Historically there have been two different types of grain modelling: One of these basically uses particle populations which evolve essentially by coagulation (e. g., the MRN model: Mathis, Rumpl and Nordsieck, 1977); the other considers the physical and chemical evolution of the particles with a particular emphasis on changes not only in sizes but also in chemical and morphological structure (e. g. Greenberg, 1978; Williams, 1989). The model of Oort and van de Hulst (1946) was the first to consider that grains must evolve in interstellar space by treating both growth and destruction in clouds. The chemical properties had already been derived by van de Hulst (1946) and then later described as the dirty ice model which consisted of the saturated molecules H 2 O, CH 4 and NH 3 with trace constituents of other atoms and molecules resulting from surface reactions of atoms on the grains. How such grains could nucleate was left as an unsolved problem but the fact that, once formed, there did not seem to be any reason why they should not grow until they exhausted the condensable atoms in the gas led to the suggestion that a limiting destructive mechanism must be provided. This was assumed to be by grain-grain collisions within clouds moving at relative speeds of 10 km s−1. We thus had the first dynamical theory leading to a steady state distribution of grain sizes. This model provided for me the starting point of the core-mantle model of grains. The observations of the 60's and henceforth clearly showed the existence of other types of small particles, which have been invoked to explain the 2200 å hump (Stecher and Donn, 1965), the far ultraviolet (FUV) extinction (Greenberg and Chlewicki, 1983), and now certain infrared emission features. These other components notwithstanding, grains still account for the major fraction of the solid particle mass in space.