A second surface antigen, B L , lethal in the presence of specific antibody and complement has been identified on some human cells and shown to behave as a good genetic marker. It is autosomal, unlinked to the human A L antigen previously described, and unlinked to 15 other human genes. The A L antigen, which is linked to the lactic dehydrogenase A gene, is found on the HeLa, the cultured human fibroblast, and in small amounts on the human lymphocyte. B L occurs on HeLa cells, on cultured human fibroblasts, and on human lymphocytes, but not on human RBCs. Hybrid cells formed by fusion of human and Chinese hamster cells have been prepared containing each of the four possible combinations of these two markers. Highly selective antisera sensitive to each marker separately can be obtained. The use of single-cell plating to demonstrate the presence of the antigens and of hybrid cells containing desired combinations of the markers facilitates study in this system.

Two purified serum protein fractions, fetuin and serum albumin, will replace whole or dialyzed serum in supporting the growth of single S3 HeLa cells in an otherwise chemically defined nutrient solution. In the serum-free medium, single S3 cells will form macroscopic colonies with essentially 100 per cent efficiency. The generation time of S3 cells in the serum-free medium is approximately 50 per cent greater than that observed in an optimal, serum-containing medium. All components of the serum-free medium are available commercially, except fetuin, which can easily be prepared in substantial quantities. The problem of the purity of the protein preparations and of their possible roles in promoting cell growth is discussed.

A methodology designed to eliminate mitotic inhibitor action and involving use of pretested fetal calf serum and careful pH and temperature control has been described by which cells from normal human and animal tissue can be maintained in active growth for long periods in vitro without development of aneuploidy. By means of this procedure, it is possible reliably to establish cell cultures from minute skin biopsies which can be taken from any individual. Clones of mammalian cells with chromosomal markers have been isolated by this means from x-irradiated non-irradiated cell cultures. Application of these techniques to chromosome delineation in large numbers of human subjects; determination of chromosomal sex in patients; spontaneuos and induced genetic changes in somatic mammalian cells in vivo and in vitro ; comparison of metabolic differences between normal and cancerous cells and other problems have been indicated.

A convenient, reliable method for chromosome delineation of animal cells grown as monolayers on glass has been applied to human, opossum, and Chinese hamster cells. Tissue cultured cells from 5 different, normal organs of 7 different human subjects uniformly displayed the expected chromosome number of 46 and showed no variations in morphology or number other than the expected sex differences and a small incidence of polyploidy. The chromosomes of normal cells from the American opossum were as uniform as those of human cells. Cells of the inbred Chinese hamster demonstrated appreciable karyotype variability, the cause of which is under investigation. The chromosome number and morphology of cells from normal human tissues have remained constant after more than 5 months of continuous, rapid growth in tissue culture involving scores of vessel transfers and a number of generations equivalent to many billions of progeny. By the use of routine recloning, even cells of malignant, aneuploid constitution have been maintained in active growth for 3 years and hundreds of generations, with stable chromosomal and metabolic characteristics. The cells of the American opossum and Chinese hamster which possess only 22 chromosomes have been established in vitro and are especially suitable for genetic studies. The readily recognizeable Y and X chromosomes of the male opossum are particularly favorable as cytological markers. Photomicrographs of the chromosomes of the various cells employed are presented.

Survival curves of normal human cells from a variety of tissues exposed to varying doses of x-irradiation have been constructed, which permit definition of the intrinsic radiation sensitivity of the reproductive power of each cell type. The mean lethal dose of x-irradiation for all the cells employed, including those from normal and cancerous organs, those exhibiting diploid and polyploid chromosome number; those from embryonic and adult tissues, including recently isolated cells and cultures which had been maintained in vitro for many years, and cells exhibiting either epithelioid or fibroblastic morphology, was found to be contained between the limits of 50 to 150 r. Other similarities in the pattern of radiation effects, such as giant formation and abortive colonial growth, in these cells and that of the HeLa S3, previously studied, confirm the hypothesis that the pattern of reaction to x-irradiation previously elucidated, is representatative, at least in over-all outline, for a large variety of human cells. While the radiation survival curves of various human cells are similar in the gross, small but important characterizing differences have been found. All epithelioid cells so far studied are approximately 2-hit, and more radioresistant than the fibroblast-like cells whose survival data correspond to a mean lethal dose of around 60 r, and which so far can be fitted by either 1-hit or 2-hit curves. The earlier prediction that the major radiobiologic damage to mammalian cells is lodged in the genetic apparatus was confirmed by the demonstration of high frequency of mutants among the survivors of doses of 500 to 900 r. All the data on the x-radiosensitivity of these cells can be explained on the basis of a defect resulting from primary damage localized in one or more chromosomes. These considerations afford a convincing explanation of several aspects of the mammalian radiation syndrome.

A methodology has been described for reliable cultivation in vitro of dispersed fibroblastic cells obtained from normal human organs. The procedure has permitted establishment of stable cell lines from almost every sample taken, among which the following organs were represented: skin, spleen, amnion, lung, liver, bone marrow, brain, muscle, and heart. Equally good growth has been achieved with cells from embryonic or adult tissues. The methods previously developed whereby single cells plated in Petri dishes grow into isolated macroscopic colonies can successfully be applied to the plating of human fibroblastic stocks. Plating efficiencies in the neighborhood of 50 to 60 per cent are readily achieved with such strains. The resulting colonies can be picked and clonal stocks established. Fibroblastic morphology is maintained in the colonies arising from every single cell of such clonal stocks. All of the single cells from epithelioid clonal strains also maintain their integrity throughout repeated subculture. Since the difference between clonal stocks of these two types is always maintained whenever the respective single cells are plated in the same medium, regardless of the previous history of these stocks, it may be concluded that a true genetic difference exists in these cell lines. In addition to the morphological differences between epithelioid and fibroblastic cell strains, the latter have more demanding nutritional requirements for single cell growth. Thus, single cells of fibroblastic lines almost never produce colonies with high efficiency unless the growth medium which is sufficient for epithelioid cells is supplemented with embryo extract, or a cell feeder layer. Fibroblastic cells are also more resistant to tryptic digestion of the bond uniting the cells to glass surfaces. By use of differential media, growth of both fibroblastic and epithelioid cells, respectively, has been obtained, from dispersed single cells obtained by trypsinization of a specimen of human embryonic lung.

Tissue culture strains of cells from four different normal human tissues—liver, conjunctiva, kidney, and appendix—have been grown by the plating procedure previously developed for the HeLa strain of cervical carcinoma cells. This technique results in colony formation from isolated single cells, in a manner completely analogous to the plating of bacteria in semisolid nutrient media. Clonal cell strains have been isolated from each cell type. All behaved exactly alike in all properties studied except that some differences in plating efficiency were displayed in some of the growth media employed. The cells from normal human tissues resembled the HeLa S3 carcinomatous cell in the following properties:— ( a ) Single cells displayed a plating efficiency close to 100 per cent in an appropriate medium. ( b ) They all grew as an epithelial sheet on glass, the cells being closely packed and polygonal in shape. ( c ) They had mean generation times of 20 to 23 hours in the nutrient media employed, ( d ) The mitotic frequency was constant, and therefore the duration of mitosis was the same for all the strains studied, ( e ) The incidence of multinuclearity and giant formation was very low and similar in both types of cells. ( f ) Both classes of cells had the same total volume, and the same nuclear cross-sectional area. ( g ) Both also showed a tendency to spread more in the presence of human serum (concentration of 20 per cent or more) than in porcine serum. However, this differential morphological response was much more marked in the HeLa cell than in those from normal tissues. The only difference noted in the behavior of these two groups of cells lay in the tendency of the cells from normal tissues always to exhibit a greater cross-sectional area when spread on glass than the HeLa cell in the same medium. The frequency of occurrence of different types of multinuclearity in the HeLa cell and cells from normal tissues has been measured. The data suggest that multinuclearity depends on two factors: a necessary, predisposing state in the cell, and a random, independent event causing the appearance of an additional nucleus in such a prepared cell.

The effects of x-irradiation have been quantitatively studied on single cells of a human cervical carcinoma (HeLa) under conditions such that 100 per cent of the unirradiated cells reproduce in isolation to form macroscopic colonies. This technique eliminates complexities due to interactions of members of large cell populations. Survival of single cells (defined as the ability to form a macroscopic colony within 15 days) yields a typical 2 hit curve when plotted against x-ray dose. The initial shoulder extends to about 75 r, after which a linear logarithmic survival rate is obtained, in which the dose needed to reduce survivors to 37 per cent is 96 r. This radiation sensitivity is tens to hundreds of times greater than that of any microorganism for which the equivalent function bas been studied. Evidence, though not proof, is presented that the lethal effect is due to a radiation-induced genetic defect which, however, cannot be a simple single gene inactivation. The locus of the action could be chromosomal. Beginning at doses of 100 r, or possibly earlier, growth-delaying effects of radiation are visible. Cells in which the ability to reproduce has been destroyed by doses below 800 r, can still multiply several times. At higher doses even a single cell division is precluded. A large proportion of the cells killed by radiation at any dose gives rise to one or more giant cells. These metabolize actively, grow to huge proportions but never reproduce under the experimental conditions employed. Methods of preparing large populations of giant cells are described. These giants are particularly susceptible to virus action. Some of the irradiated cells disappear from the plate, presumably by disintegration. This action of radiation is by far the least efficient, since even after 10,000 r, 5 to 10 per cent of the original cell inoculum is recoverable as giants.

Two methods for simple and rapid plating of single HeLa cells, human, carcinomatous cells, are described. These result in growth and formation of colonies from each single cell. One of these procedures uses irradiated, non-multiplying "feeder" cells to condition the medium. The second requires more gentle handling of the cells, but otherwise is virtually the same as that used in plating bacteria on semisolid, nutrient media. By extension of these methods, it is possible to isolate single mutant colonies and grow pure clonal stocks of animal cells. These genetically uniform strains are much more homogeneous in their behavior than the parental HeLa cell population. Growth curves obtained from developing colonies are highly reproducible. The most active mutant stocks so far isolated display a generation time of 18 to 20 hours. In pooled human serum HeLa cells assume a highly stretched, ameboid form, with marked motility; whereas growth of the same cells in a variety of non-human sera results in tightly packed, columnar, epithelial-like morphology. The two cell types possess volumes, nuclear cross-sections, plating efficiencies, and generation times which are identical within experimental error, but display widely different cross-sectional areas, suggesting that the basic change occurs in the cell surface. It is conceivable that this change may be related to that which enables the cells of a compact tumor to become invasive. Animal cells subjected to the standard trypsinization procedures which involve mechanical trauma and repeated washings in incomplete media leak large amounts of P and suffer impaired ability to reproduce as isolated cells. Application of the methods described in this paper as a tool for quantitative study of normal mammalian cell growth, physiology, genetics, and biochemistry, and the response of cells to drugs, viruses, high energy radiation, and other agents have been indicated.

The virus-induced leakage of host-cell constituents represents a true increase in cellular permeability rather than an unpeeling of cell surface components, since an intracellular enzyme participates in the leakage. All of the T-system bacteriophages exhibit this leakage. The leakage does not occur with salt concentrations which permit only reversible virus-cell attachment but no penetration. These facts support the idea that the reaction underlying cell leakage is a part of the invasive mechanism. With increasing multiplicity of T2 infection of young, fresh Escherichia coli B cells, progressively larger molecules leak out of the cell. Acid-soluble P 32 appears in large amounts with single infection. Appreciable amounts of galactosidase enzyme and RNA do not leak until multiplicities of 5 to 30 are attained. Cellular DNA is not liberated unless sufficiently high multiplicities are used to cause the extensive cell destruction and clearing of the suspension characteristic of lysis-from-without. This progression is interpreted as an increase with T2 multiplicity in the maximum hole size produced in the cell membrane. Calculation shows that this increase in hole size must result from a spreading change in the character of the cell wall, rather than the coincidental juxtaposition of 2 or more viruses at adjacent attachment sites. T1 virus liberates less macromolecular constituents than T2 from E. coli B. The following experimental results constitute evidence that in the course of normal virus infection, a resealing reaction is rapidly instituted in the cell wall which reverses the effect of the original permeability increase, and renders the cell refractory to a second lytic reaction by a homologous virus: ( a ) Cell leakage induced by T2 virus in the course of normal infection markedly slows down or stops within a few minutes, even when only a small fraction of the material potentially available for leakage has been released, ( b ) Superinfection after 8 minutes at 37°C., of a cell previously infected with a homologous virus causes little or no appearance of a second leakage of cell constituents. This experiment also leads to the conclusion that the sealing reaction, like that which causes the leakage, also involves a disturbance which spreads over all or most of the cell wall. ( c ) If a multiple virus infection is allowed to occur at 0°C. and then the cells are placed in a 37°C. bath after completion of attachment, a much greater cell leakage results than if the entire course had occurred at 37°C., as would be expected if a resealing reaction comes into play at 37°C. within a time less than that required by the completion of attachment. The virus particles attaching secondarily at 37°C. are prevented from exercising their permeability-increasing effect by the sealing reaction of the virus which had penetrated first. Although a second homologous cell infection with T1 or T2 phages after a 37°C. incubation fails to yield a second leakage, a second heterologous infection always causes exacerbation of new leakage, which, especially if T1 has preceded T2, may be much greater than the sum of those produced individually by each virus in separate cell suspensions. This phenomenon may be the action responsible for the "depressor" effect which occurs when 2 unrelated viruses attack the same cell. The properties of the sealing phenomenon are such as to make it appear a logical candidate for the mechanism underlying the exclusion of a superinfecting phage from participating in reproductive processes in a cell previously infected with a homologous virus, since the DNA of the second virus would be unable to penetrate the new barrier. Experiments to test this hypothesis revealed that the DNA from such superinfecting virus is completely extractable from cells by washing in dilute buffer, whereas about 40 to 50 per cent of the attached DNA of virus which has invaded virgin cells remains bound to the cells. Most of the viral DNA which appears in the original supernatant when P 32 -labelled T2 invades E. coli B in a multiplicity less than one, does not represent inert material but rather virus DNA which has been split, or split and hydrolyzed as a result of its interaction with the cells, as judged by the altered susceptibility to hydrolytic enzyme or to TCA precipitation. This suggests that 25 per cent or more of the virus DNA may be expendable, at least after the penetration stage of the infection cycle. Mg ++ which strongly depresses the amount of cell leakage attending T2 infection, does not prevent T2 penetration nor does it block the appearance of the exclusion reaction. Hence, if the initial leakage does mirror the lytic process by which a hole for the DNA injection is provided, the Mg ++ does not function by preventing this hole formation. Its effect would have to lie in prevention of the spreading lysis-potentiating reaction or in augmenting the sealing mechanism. A large number of independent lines of evidence indicate that the phenomenon of lysis-from-without exhibited by the T-even coliphages is the result of failure of the sealing mechanism to keep pace with the lytic reaction. This can result from an excess of infecting phages or inhibition of the cellular energy-liberating reaction required by the sealing mechanism. The complete parallelism between the development of refractoriness to lysis-from-without and development of refractoriness to the production of a new leakage from a homologous superinfection is especially convincing in this connection. It is proposed that the early phase of bacteriophage invasion involves the following steps: reversible electrostatic attachment; splitting of the viral DNA from its protein coat; initiation of a lytic reaction in the cell wall at the site of virus attachment; injection of the DNA through the hole so produced; a spreading disturbance over the cell surface which makes it momentarily more susceptible to the lytic reaction; sealing of the hole and a concommittant spread over the cell wall of a reaction making the cell refractory to initiation of a second lytic reaction. Na + , K + , and Mg ++ all behave differently in their effect on the leakage produced in the course of T2 invasion of E. coli .

Treatment of radioactively labelled host cells with T1 or T2 bacteriophages induces a leakage of cellular P and S into the medium. Evidence is presented showing that this increased cell permeability is not the result of complete lysis of a small fraction of the cells, but rather is made up of contributions from all or most of the infected population. This leakage of cellular constituents exhibits the following characteristics: ( a ) Infection of a cell with a single virus suffices to evoke the reaction; ( b ) Increasing the multiplicity up to 7 to 8 virus particles per cell does not affect the extent of leakage produced; ( c ) Some leakage does occur at 0°C., but much less than at 37°C.; ( d ) Infection by T1 virus results in a smaller amount of leakage than in the case of T2, but the pattern of response to varying virus multiplicity is the same; ( e ) The P resulting from such leakage contains no DNA and chemically resembles that which elutes in smaller amounts from uninfected cells; ( f ) At 37°C. the virus-induced leakage reaction appears within a matter of seconds, and usually decreases after 2 to 3 minutes; ( g ) The reaction is inhibited by 0.025 M Mg ++ . Theoretical considerations are presented suggesting the place of this reaction in the sequence of events constituting the virus penetration reaction; its relationship to the phenomenon of lysis-from-without; and its resemblance to the leakage reaction produced by electrostatic binding of ionized compounds to cell surfaces. The existence of similar effects in avian-mammalian virus systems is noted.

A rapid and precise method is described for determination of absolute numbers of hemagglutinating particles present in virus suspensions. The method involves spectrophotometric measurement of the number of red cells dimerized by the virus under conditions in which formation of larger aggregates is precluded. Results of this procedure are compared with those of titrations based on egg infectivity and the polystyrene microsphere electron micrography for influenza virus; and with plaque count titrations on chick embryo cell monolayers in the case of the virus of Newcastle disease. The agreement is well within the limits expected in view of the different quantities which the various procedures measure. The conversion factor for transforming titre in hemagglutinating units based on the pattern procedure into absolute number of agglutinating particles is the same for influenza and Newcastle disease virus. The precipitation of red blood cells by Newcastle disease virus at 23°C. is optimal at pH's between 6.2 and 5.8.

At 37°C., the attachment of T1 virus to its host cell in solution containing 10 –3 M CaCl 2 or 10 –2 M NaCl is extremely rapid (in the neighborhood of 100 per cent collision efficiency) and irreversible. At 1°C., the attachment rate is almost equally rapid but largely reversible. If a suboptimal concentration of the necessary ions is employed when T2 virus attaches to host cells, the resulting binding is largely reversible, even at 37°C. Reversible T2 attachment to host cells leaves the cell undamaged and capable of normal reproduction. Irreversible attachment results in death of the cell. Zn ++ exercises a specific inhibitory action on the invasion of E. coli B by T1 virus. The virus can still attach to the host cell at a rate closely approximating the maximum value, but the reaction remains reversible and the cell is protected against permanent damage. The protective action of the Zn against T1 invasion is exerted through an action on the cell, rather than on the virus. Studies of the uptake of radioactive Zn 65 show that cells become completely immune to T1 invasion when, on the average, 4 x 10 7 atoms of Zn have been taken up by each cell. Cells killed by ultraviolet irradiation still bind T1 at the maximum rate, but the reaction is reversible even when taking place at 37°C., in optimum salt concentration. The tryptophane-deficient mutant of T4 bacteriophage requires its specific cofactor for the initial step of attachment to the host cell. These experiments support the picture previously developed that virus invasion of host cells consists in an initial, reversible attachment whose properties are those to be expected from the operation of electrostatic binding forces. The step is followed by an enzymatic transformation which is irreversible, strongly temperature-dependent, and in the case of T1 virus, susceptible to inactivation by ultraviolet radiation. The resistance of mutant cells to specific bacteriophages is of two types, depending on whether the first or second of these steps is blocked.

T1 virus does not attach to its host cell, E. coli B, in distilled water. By the proper addition of salts the rate of attachment can be adjusted to any desired value up to the maximum limit set by the diffusion rate of the virus. Salts of Ca ++ , Mg ++ , Ba ++ , and Mn ++ bring about a reaction rate representing 100 per cent collision efficiency in a concentration of 5 x 10 –4 M . Both greater and smaller concentrations depress the attachment velocity. Salts of Na + , K + , NH 4 + , and Li + display a similar pattern but require a tenfold greater concentration than that of the previous group to produce the same effect. Moreover, the maximum velocity attainable in solutions containing only monovalent cations is only half that achieved by the divalent salts. The trivalent cations Al 3+ , Cr 3+ , Fe 3+ permanently inactivate the virus. Activation by Mg ++ of an inert mixture of virus and host cells in distilled water is so rapid as to be beyond the limit of the resolving time of the experimental procedure, which is 20 seconds. The temperature dependence curve of virus-cell adsorption exhibits a maximum at 37°C. and falls to a value representing approximately 3 per cent collision efficiency at 1°C. Identical curves are obtained in nutrient broth and in 10 - M MgCl 2 solution. Bacteriophage can be quantitatively adsorbed on to glass filters. In a study of several viruses this attachment reaction was found to require the same cofactors—both organic, like l tryptophane, as well as inorganic—which each specific virus required for its attachment to its host cell. The suggestion is made that the attachment of viruses to these filters is a useful model for their attachment to host cells. Virus attachment to glass filters is reversible. Such adsorbed virus can be recovered almost quantitatively by washing the filter with a solution in which the attachment reaction does not occur. Virus attachment to host cells is similarly reversible at least in its primary step. Distilled water at 0°C. can produce almost complete liberation of T2 virus from host cells infected in 0.02 M NaCl solution. Two significant differences between the behavior of glass filters and host cells toward T1 virus are: ( a ) an excess ion concentration fails to inhibit virus attachment to the glass as it does to the host cell; and ( b ) no decrease in efficiency of attachment to glass occurs at low temperatures. These facts suggest that the inhibiting action on the infective process of excess cations and low temperatures involves chemical groupings on the cell surface, rather than on the virus. There is no detectable attachment whatever of T1 virus to E. coli cells specifically resistant to it, though still susceptible to other viruses. This experiment indicates that the ion-controlled attachment forces here considered are involved in the host-virus specificity. This conclusion is strengthened by the fact that several different viruses with different host-cell specificities have different ionic requirements for cell attachment. All of these observations lend themselves to explanation by a mechanism which pictures an initial addition reaction of cations to specific sites on the surface of the virus in particular, and possibly also of the host cell. Two complementary electrostatic configurations are so produced which can unite in a reaction with a high biological specificity, which yet exhibits 100 per cent collision efficiency. An excess of ions may cover up some of the attachment sites and so inhibit the reaction. By this picture the specificity of virus-cell invasion depends upon the binding energies of sites on both bodies for various ions, and the distribution of these sites over the two surfaces. Possible relationships of such a process to other biological systems are discussed.

The viability of pneumococcus, Type I, sprayed into the atmosphere from a liquid suspension was measured as a function of the relative humidity. When broth, saliva, or 0.5 per cent saline solution is employed as the suspending medium, a very high mortality rate is observed at relative humidities in the vicinity of 50 per cent. However, at humidities above or below this value the microorganisms survive for long periods. Measurement of the rate of settling of droplets employed in these experiments demonstrated that the disappearance of microorganisms from the air is a true lethal process, rather than a manifestation of aerosol collision processes. When a saline-free fluid was used, the sharp peak in death rate at intermediate relative humidities disappears. The lethal effect of intermediate relative humidities on pneumococci atomized from a saline-containing suspension is increased when the particle size of the atomized droplets is increased or when the temperature is raised. Cultures of hemolytic streptococcus group C and staphylococcus sprayed from a broth medium exhibit the same general survival pattern as a function of relative humidity although the mortality rates are smaller than that of the pneumococcus. These effects can be explained by assuming the existence of a critical degree of cellular dehydration at which microorganisms become much more sensitive to toxic agents than in states where either more or less water is bound to the cell. The results presented here may be significant in elucidating certain aspects of the epidemiology of air-borne infections.

Theoretical analysis of the mechanism of action of chemical aerial disinfectants reveals that the rapid killing action which is obtained cannot be accounted for by a collision process between germicidal aerosol particles and the air-borne bacteria. However, a mechanism involving condensation of germicide molecules in the vapor state on to the bacteria-containing droplets results in a theoretical velocity of the correct order of magnitude. Experimental tests of this theory show that pure germicide vapors free of aerosol droplets are almost instantly lethal to air-borne bacteria. Conversely, pure germicidal aerosols in the absence of vapor, had no effect on air-borne bacteria within 20 minutes or more. Therefore, it may be concluded on both theoretical and experimental grounds that rapid air sterilization requires the existence of the germicide in the vapor state.

The effectiveness of any compound as an aerial germicide depends upon the extent of condensation of its vapor on air-suspended bacteria, and on the rate at which the resulting concentration of germicide can produce death of the microorganisms. The properties of any compound conducive to production of the highest rate of kill of air-borne microorganisms by means of the smallest possible concentration of germicide vapor, are as folows: (a) a low vapor pressure, but not lower than 0.001 mm. Hg at 25°C.; (b) high hygroscopicity; (c) toxicity for bacterial metabolism—a high degree of potency is not necessary although the killing action will be more efficient the higher the antibacterial activity of the compound employed. For any compound the killing action is always a direct function of the concentration of its vapor in the air. The maximum amount of a hygroscopic substance which can exist in the vapor state decreases as the relative humidity increases. Hence, at high relative humidities the bactericidal efficiency is lowest. At lower relative humidities the air can contain more vapor, and hence a greater effect is possible. At any relative humidity, the killing action is greater, the more closely the germicide vapor concentration approaches the saturation point. The presence of soluble compounds in droplets containing bacteria promotes more extensive condensation of the germicide than would otherwise occur, and so enhances its effectiveness. In the absence of such soluble substances, low atmospheric humidities may cause complete desiccation of a bacterial particle. Under these conditions its surface may become resistant to the condensation of the vapor, and thus prevent effective germicidal action. The influence of temperature changes on the killing efficiency maybe correctly deduced from a consideration of the effect of a rise in temperature on the vapor pressure of the germicide and on the rate of its bactericidal action in vitro . Equations are presented for estimating quantitatively the magnitude of some of the effects discussed.

This communication describes the construction and operation of two identical experimental rooms in which it is possible to produce and maintain a wide range of temperature and humidity with or without exchange of the room air. The ability to maintain a large air mass under constant conditions makes it possible to study the effects of different atmospheric states on air-suspended bacteria and viruses in relation to their survival, particle size, humidification, killing by lethal vapors, and host susceptibility. A brief description of the functioning of the rooms under experimental conditions is given.

A device for automatically maintaining any desired concentration of triethylene or propylene glycol vapor in air is described. It can also be used to measure very rapidly the glycol content of room air. This instrument should greatly facilitate tests of the efficacy of glycol vapors in preventing air-borne infections.

A study of the conditions which affect the bactericidal action of propylene glycol vapor on air-suspended microorganisms has been carried out. The killing process was found to be more effective when both the total number of air-borne droplets and the number of organisms in the bacterial suspension are small. A temperature below 80°F. and an atmospheric relative humidity between 45 and 70 per cent were found to constitute the most favorable conditions for the lethal action of the vapor. Experiments were performed to test the bactericidal efficiency of propylene glycol vapor in both small and large enclosed spaces. These studies revealed that equally marked bactericidal action is obtained when propylene glycol is dispersed in an 800 cubic foot room as occurs in chambers of 2 cubic foot capacity. The susceptibility to vapor action of bacteria re-suspended in saliva was just as great as when broth was used as the suspending medium. Both partially and completely dehydrated bacteria also succumbed to the effects of the vapor. However, when unsterile dust collected from inhabited rooms was dispersed into the air, little reduction of the natural microbic population contained in this material was observed. Data are presented showing the minimum glycol concentration necessary for effective bactericidal action on various microorganisms. Pneumococci were killed by amounts of propylene glycol as low as 1 gm. in 20 million cc. of air. Concentrations of 1 to 5 million to 1 to 10 million were required to produce the same degree of killing of streptococci and staphylococci. The observations here reported add further support to the previously proposed conception of the mechanism of the lethal action of propylene glycol vapor, namely, that a bactericidal concentration of the glycol accumulates in the bacterial droplet as a result of contact with and absorption of glycol molecules from the surrounding atmosphere.