Mario Velucchi, 29 Mar 1995
ERRATA-CORRIGE, Dawson's Book - 1915
F.Dusold N.27a - Dawson's Book 1915, p. 59 . . . . . . . . (x) Black . . P . . . . . P . . (P) . . . . (P)(P) . P . . . . (K) P . P (P) . . . . (P) K . (P) . . . . P . . P . P . . . . . . . . . White retracts his last move and self-mates in one.
Retract: 1.Pd3-d4 Play: 1.Pd4
But, in Ceriani's Book, I read, this problem is wrong (Dawson 1/29/1934). Why? [Trivial]
I am not sure what the {Trivial} means, but there may be two reasons why the problem is wrong: Not the position Dawson intended (after all, whatever the last move, if you retract it you must repeat it again) or that the position is cooked, as the following moves lead to a position in which the d pawn is on d2 not d3, and retracting and replaying allows the capture on d3.
1. a4 g5 2. Nf3 g4 3. Ng1 g3 4. hxg3 Bh6 5. a5 Bf4 6. gxf4 Nf6 7. Ra3 Rg8 8. Rb3 Rg7 9. Rb6 axb6 10. a6 Rg3 11. Rh5 Rf3 12. Ra5 bxa5 13. Nh3 b5 14. Ng5 Rh3 15. c3 Rh1 16. Qb3 Ra7 17. Kd1 Rxf1+ 18. Kc2 Rxc1+ 19. Kd3 Rh1 20. Kc2 Rh3 21. Qa2 Re3 22. fxe3 e5 23. e4 Ke7 24. Na3 Kd6 25. Qe6+ Kc5 26. Nb1 Ng8 27. Qf5 Kc4 28. Kc1 Kb3 29. Qe6+ Ka4 30. Na3 Nf6 31. Kd1 Bb7 32. Qe7 Bd5 33. exd5 e4 34. Qe8 Ng4 35. Qe7 Ne5 36. fxe5 c5 37. Kc1 Nc6 38. Kc2 Nb4+ 39. cxb4 c4 40. Kc1 Qh8 41. Nc2 Qg8 42. Na1 Qh8 43. Kc2 Qg8 44. Nb3 cxb3+ 45. Kc3 e3 46. Nxh7 f5 47. Ng5 Qg6 48. Ne4 fxe4 49. Qd6 Qxd6 50. exd6 Rc7+ 51. dxc7 d6
I don't claim this sequence is best possible, but it does prove that the last move might NOT have been d3-d4.