1. A spontaneous epidemic of Gaertner infection among rats was found associated with methaemoglobinaemia and, in some cases, anaemia.

2. Strains of Gaertner's bacillus isolated from these rats reproduced methaemoglobinaemia in fresh rats but not in rabbits, guinea-pigs or mice.

3. Other strains of Gaertner's bacillus from rats, guinea-pigs and human sources also caused methaemoglobinaemia either before or after passage through rats.

4. Other organisms pathogenic for rats did not produce methaemoglobinaemia.

The foregoing experiments conclusively show that a broth medium is quite unsuitable as a test culture medium to determine the vitality of anthrax spores in disinfection experiments, whereas agar is a suitable and delicate medium for the purpose, even when considerable traces of the disinfectant are carried over with the inoculation.

The reason for this inefficiency of broth is not obvious. We thought that it might be due to the absence of bacillar forms in the sporing material, but the emulsion of spores heated to 80°C. for 15 minutes and then inoculated directly into broth gave good growths.

Absence of oxygen might be another factor, but the results were the same when splinters of sterilised wood infected with anthrax spores were treated. The wood floated on the surface of the broth and so was subjected to a free supply of oxygen, yet no growths were obtained in broth when the splinters were soaked in the disinfectants, while good growths were obtained on agar. The control splinters gave good growths in broth. It may be that the anthrax spores are partially de-vitalised by the action of the disinfectant and that in this condition broth is a comparatively unsuitable culture medium for them. Prolonging the time of incubation of the broth cultures up to 10 or 14 days makes no difference. If a culture in broth shows no growth in 48 hours, a growth hardly ever appears with more prolonged incubation. Nor is this superiority of agar over broth as a culture medium confined to the emulsified disinfectants employed in these experiments, for similar results have been obtained with phenol, and with formaldehyde, the latter both in the fluid (formalin), and in the gaseous, conditions.

Infusoria have the same bactericidal power as flagellates. Emulsions containing Bacillus typhosus, Vibrio cholerae, V. Dunbar, V. El Tor, B. megatherium and Spirillum volutans, to which Colpoda cucullus is added, are soon cleared. Before the clearage the Colpodae multiply actively.

This bactericidal effect does not depend upon the production of toxic substances by the Infusoria. The fluid obtained from filtered cultures of the Infusoria exerts no bactericidal effect. Only living Colpodae are able to clear the emulsions.

Direct sunlight does not prevent the clearing of the emulsions by the Infusoria, but temperatures below 10°C. and above 30°C. and absence of oxygen (anaerobic culture) are unfavourable to their exerting a bactericidal effect.

The polluted water of the canals of Amsterdam slightly delayed the clearage of the emulsions; the sewage of starch factories and gas factories completely prevented this clearage.

Consequently under natural conditions Infusoria will only play a part in purifying river water (1) if the temperature is above 10° and under 30°C., (2) if the aquatic vegetation is rich enough to supply the necessary quantity of oxygen, (3) if the water is not highly polluted by adjoining factories.

In a recent communication (Penfold 1911) I described experiments with three organisms, B. coli Escherich, B. enteritidis Gaertner and B. Grünthal, showing that limitation or complete abolition of certain gas-forming powers normally possessed by these organisms, could be brought about by their growth on chloracetic acid-agar, and further, that in the case of the latter organism, resistance to this medium was associated with such limitation or abolition.

A short time ago I (1911) described a method for the quantitative estimation of indole. It may be of interest to reproduce the curves obtained by the use of this method, in the study of several strains of Bacterium coli, as curves representing indole-production have seldom been published.

The results which were obtained in the foregoing six experiments and are detailed in Tables I to VI may be summarised as follows.

It is seen then that any attempt to classify these two groups of cocci by means of complement-binding reactions would arrange them into more or less well marked sub-groups, some of which would contain both meningococcal and gonococcal strains, and some perhaps strains from only one of these groups.

It will be noticed that extracts of three strains of Meningococcus (119, 162 and 164) showed an especial tendency to give negative or feeble complement-fixation reactions with heterologous meningococcal sera, and two of these strains (119 and 162) also gave negative reactions with some of the gonococcal sera. Though the complement-fixation reactions of the strains used in this research were not fully worked out on account of the difficulty in obtaining satisfactory sera, nevertheless the following classification appears to be indicated having regard to complement-fixation alone.

Sub-group II has affinities for sub-groups I and III, but there is little, if any, affinity shown between I and III directly. M. 119 shows slight affinity for M. 141, but for no other strains.

The sub-groups are not clearly defined, but overlap and are connected with each other in various directions. For instance G. 1, M. 135 and G. 2 all appear to have common receptors, and M. 135 and G. 2 also have receptors in common with G. 4 and M. 162, but G. 1 shows no affinity for these two latter strains.

The explanation of these facts is not quite simple but they may be explained by assuming, (1) that several group antigens occur which are common to the Meningococcus and the Gonococcus, but only some of which are present in any given strain of coccus, and (2) that specific antigens which are peculiar to the Meningococcus on the one hand or to the Gonococcus on the other hand do occur, but are often absent in the case of any given strain.

The second assumption is perhaps unnecessary, and the first is almost equivalent to affirming the occurrence of special antigens peculiar to certain sub-groups which contain strains of both Meningococcus and Gonococcus.

The evidence, then, from complement-fixation experiments as also from other serum tests as far as they are of any value, appears to point to a closer relationship between some strains of Gonococcus and some strains of Meningococcus than between different sub-groups of Meningococcus. In fact rather the unity of these two groups than any essential difference between them, is suggested by these facts.

These considerations lend further support to the view that the most constant bacteriological characters available for differentiating the Meningococcus from the Gonococcus are the cultural characters seen when the organisms are grown on agar of different degrees of alkalinity.