1. Avian tubercle bacilli were cultivated in centri- fuged embryo extract in large hanging-drop preparations.
2. Small groups of individual bacilli were studied undisturbed on a warm stage for several days.
3. The cultures grew very actively and several types of reproduction were observed, all of which were modifications of what was arbitrarily termed the ‘standard’ type.
4. In the ‘standard’ life cycle, which was that most commonly seen, the bacillus elongates during the first 24 hr. to form a filament several times its original length; this divides repeatedly to form a bunch of 20–30 filaments which on the 2nd or 3rd day break down into short rods, which continue to multiply slowly by ordinary binary fission. If the bacilli are subcultivated in fresh embryo extract the cycle is repeated but not otherwise.
5. Three forms of slow growth were seen: (a) The bacillus elongates, though very slowly, to form a long filament which does not divide. (b) The bacillus elongates to about twice its original length and then divides by simple binary fission, (c) The bacillus elongates and segments into short rods, but the intermediate stage of filamentous proliferation is greatly restricted.
6. Bacilli floating freely in the surface film of the drop often form what we have termed a ‘raft’ colony. In this form of development elongation of the rods and filamentous proliferation occur, but the return to the original short form is gradual and incomplete. The organisms grow in one plane and are arranged in a tidy mosaic.
7. The development of ‘mycelial’ forms was studied. A minority of the long filaments formed by the elongation of the rod-shaped bacilli do not break up into strings of daughter rods but continue to elongate rapidly into abnormally long threads; small buds appear at intervals along the thread and grow into primary branches on which secondary and tertiary branches form in the same way: finally a complex branching ‘mycelium’ is produced.
8. The ‘standard’ life cycle appears to be characteristic of rapid initial growth. The wide differences between our results and those of Kahn are probably due, at least in part, to the different cultural conditions used in the two investigations.