In this work Ken Safir develops a comprehensive theory on the role of anaphora in syntax. First he contends that the complementary distribution of forms that support the anaphoric readings is not accidental contrary to most current thinking but rather should be derived from a principle one that he proposes in the form of an algorithm. Secondly he maintains that dependent identity relations are always possible where they are not prohibited by a constraint. Lastly he proposes that there are no parameters of anaphora - that all anaphora-specific principles are universal and that the patterns of anaphora across languages arise entirely from a restricted set of lexical properties. This comprehensive consideration of anaphora redirects current thinking on the subject.