Robertsonian rearrangements demonstrate one-break chromosome rearrangement and the reversible appearance and disappearance of telomeres and centromeres. Such events are quite discordant with classical cytogenetic theories, which assume all chromosome rearrangements to require at least two breaks and consider centromeres and telomeres as immutable structures rather than structures determined by mutable DNA sequences. Cytogenetic data from spontaneous and induced telomere-telomere fusions in mammals sup ort a molecular model of terminal DNA synthesis in which all ttelomeres are similar and recombine before replication and subsequent separation. This, along with evidence for a hypothetical DNA sequence, the kinetochore organizer, readily explains latent telomeres, latent centromeres, and re- Robertsonian rearrangements between rod chromosomes ( Fig. 1 upper) to produce metacentric biarmed chromosomes (Fig. 1 lower) are a common mechanism of karyotype evolution and occur spontaneously at an appreciable frequency in mammalian tissue culture (5) or even in the somatic tissue of certain fish (6). Reciprocal translocations (Fig. la) are consistent with Muller's rules. The reverse exchanges, Robertsonian fission of a metacentric into two rod chromosomes, have been observed, and some ( Fig. 1 b and c) appear as one-break rearrangements which do not require a centric fragment to supply a new centromere and telomeres to the new chromosomes (5,7,8). In addition, Robertsonian metacentrics generally possess twice the centric structure of rod chromosomes (7-10). In the grasshopper Neopodismopsis, Moens' (11) electron micrographs showed this doubled "knob"-like structure to be penetrated by twice as many microtubules as the single centric knob of rod chromosomes. Thus, a metacentric's centric region often appears doubled and capable of splitting by fission, each half becoming a functional centromere (8, 10).Robertsonian rearrangements, especially the fissions, reveal the inadequacy of Muller's rules, especially his concept of centromeres and telomeres as immutable structures (1), and imply some or all of the following: (i) dicentrics can be stable; (ii) fissions can result from one-break rearrangements; (iii) centromeres and telomeres can reversibly appear and disappear; and (iv) centromeres and telomeres can be terminal coincident structures.Dicentrics can be stable, showing parallel chromatid separation when the two centromeres are close together. Hair (12) observed an isodicentric through many vegetative generations in the plant Agropyron. The original dicentric was unstable at mitosis; criss-cross and interlocking separation produced a breakage-fusion cycle that resulted in shorter intercentric distances. Dicentrics with short intercentric regions, however, were mitotically stable, both centromeres on one chromatid separating to the same pole. Dicentrics can also be stable when one centromere is latent (see review, ref. 13). In humans, most Robertsonian metacentrics are dicentric (14) in that they show pericentric h...