Muir and Browning (1909) were the first to show that alexin is retained by the filter when fresh serum is passed through a Berkefeld candle. Several filtrations are generally required to remove all traces of the alexic activity, although when a fresh candle is employed for each filtration, the third or fourth filtrate is usually devoid of alexic power (Schmidt, 1914). It is not established whether or not the failure of later filtrates to haemolyse sensitised erythrocytes results from the removal by the filter of some one or of all of the known components of alexin from the serum. We have, therefore, studied the effect of Berkefeld filtration upon the alexic activity of fresh human serum with respect to the thermolabile elements, and the thermostable third and fourth components (Whitehead, Gordon and Wormall, 1925; Gordon, Whitehead and Wormall, 1926).