A method is proposed for accessing Coda s y I and functional data model databases from prolog it uses tokens or references to objects as values, instead of the more usual method of using relational attributes as values it is shown how to implement an n-ary relational view based on such tokens, and how this view relates to code used with codas sy I and FDM databases a method for improving the efficiency of joins in this formalism is given based on that implemented and tested in the ASTRID relational algebra system. directions for future work are given including suggestions on storage of more general AI structures.