1. Pieces of the anterior end of Allolobophora foetida containing less than thirteen segments rarely, if ever, regenerate posteriorly, yet such pieces can regenerate very quickly anterior segments if these are cut off. The result shows that the lack of power of the anterior pieces to regenerate posteriorly does not depend, directly, on the size of the piece. 2. Anterior ends containing from thirteen (?) to thirty segments sometimes regenerate posteriorly, but only after a long time and in general, the shorter the piece (i. e. the nearer the cut to the anterior end) the longer the interval before it begins to regenerate, and the fewer the pieces that regenerate at all. 3. Similarly very short posterior pieces do not regenerate anteriorly, longer pieces from the posterior end regenerate occasionally, but only after a long interval of time. In general the shorter the posterior piece the longer the time before the piece begins to regenerate anteriorly. 4. The experiments show thatWeismann's hypothesis of latent cells is insufficient to explain the phenomena of regeneration, because it can not account for the delay in regeneration of a lost part under the circumstances given above. 5. Short pieces from the middle of the worm sometimes regenerate both anteriorly and posteriorly. 6. If a worm be cut in two pieces and then the anterior end be cut off again from the anterior piece, the middle piece will regenerate posteriorly at the same rate and time as though the anterior end had not been cut off. 7. If posterior ends of two worms be sewed together and if then one of the ends has a part cut off, the part that regenerates is like the part removed; i. e. a new posterior end, and not a new head, regenerates.