The work reported in the preceding sections justifies, we think, a number of definite conclusions. In addition to this, some of the experiments indicate a line of thought which may lead to considerable alteration in our conceptions, both of phenomena of bacterial hypersensitiveness and of infection. 1. In guinea pigs two fundamentally different types of intradermal reactions may be observed. One of these is the immediate, transitory reaction which develops in animals sensitized against proteins (horse serum, etc.) and may be regarded as one of the manifestations of general protein hypersensitiveness, or anaphylaxis; the other is the tuberculin type of skin reaction which develops more slowly, leads to a more profound injury of the tissues and is independent of anaphylaxis as ordinarily conceived. 2. The tuberculin type of hypersensitiveness (as well as probably the typhoidin, mallein, abortin reactions, etc.) does not develop at all in guinea pigs sensitized with proteins, like horse serum, etc. While this form of hypersensitiveness may eventually be induced with materials not bacterial in origin, it has been observed up to date only as a reaction of bacterial infection. 3. Methods of treatment with protein material from bacterial cultures which sensitize guinea pigs to anaphylactic reactions with the bacterial extracts, do not sensitize them to the tuberculin type of reaction. Such sensitization is easily accomplished only by infecting the animals with living organisms. No reliable method of sensitizing guinea pigs to such reactions with dead bacterial material has as yet been worked out, though a few hopeful experiments have been obtained with massive injections of large amounts of the acid-precipitable substances (nucleoproteins?) from bacterial extracts. 4. In animals made hypersensitive to the tuberculin type of reaction by infection with living bacteria, the reaction may be elicited by intradermal injections of bacterial extracts from which all coagulable proteins, nucleoproteins, and Bence-Jones proteins have been removed, as well as this can be done by boiling with acid, etc. This proteose residue alone suffices to elicit such reactions. The exact chemical nature of the so called proteose residue must be further studied and analyzed when we have had opportunity to produce bacterial extracts in large quantity. These points seem incontrovertible on the basis of our own experiments, as well as those of other workers. There thus seem to develop two definite forms of hypersensitiveness in guinea pigs infected with bacteria, typical anaphylaxis in which the protein material of the bacterial cells is concerned, which develops late and which can be induced by repeated injections of dead bacterial material, and a hypersensitiveness to non-protein constituents which differs from the former, both in the laws that govern sensitization and in the manifestations which follow injections into the sensitized animals. While there is virtual agreement among immunologists concerning the essential mechanism of protein anaphylaxis, its dependence upon an antigen-antibody reaction, and the dominating rôle played by the sessile antibodies, the mechanism of hypersensitiveness to tuberculin and similar bacterial substances is still a problem of much uncertainty. The most striking difference between the two phenomena lies, as we have seen, in the criteria of sensitization, in that hypersensitiveness to the tuberculin type of reaction can hardly ever be induced by any of the ordinary methods of preparation with the constituents of dead bacteria, but develops promptly (7 to 10 days) in the course of actual infection with living organisms. The considerable specificity of such reactions forces the conclusion that the sensitizing substance must, in some way, be derived from the infecting microorganisms. The idea that the failure of sensitization with dead culture materials is perhaps due to the elaboration in the body of infected animals of bacterial products not represented in extracts of test-tube cultures is rendered unlikely by the fact that in the tuberculin-sensitive, infected animals, we can produce the reactions by the application of such dead extracts. It is neither logical nor in keeping with biological experience to assume that one substance will sensitize to reaction with another. This mistake was made early in the study of anaphylaxis in another connection and caused considerable delay of progress. Krause has shown that tuberculin sensitiveness may be blunted in infected animals by massive, but sublethal injections of tuberculin, and we have obtained some indications of the same thing. Moreover, others as well as ourselves have seen tuberculin reactivity decline in guinea pigs and in man in the stages of very severe infection. These facts would eliminate any assumption of mere cumulative injury as explaining this type of reaction, and stamp it as a mechanism at least analogous to ordinary anaphylaxis. The only remaining possibility to explain the difference between infected animals and those treated with dead bacterial constituents would be to assume that the difference must lie in the manner in which the sensitizing substance is administered to the animals, and that sensitization with the proteose residue materials depends upon criteria of sensitization differing in regard to the time and quantity factors from those governing protein sensitization. If one considers the relatively simpler chemical structure and perhaps physically greater diffusibility of the materials concerned in this reaction, one might readily expect such differences in the methods needed for sensitization. In keeping with such a line of reasoning our experiments have shown that the tuberculin active materials are constantly and rapidly being diffused out into the culture fluid from growing organisms, in quantities greater than can be extracted from similar amounts of the dead bacteria. It seems reasonable to assume from this that the same thing may happen in the animal body harboring a growing focus. And it would seem quite likely that the association of the tuberculin type of reaction with actual infection may depend upon the fact that sensitization to these non-protein substances depends upon a constant steady absorption of large amounts of the material. Moreover, the only hopeful experiments on the artificial production of tuberculin sensitiveness in guinea pigs obtained by us were those in which massive doses of the nucleoprotein material injected into guinea pigs gave rise to a moderate skin sensitiveness. Does the so called proteose residue form antibodies, and, if so, are substances analogous to antibodies involved in the tuberculin type of hypersensitiveness? The failure to transfer passively this form of hypersensitiveness to normal animals with the blood and tissues of tuberculin-sensitive ones would suggest that no antibodies are involved. But this is not conclusive on the basis of available experimental facts. We are inclined to believe that antibodies of a sort are involved, for the following reasons: (a) In our experiments with the uteri of highly sensitive extract-treated guinea pigs and of tuberculous guinea pigs, we have occasionally had positive reactions when the proteose residue alone was used. (b) We believe that these proteose substances are entirely analogous to the substances studied by Avery and Dochez (22) in the urine and blood of typhoid and pneumonia patients. They obtained precipitin reactions against homologous immune sera with the urine of infected cases concentrated by evaporation after boiling with acetic acid to remove coagulable proteins. (c) Petroff, with whom we discussed this proteose residue early in our work, has produced it, and tells us that he has obtained precipitin reactions with it by titrating it against the serum of a sheep treated for a long time with tubercle bacillus products. In suggesting an antibody response to a non-protein antigen we are aware that we are opposing what has been regarded as a well established doctrine in immunity; this is justified, or at least mitigated, we believe, by the consideration that reactions of the antigen-antibody type are the only explanation of specificity; and tuberculin, mallein, and typhoidin reactions are to a considerable degree specific. If such reaction bodies cannot be produced by precisely the same methods of administration as to time and quantity which are successful in calling forth protein antibodies, this should not astonish us, since, after all, the substances that we are dealing with are simpler in chemical structure than are the proteins, and physically are probably of relatively greater diffusibility. It may be that the greater diffusibility of the proteose-like substances transfers much of the actual reaction phenomena to an intracellular location, and that this to some extent influences the presence of circulating antibodies. It may also be that these more diffusible non-protein antigens are more rapidly eliminated from the animal body than are the proteins. Indeed, the above mentioned observations of Avery and Dochez, and the recent work of Wildbolz (23), Lanz (24), Imhof (25), and Gibson and Carroll (26), who demonstrated tuberculin active antigens in the urine of active cases, would corroborate such a view. The evidence available at the present time, however, concerning antibody formation to these non-protein substances is, we recognize, largely indirect, at least as far as our own work is concerned, and we present it in the present connection purely as a working hypothesis. Finally, perhaps the most important theoretical consideration indicated by our experiments is the following. We have in the tuberculin reaction a form of hypersensitiveness which seems to be (in guinea pigs, at least) analogous entirely to the typhoidin reaction, the mallein reaction, and the abortin reaction. Whenever reactions of this type have been carefully studied, whatever the bacteria involved, they have been associated with infection as in tuberculosis, and have been followed by analogous clinical manifestations. It would seem perhaps that we are dealing with a law applicable to bacterial infection in general. It would appear that certain non-coagulable substances of uncertain chemical constitution are being constantly elaborated in the course of bacterial growth, and passed into the circulation of infected animals. As a result of this, infected animals become sensitized to these heat-and acid-resistant materials, in tuberculosis in the course of I to 2 weeks, in the case of more rapidly growing bacteria perhaps sooner. Early in the course of infection, the animal becomes sensitized and subsequently the further elaboration and distribution of these materials from the bacterial focus plays a fundamental part in the injury of the animal. These proteose-like substances, like tuberculin, possessing but slight toxicity for the normal animal, become highly toxic to the sensitized one. Thus, these substances, while not being true exotoxins in the ordinary sense, would still represent a highly toxic bacterial product comparable in its injurious effect to toxins when produced in the body of an animal thus sensitized. If there is any value in these deductions the attention of bacteriologists should be turned to the non-protein constituents of bacterial cells in their further immunological studies, as well as to the protein materials. It is obvious that the next step in our investigations must consist in producing the non-coagulable material from bacterial extracts in considerable quantity, to determine their antibody-forming properties in detail, and elucidate, if possible, the laws which govern sensitization with them. This work has been begun, but it has seemed advisable to publish this as far as we have gone because it will take a long time before it can be completed.