This thesis consists of three parts. In the first part (chapter II) we modify Deo and Mahapatra's analytic model of the forward virtual Compton amplitude to get rid of the internal inconsistencies existing in the original model. This model has provided a good fit to the electroproduction and the photoabsorption data, and, together with an extra ansatz, has provided a theoretical prediction of the value of the p-n mass difference. However, we have found that the virtual compton amplitude of this model does not have the correct -plane analyticity and that the virtual Compton cross sections as predicted by this model are not positive definite. In this work we modify their model to remedy the above defects. and by so doing we even **age to obtain a better fit to the same set of the electroproduction data than the original model. Furthermore, we feel that the ansatz they have adopted to calculate the p-n mass difference does not seem to have any firm physical basis. To replace the ansatz, we thus propose the use of the unified gauge theories to dispose of the logarithmic divergence associated with the Bjorken scaling phenomena, and the use of a semiphenomenological approach to calculate the subtraction term of the Compton amplitude. We demonstrate that such a Scheme is capable of producing satisfactory results for the p-n mass difference.
In the second part (chapter III) we examine the sign of the proton-neutron mass difference in some gauge models based on the SU(2)×U(1)×U(1) or SU(2)L×SU(2)R×U(1) group. The point of using gauge models in such calculations is that some of the models can be so arranged that the mass difference automatically comes out finite. However, the sign of the mass difference is not so easy to control. We are able to show in this work that a few specific elements of the neutral-vector-meson mass matrix can determine the sign of △m｜p-n．Using this result we reexamine several existing gauge models and confirm that the sign of △m｜p-n is indeed wrong in all these models. We then present an SU(2)×U(1)×U(1) model which can give the correct value of △m｜p-n, provided we choose appropriate parameters in the model.
In the final part(chapter IV) we extend Joglekar's analysis of renormalizability in a bilinear gauge to the case of an anomaly free axially coupled spinor "electrodynamics". This generalization is completely straightforward, except that there are two additional problems that must be dealt with separately. These problems are (1) how is the action transformed under gauge transformations in view of the potential trouble associated with the presence of the axial-vector currents? and (2) how to treat 5 matrix in the dimensional regularization scheme, which will be needed in the analysis of renormalizability. Fortunately these possible difficulties turn out not to arise. Then we present a detailed proof of renormalizability of the theory in a bilinear gauge. closely following Joglekar's method. In our analysis we also manage to straighten out a few small but confusing errors and clarify a few subtle points in Joglekar's original work.