Most current hypotheses of mitotic mechanisms are based on the "PAC-MAN" paradigm in which chromosome movement is generated and powered by disassembly of kinetochore microtubules (k-MTs) by the kinetochore. Recent experiments demonstrate that this model cannot explain force generation for anaphase chromosome movement [Pickett-Heaps et al., 1996: Protoplasma 192:1-10]. Another such experiment is described here: a UV-microbeam cut several kinetochore fibres (k-fibres) in newt epithelial cells at metaphase and the half-spindle immediately shortened: in several cells, the remaining intact spindle fibres bowed outwards as they came under increased compression. Thus, severing of k-MTs can lead to increased tension between chromosomes and poles. This observation cannot be explained by models in which force is produced by motor molecules at the kinetochore actively disassembling k-MTs. Rather, we argue that tensile forces act along the whole k-fibre, which, therefore, can be considered as a classic "traction fibre." We suggest that anaphase polewards force is generated by MTs interacting with the spindle matrix and when k-MTs are severed, polewards force continues to act on the remaining kMT-stub; spindle MTs act as rigid struts concurrently resisting and being controlled by these forces. We suggest that the principles of "cellular tensegrity" [Ingber, 1993: J. Cell Sci. 104:613-627] derived from the behaviour and organization of the interphase cell apply to the spindle. In an evolutionary context, this argument further suggests that the spindle might originally have evolved as the mechanism by which a single tensegral unit (cytoplast) is divided into two cytoplasts; use of the spindle for segregating chromosomes might represent a secondary, more recent development of this primary function. If valid, this concept has implications for the way the spindle functions and for the spindle's relationship to cytokinesis.