Karpova: The annotator (Riga playing committee) explains that White abstained from <9.dxe5 dxe5>, since the ♙c6 cannot be attacked in the near future and the ♗e7 gets more freedom of action.<13.Nxc6 Qxe4 14.Nxf6+ Nxf6 15.Nd4> leads to nothing for White.
14...f5: Black goes on the offensive. If <15.Qc4+ Kh8 16.Qxc6 Nc5 17.Qxa8? Bb7> and <...Ra8> - however, Georg Marco points out <18.Qxf8+ Qxf8 19.exf5>.
Instead of 16...fxe4, <16...Ne5> was suggested as giving Black excellent play, and thoroughly annotated.
17.Qxg7+: Not <17.Rxe4 Ne5! 18.f4 c5!> and White stands worse.
After 22.f3<!>, Black is forced to give up a ♙ (<22...exf3 23.Re8+ Kf7 24.Rae1> (further analysis follows)).
This game was played by telegraph at the same time as Stockholm vs Riga, 1904 and was part of a 2-games match. Since Stockholm had lost the other game and was down a ♙ in this game, the game was not continued. Stockholm saw probably no chance to level the score.
Source: 'Wiener Schachzeitung', September-October 1905, pp. 318-319
Both games were played from January to October 1904 by telegraph between <Stockholm> and <Riga> (same source, p. 316).