For the past 20 years, the consensus among researchers has been that, at any given time, a molecule of water typically forms three or four hydrogen bonds - 3.5 on average. ''What we find,'' Bergmann says, ''is that there are not 3.5 hydrogen bonds, but only 2.'' Each molecule could still form up to four bonds, the research suggests, but two would be of different, much looser kinds.
The authors point out that the earlier estimate of 3.5 was based on theoretical assumptions that became commonly accepted because, when applied in computer simulations, they gave results consistent with known properties of water, such as the unusually high amount of energy that is required to heat it up. ''Nobody had anything to object to the prevailing model, so it became the truth,'' Nilsson says.
But the difficulty of ''seeing'' the actual molecules in action meant a dearth of real data. ''There has not really been new experimental information about water in the last 20 years, except for data from neutron studies,'' Nilsson says. ''The amazing thing is that hardly anything is known about the unique properties of liquid water.''
The new result now reopens the hunt for the structure of liquid water. ''It resurrects models that were considered inappropriate,'' Bergmann says. One possibility, he says, is that water molecules could arrange in chains or even in closed rings. Eventually, the outcome could be a better understanding of
Contact: Neil Calder