This paper describes a re-investigation into the mechanism of the selective action of local anaesthetics in paralysing muscle reflexes before affecting muscle power when they are injected into a muscle or applied to nerve. The problem has attracted little interest for the last 25 years, though it has important theoretical and practical implications for the understanding of muscle tone and movement.In 1919, Liljestrand & Magnus showed that the injection of 0 5-1 ml. of 1 % procaine into the triceps muscle of a decerebrate cat almost abolished the rigidity of that muscle without altering its response to stimulation of the brachial plexus or removing reflexes produced by turning the animal's head. Larger doses (4-8 ml.) were required to remove the last traces of tone and these also abolished the response to nerve stimulation, but not the response to direct faradic stimulation of the muscle. The experiments were performed as a preliminary to an investigation of the muscular spasm produced by tetanus toxin, and this was similarly abolished by small doses of procaine in the early stages of poisoning at a time when the spasm could still be removed by dorsal root section. They concluded that the action of procaine was to paralyse the muscle proprioceptor fibres before the motor fibres, as they, like Sherrington (1898), had shown that decerebrate rigidity was largely abolished by dorsal root section.Walshe (1924) used this method of intramuscular procaine injection to analyse the rigidity in patients with paralysis agitans. Appropriately graded injections into biceps or triceps muscles abolished the rigidity in these muscles alone, leaving them 'atonic, flabby and offering no resistance to passive stretch' and without a jerk. Meanwhile, the voluntary power of these muscles was unchanged and the freedom of voluntary movement greatly increased, though the tremor persisted. There were signs of slight ataxia and the injected muscles became 'painless [sic] to deep pressure or pinching'. Similar effects were 16-2