Immunoglobulin kappa chains from various inbred strains of rats have two serologically detectable forms that segregate in a Mendelian fashion (allotypes a and b of the RI-1 locus). Partial amino-acid sequences from the constant regions of these two forms have been compared. Of the 81 residues of the constant region studied, 10 amino-acid substitutions as well as one size difference (sequence gap) were found. This large number of sequence differences among alternative forms of the K allotype raises provocative questions as to the genetic and evolutionary implications of these light chain allotypes. We designate allotypes whose alternative forms differ at multiple residue positions as complex allotypes. There are basically two genetic models that might explain complex allotypes. First, these allotypes are alleles of a single structural gene with an unusual evolutionary history. Second, all rats have genes that code for each of the light chain allotypes and a control mechanism that permits them to be expressed so that they mimic a Mendelian pattern of segregation. We discuss evidence from other immunoglobulin systems that is compatible with this second model.The immune system is one of the most complex physiological systems that has been studied at the molecular, genetic, and cellular levels. Fig. 1 with the previously published CK sequences of myeloma (Bence-Jones) K chains from the LOU strain of rat (S211) (9) and the BALB/c strain of mouse (M321) (10). The LEW and LOU strains are of the b serotype, whereas DA belongs to the a serotype. The LEW and DA CK regions differ by 10 residues plus one sequence gap, whereas the LEW and S211 CK regions differ by only two residues. This is a minimum estimate of the total number of differences for several reasons: (i) only 81 of the 108 residues of the C region were compared; (ii) the acid and amide forms of aspartic and glutamic acid were not distinguished; and (iii) the V regions were not examined.These allotype-associated differences are distributed in a nonrandom manner. Ten of the 11 differences occur at positions where the LEW sequence differs from that of the mouse, indicating that certain positions are more likely to accumulate changes than others. The fact that the DA sequence is identical to that of the mouse at five of these positions is a puzzling point which may indicate a more rapid accumulation of changes in the LEW gene. Further, the distribution of the substitutions in the tertiary structure of the light chain is not random. Most of the substitutions lie on the external portion of the polypeptide chain (only position 136 is internal) (ref. 11; R. Poljak, personal communication). Two clusters of differences (one including 153 and 155, the other 184, 185, and 188) are external, and both lie very close to one another in a region already known to encompass the serological markers Oz and Inv, and the sequence marker Kern. Since it is known the RI-1 determinants lie exclusively in the C-region (12), it seems likely that one or more of these external subst...