Considerable difference of opinion has existed as to whether the incidence of cancer is at all affected by the habits of life of different classes of the community. There is of course no doubt that in certain occupations particular forms of malignant disease are exceptionally prevalent, the case of chimney sweeps being notorious; but the more general question as to the incidence on large groups of occupations has not been so definitely determined. The present research was suggested by the apparently contradictory results obtained by Heron from his study of London statistics, and by Maynard from an analysis of United States data.

It appears to be widely believed that normal human urine as voided is sterile, and that the fact of sterility is capable of ready demonstration.

It is the custom in most laboratories where infected material is used during the progress of experimental work to have receptacles containing a disinfectant of some kind, placed so that the worker may drop any small piece of apparatus or culture into it which he has finished with, in order that such material should not be a source of danger to himself and to others in the laboratory. The disinfecting agent is more often than not some saponified tar-acid product, which, although lethal to naked bacteria may, or may not, destroy infection under the circumstances in which it is used. The fluid is also somewhat costly, apart from being uncertain in its action when resistant spores are being dealt with.

Several attempts have been made at various times to obtain a mathematical expression of the relation between dosage, body weight and death time. The subject has recently gained importance through the interesting publications of Dreyer and Walker, who have put forward such a formula, and have illustrated its application to toxicity determinations for a number of substances, and, in particular, for different samples of diphtheria toxin—a matter of considerable practical importance. I propose to examine the theoretical basis of their formula, and its relation to formulae which other workers have suggested, and to criticise the evidence with which they have illustrated its application to experimental material.

Both the sensitation of typhoid bacilli by serum and their ingestion by leucocytes when sensitised, have been shown by Ledingham (1912) to take place in accordance with the laws regulating an adsorption process, the removal of opsonin from a serum by bacteria, and of sensitised bacteria from an emulsion by leucocytes following very satisfactorily the adsorption equation y=kxn, where y= the amount adsorbed, x= the amount left free and k and n are constants.

In order to determine whether non-lactose fermenting bacilli are numerous in fresh milk, a series of experiments was made in the early part of this year on samples of fresh dairy milk, taken in Dumfriesshire, for routine bacteriological examination. The samples were brought to the laboratory with the least possible delay, and in no case did more than seven or eight hours elapse between the time of milking and the inoculation of plates. In the majority of instances, the interval was much shorter.

Experience gained from a previous enquiry on this subject in human tuberculosis led to the belief that much information might be derived from an investigation of the complement fixation reaction in animals experimentally infected with tubercle bacilli. Accurate dosage can be measured and the true path of infection is definitely known. Various enquiries were suggested and investigated by the detailed examination of rabbits and guinea-pigs; the latter class of rodents were used in batches of six to twelve in number, as otherwise the individual differences between animals in the same group of experiments are entirely overlooked. My cultures of the human tubercle bacillus were obtained by inoculating guinea-pigs with the sputum from typical cases of pulmonary tuberculosis at the Brompton Hospital Sanatorium at Frimley. Pure cultures of the bacillus were obtained from the infected guinea-pigs and cultivated on Dorset's egg medium, so that within a period varying from 14 to 21 days an abundant growth was obtained. The culture of the bovine bacillus was supplied to me by Professor Delépine who obtained it directly from the tissues of an infected cow, and subcultures were kept going on Dorset's egg medium. In every experiment without exception the animals were infected with definitely known quantities of the human or bovine bacillus. These were obtained by carefully scraping the growth off the surface of the egg medium and weighing it on sterile platinum foil, while in some cases (for comparison) a portion of the growth was dried in a desiccator before it was weighed. The untreated or dried bacilli were then shaken in a known quantity of sterile saline, so that a perfect emulsion free from clumps was obtained. The bacilli were kept in the dark in brown stoppered bottles and were always employed within a few days of their preparation.

(1) If B. coli be subcultured into another sample of the same medium when growing at full pace, it will continue to grow at the same pace.

(2) If the maximum rate of growth be interrupted by a short application of cold, growth will recommence without lag on the temperature being raised. If the cold be long continued, lag will tend to reappear.

(3) Differences in the size of inoculum have practically no effect on lag in the case of large inoculums, in the case of small ones, on the other hand, diminution of the seeding has the effect of lengthening lag, and this lengthening effect is more marked the smaller the seedings become.

(4) Lowering the temperature lengthens the lag. The effect is very similar to the effect on growth.

(5) The older a parent culture (within limits) the longer the lag.

(6) The length of lag varies with the medium even if adaptation has been arranged for beforehand.

(7) Heat-stable products in B. coli cultures on peptone water have, in the case of overnight cultures, but little effect on lag.

(8) After washing the bacteria for two hours with saline in order to remove possible inhibiting agents, it was found that the lag, on subculture, still occurred and was indeed slightly longer.

(9) If a peptone water culture of B. coli be centrifuged, it is found that the few bacteria remaining in the supernatant commence to grow again at a quick rate but not without a period of lag.

In 1909, in a paper published from the Federated Malay States Institute for Medical Research, Fraser and Stanton in discussing the etiology of beri-beri found that polyneuritis could be produced in fowls by a diet of polished white rice and that the fowls remained healthy if, in addition to the polished rice, they received a sufficient quantity of rice polishings or a quantity of an alcoholic extract (evaporated at a temperature of 52° C.) of whole rice. They established the fact that the essential cause of beri-beri was to be sought for in a nutritive defect, and they endeavoured to determine by chemical methods precise differences between rices capable of causing polyneuritis in fowls and rices which did not cause the disease. Their experiments had shown conclusively that highly polished rice gave rise to the disease but whole rice did not and that the disease could be cured by adding rice polishings to the diet. The chemical examination showed that, whereas whole rice contains about 0·469 per cent, of phosphoric oxide, polished white rice contains only about 0·277 per cent, and the polishings contain about 4·2 per cent.

In a previous paper (Lancet, 4·I·1913), we published a preliminary report on this subject and are now able to give further particulars of our results and to discuss in detail certain points arising from them.

The object of this portion of the investigation was to determine

(a) Whether the amount of immune-body present in the sera of cases of pulmonary tuberculosis, bore any constant relation to the stage of the disease and the condition of the patient.

(b) Whether well-marked clinical phenomena, e.g. periods of fever, attacks of pleurisy etc., were associated in any definite manner with variations in the amount of immune-body in the serum.

(c) Whether gradual definite and sustained improvement or deterioration in a patient's condition was accompanied by increase or decrease of the specific immune-body in the serum.

Of the large number of experiments recorded by Penfold (1914) in connexion with the question of lag in bacterial growth, a certain proportion have been carried out in such detail as to render them eminently suitable for mathematical analysis. In all work of this kind, which is intended to throw light on the numerical aspect of bacterial growth, it is desirable to take observations at frequent and, if possible, regular intervals, during the period of lag, and to count a sufficiently large number of colonies on the plates. To secure this last and most important desideratum, orientating experiments must be performed, from the results of which one is enabled to calculate what dilution of the culture at any given period of growth will yield a tolerably large and accurately countable plate population.

Owing to the difficulty of satisfactorily explaining many facts in serological work, there has been a tendency in the last few years to consider serological phenomena more and more from a physico-chemical point of view.

The experiments detailed above have already been discussed in the sections in which they have been recorded. The following is a brief summary of the results obtained.

(1) In many ways a clear watery extract of B. typhosus behaves like an ordinary emulsion of the bacilli. This is true in a general sense in respect of agglutination by serum, and the property of agglutinability by weak salts, which some emulsions of B. typhosus possess, is also exhibited by a watery extract of the same strain.

(2) The optimum hydrogen-ion concentration for the agglutination of an emulsion is also approximately the optimum for the agglutination of a watery extract made from the same emulsion. The washed bacilli and the original emulsion show a second optimum. Agglutination appears later at this secondary optimum which is in stronger acid. There is a zone of absent or deficient agglutination in solutions of intermediate strength. The lower [H·] of the two is the characteristic optimum for B. typhosus, i.e. about 3·6×10−5. The second optimum [H·], i.e. 1·1 × 10−3 to 2·2 × 10−3, is approximately the same as the optimum for the agglutination of emulsions of many different bacteria, e.g. some strains of B. coli, staphylococcus, etc., and moreover is about the optimum for the agglutination of an emulsion of B. typhosus after it has been heated to 100°C.

These two optima appear to indicate the presence of two acid agglutinable substances in an emulsion of B. typhosus, (a) which is characteristic of this bacillus and is present in a watery extract (this substance is rendered inagglutinable by heating to 80° C. for 30 minutes or to 100° C. for five minutes, or longer, e.g. one hour), and (b) which is present in the bodies of the bacilli after repeated washing and which is not characteristic of B. typhosus, but has characters shared by similar substances in other bacilli; it resists heating to 100° C. for five minutes and is apparently little, if at all, affected by heating to 80° C. for half an hour.

(3) The substance agglutinable by serum and that agglutinable by acid No. 5, i.e. [H·] 3·6 × 10−5, are both present in a watery extract and can both be completely removed from the bacillary bodies by washing with distilled water. The serum-agglutinable substance is however less easily completely removed. This apparent difference is perhaps due to the fact that serum is a more delicate reagent for testing for the presence of agglutinable substance. If an extract is diluted it will give a precipitate in higher dilution with serum than with acid and the bacillary bodies when repeatedly washed often react with serum after the acid agglutination in solution 5 or 6 has ceased to appear. These substances agglutinable by acid and by serum are however closely associated and are both precipitated by acid No. 5 or 6.

The identity of the two substances has however not been proved and the following considerations are very strongly opposed to this assumption, (a) a considerable number of strains of B. typhosus are agglutinable by serum and not by acid; (b) strains have been described which are agglutinable by acid No. 5 or 6 but not by serum; (c) the diminished agglutinability by serum exhibited by some strains which are inagglutinable by acid is apparently due to a total diminution of the “soluble” or extractible substance; (d) the substance agglutinable by acid, tube 5 or 6, i.e. [H·] 3·6 × 10−3, is rendered inagglutinable by heating to 100° C. for five minutes, whereas Porges and Prantschoff (1906) have shown that the serum agglutinable substance is still active after being boiled.

It is not possible at present to say for certain whether the substances agglutinable by serum and acid are identical or distinct. If they are identical and there is only a single substance, its agglutination properties vary independently in different strains of B. typhosus.

(4) The importance for agglutination by serum of the substances in “solution” in the liquid part of a bacterial emulsion, and the impossibility of distinguishing between agglutination and the precipitation of bacterial extracts by serum, were first pointed out by Kraus and Paltauf, and by Kraus and v. Pirquet.

Their view is confirmed and extended to the agglutination by acids Nos. 5 and 6 by the result of some of the experiments described above. The facts adduced in support of the view that agglutination is essentially the formation of a coagulum in the liquid part of an emulsion are the following:

(a) The property of agglutination may be completely removed from B. typhosus by washing with water; the resulting washed bacterial bodies no longer agglutinate with specific serum or in acid solution Nos. 5 to 7.

(b) The washings of the bacteria or watery extracts are agglutinable by specific serum or by acid solution No. 5 or 6.

(c) The watery extract even in very small quantities when added to the washed bacilli, indifferent bacilli or other finely divided particles, confers on them the property of being agglutinated by specific serum or acid solution 5 or 6.

(d) The extract reacts with highly diluted serum, e.g. 1/27,000, and may be considerably diluted, e.g. 1/96, without entirely losing its agglutinable properties with serum or acid.

(e) In order to demonstrate the presence of agglutinable substance in very low concentration (especially if highly diluted serum or if acid is used) it is necessary to add some indifferent substance, e.g. B. coli, kieselguhr, globulin, etc. in order to increase the bulk of the precipitate.

(f) The extract or successive extracts appear to contain all the characteristic agglutinable substance of the original emulsion. This is shown by the delicacy of the reaction, and the small amount of extract required to agglutinate an indicator, e.g. B. coli. The demonstration of the reaction with highly diluted serum 1/27,000 and much diluted extract 1/96 appears to be an advance on the work of previous writers. Nicolle did not use indifferent particles with high dilutions of serum or of extract, and the only other use of an indicator under such conditions appears to be the use of conglutinin by Bordet and Gay, and Streng, and of mid-piece by Dean.

(g) The microscopic demonstration of the coagulum formed in the process of agglutination has been made easy by taking a watery extract containing very few bacilli, i.e. ordinary emulsion from which most of the bacilli have been removed by centrifuging. After adding dilute specific serum or acid the formation of the coagulum may be watched in a microscope cell with dark ground illumination or by adding dilute fuchsin or eosin to the microscope cell.

Early in 1912 Professor E. C. C. Baly, F.R.S., acting on behalf of Mr Jessup, brought one of us a collection of Swiss statistics and requested that it should be analysed, with the object of measuring the relation between the death-rates of cancer and diabetes. In the course of this work we have been led to consider many problems of theoretical and practical interest, bearing partly on the method of analysis and partly on the results yielded; it would not be possible to deal with all these in a single paper and the present memoir is devoted to a study of the problem first suggested, viz. the relation between the cancer and diabetes death-rates. Since, however, this may be the first paper of a series, it will be convenient to deal at some length with the nature of the investigation and the motives which led us to undertake it.