(1) B. typhosus ferments dulcite-peptone-water in about 10 days. By sub-culture this is readily shortenedto one day. It produces nonfermenting colonies on dulcite-agar plates. The colonies show after five days and upwards, secondary colonies as papillae, which may or may not have an acid reaction.

Growth of B. typhosus on dulcite media readily gives rise to bacilli forming, from the commencement, acid colonies on dulcite plates. The dulcite character has a tendency to atavism in the early stage which further growth on dulcite removes.

(2) B. typhosus can be trained to ferment lactose, only with very great difficulty. Definite and conclusive results have been obtained by meonly with the white atavists of Twort's trained lactose strain. Other strains (20) have not yielded to such training methods, although many of them have been trained for over a year. The lactose character when acquired is very unstable, and is soon lost when growth takes place in a lactose-free medium. A fermenting colony on replating gives rise to a mixture of fermenting and non-fermenting colonies.

(3) All strains of B. typhosus which I have examined show the secondary colonies on isodulcite described by Reiner Müller. Growth on isodulcite yields directly a strain without this papillae-forming power. So-called “acid” papillae as described by Reiner Müller have not been observed in the strains (20) investigated by me. This new strain does not regain the power to form isodulcite papillae by frequent sub-culture apart from isodulcite. Apparently the acquired character has arisen by mutation. Fermentation of isodulcite requires a considerably longer process of training.

(4) The above three characters vary in permanency. The permanency varies inversely as the time required for the selection or training.

(5) Papillae-formation arising on carbohydrate media on colonies of members of the typhoid-coli group may indicate variation by slow degrees just as well as by mutation. I have observed it in normal typhoid varying towards dulcite and lactose as well as isodulcite. In the case of Twort's strain which fermented sorbite-agar.

Formation of papillae simply indicates apparently that the variation only affects very few individuals of the colony to a great extent, and does not guarantee permanency in the character.

If the papillae arise early and without sub-culture, the evidence adduced so far suggests that the character will have considerble permanency. In the case of dulcite the results show that the formation of papillae on certain carbohydrate media cannot, as Reiner Müller stated, be regarded as definitely specific for the variety of microorganism exhibiting them.

The variations detailed in the above communication do not in our view invalidate the sugar tests as aids to differentiation, since these chemical functions are subject to variation probably to no greater extent than other functions of bacteria. Further, the ability of an organism to vary in any particular direction may be of considerable value in differential diagnosis.

This instability makes it imperative to have as large a number of differentiating characters as possible for every species, and also to take care in laboratory culture that we do not grow bacteria under conditions likely to alter their characters. It would be inadvisable for example to propagate the B. typhosus from a dulcite culture a few days old if one desired it to conform to the normal typhoid type in respect of dulcite.

In sending this paper to the Press I would like to acknowledge my indebtedness to Dr Martin, who has supervised my work on behalf of the British Medical Association, and to Dr Ledingham, who suggested the work, from both of whom I have received much valuable help and criticism. Many other colleagues at the Lister Institute have helped me from time to time in points of difficulty, and I take this opportunity of thanking them.