Case agreement in Ancient Greek has attracted a small but varied set of treatments in the generative tradition (Andrews 1971, Lecarme 1978, Quicoli 1982). In this literature the problems were framed and solved in transformational frameworks. In the present paper we wish to consider the data from the point of view of the problems they pose for a theory of case assignment and phonologically empty elements in a modern, declarative framework - Word Grammar (WG)(Hudson 1990). We present an analysis of empty elements which exploits a feature unique to WG, the separation of existence propositions from propositions dealing with other properties; and we contrast it with earlier WG analyses in which these 'empty' elements are simply absent and with Chomskyan analyses in terms of PRO which is present but covert - a subtle but important difference. The proposed analysis is similar in some respects to the one proposed by Pollard and Sag (1994) for HPSG, though they are rather different in spirit.