Preparation of antigen.-Cultures for the preparation of antigen were grown on agar in Blake bottles. The addition of 1 per cent glucose greatly enhanced the vigor of growth. The best growth, however, was obtained on agar, of a hydrogen ion concentration of pH about 6.8, prepared with liver in place of the meat of ordinary infusion agar. The cultures were incubated at 37° C. for two days. The growth from each bottle was washed from agar in about 40 c. c. of normal saline solution. In the latter part of the work here reported the suspension was then heated to 60° C. for 30 minutes. The absorbing capacity of the antigen was slightly injured by the heat. The common experience in this and other laboratories, however, in finding Brucella melitensis peculiarly infectious to laboratory workers, made it inadvisable to handle large quantities of living antigen. The suspensions were centrifugalized, the supernatant fluid was removed, the sediment was taken up in normal saline solution, and the density was adjusted to 20,000 parts per million of the silica standard of the American Public Health Association. This antigen kept without deterioration for weeks in an ice box at about 4° C. For the simple agglutination test, the stock antigen was diluted to 1,000 p. p. m. with buffered saline solution (described by the writer in an earlier publication) of a hydrogen ion concentration of pH 7.0. An antigen of double, triple, or quadruple density, for absorption tests, was prepared from the 20,000 p. p. m. antigen by centrifugalizing and removing enough of the clear supernatant fluid to obtain the desired density. Preparation of serums.-Rabbits were used for the preparation of A titer of 1:640 was found to be the most convenient for the absorption tests. Intravenous injection with 2 c. c. of a living antigen of a density of 2,000 p. p. m. sometimes produced a serum of the desired titer after 7 days. More frequently it produced an antigen of too high titer. In that case another serum of low titer was produced by inoculating a rabbit with 2 c. c, of living antigen of a density of 1,000 p. p. m. and drawing the blood on the fourth day. The two serums were then pooled in the proper proportion to give a titer of 1:640. The serums were designated by the number of the strain used for their preparation.

serums.

1

Absorption of agglutinin tests.—It was found that a living antigen of a density of 60,000 p. p. m. will absorb all agglutinins from its homologous rabbit serum of a titer of 1:640 when the absorption is carried out in a 1:5 dilution of the serum. (After the addition of the serum the actual density of the antigen is 48,000 p. p. m.) If a heat-killed antigen is used, the absorption is not always complete under those conditions. It is therefore necessary to compare the

It has been found that after intravenous inoculation with a heavy dose of living Br. melitensis a rabbit is a dangerous disseminator of infection. In further work that is contemplated a trial will be made of killed culture for agglutinogen.

absorbing capacity of an unknown strain with the absorbing capacity of the homologous strain with identical treatment in the preparation of the antigens and with identical absorption technique.

The protocol for a typical absorption experiment is given in Table II. In the case of each test, 6 c. c. of antigen of a density of 20,000 p. p. m. were placed in a centrifuge tube, the antigen was thrown down by centrifugation, and 4 c. c. of the supernatant fluid were removed. The sediment was emulsified in the remaining fluid, and 0.5 c. c. of serum was added. The tubes were incubated in a water bath at 37° C. for 4 hours, then removed to a cold room to be kept until the following day, when the antigen was again thrown down by centrifugation, and the simple agglutination test was carried out on the serums thus diluted and absorbed.

The simple agglutination test was with the homologous antigen.

4, complete sedimentation; 3, supernatant turbidity as in a control tube containing 25 per cent as much antigen as in the tubes in which the test was carried out; 2, supernatant turbidity as in a control tube containing 50 per cent of the antigen; 1, supernatant turbidity as in a control tube containing 75 per cent of the antigen. The table shows that the type of strain 466 is identical with that of strain 426, whereas strains 427, 480, and 489 belong to other serological types.

Simple agglutination test.-Serum dilutions were made with the buffered saline solution (pH 7.0). One-half c. c. of diluted serum and an equal quantity of antigen of a density of 1,000 p. p. m. were incubated together in a water bath at 37° C. for 4 hours. The racks were then removed to a cold room of a temperature of about 15° C., where they stood overnight, and readings were made on the following day. A reaction was considered positive only when sedimentation of 75 per cent or more of the antigen occurred.

SEROLOGICAL CLASSIFICATION OF STRAINS.

The classification of the strains was made according to the following principles:

1. Any strain which absorbs agglutinin from the test serum to the same degree as the homologous strain belongs to the same serological group. A strain which absorbs agglutinin from the test serum to a degree slightly different from the homologous strain may or may not belong to the same serological group; but a marked difference in absorbing capacity indicates a difference in serological grouping.

Equal absorbing capacities of two strains from a heterologous serum does not signify that the two strains belong to the same group.

2. Every strain belonging to the same group as the strain used in the preparation of a given serum will completely absorb the agglutinin from the 1:5 dilution of that serum as used in these tests if an absorbing antigen of sufficient density is used; and, vice versa, every strain belonging to another serological group will fail to completely absorb the agglutinin from a given serum under the same conditions. Through the kindness of Dr. K. F. Meyer in sending strains representing Feusier and Meyer's groups 1, 2, and 3, it was possible to correlate this study with theirs and to use their strains for the starting point in making this classification. (Strain 426 represents their group 1; strain 427 represents their group 2; and strain 428 represents their group 3.) Certain strains were found to differ from Feusier and Meyer's type strains in their absorbing capacities. Whenever a strain was found which failed to correspond with the serological types already established, a serum was prepared with the new type strain and the relationship between it and the other type strains was determined by agglutinin absorption. Altogether seven groups have been found. Some of the small groups are so closely related to the large groups, however, that they should be considered as subtypes. The distribution of the strains in the various groups, together with the animal species from which they were isolated, is given in Table III. The strains with which the serums were prepared for the classification are in heavy type at the head of the columns.

TABLE III.—The seven serological groups into which the strains fall.

The relationships between the various groups as shown by agglutinin absorption tests are given in Table IV. The same strains used for the preparation of the serums were used for absorbing antigens, except in one case. Strain 428, which was found to be identical with strain 451 in its absorbing capacity, was substituted for strain 451 in the absorption tests, because the latter, having been recently isolated from a human case, was considered more dangerous to handle in large quantities than strains which had been grown for a long time on agar. In every one of the absorption tests summarized in Table IV a serum of a titer of 1:640 was absorbed with an antigen of a density of 60,000 p. p. m. (48,000 p. p. m. after the addition of the serum) in a 1:5 dilution of the serum. Most of the absorption tests recorded in Table IV were carried out with a living antigen. Whenever a heated antigen was used, it is indicated in the table. All the absorptions from serum 480 were made with heated antigens. The protocol shows, however, that absorption by the homologous heated antigen was complete.

TABLE IV.—The relationship between the strains representing the various serological groups, as shown by the agglutinin absorption reactions.

Serum 480, diluted 1 to- Serum 426, diluted 1 to- Serum 457, diluted 1 to

10 20 40 80 160 320 640 1280 10 20 40 80 160 320 640 1280 10 20 40 80 1603206401280

10 20 40 80 160 320640 1280 10 20 40 80 160320 640 1280 10 20 40 80 160.320 640 1280

The absorptions were therefore made with antigens of a density sufficient to absorb the agglutinins completely from the homologous serums in the dilution used. Furthermore, each heterologous absorbing antigen was of a density sufficient to remove from this dilution of the serums all the agglutinins which that particular antigen could remove under the conditions of the experiment. In

Table V a protocol is given which demonstrates that a living antigen of 40,000 p. p. m. (32,000 p. p. m. after the addition of the serum) was sufficient for the desired purpose, for no more agglutinins were absorbed in any case when the density of the antigen was raised to 60,000 p. p. m. Antigen of a density of 60,000 p. p. m. was chosen for these experiments, however, in order that there might be no question that the absorption had been complete.

TABLE V.-Protocol showing that an antigen of 40,000 p. p. m. absorbed agglutinins from serum 104 as completely as possible for the given antigen.

• See footnote to Table II for the significance of the figures.

A graphic presentation of the relationship between the several serological groups is given in Chart 1. The length of the columns, calculated from the data in Table IV, represents the percentage of agglutinin for the homologous antigen remaining after absorption with an excess of the antigens representative of the several serological groups.

It may be noted that the serological groups represented by strains 426, 457, and 427 are very closely related, as judged by their absorption of agglutinins from serums 426 and 427; and no differences can be observed between these three strains when their absorption of agglutinins from serum 457 is considered. Table III shows that strain 426 represents a large group of strains, including a majority of those from bovine and porcine sources, and it also includes two strains of human origin. On the other hand, strains 457 and 427 represent small groups of one and two strains, respectively. They are so closely related to the group represented by strain 426 that they may be considered as subgroups of that large group.

The remaining groups diverge in two directions from the serological group represented by strain 426, with the group represented by strain 481 at one extreme and the group represented by strain 104 at the other extreme.