Solubility of narcotics in lipids has promoted the quest for non-aqueous and lipidic models of cell membranes. Artificial phosphatidic bilayers have been proposed. They display ionic conductance and excitability only if they are in contact with cyclic ion-carrier or specific substances, such as the protein fraction EIM. However many lipidic substances form non-bilayer membranes ion-conducting and excitable, without any specific additive. Only a small amount of free fatty acid is necessary. This is the condition for penetration through cation exchange. Coloured cations and cationic drugs undergo large exchange. Cu++, Hg++, emetine ++ cations have very high exchange coefficients which can be experimentally measured and which explain their respective antifungal, antibacterial and antiamoebian actions. The possible processes of membrane excitation are discussed. First the classical pores, specific of K+ and Na+ transfers and their "gating" mechanisms, because cell membranes are bi-ionic systems. Artificial membranes, are mono-ionic systems. But recent work shows that the axon membrane can be transformed into a monoionic system with Co++ as the only cations inside and outside the axon. Suggestions for the excitation processes are proposed. a) The assumption of a single energy barrier corresponding to minor conformational changes of structure. b) The membrane may be thixotropic. An outside cation penetrating the membrane would leave behind itself a wake of fluidity into which the next cations could penetrate if they follow each other closely. If they progress widely apart (under a small field), the ionic current would soon stop as the structure solidifies. c) The most promising suggestion is that anionic fixed charges in the membranes and cations form electrostatically bound ion-pairs. Dissociation of such pairs, that is conductance, augments markedly when dielectric constant increases. This process could be produced by water carried by incoming cations, that is by electro osmosis. This is exactly what occurs in Teorell's membrane oscillator in which a model membrane of fritted glass displays, under a weak current, oscillations of water flux and of potentials. The calculations pertaining to this model can be generalised if the electroosmotic water flux is assumed in increase the dielectric constant of the lipidic membrane. Thus the notion of an electroosmotic increase upon the dielectric constant of the membrane offers an alternative to the pore theory. Besides other phenomena show the role of low dielectric constants. The conductance of lipids containing coloured cations increase when subjected to illumination. The radiant energy absorbed then surpasses the association energy of ion pairs.