The concept of a dual spectrum facility for the assay of reactor fuel is extended here to include the assay of thermal reactor fuel as well as LMFBR fuel, and the problems of neutron filter choice and moderator choice are examined. Finally, an estimate is made of the resistance of the concept to errors introduced by tampering with the plutonium distribution. The dual spectrum concept involves oscillating fuel against a standard in a hard spectrum in a dilute fast critical assembly, then softening the spectrum by introducing a moderator into the voids in the assembly, and oscillating the fuel inside various neutron absorbers or filters. The choice of moderator and of neutron filter is a set of design variables; the objective of this work is to determine that set of design variables which yields the lowest estimated bound on the error inherent in the fuel assay, and to extend these considerations to the assay of thermal reactor fuel. It is concluded that the filters hafnium, boron, and cadmium, together with water or graphite moderator, offer an optimal design set. The errors in the assay of fuel are, roughly, inversely proportional to the mass content of the fuel; the resistance to tampering is found to be high.