“…The argument of the δ and the inverse of the propagator should be combinations of the operators used in the free equations (at last approximately and for the positive-energy solutions). Ther exists an infinity of possible combinations [8,12,19,21,22,23,24,25,26,27,28,29,30,31,32,33,34]. The best choice depends on the quantities one wants to compute (energy of the lowest state, hyperfine splitting, recoil of a nucleus, etc...) and on the properties one wants to preserve exactly in the first approximation (cluster separability, Lorentz invariance, heavy mass limits, charge conjugation symmetry...).…”