smear, as observed, was to be expected. In the case of Bacillus cereus gentian violet is more strongly retained at a pH of 6, which is distinctly on the alkaline side of its isoelectric range.

In differential staining methods it is obvious that the dominant factor in the results (and ideally the only factor in a well designed general method) should be the differential retaining power of the organism for the stain. From the chemical point of view this amounts to saying it should be the differential stability of the dyebacterial compound. In this work, it should be noted, all of the smears prepared for decolorization were prepared under conditions such that the dye-bacterial compound was, from the point of view of ideal differential staining methods, stable. The ideal decolorizer should, therefore, show this to be the case, but should likewise effectively decolorize smears in which the dye-bacterial compound is unstable. Most of the decolorizers used actually decolorized one or the other stain, and among the common substances tried acetone seemed to come nearer fulfilling the requisites of an ideal decolorizer than any other liquid or solution.

There may be occasions where there is special reason for choosing a particular decolorizer which might not be acceptable for general use and it is hoped that the data here given may furnish a basis for an intelligent choice in such a case. For example, if one had the specific problem of getting sharp differentiation between a strongly Gram-positive and a weakly Gram-negative organism, a slightly basic decolorizer might yield very striking results; or for differentiating between a weakly Gram-positive and a strongly Gram-negative organism a weakly acidic decolorizer might be preferred. For general work, however, a non-polar liquid is most generally satisfactory, and of those tried acetone is by far the best. While the enolic form of acetone is no doubt acidic in character, as evidenced by the liberation of hydrogen upon treatment with sodium, this form does not seem to be present in the ordinary pure liquid to an appreciable extent, and we have in it probably a substance which is neither sensibly acidic nor basic. Thus two of the three factors mentioned at the beginning of the section tend to be eliminated by its use. It is also a decolorizer in which most dyes are sufficiently soluble for efficient removal. That is not true of either ether or carbon tetrachloride, in which liquids both gentian violet and acid fuchsin are only extremely slowly soluble if they are dissolved at all. In agreement with the general experience of the present authors, Stitt11 makes the statement that, as a decolorizer, "acetone is preferred by many due to the fact that it is more rapid in its decolorizing action on Gram-negative organisms, yet slower to decolorize positive organisms”.

IV.

GENERAL DISCUSSION OF THE GRAM REACTION.

Throughout the large number of suggested modifications in technique for Gram staining a few conditions have always, whether intentionally or not, been kept constant. The first and most important is the character of the primary stain. Many individual dyes have been suggested, for example pararosaniline and the basic rosaniline dyes in general, pyronine, rhodamin and safranine. All the dyes which give satisfaction, all in the above list, are basic in character. That these are lipoidsoluble has been deemed important also.

Another condition invariably adhered to is the use of a mild oxidizing agent as mordant. This was noted and discussed in a preceding section.

Still another observation upon which all workers are agreed is that of the decolorizing power of acids for the primary stain. Most papers do not differentiate between the action of acids on tissues stained with basic dyes and those stained with acid dyes. The present authors have shown that tissues stained with acid dyes are less easily decolorized in the presence of acids than in neutral solution, but, as was stated above, the primary stains used in the proposed methods are all basic, so that the observation that acids have a decolorizing effect is quite general. In the same way workers are agreed that the addition of mild alkali such as sodium bicarbonate will render results much more clear cut.

This may be summed up by stating that, so far as technique is concerned, good differential results can be obtained by employing a deep colored basic primary stain, mordanted by a mild oxidizing agent in slightly alkaline solution.

In a previous section a physical theory of the mechanism of the Gram reaction was outlined, and certain general facts pointed out which tended to indicate the questionableness of its validity. The experimental results presented in this chapter increase the evidence against such a theory, and this should be pointed out by way of summary before giving a general consideration to the actual chemical differences upon which Gram differentiation is based.

These additional lines of evidence are three in number.

1. The physical theory says that the retention of primary stain is due to the formation of a dye-mordant precipitate whose particles are sufficiently large to be unable to escape through the pores of a postulated cell membrane. This means that an unmordanted smear should not be able to retain the stain, a prediction which has been repeatedly shown to be contrary to experimental findings for organisms which are Gram-positive to a sufficient degree. That is, it has been shown that the mordant is a deciding factor in only a limited number of cases which were spoken of as "border-line" organisms.

2. The truth of the physical theory as an explanation of the mechanism of the reaction should also require that any substance which has the power of precipitating the dye should act as a mordant and any substance which cannot precipitate the dye should not show mordanting properties. Both these predictions have been shown not to hold. Stannous chloride, which forms a precipitate with gentian violet, was shown to affect a preparation, which otherwise retained the stain in the presence of acetone, in such a manner that it was readily decolorized by acetone; whereas certain oxidizing agents which do not precipitate gentian violet have been shown to act in a manner similar to Lugol's iodine though to a different degree107.

3. Finally, if the reaction is one of the cell membrane exclusively, the Gram character of an organism should not be altered by a process which leaves the cell wall intact. This has been definitely shown above to be contrary to experimental fact for certain organisms.

The authors are unable to see any valid evidence supporting this theory, and their main interest has been not so much in the question whether there is a chemical differentiation at the basis of the Gram classification, since this much seemed almost an inevitable conclusion, as in the question, In what does this chemical difference consist?

There seem to be three essential differences between what are usually classified

as Gram-positive and what are usually classified as Gram-negative organisms. All of those properties show a gradation, or a gradual variation in magnitude from organism to organism. There is no abrupt jump from one class of organisms to another, but taken together, the three properties in which the different organisms show chemical differences offer a practically useful basis for classification except in a few cases.

Two of these properties have been already named and briefly discussed. It has been shown that Gram-positive organisms have, in general, an isoelectric point at a lower pH value than do Gram-negative organisms, and thus at a particular pH value would under any circumstances tend to retain basic dye more strongly. At present the authors consider this the most fundamental difference, though the other two to be mentioned are important, and it may be found that one of them is even more fundamental than the inherent isoelectric point.

It has also been shown that those organisms commonly classified as Grampositive are affected by the mordanting process to a degree significantly greater than the Gram-negative organisms.

Another very significant difference between the various organisms may be seen by reference to Figures III and IV as well as to analogous ones worked out previously.109 In the graphs here presented, a vertical broken line is in all cases dropped from the intersection of the gentian violet and acid fuchsin curves to the pH axis. It will be noted that the length of this line varies greatly from organism to organism, but that it is invariably longer in case of Gram-positive organisms than in case of Gram-negative ones. While no quantitative argument can be based on the relative length of these lines, their qualitatively comparative length may be of great importance as offering an explanation of another significant difference between Gram-positive and Gram-negative organisms.

To get a qualitative idea of these comparative lengths on an arbitrary scale for purposes of picturing their significance more easily, rough values for certain organisms are collected in Table VIII. Here again the organisms are grouped into three classes as denominated.

In the discussion of the isoelectric point of a complex system of more than one ampholyte it was pointed out above that the isoelectric point does not necessarily represent a condition in which no combination with anions or cations takes place, as is postulated for the isoelectric point of a single ampholyte, but rather repre

sents a condition in which combination takes place to an equal extent with both negative and positive ions.

In the simple case of a mixture of two ampholytes, it will be remembered that the chemical isoelectric condition was given by the equation

(AOH−) = (HB+)

(a) that is, the anion concentration of the more acidic component is equal to the cation concentration of the more basic component. Under such conditions one may expect the reaction

to proceed to a certain extent. It may, in fact, proceed nearly to completion, leaving only a few free ions which will be in a condition to combine with added anions or cations, or, on the other hand, it may proceed only a small fraction of the way to completion, leaving a larger number of the free ions. However far it proceeds, the concentration of the HB+ ion will in all cases be equal to that of the AOH-ion.

If we apply this same type of consideration to the bacterial system we shall see that even at the isoelectric point certain organisms will be able to retain basic dye ions to a greater extent than others. That is, there are more free AOH- ions which are available to combine with positive basic dye ions. Since the ordinates of the graphs in Figure II measure the intensity of retained color after standard treatment they are a measure of the extent of combination with the dye. The lengths, therefore, of these broken lines dropped from the isoelectric points to the pH axis are very rough measures of the extent of combination with dye at the isoelectric point.

The Gram-positive organisms seem to combine at this point much more than the Gram-negative ones, as is brought out by the values given in Table VIII. Thus, at its isoelectric point, B. aertrycke is almost completely decolorized while B. subtilis retains both dyes to a considerable degree. This indicates directly, of course, only that there are a greater number of ions present in the one than in the other, but the comparative results are in line with the comparative isoelectric ranges. The Grampositive organisms are inherently more acidic in character than are the Gramnegative ones, so that, even for the same fraction of mutual combination at the isoelectric point, according to equation (b) above, there may be expected to be more ions left free to retain added cations and anions by combination.

The Gram-positive organism does therefore seem to have certain physicochemical differences from the Gram-negative. And the Gram reaction makes use of all of these. It is not a single part of the cell which reacts, as for example the wall or the cell content from the center, but it is the entire bacterial system, as is brought out in the graphs of Figure IV.

I.

CHAPTER IV

EXPERIMENTS PERFORMED LARGELY ON LIVING ORGANISMS A. BACTERIOSTASIS

SIGNIFICANCE OF THE TERM.

The term bacteriostasis was introduced by Churchman in 191219 specifically to describe the effect of dyes on bacteria. This effect is not necessarily bactericidal, but is primarily a general inhibition of vital activity. According to Churchman2o, "the dye may only stop motility, it may check sporulation, it may prevent fission, it may produce a curious condition of the bacteria which can only be described as suspended animation, a condition in which they lie dormant for long periods of time, finally becoming active and virulent again; or it may cause the actual death of the organism".

The term bacteriostasis, when referred to dyes, then, seems to represent a condition of equilibrated chemical combination of the bacterium with dye. Unless the bacterial system is too badly shaken up we might expect any factor which would tend to shift this equilibrium in such a way as to remove the dye to again liberate a virile and viable organism. Engelhardt, working with mercury in place of dyes, has demonstrated specifically that such is the case. He showed that staphylococci, which had been treated for 72 hours in a 1% mercury bichloride solution, would, after removal of the mercuric ion by precipitation as sulfide, grow. Similarly apropos in this connection are the results of Churchman21, obtained on the injection of stained Bacillus anthracis. This organism was found to remain apparently innocuous for periods from ten to twenty times longer than required by unstained organisms to produce death, but finally they suddenly revived with fatal results.

In bacteriostatic studies in vitro the time element seems important in deciding the effectiveness of an agent. In the present study results are ordinarily not given for incubation periods of more than 72 hours. Failure of an organism to grow at the end of this period does not by any means indicate that it has been killed. This fact is shown by certain results presented in Table IX in which comparison is made between growth results for two strains of Bacillus subtilis after 72 hours' and after 19 days' incubation respectively. Inoculation was made into buffered nutrient broth containing varying amounts of gentian violet. Analogous results are included for a fungus which is very sensitive to gentian violet.

The point of view put forth here, then, is that the same fundamental chemical properties of bacteria which govern their staining behavior also govern their bacteriostatic behavior toward dyes.

II. THE PH EFFECT.

Probably the most obvious question which will arise is whether a basic dye, which is more effective as a staining agent at a high pH value than at a lower one, is also more effective as a bacteriostat under the same conditions. Tables X, XI and XII give data bearing on this point. They are representative results from a large amount of similar data at hand, and they are given for three strongly Grampositive organisms (Table X), for one strongly Gram-negative organism (Table XI),