Beryllium is a well-known structural and functional material with unique properties. There are plans to use beryllium in a thermonuclear reactor in the future. The properties of beryllium, specifically, the mechanical properties, depend on the presence of impurities, not only their quantity but also on their form. The process controlling the distribution and redistribution of impurities during heat treatment of a material is diffusion. Iron, which is one of the main technological impurities, can influence the mechanical properties of beryllium and its alloys [1]. Data on iron diffusion, which have been obtained by different methods, are available [1][2][3][4]. These data are, in the main, contradictory, and their use in modern engineering calculations could be invalid. The present work is devoted to determining the diffusion mobility of iron in hot-pressed beryllium with the typical chemical composition found today.The experiment was performed on two technical-grade materials -hot-pressed and cast beryllium, which is the initial material for preparing the hot-pressed beryllium -to determine the effect of the fabrication technology on impurity diffusion. The cast material was prepared by induction remelting of magnesium-reduced beryllium; the hot-pressed material was obtained after the cast material was milled. The cast material differs from the hot-pressed material mainly in that the grain size in the former is incomparably larger: 0.5-10 mm versus 3-15 µm. The chemical composition of hot-pressed beryllium is always different from that of the initial cast material (Table 1). This especially concerns iron and oxygen. Iron enters beryllium during milling in ball mills, and even though the products of milling pass through magnetic separation and special chemical tretament, some of the iron that entered the beryllium from the steel balls remains. The solubility of oxygen in cast beryllium is negligibly low, so that it can be ignored, while in the hot-pressed material its content in the form of BeO reaches 0.5% and higher.The radioactive isotope 59 Fe in the form of iron (III) sulfate was deposited on the surface of beryllium. The total activity of the remainder of the sample was measured as successive layers were removed (Fig. 1). Diffusion annealing was performed at 1148-1393 K. In the layerwise analysis, after diffusion annealing, special attention was devoted to the accuracy with which the total activity and the depth of the removed layer were determined. This was done in order to be able to choose the correct model for the boundary conditions realized in the experiment, since the wrong model can change the diffusion mobility by a factor of 3-8. The experimental error attained for the total activity N and the penetration depth x was 2-5 times smaller than the spread between the conventional models used to describe diffusion experiments. On this basis, a new model, which takes account of the facts that the source of diffusion weakens with time and that the impurity solubility has a limit, was proposed for analyzing...