Summary Lactose solutions may be supercooled greatly without crystallization taking place. The degree of supercooling necessary for crystallization to occur, in the absence of agitation, is less for concentrated solutions than for dilute solutions. There is no sharp line dividing the metastable and labile zones in the case of supersaturated lactose solutions. As a solution of lactose is supercooled, the rate of nuclei formation passes through a maximum. The temperature of most rapid nuclei formation is higher in the case of concentrated solutions. The rate of crystal growth passes through a maximum as the temperature is lowered. At low temperatures, the rate of crystallization, in the absence of agitation, is so slow that mutarotation cannot be the limiting factor. Lactose solutions may be prepared which are so greatly supersaturated as to resemble solids. Such glasses are stable at room temperature, if protected from moisture. Lactose glasses are hygroscopic, absorbing moisture from the air until sufficiently dilute for crystallization to take place. Lactose glasses are supersaturated with respect to both alpha hydrate and beta anhydride. Either modification may appear when such glasses crystallize. When desiccated, lactose glasses lose a part of their moisture quickly, coming to an apparently constant weight. This probably accounts for the success of the Mojonnier method of determining total solids in milk. The removal of the last portions of water from such glasses is very slow. Lactose precipitated by alcohol is not equilibrium lactose, though it may resemble it closely. Alpha lactose is more readily precipitated from solution by alcohol than is beta lactose. Consequently, precipitated lactose varies in composition. It usually contains a greater proportion of alpha lactose than is present in the equilibrium mixture.