After 16. Qd2:

<Surveying the position now, it is evident that White has no advantage from the opening. As a matter of fact, Black has a better position than in any of the Qe2 variations. White consequently has to resolve now upon his future plan. Pillsbury selects the bold course of allowing 16....Bxf3, which means an attack at any cost; for if it comes to an end game Black's superior pawn position must tell.>

After 19....Rxe1 (instead of 19....Nc5) 20. Rxe1 Qxb2, Shredder thinks White is just losing.

After 20....Ne6.

<A simple and sound continuation would have been 20....Bd6 21. Bxd6 Qxd6, followed by ...Ne6, and Black would have had a marked superiority of position, with no attack threatened.>

After 21. Be5

<Pillsbury, with his usual judgment, immediately seizes the opportunity of occupying the clear diagonal with the Bishop; and in bringing the Knight into play he drives the Queen back, thus getting with a few strokes a powerful attack.>

After 27....Qh5

<Lasker pointed out afterwards that 27....Rad8 should have been played.>

Note that after 27....Rad8 28. Nh6, ...Qh5 is pretty much forced anyway. But White doesn't have a convincing followup.

After 28. Qh6

<A beautiful conception. The move itself is, of course, simple enough. Obviously the Queen cannot be taken; but the idea must be considered with all its ramifications, as it involves the sacrifice of the knight and bishop to make a breach in Black's position.>

Similar, Shredder observes, is 28. Bxf6 Qxf5 29. Qh6, which also leads to a draw after 29....Bh4! 30. R1g2 Qxh3+.

After 32....Qd1+:

Lasker played the same variation of the Berlin Defense to the Ruy Lopez that Tarrasch played in his second tie-break match game with Pillsbury in the Vienna 1898 tournament, and Pillsbury followed the same line he used to defeat Tarrasch in that game. On move 10, however, Lasker--following a suggestion he made when commenting on the Pillsbury-Tarrasch match--varied with 10...Re8 instead of Tarrach's 10...Nc5.

Pillsbury was fine until his 15th turn, when he played 15. Ne4 instead of the better 15. g4 followed by Qa6 (which would have given him a small edge). Pillsbury, of course, was not looking for a small edge but for a knock-out. He was--to the best of my knowledge--the only player to have a winning record against Lasker (3 wins, 2 loses, and three draws) as of the time of this game (their lifetime record ended at 4 victories each and 4 draws). And Pillsbury was one of the few players who didn't fear a tactical battle against the mighty Lasker.

With Pillsbury's 16. Qd2, he crossed the Rubicon. He allowed Lasker to wreck his pawn structure and relied on what he hoped would be a winning King's side attack. (16. Qd3 would have been safer, but who can imagine Pillsbury making that move). Pillsbury's 16. Qd2 ushered in a maze of thrilling complications in which Pillsbury's ingenious attack met equally brilliant defense from Lasker.

According to keypusher, Shredder thinks that Lasker would have been close to winning had he played 19...RxR followed by 20...Qxb2. Fritz thinks that both 19...RxR and Lasker's actual move (19...Nc5) are about equally good (Fritz keeps changing his/her mind about which is best depending on how long I allow the search to go) and maintain an edge for Black, but that neither yields a win.

Fritz also thinks...contrary to the Tournament Book and to the analysis of the game by Sergeant in his book on Pillsbury...that Lasker's 20...Ne6 is better than the proposed 20...Bd6, but that 20...h6 is slightly better than the text.

According to Fritz, Pillsbury's "losing" move was 26. Rg3. For reasons I still cannot fathom, Fritz thinks that Pillsbury could have held the game against best play with 26. Rg4 but not with 26. Rg3. Another enigma from our silicone friends.

The major crisis of the game came on Lasker's 27th turn when he faced the following position. According to Fritz, Lasker had played nearly flawlessly to this point:

click for larger view

After the game, Lasker claimed that he should have played 27...Rad8. Fritz disagrees, and says that Lasker's one and only winning move in the diagrammed position was 27...d4!

In any case, Lasker played 27...Qh5 and Pillsbury then forced a draw with the super-brilliant 28. Qh6 !!! which yielded the following position:

click for larger view

As is obvious, Lasker could not take Pillsbury's Queen with his pawn, and either 28...QxQ or 28...QxN (Lasker's actual move) lead to the same result--a draw.

I guess our computer friends are as divided on this game as are the human commentators. I ran a 20-ply search on Fritz, and got rankings for White's 26th move as follows:

1) 26. Rg4 (-0.92)
2) 26. Rg3 (-1.13) [Pillsbury's move)
3) 26. h4 (-1.24)
4) 26. Kg2 (-1.40)

In any case, Black (Lasker) is better, but--according to Fritz--the only move move that would have kept the differential below -1.00 (one criterion for whether Black has a won game) was 26. Rg4.

It was 3-3 in decisive games at this point and ended 5-4 to Lasker. You may not be familiar with Lasker's win the the so-called Impromptu tournament in 1893.

Your search was deeper than the one I got from Fritz, so it is presumably more reliable. Both Fritz and Stockfisch have Black better in all variations, and the rankings for the three moves are pretty close on Stockfisch. It thus appears that Lasker was better but did not have anything approaching a clear or easy win.