Abstract
The question of the statistical consequences of cooperative hydrogen bonding in water is examined. Methods familiar in the theory of ferromagnetism are employed to deduce the most probable spatial configurations resulting from a simple mathematical model for the cooperative nature of the hydrogen bond, of the type suggested by Frank and Wen. Contrary to the conclusions of these authors, this so-called significant structure turns out to be an extensive bond network, interrupted by regions containing relatively few bonds, the exact converse of that originally conjectured by these two authors in their famous flickering cluster hypothesis.