We have proposed a cluster mean field theory of the spin glasses in which dynamical clusters of spins are the fundamental entity. We review experimental evidence for spin clusters and show that within our model it is possible to understand (i) why the specific heat and resistivity have a broad maximum at temperatures above TSGat which the magnetic susceptibility cusp appears and (ii) why the total neutron cross section d&sgr;/d&OHgr;khas a k‐dependent maximum at temperatures below TSG, where k is the momentum transfer to the neutron. These results suggest that there may, in fact, be a sharp transition to the spin glass phase, which turns out to be more apparent in susceptibility than in specific heat, resistivity, and neutron measurements.