That very small gas bubbles can serve as nuclei for the formation of cavities in liquids is well established. That experimental observations on the rupture of liquids can be interpreted only on this basis is perhaps not so obvious but none the less true. In the present series of experiments on acoustic cavitation in water, two distinct types of bubble formation occur. One, the violent formation and collapse of vapor‐filled cavities, results from mechanical instability of gas nuclei. The other, relatively quiet gas bubble formation, occurs as a consequence of the slow growth of nuclei by “rectified” diffusion of dissolved gas from the surrounding liquid. Each process has a threshold of excitation by a sound field; which has the lower threshold in any given case depends largely upon the concentration of dissolved gas in the liquid. Measurements of the acoustic cavitation threshold in conventionally “deaerated” water, as a function of temperature and ambient hydrostatic pressure, reveal how the equilibrium size of gas nuclei depends upon these variables. Observations on sonically induced effervescence in saturated solutions provide at least a qualitative explanation for the pulse length and viscosity effects observed elsewhere. Cavitation at the surface of a sound projector apparently is profoundly affected by adsorbed gases. The conclusion that gaseous nuclei exist more or less in equilibrium with solutions not supersaturated with gas is contrary to the conventional theory of gaseous solutions. Stabilization of nuclei in the surface cracks of suspended solid particles is a very plausible but not entirely satisfactory explanation. Revision of the theory is a tempting subject for speculation.