Traditionally, the implementation of isoneutral diffusion has been with the small slope approximation made to the full tensor (Cox 1987, Gent and McWilliams 1990). In MOM, both the small and full Redi tensors are discretized in this appendix. At this time (June, 1997) there is no compelling reason to employ the full tensor. As a result, little effort has been made to optimize the computational efficiency of the full tensor diffusion operator. Currently, it runs roughly six times slower than the small angle tensor. It remains as a frozen part of the code. It remains in this appendix since it turned out that the derivation was a easier to check when using the full tensor due to the greater amount of three-dimensional symmetry than present with the small tensor.