There are many difficulties in the way of determining the origin of the cells that make up the new part. The only means at present at our command for studying their source is by serial sections of a number of different stages taken at intervals from different animals. Since there may be differences between the processes in different individuals, and since we can only piece together the information gained from successive stages, much uncertainty exists in regard to the changes that take place during regeneration, even in some of those forms that have been examined over and over again. Were it possible actually to follow out the movements of the living cells in one and the same animal, the problem would offer fewer difficulties, but this cannot be done. It will be more profitable to consider first the better-known and simpler processes, and afterward those that are less well-known.

The regeneration of the head and tail of lumbriculus and of certain naids is a comparatively simple process, and has been studied by several investigators, whose results agree, at least in regard to the most essential features. Semper (’76) described the origin of the new organs in the formation of new individuals by budding in nais. He found that the new brain and nerve-cord develop from the ectoderm, the new mesoderm also from ectoderm, and the new digestive tract from the old one, except the pharynx, which arises by the fusion of two mesodermal “gill-slits.” BÜlow (’83) studied the regeneration of the tail of lumbriculus. He found the ventral cord in the new part arising from a paired ectodermal thickening, the mesoderm arising from a proliferation of cells. These cells are invaginated in the region between ectoderm and endoderm—the in-turning of the proctodÆum being looked upon as an endodermal invagination.[91] The more recent work of Randolph, Rievel, Michel, Hasse, Hepke, and von Wagner on the same or related forms has served to point out certain errors in the earlier work of Semper and BÜlow, and has added some new and important facts, especially in connection with the origin of the mesoderm in the new part. Without attempting to give a detailed account of these results,

I shall describe the principal changes that have been found to take place. When the anterior end of lumbriculus or of tubifex is cut off, the cut-surface very quickly closes, as a result of the contraction of the body wall. According to some investigators, the circular muscles are chiefly concerned in the closing, but according to others the longitudinal muscles bring about the result. The cut-end of the digestive tract is pulled a little inward, and its end also closes (Fig. 57, A). For a day or two no important changes can be observed to take place, but new ectoderm soon appears over the cut-surface. This ectoderm arises in all cases from the old ectoderm, and as it increases in amount the old ectoderm is pushed back from over the cut-end, leaving a layer composed of a single row of cells over the end. Since nuclei in process of division are rarely present before these initial processes begin, it is probable that the changes are due, in large part, to an out-wandering of ectodermal cells, or, what amounts to the same thing, to the leaving behind of cells as the old ectoderm withdraws from the cut-end. In the new ectoderm over the end, an active process of proliferation takes place (Fig. 57, B), that leads to the production of a large number of cells lying within the new part. The ectoderm has at this time begun to bulge outward, so that the proliferated cells come to lie within the dome-shaped beginning of the new head. There appears to be some difference in the number and in the location of the proliferations in different species. In general, the new cells arise from the ventral and ventro-anterior region of the dome-shaped ectodermal covering of the new part. Most of this new material gives rise to the brain, commissures, and ventral nerve-cord (Fig. 57, C). The cells giving rise to these structures in tubifex come from two ventral regions of proliferation that extend along the sides and dorsally to the anterior end in front of the digestive tract. Where the two masses meet above and in front, the brain is formed.[92] The cells that do not take part in the formation of the nervous system give rise to the muscles and connective tissue of the new head. These cells lie especially at the outer sides of the proliferated mass. The origin of the new muscles from ectoderm stands in sharp contrast to the current ideas in regard to the origin of new tissues, and yet it is a point on which the more recent investigators are entirely in accord. Michel, Hepke, and von Wagner have arrived at the same conclusion after a careful examination, and there seems to be no reason for refusing to accept their results. The theoretical importance of this discovery will be discussed later.

Soon after the proliferation from the ectoderm has begun, the blind end of the digestive tract starts to push forward (Fig. 57, D). The cells in the most anterior part of its wall begin to divide, and the end grows in an anterior direction as a more or less solid rod. This rod extends, in some species, as far forward as the ectoderm, meeting the latter on the inner side of its antero-ventral surface. At this point an in-turning of ectodermal cells, in the form of a blind pit, develops, and later this pit, deepening to become a tube, forms the mouth cavity. Its inner end is from the beginning in contact with the anterior end of the digestive tract, or else it connects with the latter soon after its formation. The two flatten against each other, the cells draw away in the middle of the region of contact, and the cavity of the new mouth becomes continuous with the cavity of the old digestive tract. The mouth lies at first nearly terminal in position (Fig. 57, E), but by the forward growth of the body wall over and in front of the mouth to form the prostomium, the mouth comes later to lie more on the ventral surface. The short tube produced by the in-turned ectoderm forms only a short part of the digestive tract. It leads from the mouth opening to the new pharynx, and forms, therefore, only the buccal cavity. A similar ectodermal tube, the stomodÆum, which develops in the egg-embryo, becomes not only the buccal chamber, but also the lining of the pharynx. The latter is, therefore, considered an ectodermal structure in the embryo. On the other hand, in the regenerated head the lining of the new pharynx arises from the anterior part of the endodermal digestive tract. We find, therefore, that the same organ, the pharynx, may arise in the same animal from distinct “germ-layers.” This result also has an important bearing on our ideas concerning the value and meaning of the so-called “germ-layers,” and has helped to bring about a revolution of current opinion as to the importance of these layers.

The preceding account of the development of the head has shown that while certain of the new organs and layers arise from the same organs of the old part, yet this is not true for all of them. Thus while the ectoderm gives rise to ectoderm, the new muscles do not appear to come from the old ones, or even from other mesodermal tissues, but from the ectoderm. The old digestive tract gives rise to the greater part of the new one, but the new pharynx comes from the old endoderm, and not from the in-turned ectoderm. The nervous system does not arise from the old ventral cord, but from a proliferation of ectoderm. It has, thus, the same origin as the nervous system of the embryo. The origin of the new blood vessels has not been satisfactorily made out. The seta sacs arise from ectodermal pits as in the embryo.

In regard to the origin of the new mesoderm, the evidence is still insufficient, I think, to show that cells derived from the old muscles or peritoneum take no part in the formation of the new muscles and peritoneum; but that the greater part of the new muscles, etc., comes from the proliferated cells can scarcely be doubted. This latter discovery loses none of its significance, however, even if it should prove true that the old muscles, etc., contribute something to the new part. It is also not entirely disproven that the ventral nerve-cord does not take a small share in the development of the new cord.

The regeneration of a new tail-end in these same forms appears to take place in much the same way as the head. The cut-end quickly closes; later a layer of ectoderm appears over the posterior surface, and the new part bulges out and becomes dome-shaped. A paired, or in some species a single, region of proliferation develops from the ectoderm, that gives rise to the new ventral nerve-cord. Lateral proliferations of ectoderm produce, according to some writers, the material out of which the mesoderm of the new tail is formed. Randolph, on the other hand, has described the new mesoderm as arising from the old, especially from certain large peritoneal cells that are found throughout the body. The cut-end of the digestive tract closes, and later new cells develop at its posterior end. An in-turning of ectoderm, in the form of a pit, fuses with the posterior end of the digestive tract and establishes communication with the outside.

The regeneration of the anterior end of the earthworm has been carefully worked out by Hescheler, and although on account of the greater complexity of the process the results are not so decisive as those just described, yet in many respects they are in agreement. In Hescheler’s experiments only four or five anterior segments were cut off. The closing of the cut-end is somewhat different from that in lumbriculus. A plug of cells soon forms over the end (Fig. 58, A). The new cells appear to be lymph cells. Although this mass of cells may be quite large, the cells do not seem to form later any of the organs in the new head. The presence of these cells makes it very difficult to work out the origin of the other cells that appear later. Owing to the absence of this lymph plug in lumbriculus and nais it is easier to follow in them the regenerative processes. In the midst of these lymph cells spindle-like cells soon appear whose origin is obscure, but Hescheler thinks it improbable that they are transformed lymph cells, although they are completely intermixed with the latter. The spindle-cells arrange themselves later in regular bands, that appear to be extensions of the longitudinal muscles. A few days after the operation, the lymph plug is covered over, beginning at the edge, by the ectoderm. The new ectodermal cells arise from the old ectoderm, and seem to extend over the lymph plug by a sort of migration process. Division of the cells does not occur at this time. These covering cells are at first all alike, the characteristic gland cells of the ectoderm being absent. The digestive tract withdraws somewhat from the outer cut-surface, and its end closes. The closed end abuts against the inner surface of the lymph plug. The next changes are initiated by the appearance of karyokinetic divisions in all the tissues of the new part, which lead to a rapid growth and elongation. Dividing cells are found in the new, as well as at the border of the old, ectoderm, where the new and the old parts are continuous. At this stage there appears in the lymph plug another kind of cell, that seems to arise, in part at least, from the ectoderm by an in-wandering of new cells. Other new cells may come from the edge of the old muscles, but it is not clear whether they come from a transformation of muscle cells, or from undifferentiated cells lying in the old muscles. In addition to these sources of new cells, it appears not improbable that cells may separate from the end of the digestive tract.

Nerve fibres push out from the end of the ventral nerve-cord into the new part, and groups of cells, often in process of division, appear in the old ganglia, even in those that lie a long distance from the anterior end. It is not improbable, Hescheler thinks, that new cells, as well as fibres, grow forward from the most anterior end of the nerve-cord into the new part. A mass of nerve cells and fibres appears in front of the old nerve-cord, and extends upwards and around the digestive tract, to meet over the anterior end of the latter in another mass of cells that have arisen from an early in-wandering of ectodermal cells. It is not improbable that the masses around the digestive tract (the commissures) and also the new ventral cord may also include cells that have had the same origin.

A tubular invagination of ectoderm is formed at this time at the anterior end. It meets the anterior end of the digestive tract; the two fuse, and the communication of the digestive tract with the outside is established. The pharynx develops from the anterior part of the digestive tract, which after Hescheler’s operation may contain some of the original ectodermal stomodÆum, since only five of the anterior segments were cut off, and the embryonic stomodÆum extends somewhat behind this region. In another experiment, carried out by Kroeber, somewhat more of the anterior end was removed, but the result was the same (Fig. 59), so that it is clear that the new pharynx may be formed from the old endoderm.

Hescheler leaves several points still unsettled, more especially the origin of the cells that give rise to the new musculature, but it is almost impossible to make out their origin in this animal, owing to the presence of the lymph cells. Hescheler’s discovery that the cells of the lymph plug do not themselves, in all probability, contribute to the new part, is an important result, and shows that these seemingly undifferentiated cells do not possess the power of giving rise to the different kinds of new tissues. The in-wandering of cells into this solid plug from the ectoderm, and perhaps also from other sources, and their subsequent union to produce the definitive organs, is also a point of capital importance, especially as it puts us on our guard against a too ready acceptation of the view that all cells in a mass that have the same general and undifferentiated appearance have had a similar origin, and in showing that apparently indifferent cells may really carry with them into the new part those characters that determine their fate. Other cells, apparently equally undifferentiated, and lying in the same position, may have quite different possibilities.

In the vertebrates, the regeneration of the tail and limbs of amphibia and of the tail of lizards has been studied by a number of investigators. The regeneration of the tail of several urodeles and of the larva of the frog was investigated more fully by Fraisse (’95) and by Barfurth (’91). If we examine first the results of Fraisse’s study of the tail of urodeles, which have bony vertebrÆ, we find the following changes take place. The cut-surface is covered by the skin bending over the exposed part, accompanied by a migration of cells from the edge of the ectoderm. Only the unspecialized cells leave the old ectoderm to wander out over the cut-surface; gland cells and sense cells are entirely absent from the new ectoderm. These kinds of cells develop later out of the undifferentiated cells over the new part. The development of new vertebrÆ does not follow the embryonic method of development. In the embryo the endodermal notochord is first laid down, and around this and the nerve-cord mesodermal cells accumulate to form the skeletal tissue. Later the notochord is largely obliterated, as the vertebrÆ develop, pieces of it being left along the vertebral column. In the regeneration of the tail of the adult animal, the remnants of the old notochord (even if exposed by the cut) do not take any part in the formation of new tissue. In fact, there is no notochord formed at all. From the injured vertebrÆ, or at least from their covering of skeletal tissue, cells are proliferated, out of which a cartilaginous tube develops, enclosing the new nerve-cord, which is growing out from the cut-end of the old cord. In this tube centres of deposition of calcareous material are formed, and the new vertebrÆ are produced in this way. The new nerve-cord develops from the cut-end of the old cord, and more especially out of the cells of the lining epithelium of the canalis centralis. The new muscles develop from cells that arise from the old muscles.

In the tadpole of the frog the regeneration of the tail takes place essentially in the way just described for the adult urodele, except that, there being only a notochord in the tail, only a notochord is regenerated. According to Fraisse, the new notochord develops from cells that arise from the sheath of the old notochord, and not from the vacuolated cells of the notochord itself. The notochord cells are, he states, derived from the endoderm of the embryo,[93] while the sheath arises from the mesoderm; hence the newly regenerated notochord that arises from the sheath of the old one comes from a different germ-layer. Exception may be taken to this statement, because in the frog’s embryo the notochord develops from tissue that is at first perfectly continuous with the mesoderm, and, in fact, may be called mesoderm; also because it is probable, in the light of more recent research, that both the notochord and its sheath have exactly the same origin.

It is known that the tail of lizards breaks off generally at a definite region near the base, and that the break does not occur between the vertebrÆ, but in the middle of a vertebra—in some species the seventh caudal. The vertebrÆ are thicker at their ends than in the middle, and are firmly held together by intervertebral cartilages. The centres of the caudal vertebrÆ are the weakest links in the chain, or at least the place at which the vertebral column is most easily broken in response to the contraction of the tail-muscles.[94] Fraisse and others speak of this arrangement as an adaptation for breaking off the tail.

The new tail that regenerates does not contain a new series of vertebrÆ, as does the new tail of the salamander, but, instead, a cartilaginous tube that is attached to the half of the broken seventh caudal vertebra.

The regeneration of the new tissues of the tail of the lizard takes place as follows: A scab forms over the cut-surface, composed in part of clotted blood, in part of broken-down tissues from the injured cells. In the course of a week the necrotic tissue falls off, and a smooth surface of ectoderm is found covering the end of the tail. The new ectoderm appears to come from the old, but its method of development has not been studied. The deeper layer of the skin of the lizard is composed of mesodermal connective tissue, and in the new part this layer arises from the connective tissue of the old part. The tissue that forms the cartilaginous tube of the new tail develops from the skeletal tissue of the broken vertebra. The remnants of the old notochord that are present in the vertebra, have nothing to do with the new structure, nor does the new tube represent in any way a notochord, but it appears to be a structure sui generis. In later stages, osseous plates may be formed in the cartilage, but these are too irregular to be compared to vertebrÆ. A tube grows out from the cut-end of the nerve-cord, which in some forms, as Fraisse shows, is only an extension of the lining epithelium of the nerve-cord. In other forms it is possible that other cells of the old cord may also grow backward, divide, and produce new cells. The fine thread that is formed in this way does not send out any nerve fibres into the surrounding parts. In Anguis fragilis, however, a few ganglion cells are present in the new cord. It is probable, Fraisse states, that while the new tube is morphologically a nerve-cord, yet physiologically it is not functional in any of the reptiles.

The new muscles come from the old ones. Fraisse thinks that the new muscle fibres come from the so-called “spindle fibres” that split off from the primitive muscle bundles. These fibres, Fraisse believes, originate normally during the process of physiological regeneration of the muscles, and also after injury to the muscles. From these spindle cells the new muscle fibres develop in the same way as the muscle cells of the embryo.

Fraisse sums up the results of his studies of regeneration as follows: (1) Both in amphibians and reptiles, injured tissues can only produce new tissues like themselves. The leucocytes assume only the function of nutrition and of devouring the broken-down parts of tissues. They never become fixed tissues—neither connective tissue nor any other sort. (2) All tissues are capable of regenerating themselves, either directly out of their differentiated elements, or out of a matrix. As a matrix for the epidermis, there is the Malpighian layer of the skin; for the central nervous system, the epithelium of the central canal of the nerve-cord; and for the musculature, the spindle fibres.

Fraisse also formulates the following general statements: (a) Regeneration is neither a pure recapitulation of the ontogeny nor of the phylogeny. The process is rather a hereditary one, with which complicated adaptations of the tissues are often involved that follow the laws of correlated development. (b) We cannot explain the phenomenon of regeneration, as the result of wounding the tissues, or as the outcome of an increase in the food supply, or as due to the removal of a resistance to growth. Far more important are the principles covered by the former paragraph, (a).

Barfurth has studied in detail the regeneration of the tail in some amphibia; and his results, while not covering as much ground as do those of Fraisse, yet give a more detailed account of the origin of the new tissues. Barfurth’s results on triton and siredon are not essentially different from those of Fraisse. In the tadpole of the frog, Barfurth finds that the notochord regenerates from the sheath of the old notochord. In the larval urodele, he finds that the new notochord arises as in the tadpole, and not from the skeletal sheath, as Fraisse maintains. In very young larvÆ of siredon the chordal cells themselves seem to give rise to the cells of the new notochord. In older larvÆ, in which the skeletal tissue is developed around the notochord, regeneration takes place both from this tissue and also from the sheath of the notochord. He concludes that in the regeneration of the new notochord, and also of the skeleton, the origin of the cells depends upon the developmental stage of the supporting tissues.

In regard to the regeneration of the muscles, Barfurth comes to the following conclusions: In very young larvÆ of siredon, the degenerative changes in the muscle cells are often very slight. Regeneration takes place by growth from and the displacement of the old muscles. During this time bud-like terminal and lateral formations occur in the muscle fibres. These outgrowths contain nuclei and form sarcoblasts; and these pass into the new part, where they make the new muscle fibres in the same way as do the cells of the embryo. In older larvÆ of the frog, and in mature animals in general, the changes are more complicated. Two processes can be distinguished: (a) degenerative and (b) regenerative. (a) Broken-down muscle fibres that have been cut, and torn-off pieces of muscle fibres, are found present. There follows an accumulation of leucocytes and of giant cells. The nuclei in the degenerating muscle fibres atrophy, and the substance of the fibres breaks down. (b) The muscle fibres split lengthwise to form spindle fibres, and there is an increase in the number of nuclei at the same time. Sarcoblast-like outgrowths of the old muscle fibres are formed, which produce the sarcoblasts that become new muscle fibres.

Barfurth agrees with Fraisse in two main points, viz. that all the tissues of the tail have the power of regeneration, and that each tissue produces only tissue like itself. The law which KÖlliker attempted to establish, viz. that the elements of the formed tissues have lost the power of producing other kinds of tissue,—the law of the specification of the tissue,—is supported by these results of Fraisse and of Barfurth, but is contradicted, as has been shown above, by the results on the earthworm, and also as we shall see even in the amphibia, as for instance in the regeneration of the lens of the eye.

Spallanzani[95] was the first to study the regeneration of the limb in salamanders, and found that the skeleton in the new part is like that in the normal limb. Bonnet, Philipeaux,[96] as well as other naturalists,[97] also examined the regeneration of the limbs of salamanders. GÖtte (’79) has studied the embryonic development and the regeneration of the limb of triton, especially in regard to the origin of the new bones. He found that the skeleton develops in much the same way in the embryonic limb and in the regenerated limb, and the process in the latter may be said to repeat that in the former. This is especially true for the regeneration of the limb of a very young larva, but the older the larva the more it departs from the embryonic type of development. If the limb is cut off through the upper arm, or through the thigh, new tissue develops over the cut-end. If the larva is quite young, so that formation of the cartilages in the leg has not gone very far, the new tissue differs very little from the old; but if the leg of an older larva is amputated, the difference between the old and the new parts is more striking. If the bones of the leg have become ossified, the transition from the old to the new part is at first very sharp. The new tissue, that will make the new cartilages of the new limb, develops as a cap over the cut-end of the old bone. GÖtte does not give an explicit statement in regard to the origin of the new cartilage, but his account leads one to suppose that it develops from the old cartilage or from some part of the bone. This is, in fact, the case, as I have observed in preparations of the regenerating leg of Plethodon cinereus, in which the new cartilaginous tissue comes from the periosteum of the old bone. GÖtte shows that two long rods of tissue are formed, that are separate for the greater part of their length. They give rise to the two bones of the lower leg, or forearm, as the case may be. The broken end of the femur or humerus also completes itself by a short cartilaginous cap, which is at first continuous with the two rods just described. The ends of these two rods break up into a series of pieces that form the tarsalia, or the carpalia, and the digits. Two digits are first formed, and the others are added as outgrowths from the side of one of the two rods. It is important to note that the new cartilages are formed, in large part, out of a continuous substratum (or rather of two) which separates into proportionate parts to produce the elements of the new limb.

The regeneration of the muscles of the limb of an adult animal, plethodon, has been recently worked out by Towle. The leg was cut off in the middle of the forearm. Extensive changes take place in all the muscles that extend across the level of the cut. The old fibres in the lower end of the muscle, i.e. those near the cut-end, disintegrate, and the number of nuclei greatly increases. The division of the nuclei seems to be direct, each retaining some of the old muscle substance about itself. From some of these cells the new muscle tissue is formed in the new part. Higher up in the forearm the muscle fibres break down to a smaller extent, and still higher up some of the old fibres may remain intact. New muscle fibres are also formed in the old muscle, especially in the region near the cut-end.

The process of regeneration has not been so fully worked out in any other vertebrates as in those described in the preceding pages, although the regeneration of single tissues or organs in the vertebrates has been extensively investigated. In all such cases it is found that like tissues give rise to like.

In the planarians it has been found that during regeneration the ectoderm covers the exposed surface, and from it arises the new ectoderm; the digestive tract appears to come in part from the old tract and in part from the middle-layer cells; the nervous system appears also to develop out of the middle-layer cells that are found scattered through the body. These cells seem to form a sort of reserve supply that gives rise to the digestive tract, nervous system, and middle-layer cells in the new parts. From them also arise the new pharynx, and the lining of the pharynx chambers, as well as some other structures. It is impossible to say at present whether one and the same kind of cell may give rise to all these structures, or whether different kinds of cells are present in the middle layer, that cannot be distinguished from each other by the methods at present at our command.

The changes taking place in the tissues of those animals that regenerate by morphallaxis have been only quite recently carefully investigated. Bickford stated that in tubularia the old differentiated tissue changes over directly into the tissue of the new part, and Driesch confirmed this statement. Stevens has studied by means of serial sections the different changes that take place. Division of both ectodermal and endodermal cells is found to occur, but especially the ectodermal. Whether all the ectodermal cells divide, or only some of them, is difficult or impossible to state, but whether this happens or not, all the old region goes over into the new hydranth.

The changes that take place in hydra have been recently worked out in my laboratory by Rowley, who finds that a certain amount of division takes place in the old cells, especially in the ectoderm. The division of the cells is not a very active process, and it seems not improbable that many of the old cells go over without dividing into the new part.

One of Trembley’s most celebrated experiments was that in which hydras were turned inside out (Fig. 1, A, B), so that the ectoderm came to line the inner cavity and the endoderm to cover the outer wall. The tentacles were not everted but remained sticking out of the mouth of the everted animal. Their openings, or arm-holes, therefore, appear on the outer surface of the body. In order to prevent the everted hydra from turning itself back again, as it tends to do, Trembley pushed a small bristle crosswise through the wall of the body. Finding the hydras still sticking on the bristles the next day, he concluded that they had not returned to their former condition, but that the outer layer (the endoderm) had changed its character so that it became ectoderm, and the inner layer (the ectoderm) became endoderm.[98] The experiment seemed to show that the two layers could change their specific character and be transformed into each other according to their position in the animal. These remarkable results were not challenged until 1887, when Nussbaum repeated the experiment and showed that Trembley had overlooked an important fact. It was found that even the bristle pushed through the body does not prevent the hydra from regaining its original condition, although it may delay the turning back. If the turning back can be prevented, the animal dies. Nussbaum showed how the turning-back takes place in an animal while it remains on the bristle. The everted foot-end begins first to turn back, pushing into the central cavity. When it comes to the bristle it passes to one side of it, and continuing to turn back the foot passes out of the mouth, drawing the rest of the body after it.[99] The last act of the turning can take place only by tearing away through one or both sides, and this is often done. The bristle may still remain sticking to the body through one side, or even remain through both sides if the body has, after tearing through, healed up around the bristle. The process of turning back may take place quite quickly, and had been overlooked by Trembley, who trusted too confidently to the presence of the bristle sticking through the animal.

The method by which the turning back of the layers takes place was not, it appears, clearly described by Nussbaum in his first paper, for his account seems to imply, in certain passages, that the ectoderm may slide over the endoderm during the process, rather than that both layers always turn together. Ischikawa, who studied the problem later, gave a clearer account of the method of turning back. Nussbaum has stated in a later paper that he had described essentially the same process.

In conclusion, it can be definitely stated that a transformation of ectoderm into endoderm cannot take place in hydra. Ischikawa also tried removing the endoderm from a piece by spreading it out and then killing the inner layer by weak acid applied with a brush, but pieces of this sort failed to regenerate a new endoderm.

Tower has recently stated that if a living hydra is put into a strong light from an arc lamp of 52 volt 12 ampere capacity, that is focussed on the animal (after passing through an alum cell), the ectoderm cells fly off, but if the animal is kept, it subsequently produces a new ectoderm. Whether all the ectoderm is lost, or only the larger neuro-muscular cells, was not made out.

One of the most unexpected discoveries of recent times in connection with the problem of regeneration is the renewal of the extirpated eye of triton and salamandra. Colucci first discovered in 1891 that if the eye is partially removed a new eye develops from the piece that remains and that the new lens develops from the margin of the bulb. Wolff, a few years later, not knowing of Colucci’s results, also found that after extirpation of the lens of triton, by making an incision in the cornea, a new lens develops from the edge of the old iris. Wolff pointed out the great theoretical importance of this result. The experiment has been repeated and confirmed by a number of more recent workers, so that there remains no question as to its accuracy.

After the removal of the old lens the wound in the cornea quickly heals, and in the course of two or three weeks a thickening appears at one point at the edge of the iris (Fig. 60, A). The cells that produce this thickening are the ordinary deeply pigmented cells of the iris, where the outer layer of cells of the iris becomes continuous with the inner layer. The cells increase in number and produce a spheroidal ball that hangs down into the space formerly occupied by the lens (Fig. 60, E). The cells become clearer by absorbing their pigment and arrange themselves concentrically as in the normal lens. When fully formed the new lens separates from the iris and occupies the normal position.

The most surprising fact in connection with the development of the new lens is that it arises from a part of the body from which the lens of the eye never develops in the embryo of this form or of any other vertebrate. In the embryo the lens develops from the ectoderm at the side of the head and only secondarily unites with the optic cup, that has come from an evagination of the anterior wall of the fore brain. In the regeneration of the adult lens, however, the ectoderm covering the eye takes no part in the formation of the new lens,—in fact, it is separated from the eye by the thick inner, mesodermal layer of the cornea. The lens develops, as has been stated, from the already differentiated layers of the iris. It is a point of further interest to notice that the cells that form the transparent lens come from the iris cells that are in part at least filled with black pigment. If this pigment remained in the cells the new lens, while it might be structurally perfect, would be physiologically useless. The pigment disappears, however, as the lens develops. In this case we find a highly specialized organ, the lens, developing out of tissue also specialized in another direction. It does not simplify the problem to point out that the lens and the iris are both parts of the eye, since they have arisen from different parts of the body and have only secondarily come into apposition with each other. Colucci was contented to point out that both the embryonic lens and the regenerated one come from ectoderm and that the result can be brought into harmony with the “germ layer” hypothesis.

Wolff has called attention to the fact that the new lens arises from the upper edge of the iris, and that this is obviously the most advantageous position in which it could develop from the iris, since by its own weight it falls into place as it develops. If the lens had developed from any other point of the margin, its position would be less advantageous, as it might not be brought into its proper position.

Fischel, who has more recently studied the regeneration of the lens in the larvÆ of Salamandra maculata, finds that after the removal of the lens the iris is thrown into wrinkles or folds and may stick at first to the cut-edge of the cornea. After the cornea has healed, the iris returns to its normal position. He finds that the first changes are more or less alike around the entire rim of the iris and involve a partial absorption of the pigment, a separation of the inner and outer layers at the edge, and a swelling of the margin. These changes go only a little way in those parts that do not produce a lens, but at the upper edge of the iris they go farther and lead to the formation of a lens in that region. He finds also that a new lens develops in animals kept in the dark as well as in those kept in the light, and in the same way.

Fischel also tried the effect of removing a part of the upper edge of the iris at the time when the lens was extirpated, in order to see if, in the absence of this part, the lens would develop from other parts of the uninjured margin of the iris. He found that the new lens still comes from the upper edge of the iris from the part left after the operation and not from the intact edge in other parts. This seemed to show that an injury to the iris is in itself a stimulus that starts the formation of a lens. This conclusion is made probable by the results of other experiments in which the iris was stuck at several points, when new lenses began to develop at several of these regions of injury. In some cases Fischel found that two or more lenses began to develop when the iris had not been intentionally injured; but it is not improbable that some sort of injury may have been effected when the lens was removed. Fischel, as has been said, removed extensive portions of the upper part of the iris and found that a new lens could be formed at the cut-edge, even in the region of the pars ciliaris; and, even after the removal of the entire upper part of the iris, lens-like structures may appear in the inner or retinal layer of the remaining region.

If instead of removing the lens it is displaced by pressing on the cornea until the lens leaves its normal position and comes to lie in the vitreous humor, a new lens develops from the edge of the iris, as though the old lens had been entirely removed from the eye, but in the experiments in which this was done the new lens was not well developed. The result shows that it is not necessary that the old lens be removed from the eye in order to induce the regeneration of a new one, but only that the lens lose its normal position in the eye.

In regard to the stimulus that determines the development of the lens, Fischel agrees with Wolff that gravity has a share in producing the result. The absence of the old lens from its normal position, as well as the wrinkling of the cornea, may also enter in as factors. Fischel takes issue with Wolff as to the interpretation of the result as an adaptation, and states that “the organism always responds to a change of relation in only one way, whose direction is already determined by internal structural relations, without regard to whether the result is adaptive or not. The response follows each stimulus in a way determined by the limited possibilities of the cells. With such a uniformity in the reaction, the idea of a fundamental adaptability cannot be connected, since the reaction that appears to us to be adaptive in a series of complicated changes may be non-adaptive in another series.”

Whether Fischel has here really met Wolff’s argument is, I think, open to question. It does not alter the result to show that factors already existing enter into the process, so long as the organism is so constructed that just those factors are present that bring about a useful response. That the response may be sometimes imperfect does not affect seriously the argument—in fact, it makes the case all the more remarkable if these imperfect attempts are in the direction of useful responses. Fischel sums up his conclusions as follows: “It is not necessary, and it is irreconcilable with the facts, to describe the formation of the lens in a teleological sense, and to bring this case forward as a proof of the universal application of a teleological principle. As has been already stated, the facts in regard to this case show much more clearly that the organism reacts to each change always in a manner that corresponds to its limited possibilities without regard to a teleological principle. A planarian, for instance, responds to a stimulus and makes a new head, even when it possesses one or more already; a tubularian produces a hydranth at its basal end, if this end is freely surrounded by water; an actinian forms a new mouth on the side of its body, etc.; so also do the cells of the pars ciliaris, and the pars iridica retinÆ differentiate into lens fibres. Working blindly, without respect to the consequences as far as they concern the whole, the one thing only is produced for which the conditions are present that bring about its formation in the cells.”

THE PART PLAYED BY THE “GERM-LAYERS” IN REGENERATION

Our examination of the origin of the tissues and organs in the new parts has shown that in most cases the old tissues give rise to the same kind of tissue in the new part; or in some other cases, as in the nervous system, the regenerating organs arise from the same “layer” as that from which they develop in the embryo. These facts have led many writers to state that the tissues and organs in the regenerated part arise from the same germ-layers as do the same parts in the embryo. It is supposed that ectoderm gives rise to ectoderm, and to those structures that arise from the ectoderm in the embryo, as, for instance, the nervous system, stomodÆum, etc. The endoderm is supposed to give rise to endoderm, and to endodermal structures, and the mesoderm to mesoderm and its derivates. So fixed has this opinion become that it is not uncommon to find investigators proclaiming the triumphant success of their results, because they have been able to trace the organs in the regenerated part to the same germ-layers that give rise to these organs in the embryo. Before deciding as to the value of this point of view, let us examine briefly the foundations of the so-called germ-layer hypothesis.

The origin of this hypothesis goes back at least to 1759, when C. F. Wolff maintained his thesis that the digestive tract of the chick exists as a flat, leaf-like structure that subsequently rolls up into a tube. He thought it probable that other embryonic organs might arise in the same way. His views made at the time no impression on his contemporaries, and lay buried until 1812, when Meckel republished Wolff’s work in a German translation. Pander, in 1817, distinguished two layers in the early embryo, a serous and a mucous, and stated that later a third, vascular layer appears between the other two. Von Baer published in 1829 his celebrated memoir on the development of the chick, in which he made out two primary layers in the germ, the animal and the vegetative layer, and held that each of these separates into two to produce the four embryonic layers. Remak, in 1851-1855, gave a more precise description of the germ-layers, and stated that from the innermost layer, the epithelium and glandular cells of the digestive tract arise (including the lining of the glands that open into the digestive tract). From the outermost layer he showed that the integument and sense organs and the nervous system develop, and from the two middle layers develop the muscles, blood, excretory, and reproductive organs. By the term “germ-layers” was meant at this time only that the embryo is formed out of sheets.

Huxley in 1849 pointed out that a medusa is made up of two layers, an outer and an inner, and called attention to their possible equivalency to von Baer’s serous and mucous layers. This idea of a resemblance between the layers of an embryo and of an adult of a lower form furnished the starting-point for the more modern formulation of the germ-layer hypothesis. Kowalevsky’s work on the development of a number of the lower animals showed that there is present in many forms a two-layered stage, or gastrula, formed by an in-turning of the wall of the hollow blastula. In this way two germ-layers are established, an outer and an inner, that correspond to the ectoderm and to the lining of the digestive tract, or endoderm. While Kowalevsky’s work did much toward laying the foundation of the modern study of embryology, he himself indulged in very little of the sort of speculation that came into vogue a few years later. Kowalevsky’s discovery of the gastrula stage in the embryos of many different groups has been fully confirmed and extended, but the elaborate speculations that have been built up on this as a basis have gone far beyond the evidence, and, for a time, drew the attention of embryologists away from more important problems. Haeckel took a more extreme position than most of his contemporaries, and assumed that the gastrula stage that occurs in so many of the groups of metazoa corresponds to an ancestral, two-layered adult animal, the gastrÆa, from which all the higher forms have descended. The presence of the gastrula in the development was interpreted as a “repetition” of this ancestral adult stage. Thus the two primary layers are supposed to have an historical meaning.[100] Embryologists soon began a search for a similar mode of interpreting the middle germ-layer, or layers, which led, amongst other views, to the formulation of the “gut-pouch hypothesis.” From this point of view the body cavities, or coelomes, are supposed to have been originally sac-like outgrowths from the digestive tract of an ancestral adult animal. Later, these coelome sacs are supposed to have been shut off from connection with the digestive tract—their cavities becoming the body cavities, and their walls giving rise to the mesodermal organs. The formation of pouches from the walls of the archenteron of the embryo in several groups of animals has been interpreted as a repetition of the ancestral adult animal.

A comparison of the germ-layers in different forms very soon led to an attempt to “homologize” the layers in different animals. If the layers have had historically the same origin, or appear in the same way in the embryos, or give rise to the same organs, they are said to be homologous. In the absence of a knowledge of the first two of these conditions it is generally considered sufficient, if it can be shown that similar organs arise from a layer, to “homologize” that layer in the two forms. The study of embryology soon became a search for homologies. The results led to inextricable difficulties and innumerable contradictions until, a reaction setting in, many embryologists became sceptical in regard to the value of this entire method of study.

The results of a detailed study of the process of cleavage in a number of groups have helped, perhaps, to clear the way for a sounder conception. It has been found that the cleavage of the egg in members of the groups of annelids, mollusks, and turbellarians is extremely similar—so similar, in fact, that it seems hardly possible that they could be due to chance, especially as the series of cleavages is quite complicated. The discovery of these similarities led at once to comparison, and comparison to the establishment once more of homologies, and the homologies led again to contradictions, until at present scarcely any two workers agree as to a criterion of homology.[101] Leaving this question aside, however, and fixing our attention only on the similarity of the process of cleavage, we are justified, I think, in looking for an explanation of the similarity in some sort of an historical connection. We can eliminate, I think, without discussion the possibility of this type of cleavage representing an ancestral adult animal. So far as the question of descent enters the problem, we can infer with some degree of probability that the groups in question may have come from a common group in which the egg divided in much the same way as we find it dividing at the present time. As a formal hypothesis this view meets with no serious difficulty, since a chain of forms, or a continuous living substance, connects the present animals with those living in the past; and we may assume that the same factors peculiar to the egg of the ancestors are still present in the eggs of their descendants. This sort of explanation gives us no causal knowledge of the way in which the egg divides, nor does it preclude the possibility of new changes coming in that may entirely alter the form of the cleavage. Moreover, since we are dealing with a question of historical probability only, we cannot be certain that the same type of cleavage may not have arisen quite independently in each group.

The argument in favor of the gastrula stage also representing an ancestral larval stage may be admitted as a remote possibility, but on evidence even far less satisfactory than that for the similarities of cleavage being accounted for by a common descent. That this gastrula was ever an adult form we have no means of deciding, even as a matter of probability, and even if this could be made plausible it by no means follows that such an adult stage would become an embryonic stage of later forms. Consequently that part of the germ-layer theory that rests on such a supposed connection cannot be looked upon as much more than a fiction.

But even granting that there is an historical, embryonic[102] connection, its small importance for the scientific problems connected with embryonic development, and budding and regeneration has been shown by a number of recent discoveries, and nowhere more clearly than in the cases of the formation of new individuals by budding. As an example may be cited the method of development of the ascidian from the egg, and by means of buds. The work of Kowalevsky, Della Valle, Seeliger, and Van Beneden on the budding process of ascidians showed that there are some discrepancies between the bud development and the embryonic development. The more recent papers of Hjort, Oka, Pizon, Salensky, Lefevre, and others have shown very clearly that the germ-layer theory is inapplicable to the bud development in this group. The bud arises as a double-walled tube, or rather a tube within a tube, with a space between. The outer tube comes in all cases from the ectoderm of the animal; the inner tube has a different origin in different species. In perophora, didemnum, and clavellina, the inner tube comes from endoderm; in botryllus it arises from the ectoderm of the larval peribranchial or atrial cavity. In all these forms the inner tube gives rise to the new pharyngeal cavity of the bud, while this same cavity comes from the endoderm of the archenteron of the embryo. In the bud embryo the peribranchial space is also derived from the inner tube; hence it is endodermal in the first series, and ectodermal in botryllus. In the egg embryo it is ectodermal. In regard to the development of the nervous system there is some difference of opinion. A number of investigators have found that the new brain arises from the outer part of the inner or branchial tube, which has in most cases an endodermal origin. Seeliger and Lefevre believe the nervous system to arise from mesodermal cells that lie between the two tubes. It appears, nevertheless, that in several forms the brain really comes from the inner tube, which also gives rise to the branchial sac. Therefore, in those cases in which the inner tube is endodermal the brain has the same origin, and in the case in which the inner tube is ectodermal, the brain is ectodermal, but the pharyngeal sac has also an ectodermal origin. There is obviously no definite relation between the origin of these structures in the bud and in the egg embryo.

A similar difficulty is met with in the Bryozoa in regard to the development of the egg embryo and the bud embryo.

Braem, who has made a critical examination of the germ-layer theory,[103] has found it impossible to give a morphological definition of a germ-layer, and has adopted a physiological criterion. He thinks that in whatever way a germ-layer arises, whether by folding, or by delamination, etc., it exists independently of its method or place of origin. A layer is not endodermal because it forms the inner wall of a gastrula, but it is endodermal because it develops into the digestive tract. The germ-layers of different forms are only similarly placed, but whether they are homologous will depend on other things. On this view the inner tube of the ascidian bud that gives rise to both digestive tract and to the nervous system is simply an indifferent layer until it gives rise to these structures. Its cells may be looked upon as indifferent, as are those of the blastula. Thus the difficulty of the morphologist is not solved, but the knot is cut. For Braem the germ-layers are convenient terms, since he rejects any historical significance that they may have, and it is just this side of the question that the morphologist has attempted to work out. While the evidence shows that the germ-layers cannot have any such final attributes as embryologists have attempted to assign to them, and that Braem has called attention to the real and important problems connected with the study of development, yet it may still be admitted without endangering the newer point of view, that there may be also an historical question in connection with the germ-layers, if not in the sense of a repetition of an ancestral adult gastrÆa, yet in the sense that similarity in embryonic development may in some cases find its historical explanation in a common descent.

If in the light of this discussion we turn to the phenomena of regeneration, we again find evidence showing that the germ-layer theory fails to apply in all cases. It has been pointed out that in lumbriculus, and in the naids, the new mesoderm is derived from the ectoderm, and does not come from the old mesodermal tissues. The mesoderm of the embryo in annelids is derived from one, and later from two, superficial cells of the blastula,[104] that push in about the time of gastrulation. They cannot, at this time, be referred to one layer rather than to the other. It cannot be affirmed, therefore, that in regeneration, the mesoderm arises from a different layer from that in the embryo, but neither can this be denied. The most important point in this connection is that the new mesoderm comes from the ectoderm that is already differentiated, and not from the mesodermal tissues. It is clear, however, that while the lining of the pharynx in the embryo is ectodermal, it is endodermal in the regenerated part.

It is true that these cases are very exceptional, and that generally the new organs come from similar organs in the old part, but one established exception is sufficient to show that the traditional conception of the germ-layers may be of little value, and since the hypothesis itself, out of which the idea in regard to regeneration from definite germ-layers has been formed, has been proven to be insufficient in other directions, the time is ripe to look for a more secure footing. It need hardly be added that the idea of a supposed necessity for an organ to arise from a definite germ-layer is so empty of all significance that we may well rejoice to be able to set it aside as a naÏve view that has had its day. Furthermore, a new series of problems has arisen in connection with the experimental work to be described in a later chapter. If, as seems probable, the question of the germ-layers will be merged into the much broader question of the origin of the specification of the tissues, we can in the future more profitably direct our attention to the experimental evidence that bears on the latter question.

THE SUPPOSED REPETITION OF PHYLOGENETIC AND ONTOGENETIC PROCESSES IN REGENERATION

It has been claimed that at times ontogenetic, and even phylogenetic, processes are repeated during regeneration. Fraisse, for instance, who advocates this point of view, thinks that it has been too much neglected, and calls attention to several instances of what he believes to be cases in point. He thinks that BÜlow is correct in his comparison between the method of development of the new tissue at the end of the tail in certain naids, and the method of gastrulation and formation of the mesoderm in the embryo. Later results have shown, however, that in several points BÜlow’s observations are incorrect. The in-turning of ectoderm that BÜlow compares with the process of gastrulation is connected with the formation of the ectodermal proctodÆum, and is not comparable with the development of the endoderm in the embryo.

GÖtte also, as we have seen, cites a case of resemblance between the regeneration of the limbs of the salamander and their mode of embryonic development. He finds the resemblances less marked as the animal becomes older. The resemblance is, however, not very close and of a rather general sort, and since the same structures develop in both cases out of the same kind of substance, it is not surprising that there should be some resemblances in the processes. This evidence is counterbalanced by the mode of regeneration of the tail in the adult of certain forms, and in the regeneration of the lens of the eye from the iris.

CarriÈre finds that the eye of snails regenerates from the ectoderm in much the same way as the young eye develops. Granted that the eye is to come from the ectoderm in both cases, and that the same structure develops, it is not to be wondered at that the two processes have much in common.

The mistake, I think, is not in stating that the two processes are sometimes similar, or even identical, but in stating the matter as though the regenerative process repeats the embryonic method of development. If the same conditions prevail, then the same factors that bring about the embryonic development may be active in bringing about the regenerative processes. In fact, we should expect them to coincide oftener than appears to be the case, but this may be due to the conditions being different in the young and in the adult.

It has been claimed also that in some cases there is regenerated a structure like that possessed by the ancestors of the animal. The stock example of this process is Fritz MÜller’s result on the regeneration of the claw of a shrimp, Atypoida protimirum.[105] Fraisse and Weismann and others have brought forward this case as demonstrative. The animal is said to regenerate a claw different from any of those in the typical form, and one that resembles the claw of another related genus, Carodina. The value of evidence of this sort is not above question. Przibram has shown in other crustacea that when a maxilliped is cut off a structure different in kind often regenerates, but that after several months the typical structure returns. Do we find here an ancestral organ that first appears, and then gives way to its more modern representative? If it resembled the maxilliped of any other crustacean, the evidence would, no doubt, be accepted by those who accept the evidence furnished by MÜller. What then shall we say to the case, first discovered by Herbst, in which the eye of certain prawns being cut off, an antenna-like organ regenerates? Since these antennÆ are similar to those possessed by the same animal, shall we assume that it once had antennÆ in place of eyes?

Another comparison, that Fraisse has made, is worth quoting as showing how far credulity may be carried. In the regeneration of the tail of certain lizards pigment first appears in the ectoderm of the new part and then sinks deeper into the layers. Fraisse found a lizard on Capri in which the tail is pigmented throughout life, and although he did not know whether or not the pigment is in the skin he suggests that this lizard represents an ancestral condition, that is repeated by the regenerating tails of other forms.

Boulenger (’88) pointed out that the scales over the regenerated tail of several lizards have a different arrangement from that of the normal tail, and furthermore, the new arrangement is sometimes like that found in other species. He claims that this shows that such forms are related, even where no evidence of their relation is forthcoming. That the conditions in the new tail may be different from those in the normal tail is shown by the absence of a vertebral column, etc.; therefore that the scales also should have a new arrangement is not surprising, but the facts fail, I think, to show that there need be any genetic relation between the forms in question. That the conditions in the new tail might be like those in an ancestral form may be admitted, but this is very different from assuming that the results show a genetic relation actually to exist. The main point is that, even if the results should be nearly identical, it may be entirely misleading to infer that ancestral characters have reappeared.

In some cases an extra digit or toe may regenerate on the leg of a salamander, and this too has been interpreted as a return to an ancestral condition. But Tornier has shown, as has been stated, that several additional digits, or even a whole extra hand, may be produced by wounding the leg in certain ways, and these too would have to be interpreted as ancestral, if the hypothesis is carried out logically. It has been shown by King that one or more additional arms may be produced in a starfish by splitting between the arms already present, and if we accepted evidence of this sort as having any value in interpreting lines of descent we should conclude[106] that the ancestors of the starfish had six, seven, or more arms according to the number that can be produced artificially, etc. Therefore, until further evidence of a more convincing kind is forthcoming, we can safely, I think, decline to accept the results, so far known, as having any value in interpreting the relationships or the descent of the animals.