“…The experiment sets up a time scale, and thus a lower cut-off d m in the distribution of tunnelling rates, or in the splitting of the two-level systems. This, in turn, implies that the specific heat is not truly linear, but extrapolates exponentially from C = 0 at T = 0 to a shifted, linear expression C = A(T -TB), 7B 1 > 0 [4,38] where kB T1 oc Am. The fact that the anomalous behaviour is absent in some glasses like amorphous Si [5] should be related to the size of the cut-off 4~, that is to the stiffness of the network, by comparison to vitreous silica Si02, say.…”