“…10 -3 , 1% or 10%) OR by comparing zero-field σ(T) to σ(T) for non-magnetically doped analogues a-Y x Si 1-x or a-Y x Ge 1-x . 6,21 The latter method is very effective for metallic samples (x>0.14 in a-Gd x Si 1-x and a-Gd x Ge 1-x ) because it is a zero field measurement and provides an unambiguous characteristic temperature which we defined as T*, but is ineffective for insulating samples, briefly because the temperature dependence is not a simple power law and the dependence on x is far too strong to allow simple comparison to nonmagnetic analogs. The former method, based on magnetoresistance, gives a temperature which we will here call T'; T' depends on the magnitude chosen for the cutoff MR, but since MR vanishes exponentially with increasing temperature in both a-Gd x Ge 1-x and a-Gd x Si 1-x , the choice is not significantly important.…”