The well-known postulate is adopted that the duplication of chromosome parts, followed by qualitative change in the duplicated regions, provides the most important source of new germinal materials required for evolution. The general problem is discussed and evidence is reviewed from various sources, with particular emphasis on that from the giant salivary gland chromosomes of Diptera. Numerous examples of duplicated regions or "repeats" (Bridges) are known, some induced (by irradiation), others characteristic of individual species in nature. "Pattern" repeats, in which the repeated region includes 2 or more different chromatic disks and hence forms a pattern, are of various kinds, only a few of which possess the stability required for persistence in nature. Adjacent repeats found in natural populations are "reversed," while serially oriented repeats may readily be induced. The latter are unstable because of the effects of unequal crossing over. In Drosophila and Sciara, at least, duplications apparently become stabilized only if they involve relatively short segments. The problem of "single band repeats" and the significance of "doublets" and paired disks are discussed. An analysis is presented of a "triple repeat" condition in Sciara which shows evidence of qualitative change in the repeated regions subsequent to the occurrence of the repeats.