I suggest a hypothetic explanation of recently found 'memory effect' in water (water 'remembers' the presence of some biologically active molecules even when they are diluted to levels less then one molecule per total volume). My explanation is based on isotopic diversity water, namely that some positional correlations of stable isotope (H, D, 16O, 17O and 18O) might work as 'templates' of the originally dissolved molecules. These isotopic correlations are equivalent to the choice of a particular isotopic pattern out of the highly degenerate manyfold of potentially available patterns. This pattern is further reproduced at each next dilution level by some sort of 'locking-in' mechanism. Among possible physical candidates for such locking-in action is the polaronic self-stabilization in non-linear media.