Quite rightly Aaron Nimzowitsch is considered one of the role models of the 'positional' chess player. His writings and games have formed the basis for a significant amount of modern middlegame theory. So when I was looking through "500 Master Games of Chess" by Tartakower and Du Mont, I was surprised to see Nimzowitsch use the Latvian Gambit in a game against Rudolf Spielmann in a 1926 tournament. The Latvian isn't known for it's positional features, as it is more of a playground for hack and slash merchants on either side of the board.
Nontheless it is kind of refereshing to see Nimzowitsch willingly steer away from the sterile QGD's that plaugued chess before WWII, and to test his middlegame theories outside of what could be considered safe territory. As it turned out Spielmann emerged the winner, although this wasn't the fault of Black's opening. Indeed, if he had chosen 17. ... Nb4 he would have had an almost winning position instead.
Spielmann,R - Nimzowitsch,A [C40]
Semmering Semmering, 1926
1.e4 e5 2.Nf3 f5 3.Nxe5 Qf6 4.d4 d6 5.Nc4 fxe4 6.Nc3 Qg6 7.d5 Nf6 8.Be3 Be7 9.Qd4 0-0 10.Nd2 c5 11.dxc6 Nxc6 12.Qc4+ Kh8 13.0-0-0 Bg4 14.f3 d5 15.Nxd5 Nxd5 16.Qxd5 exf3 17.gxf3 Rac8 18.Bd3 Bf5 19.Bxf5 Rxf5 20.Qc4 b5 21.Qg4 Qf7 22.Rhg1 Nb4 23.c3 Nxa2+ 24.Kb1 b4 25.Bd4 Bg5 26.c4 b3 27.Ne4 Qg6 28.Qxg5 Rxg5 29.Rxg5 Qf7 30.Nd6 Qxf3 31.Bxg7+ Kg8 32.Be5+ Kf8 33.Rf5+ Qxf5+ 34.Nxf5 Rxc4 35.Rd8+ Kf7 36.Nd6+ 1-0