Fig. 9. Osmotic growths simulating segmentation. (After Leduc.)

In some respects perhaps the best models of living organisms yet made are the "osmotic growths" produced by Leduc.[1] These curious structures were formed by placing a fragment of a salt, for instance calcium chloride, in a solution of some colloidal substance. As the solid takes up water from the solution a permeable pellicle or membrane is formed around it. The vesicle thus enclosed grows by further absorption of water, often extending in a linear direction, and in many examples this growth occurs by a series of rhythmically interrupted extensions. Some of the growths thus formed are remarkably like organic structures, and might pass for a series of antennary segments or many other organs consisting of a linear series of repeated parts. In admitting the essential resemblance between these "osmotic growths" and living bodies or their organs I lay less stress on the general conformation of the growths, which often as Leduc points out, recall the forms of fungi or hydroids, but rather on the fact that the interruptions in the development of these systems are so closely analogous to the segmentations or repetitions of parts characteristic of living things (Fig. 9). In the same way I am less impressed by Leduc's models of Karyokinesis, wonderful as they nevertheless are, for the division is here imitated by putting separate drops on the gelatine film. What we most want to know is how in the living creature one drop becomes two. The models of linear segmentation have the remarkable merit that they do in some measure imitate the process of actual division or repetition. So in a somewhat modified method Leduc, by causing the diffusion of a solution in a gelatine film, produced rhythmical or periodic precipitations strikingly reminiscent of various organic tissues, for here also the process of periodic repetition is imitated with success.

It is a feature common to these and to all other rhythmical repetitions produced by purely mechanical forces that there is resemblance between the members of the series, and that this similarity of conformation may be maintained in most complex detail. When however in the mechanical series some of the members differ from the rest we have no difficulty in recognising that these differences—which correspond with the differentiations of the organic series—are due to special heterogeneity in the conditions or in the materials, and it never occurs to us to suppose that all the members must have been primordially alike. For example, in the case of ripple-marks on the sand, which I choose as one of the most familiar and obvious illustrations of a repeated series due to mechanical agencies, if we notice one ripple different in form from those adjacent to it, we do not suppose that this variation must have been brought about by deformation of a ripple which was at first formed like the others, but we ascribe it to a difference in the sand at that point, or to a difference in the way in which the wind or the tide dealt with it. We may press the analogy further by observing that in as much as such a series of waves has a beginning and an end, it possesses polarity like that of the various linear series of parts in organisms, and even the formation of each member must influence the shape of its successor. Since in an organism the beginning and end of the series are always included, some differentiation among the repetitions must be inevitable. If therefore it be conceded, as I think it must, that segmentation and pattern are the consequence of a periodic process we realize that it is at least as easy to imagine the formation of such a series of parts having family likeness combined with differentiation as it would be to conceive of their arising primordially as a series of identical repetitions. The suggestion that the likenesses which we now perceive are the remains of a still more complete resemblance is a substitution of a more complex conception for a simpler one.

The other question raised by the problem of Serial Homology is how far there is a correspondence between individual members of series when the series differ from each other either in the number of parts, or in the mode of distribution of differentiation among them. Students, for example, of vertebrate morphology debate whether the nth vertebra which carries the pelvic girdle in Lizard A is individually homologous with the n + xth vertebra which fulfils this function in Lizard B, or whether it is not more truly homologous with the vertebra standing in the nth ordinal position, though that vertebra in Lizard B is free.

In various and more complex aspects the same question is debated in regard to the cranial and spinal nerves, the branches of the aorta, the appendages of Arthropoda, and indeed in regard to all such series of differentiated parts in linear or successive repetition. Persons exercised with these problems should before making up their minds consider how similar questions would be answered in the case of any series of rhythmical repetitions formed by mechanical agencies. In the case of our illustration of the ripples in the sand, given the same forces acting on the same materials in the same area, the number of ripples produced will be the same, and the nth ripple counting from the end of the series will stand in the same place whenever the series is evoked. If any of the conditions be changed, the number and shapes can be changed too, and a fresh "distribution of differentiation" created. Stated in this form it is evident that the considerations which would guide the judgment in the case of the sand ripples are not essentially different from those which govern the problem of individual homology in its application to vertebrae, nerves, or digits.

The fact that the unit of repetition is also the unit of growth is the source of the obscurity which veils the process. When we compare the skeleton of a long-tailed monkey with that of a short-tailed or tailless ape we see at once how readily the additional series of caudal segments may be described as a consequence of the propagation of the "waves" of segmentation beyond the point where they die out in the shorter column, and we see that with an extension of the series of repetitions there is growth and extension of material.

The considerations which apply to this example will be found operating in many cases of the variation of terminal members of linear series. Some of these series, like the teeth of the dog, end in a terminal member of a size greatly reduced below that of the next to it. Even when there is thus a definite specialisation of the last member of the series it not infrequently happens that the addition, by variation, of a member beyond the normal terminal, is accompanied by a very palpable increase in size of the member which stands numerically in the place of the normal terminal.[2] So also with variation in the number of ribs, when a lumbar vertebra varies homoeotically into the likeness of the last dorsal and bears a rib, the rib placed next in front of this, which in the normal trunk is the last, shows a definite increase in development.

The consequences of such homoeoses are sometimes very extensive, involving readjustments of differentiation affecting a long series of members, as may easily be seen by comparing the vertebral columns of several individual Sloths[3] (whether Bradypus or Choloepus) to take a specially striking example.

It may be urged that no feature as yet enables us to perceive wherein lies the primary distinction which determines such variation, whether it is due to a difference in the dividing forces or in the material to be divided. If for instance we were to imitate such a series of segments by pressing hanging drops of a viscous fluid out of a paint-tube by successive squeezes, the number of times the tube is contracted before it is empty will give the number of the segments, but their size may depend either on the force of the contractions or on the capacity of the tube, or on various other factors. Nevertheless in the case of the variation of terminal members, whatever be the nature of the rhythmical impulse which produces the series of organs, the elevation of the normally terminal member in correspondence with the addition of another is what we should expect.

If the organism acquired its full size first and the delimitation of the parts took place afterwards, there might be some hope that the resemblance between living patterns and those mechanically caused by wave-motion might be shown to be a consequence of some real similarity of causation, but in view of the part played by growth, appeal to these mechanical phenomena cannot be declared to have more than illustrative value. Similarly in as much as living patterns appear, and almost certainly do in reality come into existence by a rhythmical process, comparisons of these patterns with those developed in crystalline structures, and in the various fields of force are, as it seems to me, inadmissible, or at least inappropriate.

However their intermittence be determined, the rhythms of division must be looked upon as the immediate source of those geometrically ordered repetitions universally characteristic of organic life. In the same category we may thus group the segmentation of the Vertebrates and of the Arthropods, the concentric growth of the Lamellibranch shells or of Fishes' scales, the ripples on the horns of a goat, or the skeletons of the Foraminifera or of the Heliozoa. In the case of plant-structures Church[4] has admirably shown, with an abundance of detail, how on analysis the definiteness of phyllotaxis is an expression of such rhythm in the division of the apical tissues, and how the spirals and "orthostichies" displayed in the grown plant are its ultimate consequences. The problem thus narrows itself down to the question of the mode whereby these rhythms are determined.

It is natural that we should incline to refer them to a chemical source. If we think of the illustration just given, of the segmentation of a viscous fluid into drops by successive contractions of a soft-walled tube we can, I think, conceive of such rhythmic contractions as due to summations of chemical stimuli, somewhat as are the beats of the heart. But when we recognize the vast diversity of materials the distribution of which is determined by an ostensibly similar rhythmic process it seems hopeless to look forward to a directly chemical solution. That the chemical degradation of protoplasm or of materials which it contains is the source of the energy used in the divisions cannot be in dispute, but that these divisions can be themselves the manifestations of chemical action seems in the highest degree improbable.

We may therefore insist with some confidence on the distinction between the Meristic and the substantive constitution of organisms, between, that is to say, the system according to which the materials are divided and the essential composition of the materials, conscious of the fact that the energy of division is supplied from the materials, and that in the ontogeny the manner in which the divisions are effected must depend secondarily on the nature of the substances to be divided. The mechanical processes of division remain a distinguishable group of phenomena, and variations in the substances to be distributed in division may be independent of variations in the system by which the distribution is effected.

Modern genetic analysis supplies many remarkable examples of this distinction. When formerly we compared the leaves of a normal palmatifid Chinese Primula with the pinnatifid leaves[5] of its fern-leaved variety we were quite unable to say whether the difference between the two types of leaf was due to a difference in the material cut up in the process of division or to a difference in that process itself. Knowledge that the distinction is determined by a single segregable factor tends to prove that the critical difference is one of substance. So also in the Silky fowl we know that the condition of its feathers is due to the absence of some one factor present in the normal form. We may conceive such differences as due to change of form in the successive "waves" of division, but we cannot yet imagine segregation otherwise than as acting by the removal or retention of a material element. Future observation by some novel method may suggest some other possibility, but such cases bring before us very clearly the difficulties by which the problem is beset.

Fig. 10. The palm-and fern type of leaf in Primula Sinensis.
The palm is dominant and the fern is recessive.

In another region of observation phenomena occur which as it seems to me put it beyond question that the meristic forces are essentially independent of the materials upon which they act, save, in the remoter sense, in so far as these materials are the sources of energy. The physiology of those regenerations and repetitions which follow upon mutilation supplies a group of facts which both stimulate and limit speculation. No satisfactory interpretations of these extraordinary occurrences has ever been found, but we already know enough to feel sure that in them we are witnessing indications which should lead to the discovery of the true mechanics of repetition and pattern. The consequences of mutilation in causing new growth or perhaps more strictly in enabling new growth to take place, are such that they cannot be interpreted as responses to chemical stimuli in any sense which the word chemical at present connotes. Powers are released by mutilation of which in the normal conditions of life no sign can be detected. All who have tried to analyse the phenomena of regeneration are compelled to have recourse to the metaphor of equilibrium, speaking of the normal body as in a state of strain or tension (Morgan) which when disturbed by mutilation results in new division and growth. The forces of division are inacessible to ordinary means of stimulation. Applications, for example, of heat or of electricity excite no responses of a positive kind unless the stimuli are so violent as to bring about actual destruction.[6] These agents do not, to use a loose expression, come into touch with the meristic forces. Changes in the chemical environment of cells may, as in the experiments of Loeb and of Stockard produce definite effects, but the facts suggest that these effects are due rather to alterations in the living material than to influence exerted directly on the forces of division themselves.

By destruction of tissue however the forces both of growth and of division also may often be called into action with a resulting regeneration. Interruption of the solid connexion between the parts may produce the same effects, as for example when the new heads or tails grow on the divided edges of Planarians (Morgan), or when from each half embryo partially separated from its normally corresponding half, a new half is formed with a twin monster as the result.

Often classed with regenerations but in reality quite distinct from them are those special and most interesting examples where the growth of a paired structure is excited by a simple wound. Some of the best known of these instances are presented by the paired extra appendages of Insects and Crustacea. Some years ago I made an examination of all the examples of such monstrosities to which access was to be obtained, and it was with no ordinary feeling of excitement that I found that these supernumerary structures were commonly disposed on a recognizable geometrical plan, having definite spatial relations both to each other and to the normal limb from which they grew. The more recent researches of Tornier[7] and especially his experiments on the Frog have shown that a cut into the posterior limb-bud induces the outgrowth of such a pair of limbs at the wounded place. Few observations can compare with this in novelty or significance; and though we cannot yet interpret these phenomena or place them in their proper relations with normal occurrences, we feel convinced that here is an observation which is no mere isolated curiosity but a discovery destined to throw a new light on biological mechanics. The supernumerary legs of the Frog are evidently grouped in a system of symmetry similar to that which those of the Arthropods exhibit, and though in Arthropods paired repetitions have not been actually produced by injury under experimental conditions we need now have no hesitation in referring them to these causes as Przibram has done.

At this point some of the special features of the supernumerary appendages become important. First they may arise at any point on the normal limb, being found in all situations from the base to the apex. Nor are they limited as to the surface from which they spring, arising sometimes from the dorsal, anterior, ventral, or posterior surfaces, or at points intermediate between these principal surfaces.

With rare and dubious exceptions, the parts which are contained in these extra appendages are only those which lie peripheral to their point of origin. Thus when the point of origin is in the apical joint of the tarsus, the extra growth if completely developed consists of a double tarsal apex bearing two pairs of claws. If they arise from the tibia, two complete tarsi are added. If they spring from the actual base of the appendage then two complete appendages may be developed in addition to the normal one. We must therefore conclude that in any point on a normal appendage the power exists which, if released, may produce a bud containing in it a paired set of the parts peripheral to this point.

Fig. 11. Diagrams of the geometrical relations which are generally exhibited by extra pairs of appendages in Arthropoda. The sections are supposed to be those of the apex of a tibia in a beetle. A, anterior, P, posterior, D, dorsal, V, ventral. M¹, M² are the imaginary planes of reflexion. The shaded figure is in each case a limb formed like that of the other side of the body, and the outer unshaded figures are shaped like the normal for the side on which the appendages are. On the several radii are shown the extra pairs in their several possible relations to the normal from which they arise. The normal is drawn in thick lines in the center.

Next the geometrical relations of the halves of the supernumerary pair are determined by the position in which they stand in regard to the original appendage. These relations are best explained by the diagram (Fig. 11), from which it will be seen that the two supernumerary appendages stand as images of each other; and, of them, that which is adjacent to the normal appendage forms an image of it. Thus if the supernumerary pair arise from a point on the dorsal surface of the normal appendage, the two ventral surfaces of the extra pair will face each other. If they arise on the anterior surface of the normal appendage, their morphologically posterior surfaces will be adjacent, and so on.

These facts give us a view of the relations of the two halves of a dividing bud very different from that which is to be derived from the exclusive study of normal structures. Ordinary morphological conceptions no longer apply. The distribution of the parts shows that the bud or rudiment which becomes the supernumerary pair may break or open out in various ways according to its relations to the normal limb. Its planes of division are decided by its geometrical relations to the normal body.

Especially curious are some of the cases in which the extra pair are imperfectly formed. The appearance produced is then that of two limbs in various stages of coalescence, though in reality of course they are stages of imperfect separation. The plane of "coalescence" may fall anywhere, and the two appendages may thus be compounded with each other much as an object partially immersed in mercury "compounds" with its optical image reflected from the surface.

Supernumerary paired structures are not usually, if ever, formed when an appendage is simply amputated. Cases occasionally are seen which nevertheless seem to be of this nature. Borradaile,[8] for example, described a crab (Cancer pagurus) having in place of the right chela three small chelae arising from a common base, where the appearances suggested that the three reduced limbs replaced a single normal limb. From the details reported however it seems still possible that one of the chelae (that lettered F. I in Borradaile's figure) may be the normal one, and the other two an extra pair. The chela which I suspect to be the normal is in several respects deformed as well as being reduced in size, and this deformity may perhaps have ensued as a consequence of the same wound which excited the growth of the extra pair. Its reduced size may be due to the same injury, which may quite well have checked its growth to full proportions.

Admitting doubt in these ambiguous cases it seems to be a general rule that for the production of the extra pair the normal limb should persist in connexion with the body. Moreover it is practically certain that in no case can a single, viz. an unpaired, duplicate of the normal appendage grow from it. Many examples have been described as of this nature, but all of them may be with confidence regarded as instances of a supernumerary pair in which only the two morphologically anterior or the two morphologically posterior surfaces are developed. We have thus the paradox that a limb of one side of the body, say the right, has in it the power to form a pair of limbs, right and left, as an outgrowth of itself, but cannot form a second left limb alone.

A very interesting question arises whether it is strictly correct to describe the extra pair as a right and a left, or whether they are not rather two lefts or two rights of which one is reversed. This question did not occur to me when in former years I studied these subjects. It was suggested to me by Dr. Przibram. The answer might have an important bearing on biological mechanics, but I know no evidence from which the point can be determined with certainty. In order to decide this question it would be necessary to have cases in which the paired repetition affected a limb markedly differentiated on the two sides of the body, and of course the development of the extra parts in order to be decisive must be fairly complete. One example only is known to me which at all satisfies these requirements, that of the lobster's chela figured (after Van Beneden) in Materials for the Study of Variation, p. 531, Fig. 184, III.

Here the drawing distinctly suggests that one of the extra dactylopodites, namely that lettered R, is differentiated as a left and not merely a reversed right. For the teeth on this dactylopodite are those of a cutting claw, not of a crushing claw, whereas the dactylopodites R' and L' bear crushing teeth. The figure makes it fairly certain also that the limb affected was a crushing claw. Accepting this interpretation, we reach the remarkable conclusion that the bud of new growth consisted of halves differentiated into cutter and crusher as the normal claws are, and that the extra crusher is geometrically a left but physiologically a right. Though shaped as a left in respect of the direction in which it points, the extra crusher is really an optically reversed right, while the dactylopodite R, which is placed pointing like a right, is really a reversed left (Fig. 12).

Fig. 12. Right claw of lobster bearing a pair of extra dactylopodites (after van Beneden). The fine toothing on R suggests that this is part of a cutting claw, though the limb bearing it is a crusher.

If these indications are reliable[9] and are established by further observation we shall be led to the conclusion that the bud which becomes an extra pair of limbs does not merely contain the parts proper to the side on which it grows, but is comparable with the original zygotic cell, and consists not simply of two halves, but of two halves differentiated as a right and a left like the two halves of the normal body.

Phenomena of this kind, evoked by mutilation or injury, together with the cognate observations on regeneration throw very curious lights on the nature of living things. To an understanding of the nature of the mechanics of living matter and its relation to matter at large they offer the most hopeful line of approach. I allude especially to the examples in which it has been established that the part which is produced after mutilation is a structure different from that which was removed. The term "regeneration" was introduced before such phenomena were discovered, and though every one recognizes its inapplicability to these remarkable cases, the word still misleads us by presenting a wrong picture to the mind. The expression "heteromorphosis" (Loeb) has been appropriately applied to various phenomena of this kind, and Morgan has given the name "morphallaxis" to another group of cases in which the renewal occurs by the transformation of a previously existing part.[10] But we must continually remember that all these occurrences which we know only as abnormalities and curiosities must in reality be exemplifications of the normal mechanics of division and growth. The conditions needed to call them forth are abnormal, but the responses which the system makes are evidences of its normal constitution. When therefore, for example, the posterior end of a worm produces a reversed tail from its cut end we have a proof that there must be in the normal body forces ready to cause this outgrowth. The new structure is not an ill-shaped head-end, for, as Morgan shows, the nephridial ducts have their funnels perforating the segments in a reversed direction. The "tension" of growth is actually reversed.[11] So also when in a Planarian amputation of the body immediately behind the head leads to the formation of a new reversed head at the back of the normal head, while amputation further back leads to the regeneration of a new tail, these responses give indications of forces normally present in the body of the Planarian. Such facts open up a great field of speculation and research. Especially important it would be to determine where the critical region may be at which the one response is replaced by the other. I suppose it is even possible that there is some neutral zone in which neither kind of response is made.

Physical parallels to the phenomena of regeneration are not easy to find and we still cannot penetrate beyond the empirical facts. Przibram has laid stress on the general resemblance between the new growth of an amputated part in an animal and the way in which a broken crystal repairs itself when placed in the mother-solution. That the two processes have interesting points of likeness cannot be denied. It must however never be forgotten that there is one feature strongly distinguishing the two; for I believe it is universally recognized by physicists that all the phenomena of geometrical regularity which crystals display are ultimately dependent on the forms of the particles of the crystalline body. This cannot in any sense be supposed to hold in regard to protoplasm or its constituents. The definiteness of crystals is also an unlikely guide for the reason that it is absolute and perfect, or in other words because this kind of regularity cannot be disturbed at all without a change so great that the substance itself is altered; whereas we know that the forms of living things are capable of such changes, great and small, that we must regard perfection of form, whether manifested in symmetry or in number, as an ideal which will only be produced in the absence of disturbance. The symmetry of the living things is like the symmetry of the concentric waves in a pool caused by a splash. Perfect circles are made only in the imaginary case of mathematical uniformity, but the system maintains an approximate symmetry though liable to manifold deformation.

Since the geometrical order of the living body cannot be a direct function of the materials it must be referred to some more proximate control. In renewing a part the body must possess the power of seizing particles of many dissimilar kinds, and whirl them into their several and proper places. The action in renewal, like that of original growth, may be compared—very crudely—with the action of a separator which simultaneously distributes a variety of heterogeneous materials in an orderly fashion; but in the living body the thing distributed must rather be the appetency for special materials, not the materials themselves.

If the analogy of crystals be set aside and we seek for other parallels to regeneration there are none very obvious. I have sometimes wondered whether it might not be possible to institute a fruitful comparison between the renewal of parts and the reformation of waves of certain classes after obliteration. In several respects, as I have already said, some curious resemblances with the repetitions formed by wave-motion are to be traced in our organic phenomena, and though admitting that I cannot develop these comparisons, I think nevertheless they may be worth bearing in mind. When, after obliteration, an eddy in a stream, or a ripple-mark (a more complex case of eddy-formation) in blown sand is re-formed, we have an example in which pattern is reconstituted and growth takes place not by virtue of the composition of the materials—in this case the water or the sand—but by the way in which they are acted upon by extraneous forces.

A feature in the actual mode by which ripple-marks are reconstituted may not be without interest in connexion with our phenomena of regeneration. When, for example, the wind is blowing steadily over a surface of fine, dry sand, the familiar ripple-marks are formed by a heaping of the sand in lines transverse to the direction of the wind. The heaping is due to the formation of eddies corresponding with positions of instability. When the wind is steady and the sand homogeneous, the distances between the ripples, or wave-lengths, are sensibly equal. If while the wind continues to blow, the ripples are obliterated with a soft brush they will quickly be re-formed over the whole area, but I have noticed that at first their wave-length is approximately half that of the ripples in the undisturbed parts of the system.[12] The normal wave-length is restored by the gradual accentuation of alternate ripples. Of course the sand-ripples are in reality slowly travelling forward in the direction towards which the wind is blowing, and for this our living segmentations afford no obvious parallel, but the appearances in the area of reformation, and especially the forking of the old ridges where they join the new ones, are curiously reminiscent of the irregularities of segmentation seen in regenerated structures. The value of the considerations adduced in the chapter is, I admit, very small. The utmost that can be claimed for them is that mechanical segmentations, like those seen in ripple-mark, or in Leduc's osmotic growths, show how by the action of a continuous force in one direction, repeated and serially homologous divisions can be produced having features of similarity common to those repetitions by which organic forms and patterns are characterised. The analogy supplies a vicarious picture of the phenomena which in default of one more true may in a slight degree assist our thoughts. It suggests that the rhythms of segmentation may be the consequence of a single force definite in direction and continuously acting during the time of growth. The polarity of the organism would thus be the expression of the fact that this meristic force is definitely directed after it has once been excited, and the reversal seen in some products of regeneration suggest further that it is capable of being reflected. This polarity cannot be a property of the material, as such, but is determined by a force acting on that material, just as the polarity of a magnet is not determined by the arrangement of its particles, but by the direction in which the current flows.

To some it may appear that even to embark on such discussions as this is to enter into a perilous flirtation with vitalistic theories. How, they may ask, can any force competent to produce chemical and geometrical differentiation in the body be distinguished from the "Entelechy" of Driesch? Let me admit that in this reflexion there is one element of truth. If those who proclaim a vitalistic faith intend thereby to affirm that in the processes by which growth and division are effected in the body, a part is played by an orderly force which we cannot now translate into terms of any known mechanics, what observant man is not a vitalist? Driesch's first volume, putting as it does into intelligible language that positive deduction from the facts—especially of regeneration—should carry a vivid realisation of this truth to any mind. If after their existence is realised, it is desired that these unknown forces of order should have a name, and the word entelechy is proposed, the only objection I have to make is that the adoption of a term from Aristotelian philosophy carries a plain hint that we propose to relegate the future study of the problem to metaphysic.

From this implication the vitalist does not shrink. But I cannot find in the facts yet known to us any justification of so hopeless a course. It was but yesterday that the study of Entwicklungsmechanik was begun, and if in our slight survey we have not yet seen how the living machine is to be expressed in terms of natural knowledge that is poor cause for despair. Driesch sums up his argument thus:[13]

"It seems to me that there is only one conclusion possible. If we are going to explain what happens in our harmonious-equipotential systems by the aid of causality based upon the constellation of single chemical factors and events, there must be some such thing as a machine. Now the assumption of the existence of a machine proves to be absolutely absurd in the light of the experimental facts. Therefore there can be neither any sort of a machine nor any sort of causality based upon constellation underlying the differentiation of harmonious-equipotential systems."

"For a machine, typical with regard to the three chief dimensions of space, cannot remain itself if you remove parts of it or if you rearrange its parts at will."

To the last clause a note is added as follows:

"The pressure experiments and the dislocation experiments come into account here; for the sake of simplicity they have not been alluded to in the main line of our argument."

I doubt whether any man has sufficient knowledge of all possible machines to give reality to this statement. In spite also of the astonishing results of experiments in dislocation, doubt may further be expressed as to whether they have been tried in such variety or on such a scale as to justify the suggestion that the living organism remains itself if its parts are rearranged at will. All we know is that it can "remain itself" when much is removed, and when much rearrangement has been affected, which is a different thing altogether.

I scarcely like to venture into a region of which my ignorance is so profound, but remembering the powers of eddies to re-form after partial obliteration or disturbance, I almost wonder whether they are not essentially machines which remain themselves when parts of them are removed.

Real progress in this most obscure province is not likely to be made till it attracts the attention of physicists; and though they for long may have to forego the application of exact quantitative methods, I confidently anticipate that careful comparison between the phenomena of repetition formed in living organisms and the various kinds of segmentation produced by mechanical agencies would be productive of illuminating discoveries.