The general problem approached in this thesis is that of building computer based legal advisory programs (otherwise known as expert systems or Intelligent Knowledge Based Systems). Such computer systems should be able to provide an individual with advice about either the general legal area being investigated, or advice about how the individual should proceed in a given case. In part the thesis describes a program (the ELl program) which attempts to confront some of the problems inherent in the building of these systems. The ELl system is seen as an experimental program (currently handling welfare rights legislation) and development vehicle. It is not presented as a final commercially implementable program. We present a detailed criticism of the type of legal knowledge contained within the system. The second, though in part intertwined, major subject of the thesis describes the jurisprudential aspects of the attempt to model the law by logic, a conjunction which is seen to be at the heart of the computer/law problem. We suggest that the conjunction offers very little to those who are interested in the real application of the real law, and that this is most forcefully seen when a working computer system models that conjunction. Our conclusion is that neither logic nor rule-based methods are sufficient for handling legal knowledge. The novelty and import of this thesis is not simply that it presents a negative conclusion; rather that it offers a sound theoretical and pragmatic framework for understanding why these methods are insufficient - the limits to the field are, in fact, defined.