Improving our understanding of cochlear mechanics requires the analysis of cochlear models. In the formulation of such models, it is necessary to make assumptions as to the relative importance of the many structures which comprise the cochlear partition. Data on the relative motion of these structures are virtually non-existent, and so the accuracy of many of the assumptions made is questionable. The bulk of the content of this thesis relates to the formulation and analysis of a new type of linear mechanical cochlear model. The new assumptions made in the formulation of the model are justified on the basis of the structure of the cochlear partition. The apparent realism of the model (both its structure and certain features of its response) seriously questions the hypothesis that response tuning and changes in response resulting from trauma prove the existence of active processes in the cochlea. Furthermore, arguments are presented that question the validity of using the presence of spontaneous otoacoustic emissions to justify the existence of active processes in cochlear tuning. The new model suggests that the complicated structural geometry of the cochlear partition (particularly the organ of Corti) must be incorporated in a model before conclusions relating to real cochlear behaviour can be drawn from it. In particular, the model suggests that a mechanical second filter exists in the cochlea, from rather broad tuning in the pectinate zone of the basilar membrane to sharper, neural-like tuning in the arcuate zone. It is concluded that the only way to properly check the validity of cochlear models is to obtain more experimental data pertaining to the relative motions of the various components that constitute the cochlear partition. Before this is done, we should not place too much faith in our present (alleged) understanding of cochlear mechanics. Also presented in this thesis are new modelling techniques for improving the realism of electrical transmission line cochlear models: the ability to include longitudinal and radial mechanical coupling, multidimensional fluid motion and stick-slip friction. It is shown that the inclusion of mechanical coupling and two-dimensional fluid motion in the new cochlear model has a predictable (and trivial) effect on-its response. The use of the stick-slip friction modelling technique is illustrated by means of two simple examples.