Viscosity measurements of alcohol-water mixtures and the structure of water.

In a previous paper,' a model of the hydrogen bond was proposed that gave a structure for water differing markedly from others that have been suggested.2 According to this model, the hydrogen bond would not be located between any two atoms of a given molecular system, as in the case of an ordinary bond, but it would be a collective bond linking a number of molecules together and consisting of a ring system of correlated hydrogen bridges. The bond would depend on a simultaneous displacement of the electric charge in each component molecule, all around the ring. If one of the hydrogen bridges of the ring were broken, all the others would break down at the same time. When this model is applied to water, one finds that each molecule cannot participate in more than two independent collective hydrogen bonds. Several types of molecular aggregates would then be present in the water: e.g., linear chains containing any number of molecules each of which is bound to its neighbor by two hydrogen bridges, or rings of three, four, five, and six molecules, and more complex aggregates made up of a number of rings linked together. These aggregates would not be permanent, as each of them would be able to form further hydrogen bonds, and in the course of a thermal collision some molecules would move from one aggregate to the other, the total number of hydrogen bridges remaining constant. This model was used primarily to suggest further experiments. First of all we examined the connection between this continuous transfer of molecules between aggregates and the physical properties of water, particularly as it seemed probable that it would affect the value of the viscosity of water. Let us suppose that the viscosity of any liquid is obtained by first putting it into the hollow space between two concentric cylinders, one of which is kept in uniform rotatory motion. An ordered motion of the molecules is then superimposed on their thermal motions, and a velocity gradient is maintained at the expense of the interactions between the molecules of neighboring infinitesimal annular layers of the liquid. In these interactions, a fraction of the kinetic energy of the ordered motion is transferred to the thermal motions, and the viscosity of the liquid arises from this effect. If, however, the liquid consists of large molecular aggregates, and some molecules can move from one aggregate to another at constant binding energy, then there would be a tendency to maintain the velocity gradient almost without any transfer of energy from the ordered to the thermal motion. In such cases we would expect the viscosity to be lower than in the liquid when resonance transfer of molecules did not take place. If, somehow, we stop the resonance transfer of water molecules from one aggregate to another, the viscosity should increase and such an experiment will now be described. According to our model, a molecule of a primary alcohol can only be involved in one collective hydrogen bond and Figure 1 shows how methanol and ethanol molecules should normally form dimers. These substances have a fairly