A fully covariant investigation is made of the Bethe-Salpeter equation for a pair of nucleons, with pseudo-scalar interaction. The " ladder " approximation is adopted, but pair creation and nucleonic recoil are accounted for exactly. Matrix solutions are obtained, with varying degrees of explicitness, for instantaneous and delayed interaction, vanishing and non-vanishing meson mass, and for vanishing and non-vanishing total energy. Important properties are disclosed which are either obscured or do not appear at all in non-relativistic approximation.First the Bethe-Salpeter equatiort is reduced to a pair of coupled differential equations in which the Dirac matrices appear only in the coupling. 1n the instantaneous interaction approximation, these can be reduced to a single covariant equation in a single variable, showing the radical influence of nucleon recoil on pair effects. When the instantaneous. interaction approximation is discarded, new features appear. There is a discrete infinity of stable states corresponding to each one of the nonrelativistic theories, requiring a new quantum number for their enumeration. Jastrow's hypothesis of a repulsive "core" interaction is rigorously f'Stablished, and the singularity is isolated. It is shown how to obtain solutions corresponding to states of higher angular momentum from those with ]=0, making use of the relativistic quantum enumeration. The conclusion is drawn that the relativistic quantum number is a property of the state of any pair of interacting particles, and its possible connection with the " strangeness " number is discussed. § 1. IntroductionIt has been suspected for some time that the covariant wave equations, which have been derived on the basis of quantized field theory/> have certain features differing essentially from those of corresponding SchrOdinger equations. It is true that by making non-covariant approximations it has been possible to obtain analogues of the Schrodinger equations, even to the extent of predicting effective potentials. Among others Salpeter and Bethe, 2 ' Uvy 31 and Klein 4 > have treated the inter-nucleon interaction in this way. But, on removal of the non-covariant approximations difficulties arise. It was first noticed by Goldstein, 5 > in a special (unrealistic) case, that the Bethe-Salpeter (B-S) equation appeared not to possess solutions of the type expected for the prediction of bound states. A variety of diagnoses of this trouble have been made, blaming sometimes the extreme value of the binding energy assumed by Goldstein and sometimes the highly singular nature of the field-theoretical potential. The remedy suggested by Goldstein himself was to introduce a less singular potential which by a limiting process might be made to coincide with the exact theoretical potential. Though such, a procedure does not work 6 > in the extreme case studied by Goldstein, possibly it would do for less extreme binding at