Abstract
The authors discuss the effect of a small uniform symmetry breaking field on the Sherrington-Kirkpatrick (1978) spin glass model. Both q(h)=((Si)T2)J and qij(h)=N((Si)T(Sj)T-(SiSj)T)J are given by scaling functions, f(h/h*) and g(h/h*) respectively, where h*=T/N12/. They show that g(x)=xf'(x) so that in the symmetry broken situation (h/h* to infinity but h<ij=0. The latter is in contrast to the conclusions of Young and Kirkpatrick based on numerical evidence for small samples. They discuss the reasons for this discrepancy.