Inoculative freezing of insects is distinguished from nucleative freezing and is treated from the standpoint of the delays observed in freezing. On the assumption that the fine structure of the cuticle regulates the penetration or growth of external ice into the insect's body, the cuticle was experimentally altered by soaking, boiling, and immersion in strong detergent solution. Soaking was of doubtful effectiveness, whereas boiling and detergent hastened inoculation, Detergent also caused many insects to freeze when the contact water froze, with no delay. Rate of inoculation of untreated larvae was directly proportional to the area in contact with ice, and inversely proportional to temperature down to −10 °C. No further rate increase took place down to −15 °C, possibly because the vapor pressure difference between ice and supercooled body fluid does not change appreciably in this temperature range. Larvae froze more rapidly when they had already been frozen and thawed once or twice. No relation between inoculation rate and rate of water loss in a dry atmosphere could be established.

The freezing of water in extremely small spaces is discussed and related to the process of inoculative freezing. Probable pathways of ice growth through the cuticle are considered in relation to known structural characteristics and the results of the present experiments. Two hypotheses are proposed to account for the observed delays in freezing, but neither is wholly satisfactory. The first postulates that the outer extremities of the pathways, through the epicuticle, are hydrophobic and contain air which temporarily separates contact moisture from pore liquids. After freezing of the contact moisture a vapor pressure differential results in a net flow of vapor outwards, the vapor freezing on the ice and building inwards until the ice front touches liquid in a space large enough to allow freezing. This hypothesis assumes the existence of at least one pathway large enough to support ice penetration without hindrance caused by restricted size. The second hypothesis assumes that no such unrestricted pathways are present, all being inadequate in size. Differences in chemical potential between ice and liquid cause pore liquids to move outwards to the contacting ice surface and accrete on its