Checkless chess

An old chess variant (Pritchard gives as date of origin `early 1800s?') is checkless chess.

The rules of this variant consist of the following simple modification to orthodox chess:

One may only give check when it is mate.

Keller, in his 10th issue of World Game Review mentions the following paradox: what if, say white checks black, such that blacks only move is to check white, but in that position, whites only move is to check black, and so on and so on. (Keller mentions a fantasy chess problem, using a knightrider, with this property.) He suggest to add the rule:

A check is mate, and hence legal, if the only move parrying the check is also a check.

Alternatively, one can play:

One may not check the opponent when he has a move that is legal in normal chess, hence also when the opponent can make a move that gives check.

For example, look at the following position:

White:
King a8; Queen f6.

Black:
King g8; Rook h8.

Now, when the first rule is used, Q f6-f7 is mate: black cannot take the queen, as that would mean his rook would check white. With the second rule, Q f6-f7 is forbidden, as this is not a mate position in normal chess.

Absolute checkless chess

Dr. Roger Powell proposed in 1975 the following variant to checkless chess: Absolute Checkless Chess.

Now, no piece may cross a square where it can give check. Additionally, the same restrictions as with checkless chess apply, and one should also choose one of the two options discussed there. (In the case of `Absolute Checkless Chess', I think one should be absolute, and take the first of the two options.)

Sources

Issue 10 of World Game Review, and The Encyclopedia of Chess Variants.