Karpova: The annotator notes that <7...Ne4> is correct.
<11.Bf4> or <11.Nc3> are suggested instead of 11.c3, which was played to enable Bc2 and Qd3.
The move is not annotated, but 16.Bxc5 appears to be too much of a concession. White could simply have played 16.Qe2, but he may have feared ...Ne4.
21.g4 is a serious inaccuracy and, as the annotator notes, 21...g6 was unnecessary. The ♙f5 is well-protected. <21...b5> is suggested, but 21...Qc6 may well be stronger (after 22.g5 d4, the ♘f3 is pinned).
22.h4 is again weak and was not annotated. Now, Black could have played 22...d4, but he misses the chance. Better was 22.Qe2 or maybe even 22.Kg1.
The position after 26.Kg3 is interesting, since Black seems to have a clear advantage. The annotator conjectures that <26...c4> may have already been possible, although the ♘d4 could become bothersome (but 27.Qe3 Bc5 28.Nd4 Bxd4 and Black gets rid of him).
30...Bf8 is criticized and gets a <?>, although Black seems to keep some advantage. There may have been something better than 31...Ke7 also, perhaps 31...Be7.
34...Ke8 receives a <?>, although the position remains equal after Black giving away his advantage earlier.
36.Nd4 is a speculative sacrifice and I'm not sure if it's correct. The annotator notes that Black should have taken the ♘: <36...cxd4 37.Rcc7 Kd8 38.Rb7> and then possibly 38...Qc6 39.Bb5 Qc3+ 40.Kg2 Qc8.
The text move, 36...Be7, is a mistake and loses.
Source: 'Wiener Schachzeitung', October-November 1907, pp. 335-337