Since Gruber and Nicolle expressed the opinion that the agglutination of bacteria by specific immune sera was due to the surfaces of the bacteria becoming viscous under the influence of the specific sera, many theories have been advanced aiming at a more complete explanation of the reaction. Bordet (1899) considered that the agglutinins by uniting with the agglutinable substances lead to changes in molecular attraction between the elements affected, either as among themselves, or between them and the surrounding fluid. He distinguished the two phases of the reaction, (1) the interaction between the agglutinin and the bacteria, and (2) the agglomeration of the bacteria.

1. The chain of evidence incriminating the house fly as a disseminator of typhoid fever is at present fairly complete, but many of the links are weak and not thoroughly strengthened by experimentation.

2. The bulk of experimental work has hitherto been done under highly unnatural and artificial circumstances and the results so obtained cannot be accepted unreservedly as giving a correct view of conditions in nature.

3. The experiments described in the present paper show that flies can ingest typhoid bacilli from natural matter, i.e. human faeces and urine, and carry them for a certain period of time.

4. There is no evidence to show that the typhoid bacilli multiply in the house fly. On the contrary the evidence goes to show that they are not adapted for prolonged life on or in the fly.

5. It thus follows that the house fly is a purely mechanical carrier of the typhoid bacillus and is not a natural “host” in the strict sense of the term.

6. Many bacilli closely resembling B. typhosus in cultural characteristics appear to be natural or, at least, common inhabitants of the intestine of the house fly. These are extremely likely to be mistaken for B. typhosus unless the most stringent tests are employed.

7. As might be expected there is evidence to show that a process of bacterial selection occurs in the fly's intestine. Some bacteria appear to flourish but others are rapidly eliminated. Among the latter must be numbered B. typhosus.

I have to thank my colleagues at the Lister Institute, particularly Dr Ledingham, Dr Henderson Smith and Dr Petrie, for much valuable help and advice during the course of these investigations. To Prof. Bainbridge, Dr Arkwright, Dr Penfold and Dr Macalister my thanks are also due, inasmuch as they helped to smooth over many difficulties both theoretical and practical.

Although plague broke out in India in 1896, it was not until a period of 18 years had elapsed that it reached Ceylon, notwithstanding the close proximity and the daily intercourse which takes place between the two countries. It is not surprising that this prolonged immunity, coupled with the example of the failure of the disease to thrive in such places as Madras, Singapore and other maritime towns within the tropical belt, should have created in the minds of the public in Colombo a feeling of considerable security as regards invasion by this disease, and that when it was announced that plague had at last broken out, there was the usual scepticism expressed as to the accuracy of the diagnosis. The advent of plague in Ceylon has already been announced by Castellani and Philip (1914).

Fifty-six cases examined for Dysentery bacilli. All save one were clinically dysentery, and all save two were negative for amoebae and other protozoa.

Twenty cases (35·7%) yielded dysentery-like bacilli.

From these twenty cases thirty-six organisms were isolated and studied. Of these—

Six corresponded with no familiar pathogenic organism.

Six resembled B. Shiga. Two agreed completely both by fermentation and agglutination. One also gave typical fermentation and agglutination results but was at first a motile organism, although in later generations it became non-motile. The other three are less definitely Shiga bacilli.

Five were mannite-fermenting organisms but only one agreed very closely in its fermentation reactions with B. Flexner and it was not agglutinated by Flexner serum.

Sixteen cultures from ten separate cases corresponded closely with Morgan's bacillus.

Eight of these agreed completely in their characters with Morgan's No. 1 bacillus.

Three differed only from Morgan's No. 1 in being non-motile.

Five presented characters intermediate between those of Morgan's No. 1 and Bowman's “Bac. S.”

Of six cases yielding Morgan's bacillus, agglutination with the patient's own serum was positive in five.

Three cultures showed affinities with the Bac. enteritidis.

Polyvalent anti-dysenteric serum treatment tried on a considerable number of cases was a complete failure.

The following conclusions of practical value to those engaged in serum concentration can be drawn from the experimental work described above:

(1) An end product containing not more than 20% of protein and having a potency of 8 or 10 times that of the original can be obtained as a matter of routine where the pooled antidiphtheritic sera has a unitage not greater than 500, and where the heat denaturation is of the order of 40%.

(2) By the fractional precipitation methods at present employed for the concentration of antitoxic sera for therapeutic use the degree of concentration of high potency sera cannot be taken beyond the limit of about 22,500 units of antitoxin per gramme of protein in the end product.