The D(2)O hydration of pure tricalcium silicate (Ca(3)SiO(5)) was examined at 23 degrees C with parallel measurements using deuterium NMR spectral analysis, deuterium NMR T(1) relaxation measurements at several magnetic fields, isothermal calorimetry, scanning electron microscopy, and Vicat needle tests. The results show that the maximum rate of decrease in the deuterium T(1) relaxation time of the D(2)O resonance, the maximum rate of increase in heat production, and the maximum rate of decrease in Vicat needle penetration occur simultaneously within experimental error. Together with results from other recent deuterium NMR studies in our laboratory, these observations support the new hypothesis that the formation of solid water bridges between closely adjacent developing hydrate surfaces causes initial setting. As a confirmatory experiment, it was found that microwave heating immediately after setting, i.e. immediately after zero Vicat needle penetration, softened the paste sufficiently to once again allow some needle penetration, and furthermore, that upon cooling, the paste reset. This process could be repeated several times.