Summary(i) When nuclei are irradiated by X‐rays, neutrons or radioactive radiations either clumping of the chromosomes or breaks and structural rearrangements of the chromosomes are observed at metaphase according to the division stage, respectively late prophase and metaphase or resting stage and early prophase, reached at the time of radiation. (2) Chromosome structural changes are induced at the resting stage, chromatid at early prophase. The number of primary breaks induced greatly exceeds the total number of breaks observed at the subsequent metaphase. The majority of the breaks restitute after being open for a few minutes. Exchanges are possible during the time the breaks are open. If the dose of radiation is given at a high intensity, in a short time, so that all the primary breaks coexist, the yield of X‐ray induced exchanges is proportional to the square of the dose. If the dose is given at a lower intensity, in a longer time, some restitution of the earlier formed breaks occurs before the later ones are formed, and the yield of exchanges is then proportional to a lower power of the dose. (3) Only breaks with an initial separation of not more than 1 μ have an appreciable chance of taking part in an exchange. (4) In neutron and α‐ray experiments, in which a small number of ionizing particles cross the nucleus, the yield of exchanges is linearly proportional to dose and independent of intensity. Thus the majority of the exchanges are between pairs of breaks produced simultaneously by the same ionizing particle. (5) An ionizing particle may break both chromatids of a split chromosome at the same locus and an isochromatid break usually results. (6) A proportion of the primary chromatid breaks are unjoinable; hence a proportion of interchanges are incomplete, a proportion of isochromatid breaks show non‐union and a proportion of breaks not taking part in exchanges persist as visible chromatid breaks instead of restituting. (7) The number of primary chromatid breaks can be inferred (Table 2) and are found to differ for different radiations. A proton or an α‐ray traversing a chromatid has almost unit probability of causing a break. An electron is likely to cause a break only if the last 0–3 μ of its ionized track traverses the chromatid; hence a minimum of 15–20 ionizations must be produced in a chromatid of diameter 01 μ for the breakage probability to approach unity. The relative efficiencies of different wave‐lengths and types of radiation can be explained on this basis.I am greatly indebted to Dr D. E. Lea, Strangeways Laboratory, Cambridge, for his collaboration on radiation problems and especially on the present occasion for most generous permission to use unpublished calculations without which the present account would have been