In 1951 H. London proposed the adiabatic expansion of 3He in 4He to obtain cooling in the millikelvin range. In this paper the feasibility of the technique is investigated to provide a cooling power of 1 μW at 50 mK. Injecting 4He in a fixed amount of 3He is, in principle, a single-cycle method but it can produce continuous cooling by using two such expanders working in opposite phase. It is independent of gravity, so it may be a candidate for cooling in space. In contrast to the Planck mission it is a closed system which uses fixed amounts of 3He and 4He.