__acirce__: continued:

<B) G. Levenfish and V. Smyslov suggest 39..d4!? 40.exd4 Re2 41.Rc3 Rxg2 (41..Rd2 42.Rc4) 42.Rxc6 Rxh2 43.a4 g5 (idea: ..g4; ..Rh6+). However White maintains the advantage by placing the rook behind the g-pawn: 44.Rc7+ Kb6 45.Rg7!, because his own passed pawn is quite dangerous.
Such an alternative (with consequences that can hardly be calculated and evaluated over the board) is practically still better than the passive defense with the rook on a8. Moreover, it can be improved: a third way exists, although endgame treatises do not mention it.

C) 39..g5! 40.g3

(After 40.Rc3 f4 41.exf4 gxf4 Black maintains enough counterplay, for example 42.Rxc6 Rd8+ 43.Kc5 d4 44.Re6 d3 45.Re1 Rg8=)
40..g4!

(Again, 40..d4?! 41.exd4 Re2 is dubious here in view of 42.Ra5! (42.Rc3 Rxh2 and the c-pawn is inviolable) 42..h6! 43.a4! (43.Rxf5 Rxa2 44.Rf7+ Kb6 45.Rc7 Rxh2 46.Rxc6+ Kb5) 43..Rb2 44.Rxf5 Rxb4 45.Kc5 Rxa4 46.Rf7+ and Black's position is difficult.)
41.f4 (41.fxg4 fxg4 42.Rc3 Rf8=) 41..Re4 42.Rc3 Rc4 unclear>

37.a3 Ra8

<The b6-square is perhaps even a worse place for the rook than a8.>

38.e4!

<White has achieved the maximum on the queenside and cannot improve his position in this sector anymore (38.a4? Rb8). Therefore he applies a standard method: widening the beachhead! After the exchange of the central pawns the white king attacks the kingside while the rook gets full control over the 5th rank.>

38..fxe4 39.fxe4 dxe4 40.Kxe4 Ra7?

<Black follows the same fatal policy of passively marking time. He still should have done what we have said: to release the rook from its mission (guarding the a-pawn) by bringing the king to b6: 40..Kc7! Now 41.Kf4? gives nothing in view of 41..Rf8+ 42.Kg3 Kb6=. Levenfish and Smyslov give the following line: 41.Re5!? Kb6 42.Re7 a5! 43.Rxh7 axb4 44.axb4 Ra4 45.Rg7 Rxb4+ 46.Kf3 Rh4! 47.h3 Rh6 (this is only a short-term passivity: the rook heads for the 8th rank, to take a position behind the passed pawn) 48.Kg4 c5 49.Kg5 Rh8 50.Rxg6+ Kb5 51.Rg7 c4 (idea: ..Rc8 with counterplay).
I think that White should not force events. The restraining method 41.h4!? Kb6 42.g4 (42.Kf4!?) 42..Rf8 43.h5 maintains an indisputable advantage; the question is solely whether it is sufficient for a win.>

41.Kf4 h6

<Otherwise the king passes to h6 with a decisive effect: 41..Ra8 42.Kg5 Ra7 43.Kh6 Re6 44.g4 (idea: h4-h5 ).>

42.h4 Ke6 43.Kg4 Ra8 44.h5! g5

<44..gxh5+ 45.Kxh5 Rg8 46.g4 >

45.g3!

<White has created and fixed a new weakness in Black's camp: the h6-pawn. Prior to returning his king to the center, he takes control over the f4-square. 45.Kf3 is less accurate in view of 45..Rf8+ 46.Ke4 Rf4+.>

45..Ra7 46.Kf3! Ra8 47. Ke4 Ra7 48.Kd4 Kd6 49.Ke4 Ke6 50.Re5+! Kd6

<If 50..Kf6, then 51.Rc5 Rc7 52.Ra5 Ra7 53.Kd4 Ke6 54.Kc5 .>

51.Re8 c5

<The pawn endgame after 51..Re7+ 52.Rxe7 Kxe7 53.Ke5 is absolutely hopeless.>

52.Rd8+!

<Perfect endgame technique. Flohr had calculated the following line: 52..Kc7 53.Rh8 cxb4 54.Rh7+ (54.axb4, of course, also wins) 54..Kb8 55.Rxa7 Kxa7 56.axb4 Kb6 57.Kf5 Kb5 58.Kg6 Kxb4 59.Kxh6 a5 60.h6 . Other king retreats lose the c5-pawn.>

52..Kc6 53.Rc8+ Kb6 54.Rxc5 Rh7

<The rook has changed its parking space, but the new one is as unattractive as the previous.>

55.Re5 Kc6 56.Re6+ Kb5 57.Kf5 Rf7+ 58.Rf6