During normal life destruction and regeneration of red cells is constantly going on. Why this should be so we do not know ; it is convenient to say that red corpuscles ultimately wear out, and to attribute their destruction to a process of natural senescence. How much blood is thus changed every day is unknown, nor have we any idea of the natural length of life of a red cell. The constant activity of the marrow and the steady formation of bile pigment show, however, that there is a normal destruction.

The effete red cells are probably not dissolved in the circulating blood, though the absence of haemoglobin from the plasma and urine does not necessarily prove this. In animals about 1 per cent, of all the red cells in the circulation must be dissolved simultaneously before any haemoglobin appears in the urine . Free haemoglobin is quickly removed from the plasma,, even if none is excreted by the kidneys. In this way a good many red cells could be destroyed, a few at a time, without giving rise to obvious signs. It appears, however, that red cells which are to be destroyed are eaten by phagocytic cells, the cell disintegrated and the haemoglobin decomposed in the interior of endothelial and other cells in the spleen, bone marrow, liver and lymphatic glands. The remains may be recognized for some time within the phagocytes as globules and fragments containing haemoglobin ; the fragments then begin to give the microchemical tests for inorganic iron, which indicate the disintegration of the haemoglobin. The iron in great part remains in situ as " haemosiderin " (an iron oxide protein compound), while the haematoporphyrin part is presumably excreted as bile pigment. By this process no risk is run that the valuable constituents of the cells will be excreted and lost to the body. The view that cells which are destined to shortly break up are taken out of the circulation by the phagocytes, probably with the idea of avoiding intravascular haemolysis, is supported by the observation that, if considerable destruction of red cells within the circulation is induced by such agents as phenylhydrazine or haemolytic serum (infra), extensive phagocytosis of apparently healthy red cells by endothelial cells may be found. The cells eaten are presumably not really healthy, but injured by the poison administered, and if left alone would shortly dissolve in the circulating blood.

Haemolysis may be effected in a variety of ways

• (1) Physical : If red cells are placed in distilled water, salts diffuse out and water is taken up until the cell bursts ; conversely, they lose water and shrivel if they are placed in concentrated salt solutions. These observations show that a red cell has a differentiated outer layer or cuticle. With human blood haemolysis takes place with solutions of sodium chloride of about O4 per cent, and less ; all the cells are not equally resistant, and a few are not laked even in distilled water. Mechanical violence, such as shaking with shreds of asbestos or alternate freezing and thawing, also produces haemolysis.
• (2) Fat solvents and substances which can combine with the cholesterol and " lecithin " of the red cells : ether, chloroform, bile salts, saponin and soaps. Some of these are intensely haemolytic, and show that the structural integrity of red cells depends on its lipoids.
• (3) Various chemical substances, such as ammonium salts, arseniuretted hydrogen, aniline, dinitrobenzene, pyrogallic acid, etc.
• (4) Poisons of biological origin, produced by animals (e.g., snakes, spiders, toads) or plants (e.g., ricin from castor-oil seeds, phallin from toadstools), including bacteria (e.g., tetanolysin from the tetanus bacillus). In some cases these probably act as general protoplasmic poisons, with a specific point of attack ; others by lipoid combination.
• (5) Serum haemolysis.

The serum of one species of animal is often haemolytic towards the red cells of another species ; guinea-pig serum, for example, usually dissolves ox red cells and dog serum rabbit cells. Rabbit serum, on the other hand, is not haemolytie to human red cells, but soon acquires that property in a rabbit repeatedly inoculated with human cells. The mechanism of haemolysis in these cases is dealt with elsewhere.

Extensive intravascular haemolysis seldom occurs in human pathology. Ked cells are mechanically destroyed by endoglobular parasites (malaria) ; but, apart from rare cases of poisoning with some of the substances mentioned, the only conditions requiring notice are paroxysmal haemoglobinuria, blackwater fever and pernicious anaemia. The haemolysis in this last disease is probably intracellular ; at any rate, not enough haemoglobin is set free in the circulation at once to give rise to haemoglobinaemia or haemoglobinuria. Nothing is known of the mechanism of blackwater lever,* in which attacks of haemoglobinuria occur in malarious persons. In paroxysmal hcemoglobinuria^ we have a condition in which exposure to cold, either of the whole body or of an extremity, precipitates an attack of intravascular haemolysis, which is generally of such a degree that free haemoglobin is passed in the urine . Attacks sometimes appear to be without any exciting cause. The red cells themselves are not abnormally haemolyzable ; but the serum contains using the terminology at present in vogue a haemolytic immune body which can only combine with the red cells at a low temperature. When the temperature is thereafter raised, either experimentally in the test-tube or by the cooled blood returning from the extremities to the viscera, complement comes into play, and haemolysis results. In mild attacks there may be albuminuria instead of haemoglobinuria, and this condition generally precedes the haemoglobinuria in an ordinary attack. Up to a certain point, therefore, the haemin moiety seems to be taken up by the liver. It is of interest that the attacks of haemolysis in both blackwater fever and paroxysmal haemoglobinuria are accompanied by severe general illness, in the former condition often fatal. Experimental haemolysis frequently causes death within a few minutes or hours when only a trivial quantity of red cells has been destroyed. Haemolytic immune serum and some saponins are known to be active general tissue poisons, apart from their haemolytic action, but the same thing is seen after intravenous injection of distilled water. Why haemolysis in the blood should be dangerous is not known, but the facts supply another reason for avoiding intravascular in favour of intracellular destruction of red cells.

The resistance of red cells in vitro to haemolytic agents varies in different conditions. Thus, during active regeneration the newformed red cells, which are liable to be imperfectly made, are broken up more easily than normal cells by all haemolytic agents. In jaundice their resistance to hypotonic salt solution is increased, while to saponin it is diminished, since the lipoid constituents are already to some extent occupied by bile salts. Caution is required before such results obtained in vitro are transferred to in vivo phenomena. The haemolytic action of saponin and soaps, for example, is prevented by cholesterol (or serum), and is greatly dependent on their concentration. Though saponin, sodium oleate and haemolytic immune serum all produce haemolysis readily enough in the living animal, their action is quantitatively different in vivo and in vitro.