The heterogeneous nucleation of ice from supercooled water is influenc
ed by the nature of the foreign nuclei that serve as the sites for ice
embryo formation, and by the stochastic nature of the process of embr
yo growth to critical size. The relative roles of these two factors ha
ve been the subject of some debate, especially as they influence the w
ay nucleation of ice is modeled in clouds. ''Freezing rate'' is define
d as the time-dependent rate at which a population of macroscopically
identical samples (e.g., drops in a volume of air) freeze due to the n
uclei contained in them. Freezing rate is the combined result of nucle
us content and of time dependence. The time-dependent freezing rate mo
del (TDFR) is consistent with available empirical evidence. For drople
ts cooled at rates of the order of - 1-degrees-C per min, the nucleus
content, or nucleus spectrum, predicts the freezing rate with reasonab
le accuracy. For samples exposed to a fixed temperature, the time depe
ndence of the freezing rate becomes important, but the probability of
freezing is not the same for each individual of the sample population.
Stochastic models are not supported by the results. Application of th
e TDFR model and use of measured freezing nucleus data for precipitati
on provide a basis for the description of ice formation via immersion-
freezing nucleation in cloud models. Limitations to full development o
f these models arise from inadequate knowledge about the freezing nucl
eus content of cloud water as a function of cloud evolution.