For isotopic targets we may look up the masses of the particles in the bdfls mass tables and obtain the energy due to the mass difference by calculating the difference (1). For reactions on natural targets the endep code reads the mass-difference energy from the header of the ENDL data file. For scattering reactions this number is, of course, zero. But it means different things for exothermic and endothermic reactions.
For exothermic reactions on natural targets the mass-difference energy in the header of an ENDL data file is simply a weighted average, according to the natural isotopic abundances.
For endothermic reactions on a natural target there is no single number for the mass-difference energy, because different isotopes have different thresholds. Consequently, the weighting depends on which isotopes are above threshold, and strictly speaking, the mass-difference energy is a step function of the energy of the incident particle. Specifically, there is a jump discontinuity corresponding to the energy threshold for each isotope in the natural mixture. The way that these reactions are handled in ENDL is that the mass-difference energy value given in the file header corresponds to the lowest natural threshold. That is, it is the mass difference for the natural isotope which is the first to react, and the reason for this choice is that it puts the threshold at the proper energy. This has the consequence, however, that for incident particles at higher energies, the energies computed by endep for the gammas and the residual may well come out negative. The prime example of this difficulty in ENDL is natural zinc, and for some reactions the situation is so bad that the gamma and residual average-energy files are not included in the library.