Summary Sterile culture systems for the examination of symbiotic nitrogen fixation are described, and emphasis is laid on their significance in preventing erroneous results. The hypothesis advanced by some previous workers of the change of poor bacterial strains to effective ones, and vice versa, by repeated plant passages is disproved. There is regular difference in colour between the effective and ineffective nodules. Effective nodules always contain red pigment, ineffective never. Another difference between effective and ineffective nodules is the occurrence of swollen bacteroids in the former and their non-occurrence in the latter. This difference is a striking one in the experiments carried out in our laboratory with peas but in the literature some records are found which indicate that it is not general in all legumes. The problem requires further elucidation. The red pigment of effective root nodules is haemoglobin, which is able to store and transfer oxygen. For the sake of brevity it has been named leghaemoglobin. Its divalent iron is easily changed to the trivalent form, legmethaemoglobin. Young nodules chiefly contain leghaemoglobin when growing in a good light; with decreasing light the amount of brown legmethaemoglobin increases. Leghaemoglobin or legmethaemoglobin changes gradually during the ripening of plants to a green pigment or pigments. A similar change occurs in the pigments if young plants are kept in the dark. The green pigment still contains iron though it can be easily split off with acids. The green pigment may still contain protein. It is assumed that the green pigment is similar to those formed from the haemoglobin of blood in a solution containing ascorbic acid through the agency of oxygen or hydrogen peroxide. Leghaemoglobin has been isolated with a purity of about 85 %, assuming that the haemin content of the leghaemoglobin is the same as that of the haemoglobin of blood. The molecular weight was 34,000. The observations so far made show a distinct parallelism between nitrogen fixation and the occurrence of leghaemoglobin. If the pigment is not present in the nodule, or it it is decomposed, nitrogen fixation does not occur. The formation of the leghaemoglobin is evidently brought about by the host-bacteria symbiosis, as also is the formation of bacteroiď mass in the nodules. The causes of the inability of ineffective strains to form ‘bacteroids’ and leghaemoglobin is not known. The possibility is suggested that the slimy layer surrounding the rods may isolate the bacteria and thus prevent the onset of symbiosis. Both the bacteria and the host-plant specificity are discussed in the light of the above observations. An effective bacterial strain forming red nodules on a certain host plant may on another closely related variety form white ineffective nodules. Thus the host plant largely determines the effectiveness of bacterial strains. The chemical significance of leghaemoglobin in nitrogen fixation is discussed. It is possible that its effect is restricted to the transfer of oxygen. The excretion of organic nitrogen compounds to the substrate of inoculated legumes, about which there has been much argument, has gradually been established in many laboratories. The causes of the great differences in the rate of excretion are, however, still obscure. The problem has been treated from a new aspect: the unusually high excretion of amino-dicarboxylic acid nitrogen, in certain cases exceeding the proportion of fixed nitrogen the plant has taken up, supports the idea that these amino-acids are of primary importance in nitrogen fixation. The data recently obtained with Azotobacter by the isotopic method lead to the same result. The transamination of the amino group from amino-dicarboxylic acids to aliphatic α-keto acids has been demonstrated in green plants and in root nodules. The conception that the nitrogen fixed by the root nodules is transferred chiefly by excretion to the cytoplasm seems justifiable in view of the experimental data. Not until the last stage of nitrogen-fixation does the central bacterial mass of the nodule decompose. Then the decomposition products are partly used by the host plant. The carbohydrate nitrogen ratio in symbiotic nitrogen fixation has a marked effect on nitrogen fixation. No explanation of this phenomenon is available, but the question has been considered from a new point of view. Investigations of the accessory growth factors of legume bacteria and of their bacteriophages have failed to throw light on the phenomenon of symbiosis.