Kxg7

2 d4 e6 21 Bxf6+ Kxf6

3 Bg5 c5 22 Ne4+ Ke7

4 e3 Be7 23 Qg5+ Kd7

5 Nbd2 cxd4 24 Qf5+ Ke7

6 exd4 Qc7 25 Qf6+ Ke8

7 Bd3 b6 26 Nxd6+ Kd7

8 0-0 Bb7 27 Qxf7+ Kxd6

9 Re1 d6 28 Qxf8+ Qe7

10 c4 0-0 29 Rxe7 Bxe7

11 Rc1 Re8 30 Qf4+ Kd7

12 d5 e5 31 Qf5+ Kd6

13 Nd4 Nbd7 32 Qxd3 Bb7

14 Nf5 Bd8 33 Qg3+ Kd7

15 Qf3 Bc8 34 Qg4+ Ke8

16 Rc3 e4 35 Qg8+ Bf8

17 Qg3 exd3 36 Qe6+ Kd8

18 Rcc1 Rxe1+ 37 Qf7 resigns

19 Rxe1 Nf8