[Illustration: FIG. 52.--After Joest. _A._ Union of two pieces of _Allolobophora terrestris_ in normal position. Twenty-two months after operation. _B._ Union of two pieces _Lumbricus rubellus_. Pieces turned 180° with respect to each other. _C._ Union of three pieces of _A. terrestris_ to make a “long worm.” _D._ Union of two worms (by anterior ends) from each of which eight anterior segments had been removed. After three months. Regenerating two new heads. _E._ A small piece of _Lumbricus rubellus_ grafted upon _Allolobophora terrestris_. Former regenerated an anterior end.] Joest also united two posterior ends by their anterior surfaces. In many cases no regeneration took place, and, in the absence of a head, the combination is destined to die, although it may remain alive, without food, for several months. When two very long pieces were united by their anterior ends,--only eight segments being removed from each worm,--although perfect union took place at first, later one or two new heads generally developed at the region of union (Fig. 52, _D_). When only one head developed it did not seem to belong to one of the components rather than to the other, and originated in the new tissue that appeared between the two pieces. These experiments, in which the anterior surfaces of two pieces are united, show also that the new head arises between the two pieces most often, if not exclusively, when the union is in the anterior ends of the worms. This corresponds with what is now known in regard to the development of new heads by isolated pieces, since there is less tendency to produce a head the farther posteriorly the cut has been made. At more posterior levels a tail and not a head is often regenerated, as has been stated, on the anterior cut-surface. This formation of a heteromorphic tail seems to have been suppressed in the pieces united in this region, except in one case,[82] in which it appears, from Joest’s account, that a tail probably regenerated, although Joest speaks of it as a head. It is more difficult to unite two anterior ends by their posterior cut-surfaces, not because the surfaces refuse to unite, but because the two pieces crawl away from each other and pull apart. In one case, however, union of this sort was brought about.