Variability in tension distinguishes the tension history for kinetochores that detach from those that do not detach. (A) Time series depicting sister centromeres (green) and spindle poles (red) in metaphase yeast cells (all images have been band-pass filtered in MATLAB). Top: Two separated centromere signals (green) along the axis between two spindle pole signals (red) are representative of attached kinetochores. Bottom: A detachment event is observed by a single diffraction-limited centromere signal (green) that is near to one spindle pole (red), but offset from the spindle axis, for multiple time points (blue box). (B) Sister centromere spacing is measured as the spacing between two green lacO-lacI-GFP signals. Sister centromere spacing was plotted against time for each single cell. Left: Sister centromere spacing history for a cell with no detachment event. Right: Sister centromere spacing history for a cell with a detachment event (green box). (C) Left: Combined tension history for cells without detachment events (red; n = 28 cells; binned at 1 min intervals; mean ± SEM). Right: Combined tension history for cells with detachment events (blue; n = 20 cells; binned at 1 min intervals; mean ± SEM). (D) Linear fit model for cells with and without detachment events. The variability in tension from the mean fit line appeared larger for cells with detachments (blue open circles) as compared to cells without detachments (red open circles). (E) Cumulative squared error vs. time, with cumulative squared error calculated by squaring the difference between the linear-fit predicted tension value and the actual tension value, at each time point.