While heterogeneous catalysis, and especially catalysis by metals, is concerned with the size of the particles and hence with the developed surface area, this is not only to prepare an effective product at minimum cost. The study of the relationship between the catalyst characteristics and their reaction properties has helped in some situations to make a virtually atomic description of the specific active site of each reaction. The various authors have all taken a different approach to the problem. One approach considered the crystallographic properties of the mass metals, while another considered that small particles have neither the properties nor the structure of the mass metal. The crystallographic approach first led to the consideration of the crystal parameters and to a distinction between the different atoms located at the corners, kinks and faces. However, it was found very soon that the particles did not have the expected structures. It then became necessary to calculate the energy of the particles considered to identify the way in which they could grow from a small nucleus. These calculations considered neither the atmosphere in contact with the particle, no the presence of a support, nor the disturbances caused by the reaction. Depending on the reaction investigated, `easy＇ reactions, unaffected by the structure, were distinguished from `demanding＇ reactions, of which the intrinsic rate varies with the surface structure. Many investigations were conducted on monocrystals, whose exposed surfaces had difference indexes, and others attempted to analyze these relationships with supported metals, with all the artefacts that this could imply. It was difficult to find any unity in the overall results obtained, especially since self-poisoning by the reactants (hydrocarbons or hydrogen) could readily explain the mechanisms observed. Moreover, a metal deposited on silica and the same metal deposited on alumina can display completely different behavior. This shows that some interpretations are too simplistic, and that it is inadequate to vary the particle size by any available means and to analyze the consequences on the catalytic process. The two complementary approaches, that of the crystallographer, who tries to describe the small particles from the parameters of the mass metal, and that of the chemist, who tries to determine the structure from the behavior of the catalyst observed in the reaction investigated, do not truly merge to provide a totally acceptable description of the particle structure. On the one hand, the physicochemist uses outrageous simplifications when he tries to describe his structures on the basis of functions of state which do not always have clear solutions. And, on the other, the chemist handles real objects but has difficulty in isolating the parameter that he wants to investigate. His conclusions are never safe from the artefacts generated by the operating conditions or the support effects. This dilemma is faced by the physicist, who tries to synthesize aggregates that are clearly defined in a gas stream, but far from the reality of catalysis, and likewise for the chemist, who wants to reduce the structural effects to simple comparisons between the faces exposed by monocrystals.