By means of grafting it is possible to keep alive small pieces of a worm that would otherwise perish. For instance, pieces of a worm containing only three segments are not capable of independent existence, except for a short time, and even pieces of from four to eight segments die in most cases. It is not possible to unite small pieces of this size directly upon larger pieces, since they will die, ordinarily, as a result of the operation, but larger pieces can be united and then after union has been effected, one of them may be cut off near the place of union. The same result is sometimes brought about accidentally by the worms themselves pulling apart and leaving a small piece of one component attached to the other. Joest found that in several cases these small, attached pieces regenerated. In one case, after two long pieces had pulled apart, a small piece, left by one of the two, regenerated a single new segment with a mouth at its end. In another case, after one of the components had been cut off, leaving two segments attached, a new part of seven segments regenerated.[83] Especially interesting is the case in which two individuals (_A. terrestris_) had been united to form a long worm. The anterior component extended to within two centimetres of the anus; the posterior piece had had the first four segments removed. Three days later the anterior piece was cut off three segments in front of the region of union. About a month later a small part of eight segments had regenerated from the cut-end (Fig. 53, _A_). Fifteen days later another new part of thirty-seven segments developed at the end of the first new part (Fig. 53, _B_). Joest speaks of the first eight segments as a head, and the second simply as a regenerative product. There can be little doubt, I think, that both parts represent a heteromorphic tail. The region from which the regeneration took place would make this interpretation highly probable, and Joest’s figures also indicate that the structure is a tail. The result is very interesting, if my interpretation is correct, as it shows that the major component did not influence the kind of regeneration, although the surface of regeneration was separated by only three tail-segments from the anterior end of the major component.