1. The cases already mentioned by Krug (4) of the somatic mutations of the Na-na pair of aleles of Coffea arabica L. were discussed. It is emphasized that the murta variety of this species has probably been originated from the bourbon variety by mutation of one member of the pair Na-na (Na-Na → Na-na). 2. Three new cases of somatic mutations of this pair of genes are described, namely : a) A plant of the nana variety, with the main branch supposed to be of the bourbon type; the leaves are however smaller and thicker than those of the typical bourbon variety ; now the nana part is dead, living only the mutated one. b) Another plant of the nana variety showing a lateral branch of the murta variety. c) A coffee-plant, also of the nana variety, which at different periods showed mutations affecting : entire branch, pairs of leaves, a single leaf, half of a leaf, or a little area of the leaf. 3. The most frequent direction of the mutation of the Na-na factor is from the recessive to the dominant condition as commonly happens with other mutable genes (1, 2). 4. It seems that the Na-na pair is stable in the germinal tissue, as the ratios cbtained in the progenies of murta plants are not far from the expected ones and those obtained from nana plants are all of this type. 5. It is supposed that the differences in the degree of mutability of the Na-na aleles in different plants are due to modifying factors.