I should like to enlarge on the part of our paper (3.10) in which we considered the effect of trace additions of specific elements on precipitation and on methods of controlling precipitate size. By the term ‘trace elements’ I differ from those working with steel and mean deliberate additions of specific elements in amounts from a few hundreths to about one tenth of an atomic percentage. The atoms of these elements tend to be different in size from those of the matrix and so it is natural that they should seek some site in the lattice at which they can reduce their energy. This they do by segregating preferentially to defects and interfaces and this is how their effects can be utilized. There are four typical sites: 1. Grain boundaries which have recieved most attention. Here trace elements can be used to control the rate at which boundaries can migrate. This can be at higher or lower rates than the normal migration rate of the boundaries in a pure alloy depending on the nature of the trace element. Thus they can cause grain refinement or at the other extreme, they can introduce a property akin to superplasticity for which only about 0.1 atomic percentage is needed. Unwanted trace additions can cause embrittlement but intentional additions can replace detrimental elements at boundaries and thus increase ductility. Embrittlement is often caused by discontinuous precipitation which is a consequence of grain boundary migration. The use of boron in steel to increase the depth of hardening is an example of a trace element effect at grain boundaries. By reducing boundary migration, the formation of pearlite is inhibited. 2. Dislocations which have been well treated especially in steels and aluminium alloys. 3. The matrix/precipitate interface which is now thought to be a special case of segregation to dislocations, in this case to growth dislocations at the m atrix/precipitate interface. Trace elements have been detected at this interface and as long as they remain there, growth of the precipitate is prevented. This avoids the coarsening to which Professor Honeycombe has referred. 4. Vacant lattice sites where they can control diffusion rates. This is extremely im portant as less than 0.1 % of one atomic species can control the whole diffusion process over a wide range of temperature. If the trace element forms part of the precipitate or structure which is produced by diffusion, then the trace element will accelerate precipitation; whereas if the trace element plays no part in the precipitate structure, it denies vacancies to the diffusing species. We have cases of elements added to aluminium -copper alloys delaying g.p. zone formation for more than three years, when in the absence of the trace element, g.p. zone formation would be complete in about 48 h. I would like to concentrate on the last two processes and illustrate them by an example which enabled us to develop a new engineering material from first principles using our knowledge of trace element effects.