Is it possible to disprove water memory with an entropy argument? Water memory was a controversial experiment claiming to provide an explanation supporting homeopathy. The results were largely dismissed as being tainted by experimental error.
One possible mechanism invoked was that water molecules, on account of being polar, would form a structured network, thus storing information about other molecules they had been in contact with. I wonder if it is possible to disprove this explanation with an entropy argument. If water molecules did arrange themselves in such a fashion, this would have an effect on their entropy, which in turn would impact other thermodynamic functions such as Gibbs free energy. In other words, it might be possible to look for such an entropic effect by looking at deviations from expected behavior in a steam table. Would the magnitude of such an effect be measurable ?
 A: Quite simply, no.
Water memory doesn't appear to violate any physical laws, and the claims made about it are not well-defined or specific enough to be falsified (e.g. with an entropic argument). It's revealing that while a scientist could be convinced that he's wrong, there's nothing that could change the mind of a homeopath.
The best we can do is test the specific mechanisms that have been proposed to produce a memory effect.
For example, most homeopathy theories revolve around some sort of persistent water structure, and this structure would presumably be the result of hydrogen-bonded networks. But it's well-established from experiment as well as computer simulations (e.g. ab initio MD) that hydrogen-bonded networks last for a matter of femtoseconds and therefore could not give rise to long-lasting memory effects.
Indeed, there does not currently exist any plausible mechanism (or evidence) for water memory that does not contradict well-established science. And for this reason the scientific community does not accept it. But, as was seen to be the case with Jacques Benveniste's work, if some evidence were to arise for it then the scientific community would be willing to take it seriously and investigate.
A: Entropy argument might not work here, because amount of information is going to be influenced by number of moles, not by arrangement of individual molecules. In the same fashion as 1TB hard-drive is 1TB no matter what is written onto it. However, more precise crystal structures will tell you exact picture stored in those bits. So measuring, essentially, temperature of different water samples might not be useful.
I would like to also make a note on lemon's reply. In my opinion, rate at which hydrogen bonds are broken and created cannot be addressed here. Just consider Conway's Game of Life spaceships: individual elements change value with rate of game's clock, but structure all together is stable and  information is stored.
