If 0-0-0, which rook does King's rook apply to?

A position similar to this is in Nimzowitch's My System and he says the King's rook moves. The rook on d1 makes most sense, but I assume King's rook applies to the one on the king side from the beginning.

Note the diagram below isn't the positon from the book, it's just an abstract position showing what I'm asking.

The kr would be the one on e1 but people dont use that notation much , if any, these days.

Thanks Reb. Yea, I've not seen that notation much. My System (the 21st Century Edition) says it's been updated to algebraic notation, but a bit of the older style notation still slips into his prose.

Many of my books are in the descriprive notation and its what I started out with. I still keep my score in descriptive at least one tourney per year, or try too. FIDE forbids it though so it aint easy. 

Why Do they forbid it?

I thought descriptive notation only used the squares as kingside/queenside. Surely you don't keep track of the starting squares of the rooks once they are moving all over the board, do you? Also, I don't see what that adds to the notation.

Well , in that position we know the rook on e1 came from h1 so theres no doubt its the "KR" . however if there were some doubt and you still wanted to use descriptive you would write R/k1-  etc , ofcourse in the actual position given only one rook can move so only R-    would be necessary. I think the old jaques sets even marked the king side pieces with some distinctive stamp ?  Maybe they did have to keep up with where the pieces came from ? With the bishops its not a problem but rooks and knights could get messy 

R(K) would apply to the Rook on the King file.

I must say, I actually preered this notation. In old Books they used to say that your KB2 is a vulnerable point, translated to algebraic, that would read: f2 is a vulnerable point if you are playing white, and f7 if you are playing black.... No wonder it's a hassle to "translate" the old books....

 I dont know their reason for forbidding it I just know that they require you to keep score in ONLY the descriptive notation. I even had an arbiter warn me once because I was using descriptive and he said it wasnt allowed.....

Kings rook refers to the rook on the board closest to the position of the kings rook as the board is originally set up.  King side and queenside does not change throughout the game.  If you O-O-O, and play g4 you are still making a kingside pawn move.  King/Queenside descriptions refer to the orginal piece placement.

Ah, you youngsters. For us old guys who cut our teeth on the descriptive notation, it was the algebraic notation that took a lot of getting used to. I still prefer the old books . . .

The King Rook, Queen Rook, King Knight and Queen Knight designations are only important (or used) when either of two similar pieces can move to the designated square. A similar case occurs with the usual abbreviated form of algebraic notation.

Imagine a position where white has the back rank empty except for the rooks on their original squares. He moves a rook to the d file. Which rook? In algebraic you can't just say Rd1, you would need to specify Rad1 or Rhd1. In descriptive, those same moves would be either QR-Q1 or KR-Q1. Now take the same set-up, but with a knight at b1 (QN1). The only rook that can move to d1 is the one on the h file. Writing Rd1 is sufficient. In descriptive, its R-Q1.

In other situations as the game progresses, as Reb said, when necessary, the starting square may be named rather that the K or Q designation preceeding the piece. Those situations were fairly uncommon, though they did occur a small percentage of the time, as they do with algebraic.

My guess is that, in the event they enter the games into a computer, they don't have to translate. Either that, or too many of the officials can't read descriptive.

I wouldnt want to be the arbiter that had to tell Fischer he had to use algebraic !  He always used descriptive notation his entire career.