Sulfonamide-resistant strains of Staphylococcus aureus produce greater amounts of p-aminobenzoic acid than do their parent strains. This synthesis occurs both in the absence and in the presence of sulfonamides. The quantity of p-aminobenzoic acid synthesized by resistant strains appears sufficient to account for their resistance to sulfonamide drugs. On the basis of this evidence, it is suggested that the development of ability to synthesize p-aminobenzoic acid in excess of the normal metabolic requirements, as a result of continued exposure to sulfonamides, explains the phenomenon of sulfonamide fastness in Staphylococcus aureus.

BACKGROUND AND OBJECTIVE: Although risk factors for extended spectrum beta lactamase E. coli (EBLE) infection have been explored, specific risk factors for bacteremic urinary tract infection by EBLE have been hardly analyzed. PATIENTS AND METHOS: We collected data from all patients with bacteremic urinary tract infection by E. coli attended in our hospital during 2006. Logistic regression was performed to explore predictors for EBLE bloodstream infection in this group of patients. RESULTS: EBLE was present in 19 cases (17,9%) out of 106 bacteraemia from urinary origin. Patients with bloodstream infection by EBLE were male, older, demented, living in a nursing home, with previous urologic diseases and urologic manipulation, with a higher percentage of previous urinary tract infection, previous antibiotic use, more frequent nosocomial infection, and hospital admission in the previous month. In the logistic regression analysis, only previous urologic diseases (OR 13,9; IC95% 2,5-78,2) and living in a nursing home (OR 6,5; IC95% 1,4-0,9) were associated with EBLE bacteremic urinary tract infection. CONCLUSIONS: Previous urologic disease and living in a nursing home are independent risk factors for EBLE bacteremic urinary tract infection.

The above data relating to the reaction between 16 hour cultures of S. aureus and antistaphylococcus bacteriophage in nutrient broth of pH 7.6 at 36 degrees C. and with mechanical shaking to maintain a uniform B suspension, bring out the following points:(a) B growth in P-B mixtures does not differ from growth in controls without P except in the case of a very high initial P/B ratio as noted below. There is no evidence that lytic destruction of B begins shortly after mixing P and B nor that B growth is stimulated by P, for the B growth curves in the presence of ordinary [P]'s and in controls are identical. Only at the sudden onset of the rapid lytic process does the B curve of a P-B mixture deviate from the control curve.(b) B growth is an essential conditioning factor for P formation.(c) Both B growth and P production exhibit short lags. During this time P diffuses into or becomes adsorbed to B so rapidly that by the end of the lag period only 10 to 30 per cent of the total P present is extracellular, the remainder being associated with the B.(d) During the logarithmic B growth phase, P formation is also logarithmic but proceeds at a much faster rate. That is, d P/d t is proportional to a power of d B/d t. Consequently the statement that each time a B divides a certain amount of P is formed is not correct.(e) As B growth enters the phase of positive acceleration equilibrium between the extracellular and intracellular P fractions becomes established and is maintained up to the onset of lysis, extracellular [P] representing a small constant percentage of total [P]. The distribution of P on a constant percentage basis suggests the manner in which a relatively simple chemical compound would be distributed and is not at all typical of the distribution one would expect if P were a complex organized parasite.(f) When the value of log P/B = 2.1 lysis begins. Obviously, this limiting value for any initial [B] is reached sooner the higher the initial [P]. When log P/B at the time of mixing P and B is already 2.1 or greater, there is no growth of B and lysis soon occurs.(g) While there is good evidence that lysis is brought about by the attainment of a particular [P] per B and not by a certain [P] per ml., it is not clear at this time which of the ratios intracellular P/B, extracellular P/B or total P/B is the major conditioning factor for B lysis.(h) Experimentally the maximal [P]'s of lysates made by mixing a constant initial [B] with widely varying Po's fall within a relatively narrow range. This fact is explained by the large value of d log P/d t as compared to d log B/d t. That is, the loci of points at which log P = 2.1 + log B (maxima-lysis begins) on the curves of log P against t originating in various [Po]'s will lie at a nearly constant level above the abscissa. Because of this same relationship the maximal [P]'s of such a series will be in the reverse order of magnitude of the Po's, i.e., the larger the Po the smaller will be the maximal [P] attained during the reaction (cf. Fig, 16).(i) The lytic destruction of B is logarithmic with time, in this respect being similar to most death rate processes. The value -d log B/d t for a particular initial [B] is constant for various initial values of [P]. There is good evidence that cells need not be growing in order to undergo lysis.(j) During B lysis a considerable percentage of the total maximal P formed is destroyed, the chief loss probably occurring in the intracellular fraction. The major portion (70 to 90 per cent) of the final P present after the completion of bacteriophagy is set free during the brief phase of bacterial dissolution.(k) When the entire process of bacteriophagy is completed the lysates are left with certain [P]'s determined by the foregone P-B reaction. The destruction of P during lysis is sufficiently regular to maintain the relationship established at the maximal [P]'s. Therefore the final [P]'s have the same points in common that were noted in "h" as applying to the maximal [P]'s. That is, they all are grouped within a narrow range of [P] values, those having been made with high Po's being of lower titre than those made with low initial [P]'s.(1) There is a significant difference in the temperature coefficients of P and B formation. Further, the temperature coefficients of P and B destruction during lysis differ in almost the same ratio. Consequently, while all experimental evidence postulates B growth as an essential conditioning factor for P formation, the temperature coefficient data suggest that the two processes are basically separate reactions. A similar interpretation holds in the case of B dissolution and P inactivation.(m) The major events in the complete process of "bacteriophagy" are mathematically predictable. The [B] at which lysis occurs under certain standard conditions for given values of Bo and Po may be calculated from the equation: See PDF for Equation Substitution of this value for log B in the equation: See PDF for Equation gives satisfactory agreement with observed values for t((lysis)).(n) The kinetic analysis of the P-B reaction predicts that the values of log Po plotted against t((lysis)) for a constant Bo will give a straight line. This plot is employed in a method for the quantitative estimation of P described in an earlier paper on the basis of experimental observation alone. Its use is made more rational by the facts given above.

The addition of active or irradiated T2 bacteriophage and T4 bacteriophage to E. coli B stops bacterial multiplication. The respiratory rate and respiratory quotient of the inhibited bacteria remained at the values observed just before infection. A respiratory rate decrease which occasionally appears can be roughly correlated with change of turbidity of the suspension. An intracellular inhibitor of multiplication appears to be liberated into lysates. A similar substance has been separated from normal E. coli B after sonic disintegration. These bacteriostatic preparations contain cytoplasmic granules with lactic acid dehydrogenase activity. The relationship of these phenomena to the interference effect in this system and others has been considered.

1. Mice with relatively great inherent resistance to certain bacterial infections were heavier but not more fertile than mice with relatively little inherent resistance. Mice with relatively little inherent resistance were with one exception not abnormally low in weight. 2. B. enteritidis given intrastomachally to susceptibles appeared in the blood stream more promptly, in larger numbers, and in a greater percentage of cases, and was present in feces in larger numbers, for a longer period, and in a greater percentage of cases than when given to resistants. 3. Mice relatively resistant to B. enteritidis administered by the natural gastrointestinal route were likewise resistant to the organisms introduced subcutaneously, intraperitoneally, and intravenously. Mice relatively susceptible to the organisms administered by the natural route were susceptible when the organisms were injected directly into tissues and blood stream. 4. Of four lines of mice relatively susceptible to B. enteritidis, three were likewise susceptible to Pasteurella avicida, B. friedlaenderi, and pneumococcus given intranasally. A fourth line was significantly more resistant. Lines of mice relatively resistant to B. enteritidis were likewise resistant to the three respiratory tract pathogens. 5. When Pasteurella avicida, B. friedlaenderi, and pneumococcus were injected intraperitoneally or intravenously, no significant differences in duration of life of the several lines of mice could be demonstrated. 6. Of four lines relatively susceptible to B. enteritidis, two were susceptible to an intranasal instillation of louping ill virus. Lines resistant to B. enteritidis proved relatively susceptible to the virus infection.

This paper contains the records of a motion photomicrographic investigation of the lysis of Bact. coli and B. megatherium by bacteriophage. The bacteria mixed with bacteriophage were grown on moist nutrient agar in small culture chambers on the stage of a microscope in an incubator maintained at 37 degrees C. The apparatus used permitted continuous inspection of the preparations. Photographs were made at the rates of 2 and 30 per minute and at the rate of 8 per second during the terminal stage of lysis of Bact. coli. The accurately timed films were studied by rapid projection and by the projection of single frames. Measurements of dimensions of cells, calculations of volumes, information on generations, generation times and duration spans are presented in the tables. Similar information on normal cultures grown and photographed in the same way is furnished for comparison. Groups of serial photographs are reproduced in the plates to illustrate the special features observed. These observations seem to us to warrant the following conclusions: 1. Enlargement or swelling of the cells of Bact. coli usually, but not always, precedes lysis. Some of the enlargement is an expression of increase of cell substance and is not altogether due to imbibition of water. Cells of early generations of Bact. coli enlarge to greater absolute and relative proportions than cells of later generations. Enlargement does not occur before lysis in B. megatherium. 2. The terminal stage of lysis of Bact. coli is explosive, occupying (1/2) to (7/8) second. The terminal stage of lysis of B. megatherium is a slow disintegrative process, extending over 2-10 minutes. 3. Bacteriophage inhibits fission of some cells, but does not stop the reproduction of other cells in contact with it. The genealogical records of six generations of cells of Bact. coli and of two generations of cells of B. megatherium indicate that bacteriophage may be transmitted through parents to the offspring which ultimately undergo lysis. 4. Bacteriophage spreads by contact through a group of cells and also along paths determined by genetical relationships. 5. A large amount of cellular debris remains after the lysis of the cells in both of these species of bacteria. This residue of material is in the form of irregularly shaped masses and granules. This material is not in solution at the time of lysis and appears not to be digested or hydrolized. 6. Theories of the mechanism of lysis are discussed. It is suggested that reduction of surface tension of the cells may be an important factor in the mechanism of lysis.

It has been found that although there is some parallelism between the quantity of tubercle bacilli demonstrable histologically and the number of colonies that can be isolated from a given tissue, the culture method is far the more efficient in indicating quantitative relations. Tubercle bacilli were not perceived in the organs of rabbits 1 day after infection with the modified BCG although as many as 1,500 colonies were isolated from one of them. This may be solely because it is difficult to see widely dispersed single minute acid-fast rods in the diffuse infiltrations of mononuclears with their hyperchromatic nuclei and sparse cytoplasm. Later, with the formation of tubercle, the parallelism is much closer. The culture method gives evidence concerning the number of living tubercle bacilli in the tissue. The significance of the accumulation of acid-fast particles in the tissues has been discussed. It has been seen that from the beginning this accumulation is greater in the Kupffer cells of the liver, in the macrophages of the spleen and in the reticular cells of the bone marrow than within the mononuclears of the lung, the organ where the bacilli grow with the greatest rapidity and are destroyed with the greatest difficulty. Acid-fast particles are more prominent with the bovine than with the human bacillus or the BCG, the microorganism that is destroyed with the greatest difficulty thus leaving more incompletely digested bacillary debris at a given time within the cells. Thus it seems permissible to conclude from the presence of acid-fast material that some tubercle bacilli are undergoing destruction even 24 hours after infection. The initial accumulation of polynuclear leucocytes corresponds with the subsequent severity of the infection. Despite the greater primary localization of bacilli in the liver, this initial inflammatory reaction with all three infections is much greater in the lung than in the liver. In each organ it is more intense with the bovine than with the less virulent strains. The multiplication of the bacillus and its accumulation within large mononuclear and young epithelioid cells is accompanied by an intense formation of new mononuclears by mitosis. The more rapid the growth of the bacillus, the more conspicuous the regeneration of these cells. Thus with all strains mitosis is more intense in the more susceptible organ, as in the lung compared with the liver; with the most virulent strain the most extensive and diffuse accumulation of these new cells corresponds with the greater rise in the numbers of bovine bacilli after the lag of the 1st week. With the maturation of the epithelioid cells and the formation of tubercles the bacilli have already been greatly reduced numerically and the speed of this process diminishes with the virulence of the three strains used. The faster the development of tubercle the faster the destruction of the bacillus and the earlier the resorption of the tubercle. Tubercle bacilli never accumulate in such large numbers in the mononuclears of the liver as they do in the lung. Though at first the tubercles in the liver may be more numerous than those in the lung they never attain the same size. The formation of new mononuclears by mitosis is restricted and Langhans' giant cells appear very early (1st and 2nd weeks). In the lung, giant cells are not found until much later with the BCG and the human bacillus (4th week); they were not noted in the interstitial tubercles with the bovine type, but the extension of these tubercles was accompanied by an unabated mitosis of mononuclears until the death of the animal. The liver tubercles are resorbed early even with the bovine infection. Associated with these histological differences are the slow initial growth and the early and complete destruction of the tubercle bacilli even of bovine type in the liver, and the more rapid initial growth in the lung, with the later destruction of the BCG and the human bacillus and the unabated growth of the bovine bacillus. Similar differences were observed between the splenic pulp and corpuscle. In the former the accumulation of acid-fast particles was much greater and the tubercles developed earlier. Mitosis of mononuclears was less frequent and giant cells appeared earlier. Tubercle bacilli, always intracellular, disappeared from the tubercles in the pulp sooner than from those in the corpuscle, and the tubercles themselves first disappeared from the pulp. Consequently with the persistence of bacilli mitosis continued in the tubercles of the corpuscle and these attained a much larger size. Moreover individual resistance is linked with the ability to form mature tubercles early. In two animals simultaneously infected with the same strain and killed at the same time, the destruction or retardation of the bacillus is greater in that rabbit in which maturation of the tubercle and of epithelioid cells has proceeded further (Figs. 15 and 16). These observations indicate that the mononuclears of different organs or even of the same organ, as in the different parts of the spleen, have a different capacity to destroy the tubercle bacillus, and that the transformation of the mononuclear into the mature epithelioid cell follows its destruction of the tubercle bacilli. In the lung the more virulent types of bacillus are destroyed within the epithelioid cells of interstitial tubercles but persist in foci of tuberculous pneumonia. In this organ in rabbits infected with the human strain and to a lesser degree in rabbits infected with the bovine strain, the parasite largely disappears from the epithelioid cells of interstitial tubercles. But with both strains tubercle bacilli in large numbers may accumulate within epithelioid cells lying free in the alveoli. With the human type they are numerous within the cells and free in caseous material in the localized foci of caseous pneumonia. With the bovine infection, this caseous pneumonia is more often widespread and in the areas of caseous pneumonia the greater part of the vast accumulation of bovine bacilli in the lungs is found; as many as 200,000 colonies have been isolated from 10 mg. of tissue (Fig. 11). Flooding of the respiratory passages by the caseation of tuberculous lesions into the bronchi plays an important rôle in dissemination of tubercle bacilli through the lung. The process on the contrary is predominantly interstitial when the bovine bacillus is held in check (Fig. 12). Thus there is apparently some factor acting in the alveoli that favors the growth of the parasite. The accumulation of tubercle bacilli is seen especially in the peripheral epithelioid cells in immediate contact with the alveolar space. In the same lung the bacilli are much fewer in the interstitial tubercles. The accumulation in human tuberculosis of large numbers of tubercle bacilli in the tissues lining cavities is well known. Novy and Soule (20) have shown that within certain limits the growth of the bacillus in vitro is proportional to the oxygen tension of its environment. Corper, Lurie and Uyei (21) have confirmed these observations and have noted further that a difference in the gaseous environment of the bacilli equal to the difference between the conditions existing in the alveolar air and the venous blood is sufficient to cause a considerable increase in the growth of the microorganism in vitro. Loebel, Shorr and Richardson (22) by the use of Warburg's manometer have found that the oxygen consumption of tuberculous tissue is such that a tubercle 0.5 mm. thick would completely exhaust the oxygen of the air before it reached the center. These observations suggest that a factor responsible for the greater multiplication of the bacillus in the cells of the alveoli may be the greater oxygen tension of the alveolar air. In the liver, spleen and bone marrow even with the bovine infection many instances were found of the effective destruction of the parasite synchronously with the maturation of epithelioid cells and the formation of tubercle. On the other hand, in the spleen and bone marrow of some rabbits, living bacilli persisted within the epithelioid cells of isolated tubercles even 2 months after infection, a condition never found with the human type or BCG infection. Thus the epithelioid cell is the means of defense for the rabbit against the bovine type bacillus, and as such it is usually adequate in the liver, spleen and bone marrow though ineffective in the lung and kidney. In the latter, descending infection, and the occasional colony-like multiplication of bacilli in unorganized material, tubular casts, determine the long persistence of large numbers of bacilli in this organ. In differentiating the mononuclear phagocyte of the connective tissues into the monocyte and clasmatocyte Sabin and her coworkers (23) have maintained that the clasmatocyte can efficiently destroy the tubercle bacillus but that the monocyte and its derivatives, the epithelioid and Langhans' giant cells, cannot. With the progress of the disease they have noted that the monocytes accumulate in great numbers in the foci of infection and overflow into general circulation (4). White (24) and Sabin and her coworkers have concluded that tuberculosis is specifically a disease of the monocyte, and that this cell and its derivatives act as incubators for the tubercle bacillus. Doan and Sabin (25) have therefore sought, with indecisive results, to protect the body against tuberculosis by an antimonocytic serum. However it has been shown here that although an intense multiplication of mononuclears is associated with the growth of the tubercle bacillus, their transformation into mature epithelioid cells is constantly associated with its destruction, and the rapidity of the destruction varies with the rapidity of the maturation of tubercle. Even in the bovine infection the epithelioid cells destroy the bacilli in the liver, spleen and bone marrow as a rule, and even in the lung, keep them in check in the interstitial tubercles. The appearance of giant cells is associated with cessation or diminution of mononuclear regeneration by mitosis, and is coincident with cessation of multiplication or marked reduction in the number of living bacilli. They therefore appear earlier and in larger numbers in these organs or parts of organs that first destroy the bacillus (Figs. 16 and 17). They were not observed even 2 months after the bovine infection in the interstitial tubercles in the lung. Their absence and the continued mitosis of mononuclears, which accounts for the massive pneumonic and interstitial consolidation of the lung with this infection, were associated with the failure of the lung to destroy effectively the bovine parasite. The formation of giant cells in the pneumonic foci in the bovine infection would seem to be an exception to this rule. The Langhans giant cells have often been considered an indication of the chronicity of the pathological process. It would appear that they are formed from existing epithelioid cells when the multiplication of the bacillus has ceased and the stimulus for the formation of new cells has decreased or stopped. Giant cells were most conspicuous in the liver and splenic pulp where, with the BCG infection, no caseation ever developed, and in the liver before caseation was seen anywhere in the body. In the human and bovine infections, giant cells formed in the liver before caseation appeared. Hence caseation is not a necessary requirement for giant cell formation, as maintained by Medlar (26), though these cells frequently form about caseous material. Lymphocytes and granulation tissue do not cause the destruction of tubercle bacilli, these being destroyed in their absence. They usually appear about tubercles due to all strains and in all organs, after the greater part of the microorganisms have been destroyed (Fig. 18). The bacilli are not destroyed in the lung with bovine infection where the tubercles are usually little permeated by lymphocytes and granulation tissue. There is however, no constant relation between granulation tissue and destruction of tubercle bacilli, for in the lung after the human infection and even in other organs after the bovine infection isolated tubercles may be surrounded and penetrated by lymphocytes and granulation tissue at a time when considerable numbers of living bacilli are still histologically demonstrable within the epithelioid cells. Caseation is usually not caused by the local accumulation of tubercle bacilli. At first, when the BCG (after 1 week) and the human microorganism (after 2 weeks) are present in the cells in very large numbers as demonstrated both histologically and by culture (Figs. 4 and 13) there is no necrosis of these cells. An exception to this rule found in the lung with the bovine infection is considered below. Later, after the bacilli have been destroyed to a great extent and even though the number of bacilli is small, caseation appears (Fig. 14). After this preliminary destruction the extent of caseation apparently varies with the number of residual bacilli. With the least virulent microorganism, the BCG, few bacilli remained in the liver in the 4th week and no caseation was seen. In the tubercles of the splenic corpuscle at the same time bacilli were somewhat more numerous and there was scant caseation. On the other hand with the human bacillus after 4 weeks more bacilli survived and caseation was more extensive in both organs; with the bovine microorganism tubercle bacilli were much more numerous and caseation was far advanced. In the lung, however, caseation appeared with the first considerable accumulation of the bovine bacilli present 2 weeks after inoculation. That the bovine bacillus is primarily more injurious to the lung of rabbits than the BCG or the human bacillus is suggested by the greater intensity of the initial inflammation and by the more conspicuous accumulation of cells in the alveoli evident from the very beginning of infection. Maximow (27) showed that bovine bacilli even in small numbers cause the death of cells in tissue cultures of rabbit lymph nodes whereas the BCG or the human bacillus may accumulate within the cells in tremendous numbers without injuring them. Nevertheless in the liver, spleen and bone marrow of the living animal, caseation does not appear at the time when bovine bacilli are most abundant, but after they have been greatly reduced in numbers. Large numbers of the less virulent types of tubercle bacilli accumulated in different organs a short time after infection do not cause caseation, and with the bovine infection caseation under the same conditions occurs only in the lung. Later when the animal is sensitized caseation occurs in various organs in the presence of the small numbers of tubercle bacilli that remain in the tissues after most of them have been destroyed, and the extent of this caseation varies with the numbers of residual bacilli. These observations suggest that a large number of bacilli fail to cause necrosis soon after infection whereas a few bacilli produce caseation in the animal that is sensitized. Many investigators have held that caseation is due to sensitization. Krause (28), Huebschman (29) and Pagel (30) think that caseation is caused by the action of tuberculin-like substances on the sensitized tissues of the allergic animal. Rich and McCordock (31) view the process in essentially the same light. Recently Schleussing (32) has suggested that caseation is a coagulation necrosis in Weigert's sense of an allergically inflamed tissue, and is similar to the necrosis of the Arthus phenomenon.

1. During early stages of multiplication, single cells from smooth-, mucoid-, and rough-susceptible and variant colonies show no differences in morphology or growth rate. 2. Cells from 18 to 24 hour single cell cultures of these various colony types possess similar oxygen absorption and cataphoretic migratory rates. In staining property, the cells from mucoid colonies appear larger, and those from rough colonies smaller, than the typical cells from smooth-susceptible colonies. 3. Cells from bacteriophage-resistant colonies differ from those of bacteriophage-susceptible colonies in their ability to multiply luxuriantly in the presence of bacteriophage, and in their tendency to flocculate in acid solutions at pH 3.8 to 4.1, as well as in their low degree of virulence. 4. Cells from smooth bacteriophage-susceptible colonies in contact with bacteriophage under conditions where multiplication is restrained may be altered so as to resemble the cells from the bacteriophage-resistant colonies. 5. These facts furnish evidence that bacteriophage adheres to the surface of the bacterial cell and that the various cell changes and colony alterations are of an environmental rather than genetic nature.

1. A lytic agent was not demonstrable in culture filtrates of either the parent or variant type of the scours organism. 2. The parent type was resistant to the action of a "weak" bacteriophage, obtained from the animal host, while the variant type was susceptible. 3. Exposure of the variant type to the scours bacteriophage was attended by agglutination of the cells, marked swelling, and an alteration of the contents prior to lysis. 4. The manifestations of variation which regularly occur on agar plate cultures of the scours organism do not appear to be the result of bacteriophagic stimulation.

We have been able to confirm the observations of Twort as well as of Gratia, that dead staphylococcus may undergo lysis if, in addition to a suitable bacteriophage, there is also present live staphylococcus. Moreover, we have endeavored to ascertain the mechanism of this phenomenon and have found that in order to elicit it it is necessary to control the numbers of live and dead bacteria in the mixture. An excess of dead bacteria interferes with lysis by adsorbing the bacteriophage before it has the opportunity to initiate necessary changes in the live bacteria, so that all lysis is prevented. The phenomenon is specific, that is, the lysis of live bacteria is accompanied by lysis of dead bacteria of the same species only. Lysis of dead bacteria occurs best with staphylococcus, an organism which easily undergoes spontaneous autolysis under appropriate conditions. In the case of B. coli or B. dysenteriae the lysis of the dead bacteria is uncertain. Dead bacteria need not be present in the mixture at the beginning of the experiment; they will be dissolved if added any time before, during, or after the completion of lysis of live bacteria. If the test is performed so that a suitable semipermeable membrane is interposed between the dead and live bacteria, the dead bacteria are not dissolved, in spite of the lysis of live bacteria on the other side of the membrane. The agent determining the lysis of dead bacteria is not diffusible, while the principle initiating the lysis of live bacteria diffuses freely and is demonstrably present on both sides of the membrane. The complete independence of the agent causing dissolution of dead bacteria from bacteriophage can also be shown by separating the two agents by means of filtration, or by adsorption on bacteria. The ferment-like substance responsible for the lysis of dead bacteria is different from the bacteriophage. It is not diffusible through collodion, it is easily adsorbed on clay filters, it is heat-labile, and is inactivated on standing. An agent possessing identical properties was found in cultures of staphylococcus undergoing spontaneous autolysis in the absence of bacteriophage, but in this instance the agent appeared in the filtrates considerably later than it did when phage was present.

The investigations of the influence of oxygen on the bacteriophage phenomenon recorded in this paper show that this factor plays an important role which is due exclusively to its ability to induce certain changes in the behavior of bacterial cells towards lytic principle. At certain H ion concentrations both aerobic and anaerobic cultures can be made resistant in the absence of lytic principle. The resistance thus acquired is of a stable nature under suitable conditions. If the lytic principle is added to aerobic or anaerobic types of resistant strains the cultures are able to regenerate it to a certain extent. However they do not undergo any visible lysis themselves even under the action of lytic principles which were passed through several generations of these types. The principles regenerated by both types of resistant cultures are identical in action with each other as well as with the stock lytic principle. These experiments suggest a new method of investigation into the hitherto unexplained nature of resistance of bacteria towards bacteriophage.