How Did the Rook Get There?

Starting the New Year with the seventh and last game from Petrosian's Exchange Sacrifice, it isn't easy to see the sacrifice occurring anytime soon in the diagrammed position. Petrosian wrote:

In the Sicilian Defense, an attack on the Pawn e4 along the 4th rank from the squares b4, c4, even d4, is quite a usual matter. But the idea, which I managed to discover in this game, is interesting mainly due to its paradoxical character.

Even after seeing the next move, 21...Rh5, an exchange sacrifice is not easily discerned. Petrosian again:

This move seems to be absurd. The Rook will be surrounded by White's pieces; when the Queen goes away, g2-g4 threatens, and what should Black do with the Rook.

The game continued 22.Qf3 e5 23.f5. If you showed this position to another player, I'm sure they would have some trouble explaining how the Rook got to h5.

Moscow 1971
Petrosian, Tigran

Parma, Bruno
(After 21.Nd2-b3)
[FEN "rq4k1/1b2bppp/3ppn2/pNr5/Pp2PP2/1N1B3Q/1PP3PP/4RR1K b - - 0 21"]

After 22...e5, Petrosian gave some useful advice on conducting an attack against a weakness:

This move is necessary. Black intends to attack the Pawn e4 in one form or another. An object of attack should first be fixed and deprived of mobility; then follows the attack itself.

After that hint, and the further moves 23...d5 24.Nd2 Rh4, it suddenly becomes clear that Black plans to sacrifice the Rook for the Pawn on e4. There followed 25.g3 dxe4 26.Nxe4 Rxe4. Now since 27.Bxe4 Nxe4 28.Rxe4 gives Black the better game, Parma tried 27.Rxe4. Petrosian continued 27...Qd8 and won ten moves later without having to recapture the sacrificed material.

To play through the complete game see...

Bruno Parma vs Tigran Vartanovich Petrosian, Moscow (Russia) 1971
http://www.chessgames.com/perl/chessgame?gid=1105610

...on Chessgames.com.