As Blix knew in 1895, the total tetanic tension of frog's muscle passes through a maximum at a certain resting length L', near to the body length Lo. At lengths greater than L' tension decreases through a minimum, then rises again. Blix remarks that 'the relation disclosed here gives support to the idea, which has been proposed many times, that the force of contraction depends on the attraction between longitudinal tissue structures within the muscle, which are set apart by extension and in this way weaken the attraction'. The discovery of two separate sets of A and I filaments (Hanson & Huxley, 1953) and the demonstration that the A bands do not change in width during passive stretch or active shortening of muscle fibres (Huxley & Niedergerke, 1958) provided the structural basis for a revival of this type of theory of muscle contraction (e.g. Huxley, 1957). It is natural to suppose that the shear force between actin and myosin filaments, and thus the tension developed by the whole muscle, should be a function of the area of overlap of the two types of filaments; the decrease of tension development with increase in muscle length, which proved a stumbling block to many visco-elastic theories, is then readily explained. It also seems natural to suppose that other phenomena of contraction should be similar functions of the area of overlap. The purpose of this paper is to examine the mechanical properties of whole living muscles over the range of lengths beyond L', where the isometric tension diminishes with increasing muscle length and to answer the following questions:(1) Does tension fall when the active muscle is stretched beyond L'? (2) Does tension rise, then fall again, if the active muscle is allowed to shorten at constant speed through L'? (3) Is the relation between force and speed of contraction affected beyond L'2? And (4) is the total amount of shortening against a given load altered if contraction begins from beyond L'?