An earlier paper (McConnell, 1950), outlines a structural hypothesis according to which high-temperature water occurs in the tetrahedral layers of montmorillonite. In real crystals of montmorillonite, hydroxyl ions are presumed to be distributed at random within the Si2O5 layers in such a manner that four hydrogens substitute statistically for one silicon—the number of oxygens remaining unaltered. In an idealized model the hydroxyls are visualized as forming discrete (OH)4 tetrahedra which are linked at three vertices to other tetrahedra of the sheet in the same manner as the SiO4 tetrahedra for which they substitute.

Liberation of tetrahedral water was employed as a basis for partially explaining the second high-temperature endotherm of the differential thermal analysis, whereas it was indicated that the quantity of water liberated during the first high-temperature endotherm appears to be essentially equal to the theoretical hydroxyls required for octahedral co-ordination. The number of hydroxyl ions that occur in the silica sheets is presumed to be small but, nevertheless, sufficient to cause anomalies in rigorous calculation of the contents of the unit cell.