Summary. Variable numbers of bivalents and sex chromosomes do not attach to the spindle when prophase or early prometaphase cranefly spermatocytes (2n = 8) are treated with cytochalasin D or latrunculin. The unattached bivalents lie in the cytoplasm or at the spindle pole, and they do not delay onset of autosomal anaphase; sometimes they disjoin at the same time as the attached bivalents, so they respond to the global signals that initiate anaphase. Unattached sex chromosomes do not delay autosomal anaphase, either. Of various interpretations of these data, we think the best explanation is that the checkpoint system responds to physical rather than chemical cues; we think that the spindle is a "tensegral" structure, that chromosomes need to interact with the spindle in order to be recognised by the anaphase-onset "checkpoint control", and that the physical interaction of chromosomes with spindle acts as a signalling network. Cytochalasin D and latmnculin treatments delay onset of sex chromosome anaphase (which normally occurs about 15 min after autosomal anaphase) and cause altered patterns of sex-chromosome segregation.