The concept of unified kinetochore action came from light microscopy. But when the New York team used electron microscopy they saw that the 10–30 kMTs in one fiber had a mixture of morphologies. The kMTs with straight ends at the kinetochore are believed to be polymerizing, whereas those with curved ends are the unravelling depolymerizing kMTs. Surprisingly, two-thirds of kMTs were in the depolymerizing state during metaphase, suggesting that the kinetochore restrains kMTs that are in the depolymerizing state and retards their shrinkage. This restraint should produce the tension that is central to spindle dynamics.
Some kinetochores had fewer depolymerizing kMTs and some had more, but the differences were on a continuum. Thus the kinetochore appears to change direction based on a balance of pulling and pushing forces rather than a discrete, all-or-none switch. How the polymerization bias and kMT detachment are controlled remains largely mysterious.