1. Chromosomes ofAllium cepa, Tradescantia virginiana andT. paludosa were studied at mitosis in order to investigate the behavior of telomeres at anaphase. The studies were made in untreated fixed material, in cochicine treated cells, and in living staminal hairs. 2. InAllium cepa andT. virginiana chromosomes of untreated material, when the kinetochores start separating to opposite poles at early anaphase of mitosis, the median regions of the arms separate first, the proximal regions next and the telomeres last. 3. To study which region of the arms outside the proximal regions was involved in holding the sister chromatids together, 736 chromosome arms of completely flattened chromosomes were analyzed in colchicine treated cells ofA. cepa. In 14.9% the delayed separation involved the whole arm, in 2.8% it was in the median region, in 8.9% it was close to the telomere, and in 71.0% of the cases it was well localized at the telomere. 4. The colchicine treatment causes a change in the timing of the separation process in bothA. cepa andT. virginiana. When this treatment is applied the proximal regions of the arms separate later than the ends. 5. A study was made at mitosis of the whole chromosome complement ofAllium cepa. 1472 chromatid arms were analyzed and measured. All the 8 chromosomes of the haploid set ofA. cepa can be identified in well flattened cells. To find out whetherthere is any correlation between the division cycle of the telomere region and arm length (the distance telomere—kinetochore), an analysis was made of 736 chromosome arms. The 16 arms of the complement were grouped according to their length. Group I contains the two smallest arms and group IV the three largest. A statistical analysis of the data shows that there is a significant difference between the delayed separation at the ends of small versus longer arms. A χ2 of group I versus groups II, III, and IV gives the value 12.71 with P< 0.001. The telomeres which are situated closer to the kinetochore separate earlier than those that are situated further away. This is not interpreted as being due to a mechanical effect of the kinetochore, because the proximal regions which are situated between the kinetochores and the telomeres, separate later than the telomeres of all groups of arms. 6. Studies of mitotic chromosomes were madein vivo in the staminal hairs ofT. virginiana andT. paludosa. At anaphase the telomeres can be seen to separate last just as in the fixed untreated material. 7. When the sister chromatids of an arm separate at early anaphase the telomere is considered to have the role of the fulcrum of a lever. The two arms of a chromosome are lifted as levers by the kinetochore and in the telomeres they get their point of support. In short arms such an adaptation is not necessary, as the telomeres are quite close to the kinetochore, but in long arms, as those of the species studied here, this adaptation results in an easier and more regular separation of the arms.