[A distinctive scheme of amino acid replacement was evolved for the generation of diversity, among hypervariable positions (author's transl)].

The diversity of amino acid residues, at a same position when comparing several aligned polypeptide sequences, may be translated as follows. The way along which a given amino acid, A, is replaced--in average--by another amino acid, B, is characterized by a coefficient linked with the pair A-B. Thus, one given amino acid is given a "set" of 19 coefficients, and the 20 different such sets may be analyzed. This method applies to the analysis of the diversity, among different sequences VH and VL of the variable regions of immunoglobulin heavy chains and light chains. From an observation of the alterations of those different sets, according to the sample of positions from which they were derived, it is possible to reach to the following conclusions. A) In the first approximation, all the amino acids present the same behaviour, whichever the sample. The frequency of replacement of an amino acid, A, by another amino acid B, is mainly a function of the proportion of B in the sample. B) In the second approximation, a more elaborate scheme of replacement is apparent, and is linked with an equivalent scheme in the genetic code; it is shown that hypervariable positions as well as random positions in VH and VL obey to this scheme. C) In the third approximation, a complementary structure is observed, which only pertains to the sample of hypervariable positions, and which might constitute a peculiar aspect of a selective process: this complementary structure is quite diverging from the genetic code. This analysis brings a strong argument against somatic theories, for the generation of diversity.