The relation between the frequency of discharge of a muscle spindle ending and the extension applied to the muscle has been studied under various conditions (Eldred, Granit & Merton, 1953;Granit, 1958;Granit & Homma, 1959a; Whitteridge, 1959;Eldred, Lindsley & Buchwald, 1960). There has not, however, been any comparison of the properties of the primary and of the secondary afferent endings of the muscle spindle in this respect; nor is it known whether Whitteridge's finding for spindles in eye muscles, that intrafusal fibre contraction may increase the sensitivity of an ending to stretch, also holds for either the primary or the secondary endings of spindles in limb muscles. These points have, therefore, been investigated in the present work. The frequency-extension relations were determined under dynamic conditions by recording the discharge of an ending while the muscle was slowly stretched at a constant velocity. The use of this method showed incidentally that for de-efferented spindles the response of primary endings to the phasic component of the stretch is much more marked than that of secondary endings. This confirms the findings of Cooper (1959) for spindles with an intact motor supply. METHODS Preparation. The experiments were performed on 23 cats. One was decerebrated and the rest were anaesthetized with pentobarbitone sodium (Nembutal; Abbott Laboratories) given intraperitoneally. The discharge of single afferents from muscle spindle endings lying in the right soleus muscle was recorded from thin dorsal root filaments. Soleus was completely isolated from the spinal cord by cutting the L 6, L 7, S 1 and S2 dorsal and ventral roots. The nerves to most other leg and hip muscles, including the lateral head of gastrocnemius, were cut.Myography and application of 8tretch. The right tibia was fixed rigidly by pins and the tension in soleus recorded from the cut Achilles tendon by an isometric myograph which could be moved so as to extend soleus. The movement was at constant velocity, variable from about 0 5 mm/sec to about 9 mm/sec, and was produced by a screw thread driven by a speed-controlled motor. The extension was monitored by means of a microswitch which * M.R.C. Scholar.