In Holothuria tubulosa, however, Bertolini (1932b) found that the alimentary canal was formed by the growth of blind tubular elements along the edge of the mesentery, one growing anteriorly from the cloaca and one posteriorly from the oesophagus until they unite to form a continuous structure. As this method was also described in H. floridana and H. impatiens (Kille, 1936), it seems to be general for the genus Holothuria.

Regeneration along the whole free mesenteric edge occurs in S. mollis. If there are no contributions from the oesophageal remnant and cloaca, the alimentary canal and haemal system must be formed from purely mesodermal tissue. The examination of sections of the junctional region between oesophageal remnant and regenerating alimentary canal showed no evidence of growth from the oesophagus. Specimen 87 showed the posterior end of oesophageal remnant separated for about 2 mm. from the dorsal mesentery. A solid cord of cells was present along the dorsal mesentery and continued to the point of junction laterally to the oesophagus. It was not continuous with oesophagus and there was no evidence of proliferation from the oesophageal remnant. Two other specimens showed a similar condition. In all specimens examined up to 33 days' regeneration, the junction of the two regions was distinct. In later stages of regeneration the more rapid increase in diameter of the anterior half of the alimentary canal tends to lessen the distinction between the original and the regenerating alimentary canal, but a slight constriction at the junctional region was still present after 110 days' regeneration.

The junction between the ventral mesentery and the cloaca occurs on the ventral side of the cloaca just posterior to the point of rupture. Growth in thickness of the alimentary canal along the ventral mesenteric edge is slower than along any other region of the mesentery. The alimentary canal is narrowest just anterior to the cloaca. If there were contributions from the cloaca, the alimentary canal would be expected to be thicker along the posterior end of the ventral mesentery than elsewhere, instead of the reverse. In no specimen studied was there any evidence of forward contributions to the alimentary canal from the cloaca.

There remains the possibility that the lumen and inner epithelium may extend forward from the cloaca into the solid cord of cells along the mesentery edge. Kille (1935) showed that this occurred in Thyone briareus, and at the anterior end cells invade the solid rudiment from the regenerating lantern. As no anterior remnant of the alimentary canal is left in Thyone, these cells cannot arise from the alimentary canal. Stichopus mollis shows no regular progressive invasion of lumen or intestinal epithelium either anteriorly or posteriorly. The lumen appeared first in most cases among the cells along the edge of the angle between the dorsal and lateral mesenteries and in a number of cases it appeared independently in other regions (e.g., specimen 87). Thus all the regenerated alimentary canal, including the intestinal epithelium (normally endodermal in origin), is formed from mesenchyme cells and is mesodermal in origin. Such plasticity of the mesoderm is also found in S. regalus (Bertolini, 1930b), and less completely so in Thyone briareus. It is known to occur in certain cases in other groups, such as that shown by Penner's (1937) defect experiments in Tubifex where, after the removal of the

embryonic ectodermal germ band, the mesoderm can later form all the structures normally ectodermal in origin. Plasticity in embryonic forms is normally greater than that found in adult animals. There appears to be no record of adult animals with the complexity of structure found in holothurians, retaining into the adult state such marked plasticity of the mesoderm, as that found in Stichopus.

Autodifferentiation within the mesenchyme cells forming the regenerating alimentary canal, has not been demonstrated in other holothurians. The lining epithelium in the regenerating alimentary canal of S. regalus appears in the whole canal at the same time (Bertolini, 1931). In S. mollis it is shown that the development of a lumen and the differentiation of cells to form a lining epithelium can occur in regions with a solid cord of undifferentiated cells anteriorly and posteriorly. It is therefore not controlled rigidly in early stages of regeneration by the position on the axis of the animal. After the mesenchyme cells separate to form a lumen, the latter becomes lined by a concentration of nuclei without distinct boundaries, i.e., a syncytium. Presumably the nuclei are formed by rapid division of the mesenchyme cells adjacent to the lumen. Definite cellular structure then appears, and the cells, which are at first short, elongate to form a typical columnar epithelium. A remarkable co-ordination must be exercised for these structures to form in their appropriate positions in the mesentery edge, but there is no close relationship to location in the body during early stages. A gradation in diameter from the anterior to the posterior end later develops, and a definite relationship to location is then established.

In S. regalus, Bertolini (1930, 1931) states that the alimentary canal arises as a thin transparent tube which is at first perfectly straight (“perfetamente retillineo”). It later thickens and begins the formation of the loop. This is very different from the stages found in S. mollis. In the latter species, the alimentary canal commences development along the mesenteric edge relatively soon after auto-evisceration. At this stage, the course of the mesenteric edge is still similar to that followed in normal animals. The developing alimentary canal must therefore necessarily first follow the looped course of the mesenteric edge at this stage, and only becomes straight after a later extension of the mesenteries to eliminate the angles of the loops. The method by which a straight tube can arise in S. regalus is not described by Bertolini. It is very difficult to visualise a method by which regenerating tissue along a looped mesentery edge could arise first as a straight tube. It seems highly probable to the writer that the straight tube was not the earliest stage in regeneration in S. regalus, but had been developed before the specimens were collected from the sea, by a method of development similar to that of S. mollis.

The lengthening of the mesentery in the angle between the dorsal and lateral mesenteries in S. mollis achieves two main results. It reduces the length of the free mesenteric edge to less than half the distance. The alimentary canal functions normally before again lengthening. This enables the animal to commence feeding much earlier, with smaller demands on body reserves than would be the case if the alimentary canal had to regenerate along the whole length to the state of differentiation and diameter necessary for feeding to be possible. In addition the straightening of the alimentary canal

provides the base from which the transverse connecting haemal vessel is separated off. If looping of the alimentary canal were retained in regeneration a much more complex series of events would be necessary for this vessel to develop and regain its normal relationships across the body cavity. There appears to be no record of the method of development of the transverse connecting vessel in other holothurians.

The dorsal and ventral intestinal vessels in S. mollis were found to develop from the extension of a cord-like swelling which is budded off longitudinally along the length of the alimentary canal with local condensations separating the connecting strands between the vessel and gut. In the developing haemal vessel a further longitudinal splitting parallel to the alimentary canal separates off the transverse vessel as previously described, and leaves the intestinal vessel adjacent to the gut. In Thyone briareus, Kille (1935) shows that concurrently with growth in length of the intestine in the angle between the dorsal and lateral mesentery “the intestinal haemal plexus is established within the first major loop,”… as “a crescent-shaped membrane which in texture resembles the mesentery.” This membrane at first shows its greatest distance from the edge to the intestine in the angle of the first major loop and tapers to merge with the intestine anteriorly and posteriorly. With increase in length of the intestine posteriorly in the angle between dorsal and lateral mesentery, the membrane does not keep pace and separates from the gut. Condensations form a plexus of strands, and it is left as a bridge-like plexus connecting the first two major sections of the intestine. This is described as the fore-runner of the intestinal haemal plexus.

The method of development followed by S. mollis differs in a number of details. There is evidence that the ventral intestinal vessel separates at a relatively much earlier state of development, appearing before the lengthening of the mesenteries has achieved the straightening of the alimentary canal. It arises as a cord of cells adjacent to the alimentary canal, and extends in either direction until it connects with the haemal vessel remaining along the oesophageal remnant, and posteriorly it extends towards the cloaca. The dorsal intestinal vessel arises in a similar manner, but develops later, while the transverse connecting vessel is separated off from the ventral vessel by a similar process which, however, occurs only after straightening of the alimentary canal. Except for the series of events in the establishment of the intestinal haemal plexus of Thyone across the body cavity, which partly resemble those followed by S. mollis in the establishment of the transverse connecting vessel across the body cavity, the method and relative time of origin of the haemal vessels in S. mollis differs markedly from that found in Thyone.

Rudiments of the respiratory trees in S. mollis appear first after 25–35 days' regeneration and then grow forwards as tube-like structures from the dorsal wall of the cloaca. Respiratory movements of the cloaca continue after autotomy during the period of regeneration.

Before the development of respiratory trees, sea-water is taken into the body cavity, as closure of the cloaca is sometimes delayed up till 20 days after autotomy. The respiratory trees in their early stages show distinct perforations in the elongating tubes which still allow direct communication between sea-water and coelomic fluid,