We are submitting this series of experiments as observed facts, realizing that there are so many uncertainties in this form of indirect observation that great caution must be exercised in drawing conclusions of any kind. The most serious of the possible errors involved is that the active substances which we have studied, the enzyme—the bacteriophage—and the several varieties of virus, may not be free in our suspensions, but are adsorbed to larger particles. The peculiar difficulties encountered in filtering herpes virus particularly suggest a source of error of this kind, and if we are right in assuming the intracellular position of this virus in the nervous tissue, it is more than likely that most of the virus obtained in suspension may be closely associated with protein particles derived from the cells. Keeping all this in mind, we may, nevertheless, derive a certain amount of information from our experiments as follows:

1. The order of magnitudes of the pure proteins with which we have worked,—namely, crystallized egg albumen, crystallized serum albumen and purified casein,—follows the order of molecular weights of these substances as determined by Cohn. As far as casein is concerned, the size indicated by filtration in comparison with collargol is far greater than it should be by calculations which take a molecular weight of 192,000 as the point of departure. While one cannot be sure of the reason for this, there are many possible explanations such as considerable swelling of the casein partides, aggregation of molecules and the fact that casein is not at its isoelectric point under the conditions of filtration and surely present as a salt.

2. Trypsin, even in the certainly very impure condition in which we employed it, is but very slightly larger than serum albumen and distinctly smaller than casein. In its pure form it may well be much smaller even than our filtrations indicate, but certainly not larger. This relatively small size of trypsin may have considerable bearing upon the question of whether or not the lytic agents spoken of as "bacteriophage" are substances of the nature of enzymes, or whether they are more comparable to the filtrable virus, as supposed by d'Hérelle.

3. Herpes virus, the Rous chicken sarcoma and a staphylococcus bacteriophage were all subjected to filtration at pH 7.2 and at hydrogen ion concentrations higher than 8, which is given by Olitsky and Boëz as the isoelectric point of foot-and-mouth disease, but failed to pass membranes which, at the same pressures, were permeable for casein and collargol. The bacteriophage and the Rous sarcoma with considerable regularity passed through membranes which held back colloidal arsenic trisulfide. We have cited only a few of the experiments which were actually done, every one of the tests tabulated being merely representative of a number of others that were omitted for economy of space. The herpes virus we have had greater difficulty in filtering. We cite one experiment with a 2 per cent acetic-collodion membrane and another with a 1.5 per cent membrane through which the herpes virus passed, the membrane being so controlled that gross leakage could be excluded. We believe that the difficulty here is very largely due to the fact that in preparing the herpes virus for filtration it cannot be separated from considerable amounts of brain material, from which, perhaps, it is not easily dissociated. This would be natural if the herpes virus were intracellularly located, as we believe it to be. This experiment and similar ones, however, incline us to believe that the herpes virus is not far different from the Rous sarcoma virus and the bacteriophage, as far as filtration through membranes is concerned. It certainly is not smaller than either of these substances and probably, as we judge from a few experiments carried out at higher pressures, is not much larger.

It may be assumed, therefore, that in the form in which we were able to procure the bacteriophage and the two varieties of virus investigated by us, they were of a magnitude larger than casein and collargol and smaller than colloidal arsenic. The weak point in drawing our conclusions is the fact that we were not in a position to measure for ourselves with any accuracy the actual sizes of collargol and arsenic trisulfide particles. Accepting the general views of Bechhold and others, however, our experiments would define the sizes of the separticular substances as larger than 20 mµ and probably smaller than 100 mµ.

The order of magnitudes of the substances measured by us would then be as follows:

Crystallized egg albumen

Crystallized serum albumen




Bacteriophage, Rous sarcoma virus, herpes virus

Arsenic trisulfide

Our experiments show little agreement with the work of Levaditi and Nicolau and of Levaditi, Nicolau and Galloway. In their recent filtration tests of foot-and-mouth disease this virus is reported by them as passing through membranes that held back trypsin, indicating a size much smaller than any of the viruses measured by us. Our results, on the other hand, are in actual measurements comparable to those of Olitsky and Boëz, not only in the fact that the viruses with which we worked correspond approximately to the size determined by them for foot-and-mouth disease, but that the percentage of collodion in membranes permeable for virus and impermeable for colloidal arsenic corresponds almost exactly to our own. This gives us confidence that the technique developed may be more easily standardized than we at first believed and that the method of ultrafiltration, owing to the great ease with which membranes of relatively standard size may be made, may have valuable applications in the investigation of bacteriological and immunological problems.

This content is only available as a PDF.