BIOLOGICAL BULLETIN OF THE fIDarinc Biological Xaborator^ WOODS HOLL, MASS. JEMtorial Staff. E. G. CONK.LIN — The University of Pennsylvania. JACQUES LOEI: — The University of California. T. H. MORGAN — Columbia University. W. M. WHEELER — American Museum of Natural History, Neiv York. C. O. WHITMAN — The University of Chicago. E. B. WILSON — Columbia University. Editor, FRANK R. LILLIE — The University of Chicago. VOLUME VII WOODS HOLL, MASS. JUNE, 1904, TO NOVEMBER, 1904. ft. PnESSOf THE Ne* E»A P»I«I,NC COMPANY LANCASTER. PA, CONTENTS OF VOL. VII. No. i. JUNE, 1904. PAC.K ARTHUR W. GREELEY i ARTHUR W. GREELEY : Experiments on the Physical Structure of the Protoplasm of Paramcecium and its Relation to tlie Re- actions of the Organism to Thermal, Chemical and Elec- trical Stimuli 3 FRANK R. LILLIE : Experimental Studies on the Development of Organs in the Embryo of the Fowl (^Gallus domesticits}... 33 W. K. GREGORY : The Relations of the Anterior Visceral Arches to the Chondrocranium 55 EVERETT F. PHILLIPS : Variation in Bees : A Reply to Mr. Lutz. 70 L. R. GARY: Notes on a Peculiar Actinozoan Larva 75 No. 2. JULY, 1904. J. E. DUERDEN : 77.ii? Morphology of the Madreporaria, V. Septa! Segue nee 79 BASHFORD DEAN : Evolution in a Determinate Line as Illustrated by the Egg-cases of Chi aneroid Fishes 105 WILLIAM PATTEN : New Facts Concerning Bothriolepis 113 No. 3. AUGUST, 1904. C. M. CHILD : Form- Regulation in Cerianttuts, V. 127 ALICE M. BORING : Closure of Longitudinally Split Tubularian Stems 154 T. H. MORGAN AND A. E. SCHIEDT : Regeneration in the Plan- arian Phagocata gracilis 160 ESTHER F. BYRNES : On the Skeleton of Regenerated Anterior Limbs in the Frog '. . . 1 66 ADELE M. FIELDE : Observations on Ants in their Relation to Temperature and to Submergence '. i 70 No. 4. SEPTEMBER, 1904. S. \V. WILLISTON : The Temporal Arches of the Reptilia 175 C. M. CHILD : Form- Regulation in Cerianthus, VI. 193 ADELE M. FIELDE: Portable Ant-Nests 215 IV CONTENTS. EDWIN G. CON KLIN : Experiments on the Origin of the Cleavage Centrosomes 211 No. 5. OCTOBER, 1904. ADELE M. FIELDE : Power of Recognition Among Ants 227 CHAS. W. HARGITT : Notes on a Hitherto Undescribed Hydroid from Long Island Sound 251 EDWIN C. STARKS : A Synopsis of Character of some Fishes Be- longing to the Order Haplomi 254 No. 6. NOVEMBER, 1904. C. M. CHILD: Form- Regulation in Ceriantlnis, VII 263 -/ CHARLES D. SNVDER : Locomotion in Batrachoseps with Severed Nerve-cord 280 ALEXANDER G. RUTHVEN : Butler's Garter Snake 289 ADELE M. FIELDE: Tenacity of Life in Ants 300 ARTHUR W. GREELY. Vol. VII. June, 1904. No. i BIOLOGICAL BULLETIN ARTHUR W. GREELEY. .Arthur W. Greeley died at St. Louis, after an operation for appendicitis, on March 15, 1904, at the age of twenty-eight. The following paper, which he had prepared for publication just before his death, will indicate how lamentable for American sci- ence is the loss of this enthusiastic, industrious and keen investi- gator. In this suggestive paper, explaining on the basis of modern physical chemistry the changes in structure of protoplasm ac- companying changes in function, Greeley has mapped out for others work which he had planned for himself, but which he was unable to accomplish. There has been hitherto no thorough study of the changes in structure of living protoplasm produced by salts from the standpoint of modern views of electrolytes and colloidal solutions. This paper is a most important contribution to this subject and opens up a great field for further work. It was Greeley's good fortune to be able to reduce the changes to order, and thus to supply the structural basis for observed changes in function produced by salts and other agencies. As a result of this work he was able to reduce many of the so-called " tropic " responses of organisms to a common basis ; all agencies produc- ing a certain change in the protoplasm producing also a definite response in orientation of the organism. What a great step in advance this is will be appreciated by those familiar with the con- fusion prevailing in this most difficult field. This paper, in connection with his earlier discoveries of the production of spores in infusoria by cold ; on the identity of the physiological action of dehydration and exposure to low tem- peratures, and of the production of artificial parthenogenesis in the echinoderm egg by cold, stamps Greeley as an original, 2 ARTHUR W. GREELEY. industrious and accurate investigator of great scientific ability, doing work of a most fundamental character. All of those who knew him feel that in his death a man of great promise has passed away. Dr. Greeley was born in Oswego, New York State, in 1875. He took his undergraduate degree in Stanford University in 1898, and spent one year as a graduate student in zoology, during which he went to Alaska with the fur-seal expedition and to Brazil with the Agassiz expedition. The following year he was a teacher in the State Normal School at San Diego, leaving there to enter the University of Chicago as fellow in physiology. Two years later he took his doctorate of philosophy under Loeb with a thesis on the action of low temperatures on the infusoria, and was then appointed Assistant Professor of Zoology at the Washington University, in St. Louis. For three summers he was a member of the staff of instruction in physiology at the Marine Biological Laboratory of Wood's Roll. During his two years of residence in St. Louis his enthusiasm and unusual per- sonality had already aroused marked interest in biological science in that city. Dr. Greeley was of a rare and winning personality, remarkable for extraordinary enthusiasm which inspired all with whom he came in contact. He had a happy disposition, great courage and high principles. His frank open nature, his consideration for others and his loyalty made him many friends ; and he had no enemies. He was an inspiring teacher. To his university, to his friends and to his colleagues his sad death at the outset of a most promising career is an irreparable loss. A. P. MATHEWS. EXPERIMENTS ON THE PHYSICAL STUCTURE OF THE PROTOPLASM OF PARAMCECIUM AND ITS RELATION TO THE REACTIONS OF THE ORGAN- ISM TO THERMAL, CHEMICAL AND ELECTRICAL STIMULI. ARTHUR W. GREELEY. CONTENTS. PAGE. Introduction 3 Structural Reactions of the Protoplasm of Protozoa to Physical and Chemical Changes. I. Reactions to Variations in the Temperature 8 II. Reactions to Chemical Changes 9 III. Reactions to the Electrical Current 19 Effect of these Structural Modifications on the Vital Properties of the Protoplasm. I. Growth and Cell Division 22 II. The Tropisms 24 1 . Thermotaxis 24 2. Galvanotaxis ". 25 3. Chemotaxis. 29 Conclusions 3 l INTRODUCTION. It has been known for several years that a marked similarity in physical structure exists between protoplasm and that class of chemical solutions known as colloidal solutions. This similarity was pointed out by Hardy in 1899 as a result of his investiga- tions upon the physical structure of certain organic colloids, as egg albumen, gelatin, etc., which resemble protoplasm very closely in their gross appearance, and of observations upon pro- toplasm itself under various conditions. The conclusions of Hardy and others l in regard to the physical structure of the organic tolloids are of such importance in the further development of this paper, and are so largely unappre- ciated by biologists, that they will be briefly recapitulated at this place. i. A colloidal solution consists of a fluid matrix which holds in suspension more solid or viscous granules (the colloidal par- 1 Hardy, Jour, of Physio!., 1899, XXIV., p. 172; ibid., p. 288. Mann, "Physiological Histology," Oxford, 1902. 4 A. VV. GREELEY. tides), which differ from the dissolved substance in a crystalloidal solution in that they are relatively very large aggregates of molecules of the colloidal matter. These particles do not affect the osmotic pressure of the fluid matrix ; while in a crystalloidal solution the solute exists in a molecular or ionic condition, and thus gives to the solution a definite osmotic pressure. 2. The physical state of the entire solution depends on the condition of these colloidal particles. When they are finely divided and separated from each other in the solvent, the colloid appears as a liquid or exists in the " sol ' phase. If, however, the colloidal particles become fused, and thus lose their condition of fine suspension, the colloid becomes relatively solid, or passes into the " gel " phase. 3. The physical state of the colloidal particles and hence of the entire solution varies directly with certain external conditions. Thus the passage from the " sol ' into the " gel " phase is accomplished in the following ways : (A) by variations in the temperature, (BJ by chemical .changes, (C) by the action of the electric current. A. The physical state of the organic colloids varies directly with the temperature. If the "sol" phase is constant at the normal temperature, 20° C., coagulation gradually takes place as the temperature is lowered, until at o° C. almost complete gelation has occurred. As the temperature is raised above the normal the fluidity of the solution is increased (by the subdivision of the colloidal particles and absorption of water), until a critical point is reached at which coagulation suddenly occurs, and the colloid goes into a condition of heat rigor. B. A colloid may be coagulated by adding to it a solution of any electrolyte which bears an electrical charge opposite in sign to that carried by the colloidal particles.* Thus Hardy found that a positively charged colloidal solution was coagulated by any electrolyte with a powerful anion or negatively charged ion, and the rapidity of the coagulation varied directly with the valence of the anion. A negatively charged colloid was coagulated by electrolytes of an opposite electrical character, or by powerful cathions. Similarly an electrolyte carrying a charge of the same sign as that of the colloidal particles causes them to subdivide STRUCTURE OF PROTOPLASM OF PARAMCEC1UM. 5 still further. As a result water is absorbed so that a liquefaction of the whole colloid occurs, as by a slight increase in the tem- perature. Thus cathions liquefy positively charged colloids ; anions, negatively charged colloids. C. The same conditions prevail in the reaction of colloids to the electrical current. Negatively charged colloidal particles fuse and form a " gel" around the anode, and tend to liquefy about the cathode during the passage of the current. Positively charged colloids coagulate about the cathode, and liquefy about the anode. From the behavior of these colloidal solutions under various chemical and electrical conditions Hardy concluded, that the " sol " phase is maintained under normal conditions because all the colloidal particles carry an electrical charge of the same sign, and are thus mutually repelled, and remain in a state of fine suspension. Whenever there is introduced an elec- trical charge of an opposite sign, either by means of a dissociated electrolyte or by the electrical current, the charge carried by the colloidal particles is neutralized, and fusion or coagulation occurs. The reaction to temperature variations is apparently due to the fact that, within certain limits, the kinetic energy of the colloi- dal particles varies directly with the temperature. A reduction of the kinetic energy causes a gradual fusion of the particles ; an increase brings about a still finer state of suspension, and conse- quent liquefaction of the colloid through the absorption of water. The sudden coagulation at the critical point is probably due to a chemical change in the colloid itself. The chief points of similarity between these colloidal solu- tions and protoplasm made apparent by Hardy's work are as follows : i . The elementary physical structure of the two is the same. Like the colloid, protoplasm is known to be made up of two substances — (a) the fluid cell-sap or matrix which holds in suspension ($) the more solid proteid or protoplasmic particles, granules or microsomes of the morphologists. And the physical state of the protoplasm depends on the condition of these proto- plasmic granules, just as the state of the artificial colloid depends 6 A. \V. GREELEY. on the condition of its constituent particles. It is significant to find, that all the controversies that have been waged over the various theories of a fixed morphological structure in protoplasm have centered about the supposed unchanging physical condition of these more solid or viscous elements in the protoplasm. Now in the artificial colloids we see that their physical state varies di- rectly with certain external conditions. The protoplasmic particles of course are vastly larger than the particles in an artificial colloidal solution, and we have no right to assume that the two are identical. But the protoplasmic granules bear the same relation to the cell- sap as the colloidal particles do to the fluid matrix, and they both respond to chemical and physical changes in a similar man- ner. A close comparison of the behavior of these two structural elements under various conditions will, I believe, throw a great deal of light on the physical basis of protoplasm. Hardy observed that when a colloidal solution is exposed to the action of certain so-called fixing agents, the same structures are produced as by the action of these fixing agents upon proto- plasm, viz.: a coagulation occurs in which the colloidal particles fuse in definite ways, producing a type of structure varying with the fixing agent employed. He therefore concluded that coagu- lation of the protoplasm is necessary to reveal the fixed types of structure thought by some to be characteristic of protoplasm under all conditions. Such types of structure then appeared to be in protoplasm, as in the organic colloids, merely artefacts ; and Hardy argued that protoplasm in the living condition must be identical, as far as its physical structure is concerned, with the organic colloids in the " sol " phase, and that like the colloids its structure is constantly changing with the external conditions. Undoubtedly under the same external conditions, the protoplasm of different forms will show varying types of structure, as the above statement will allow. In some cases this structure may resemble a reticulum or other types described by the morphologists, but no one will now maintain that such a type is of universal occur- rence. In the living protoplasm of Paramoscium, I have never .seen a trace of a fixed structure. These suggestions of Hardy and others upon the physical structure of protoplasm remained almost unnoticed by biolo- STRUCTURE OF PROTOPLASM OF PARAMCECIUM. J gists until, very recently, attention was directed to them by the work of Loeb,1 Mathews,2 R. S. Lillie and others. Mathews, in particular, in his work upon the chemical stimulation of the motor nerve of the frog, has arrived at the conclusion that the protoplasm of the nerve is a colloidal solution whose particles carry a definite electrical charge, and that stimulation of the nerve is accomplished by a reversible change in the physical state of the protoplasm, analogous to coagulation. He finds that this stimulation or coagulation is effected chemically only by a series of electrolytes which agree in carrying a predominant negative charge of electricity, as would be the case 'in Hardy's positively charged colloidal solutions. It is only fair to say that these remarkable investigations of Mathews have been the inspiration of this work on the physical structure of protoplasm. R. S. Lillie 3 shows, in his work on the reaction of cytoplasmic and nuclear structures to the electric current, that these forms of pro- toplasm behave exactly as would positively and negatively charged colloidal solutions under the same conditions. The experiments described above have shown that proto- plasm reacts to various external conditions in a manner strictly parallel to the behavior of colloidal solutions in the presence of like stimuli. But meanwhile we have had but little evidence of the precise structural changes, that accompany these reactions, and are, as we have seen, the leading distinguishing feature of colloidal solutions. It was with the idea of studying these struc- tural changes, which occur in protoplasm when subjected to vari- ous external stimuli, and of comparing them with the changes in colloidal solutions under the same conditions, that the following experiments were undertaken. . The material used has been the protoplasm of various proto- zoa, viz : Paramcecium, Stcntor, Amccba and others, which are ideal objects for this study because of the abundance in which 1 Loeb, as a result of his work on the stimulation of contractile tissue by ions, and on the toxic and antitoxic effect of ions on the duration of life of the Fundulus egg and frog's muscle, was the first to suggest, that the protoplasm reacts under these conditions like artificial colloids, although he has elsewhere adopted a different explanation. (See Amer. Jour. Physio!., 1902, VI., p. 411.) 2 Mathews, Science, 1902, XV., p. 492; 1903, XVII., p. 729. 3 Lillie, Amer. Jour. Physio!., 1903, VIII., p. 273. A. W. GREELEY. they can be procured, and especially because their small size makes it possible to study the structural changes in the living protoplasm under high magnifications, which is out of the ques- question in more bulky tissues. The protozoa were subjected to thermal, chemical and electrical changes, and the structural modifications produced by these means were studied. STRUCTURAL REACTIONS OF THE PROTOPLASM OF PROTOZOA TO PHYSICAL AND CHEMICAL CHANGES. I. Reactions to Variations in the Temperature. In previous papers l I have described the structural changes that occur in the protoplasm of various protozoa when exposed to variations in the temperature, so that a very brief description will suffice here. We find, exactly as in the case of organic col- loids, that the fluidity of the protoplasm varies directly with the temperature within certain limits. As the temperature is lowered below the normal, 2O°C., a very gradual coagulation occurs, be- cause of the decrease in kinetic energy and consequent fusion of the protoplasmic particles. This change is accompanied by a loss of water, so that at o° C. there is produced a nearly solid opaque, spherical mass of protoplasm, which exists only in a rest- ing condition. As the temperature is elevated above the normal, the reverse changes occur through the increase in kinetic energy acquired by the protoplasmic particles. They may be seen to subdivide further and become more widely separated through the absorption of water, so that the fluidity and motility of the protoplasm is greatly increased. These changes continue until the critical point is reached, at which coagulation suddenly occurs, and the protoplasm goes into heat rigor, probably because of some chemical change in the protoplasm itself. So closely do these structural changes agree with those that occur in artifi- cial colloids under the same conditions, that a description of one would apply equally well for the other. Indeed, it was this striking similarity between the results in the two cases, that led me to compare the reactions of colloids and the protoplasm of 1 Greeley, Amer. Jour. PhvsioL, 1901, VI., p. 201 ; BlOL. BULL., 1902, III., p. 165 ; //>/,/., 1903, V.. p. 42. STRUCTURE OF PROTOPLASM OF PARAMCECIUM. 9 various Protozoa to chemical and electrical stimuli to see what light these reactions might throw on the physical structure of protoplasm. II. Reactions of the Protoplasm of Paramcecium and other Protozoa to Chemical Changes. Paramcecia reared on cultures containing bread were used chiefly in these experiments, although other protozoa were util- ized for purposes of comparison. The method employed in the experiments was as follows : The solutions whose action upon the protoplasm was to be tested were made up in the proper dilutions, and distributed among Minot watch glasses holding about 10 c.c. of the solution. Pure, concentrated cultures of the protozoa were obtained, and a drop or two added to each of the dishes containing the solutions. At short intervals the protozoa were examined to observe any structural changes in the protoplasm. A 1/12 inch oil immersion was used to make out the finer details. Great care is needed to distinguish between chemical and o osmotic effects. In order to obviate the possibility of modifying the protoplasm through the extraction of water, all the solutions were used in concentrations of a lower osmotic pressure than that of the protoplasm, roughly equal to a #2/40— #2/50 cane sugar solution. In some cases the dilutions were very great, and, inas- much as distilled water or any solution of a lower osmotic pressure will liquefy the protoplasm through the absorption of water osmot- ically, there would be danger of confusing the osmotic and chem- ical effects of the solutions, were it not for the fact that osmotic and chemical liquefaction can be readily distinguished by the type of protoplasmic structure produced by each means.1 All the chemicals used fall into two classes : i. Those that effect the protoplasm only through the osmotic pressure of the solution. 1 Miss Towle, in investigations now in progress at the University of Chicago, has observed that paramoecia may live indefinitely in redistilled water. It is probable that in this case, as in others mentioned in this paper, the variations in the structural reactions are due to the differences in the chemical composition of the cultures in which the paramoecia were reared. But this question of the structural effect of dis- tilled water requires further investigation. IO A. W. GREELEY. 2. Those that modify the structure of the protoplasm inde- pendently of the osmotic pressure. The first class includes the non-electrolytes used, the second, the electrolytes. Non-clectrolytcs. — The non-electrolytes used, distilled water, cane sugar and urea, affect the protoplasm of paramcecia only through the osmotic pressure of the solution. There is no chem- ical effect. All those solutions hypotonic to protoplasm produce a liquefaction through the absorption of water, while isotonic solutions have no effect on the structure of the protoplasm, and hypertonic solutions coagulate the protoplasm through a with- drawal of water. These effects are shown in the following table : TABLE I. 7«/5 ;«/io in 40 ;«/i6o Cane sugar, Coagulation. Coagulation. No effect. Liquefaction. Urea, An examination of these structural changes under a high magnifi- cation makes clear the fact that these non-electrolytes modify the structure of the protoplasm solely through a change in the amount of water in the fluid matrix which holds the more viscous elements in suspension. The size of the protoplasmic particles remains unaffected. In hypotonic solutions the protoplasmic particles are more widely separated by an increase in the amount of water in the fluid matrix. In hypertonic solutions they are brought closer together, or may fuse through a withdrawal of the sur- rounding liquid. Electrolytes. - - As far as their effect on the physical structure of the protoplasm of Paramcecium is concerned, all the electro- lytes used fall into two classes : first, those that coagulate even in solutions whose osmotic pressure is far less than that of the protoplasm ; second, those that liquefy the protoplasm in any dilution. Moreover, all the members of the first class agree as far as their electrical conditions are concerned. They all have a predominantly powerful cathion,1 but resemble each other in no other particular. Likewise all the members of the second class agree in possessing a predominantly powerful anion. Thus we 1 This "predominance" may be a function of the solution tension of the ions. See Mathews, Amer. Jour, of Physiol. , 1904, X., p. 290. STRUCTURE OF PROTOPLASM OF PARAMCECIUM. I I see that anions, or negatively charged ions, liquefy the protoplasm of Paramoecium, cathions, or positively charged ions, coagulate without any regard for the supposed chemical affinities of the electrolyte. That we are not dealing with a specific chemical effect for each electrolyte is still further shown by the fact that the activity of each solution is roughly proportional to the valance of the pre- dominant ion. Thus salts containing trivalent anions or cathions are effective in much greater dilutions than bivalent or univalent salts. All these facts are indicated in the accompanying table which gives a list of the electrolytes used, their effects on the physical structure of the protoplasm, and the greatest dilution at which they are effective. The maximal dilutions can be only approximate, as the action of identical solutions is not the same on paramoecia from different cultures, because no two are exactly alike in respect to chemical composition and osmotic pressure. Especially is this true of the liquefying electrolytes, for, as we have seen, in very dilute solutions it is frequently difficult to determine the point at which the specific liquefying action of the anion ends and the osmotic absorption of water commences. As has been already mentioned, there are structural differences between these two forms of liquefaction which enable us to dis- tinguish between them with considerable accuracy, and the fact of chemical liquefaction can be easily demonstrated by using the liquefying substances in solutions about isotonic with protoplasm. Of course no such difficulty arises with the coagulating solutions, for at great dilutions the specific coagulating action of the cathion must overcome the tendency for the protoplasm to be liquefied through the entrance of water osmotically, and there is thus a well-defined point at which the coagulation of the protoplasm by the cathion ends and its osmotic liquefaction commences. TABLE II. Coagulating Solutions. Liquefying Solutions. /w/1,600 HC1, ;«/i,6oo HNOS w/i,6oo NaOH, w/i,6oo KOH w/2,400 H,SO4 w/l,6oo Ba(OH)2, mj 1,600 Sr(OH)2 w/40 KC1 w/40 NaCl, ;«/4O NH4C1, w/4O NaNO3 w/640 MgCI2, w/640 CaCl2, mfiqo mfiqo Na.2SO4, m^o (NH4)2C2O4 Ca(NO3)2, w/640 Bad, ;«/2,4OO Na3PO4, HI {2,400 Na3C6H.O7 w/320 MgS04 »i 1 1,600 A12C16, in 1 1, 600 Fe2Cl6 12 A. W. GREELEV. FIG. I. Normal protoplasm of Para- m cccii tn i. TX5 in. oil immersion. It will be seen by an examination of the table that the activity of any of the salts, as is shown by the strength of solution re- quired to modify the structure of the protoplasm of Paramoccinin, varies directly with their valence. The acids and bases are much more powerful in their action than any of the salts of a similar valence, probably because of the known disproportionate kinetic energy of the hydrogen and hy- droxyl ions. All the acids and salts found in the first column of the table, which agree in coagulating the protoplasm through the action of the predominant cathion, effect changes in the protoplasm so similar as to be practically indis- tinguishable even under a high magnification. The less active solutions, such as KC1 and MgSO4, do not produce quite so dense a coagulum as the others, and the reaction is considerably slower. But in all the bivalent and trivalent salts and the acids, a distinct clouding of the protoplasm can be observed within thirty minutes after the paramcecia have been immersed in the solution. This clouding of the protoplasm increases and is ac- companied by a shrinking of the cell owing to a loss of water, until within a few hours, the whole cell is reduced to a subspher- ical mass of densely opaque protoplasm (see Fig. 2). The changes are identical with those produced by a lower- ing of the temperature. An examination of the pro- toplasm with a one-twelfth inch oil-immersion lens re- FIG. 2. A Paramcecium in w/32O MgCl,, , . e , showing a typical coagulation of the pro veals the fact that the cloud- toplasm. ing of the protoplasm is due to a separation of the two elements of the protoplasm, the cell sap and the protoplasmic granules. These two elements lose STRUCTURE OF PROTOPLASM OF PARAMOZCIUM. their state of fine mixture or suspension, which is always char- acterisic of motile protoplasm, and the protoplasmic particles tend to fuse into a spongy coagulum or " gel" from which the cell sap continually escapes, until a complete separation of the two elements is brought about. This fusion of the protoplasmic particles may result in the formation of two varieties of coagu- lated structure. The fused particles may appear as spherical bodies of proteid material, unconnected but closely massed to- gether in such a way as to form an exceedingly dense coagulum FIG. 3. Coagulated protoplasm of FIG. 4. Another type of coagulation Paraituvciam under TV oil-immersion. A viewed under the T\ oil-immersion. The dense coagulum formed after an expo- result of a ten hour residence in w/8oo sure to w/32O CaCl.2 for twenty-four HC1, showing a trace of the reticular hours. structure. (see Fig. 3). Or the protoplasmic particles may fuse into fibrils or anastomosing threads which form an incomplete network, holding the cell sap in its interstices. The fibrils may eventually become thicker and transform the network into a relatively solid coagulum (Fig. 4). These last forms of coagulated struc- ture are very similar to the network formations obtained by Hardy in organic colloids and various protoplasmic tissues by the action of fixing agents, and have been formerly supposed to be characteristic of living protoplasm. As in Hardy's experi- ments these fibrillar or reticular structures never appear in the living protoplasm of Paramceciuiu, but are solely an incidental result of the process of coagulation by chemical or other means. The salts and bases in the second column of the table, which produce a common liquefaction of the protoplasm of Paramcecium through the action of a predominant anion, are effective like the 14 A. W. GREELEY. coagulating solutions, in dilutions which are roughly proportional to their valence. The univalent salts have a comparatively weak effect upon the protoplasm, and relatively high concentrations are necessary before we can be sure that the liquefaction is due to chemical and not osmotic means. In solutions of the bases, bivalent and trivalent salts, however, the effects are unmistakable, and are ex- actly the reverse of those initiated by the coagulating solutions. Within a very few minutes after immersion in the solution lique- faction first becomes discernible as a clearing of the protoplasm. This process proceeds until the protoplasm loses its character- istically granular appearance, and becomes semi-transparent. At FIG. 5. A Paramaecium in mj^2O Na.2SO4, showirg a typical liquefaction of the protoplasm. the same time the cell membrane becomes greatly swollen through the absorption of water, which frequently gives the protoplasm a vacuolated appearance, and gives rise to droplets which cling to the protoplasm underneath the cell wall. The result is an irregu- lar watery mass of protoplasm from which the solid elements have, to a superficial examination, completely disappeared (see Fig. 5). In the solutions of trivalent salts these changes occur with such violence that the cell membrane is disrupted, and the disintegrated protoplasm becomes scattered throughout the solu- tion. Thus the liquefying anion has the same effect upon the protoplasm as a slight increase in temperature. If, as in the case of the process of coagulation, these micro- scopic changes be studied under a one-twelfth inch oil-immersion, it will be at once seen that a change in the physical state of the protoplasmic particles is responsible for these profound structural modifications. The particles apparently continue to divide as the process goes on, until their size becomes so small that they STRUCTURE OF PROTOPLASM OF PARAMOZCIUM. FIG. 6. Liquefied protoplasm of Paranict'cium under ^ oil -im- mersion. Formed by an exposure of two hours to w/32O Na2SO4. are discernible only under high magnifications. At the same time a rapid absorption of water takes place, and the particles become widely separated in the now exceedingly fluid cell sap. We thus have a means of distinguishing between liquefaction by chemical and osmotic means. As has been described above, the central feature of the process of chemical liquefaction is a splitting up of the protoplasmic particles and consequent imbibition of water through this increase in the absorb- ing surface (see Fig. 6). In the process of osmotic liquefaction, how- ever, the size of protoplasmic par- ticles remains unchanged, and water enters the protoplasm solely because of the osmotic relations between the cell-sap and the surrounding liquid. These two processes can be distinguished microscopically as indicated even under a low magnification, for paramcecia that are liquefied osmotically never lose their granular appearance, while those liquefied chem- ically become markedly transparent. That in these phenomena we are dealing also with variations in the surface tension relations of the protoplasm is apparent from the change of form which occurs during coagulation and liquefaction. The surface tension force is neutralized either by an increase in temperature or by giving all the protoplasmic particles a like charge which tends to make them mutually re- pellent, and thus introduces a disrupting force. Hence, if we de- stroy the electrical charge, we not only allow the protoplasmic particles to fuse, but we increase the surface tension. Thus, dur- ing coagulation, the cell tends to assume a spherical form which is characteristic of all resting cells. The opposite is true during liquefaction. The disrupting force is still further increased by the introduction of a charge of the same sign as that carried by the protoplasmic particles, and the cell assumes an irregular form. The most powerful anions used, the phosphate and cit- rate, bring about this disruption of the cell with almost explosive violence, so that the cell membrane bursts and the protoplasmic 1 6 A. W. GREELEV. particles are widely scattered. After a residence in these solu- tions of ten to fifteen minutes, no vestige of the paramoecia remains beyond the scattered protoplasmic granules. The above experiments were repeated on Vorticclla, Stcntor and Hydra, and essentially the same structural changes were produced. The protoplasm of Vorticella reacts to electrolytes exactly as does the protoplasm of Paramoscium. In Stcntor the differentiation between the two layers of protoplasm, endosarc andectosarc, is more complete, and tl*e less differentiated granular endosarc responds to the solutions in the typical manner, while the clear striated ectosarc remains practically unaffected. Thus the process of coagulation results in the formation of a cell whose endosarc is shrunken into a dense, spherical mass within the ectosarc, which retains its original form and size. During lique- faction the endosarc becomes wholly transparent, and we have pro- duced an apparently empty cell which is bounded by the un- changed ectosarc as before. Hydra reacts in the same manner, the endoderm responding to the solutions as did the endosarc of Stentor ; and we have produced animals of the original size and shape, but with either a coagulated or liquefied interior. It appears from these experiments that the physical conditions already de- scribed are applicable to protoplasm only in its primitive granular condition. The differentiation into such simple structures as the ectosarc of Stcntor, or the ectoderm of Hydra has changed to some extent its elementary physical state. The high degree of viscosity of such protoplasm suggests that it is normally much nearer the " gel " phase. Like all the typical physical modifications of protoplasm, these changes of structure are reversible, if they are stopped at the proper time. Thus coagulated protoplasm may be again lique- fied by the action of a powerful anion, or liquefied protoplasm may be coagulated by a cathion. This fact seems to show with special force that the physical state of the protoplasm at any moment is the result of a definite reaction to the chemical condi- tions of the environment. The whole behavior of protoplasm under the action of dilute solutions of electrolytes suggests the conclusion, that, as far as its chemical and physical reactions are concerned, it is a colloidal STRUCTURE OF PROTOPLASM OF PARAMCECIUM. I/ solution whose particles carry a definite electrical charge. The colloidal particles in an artificial colloidal solution are apparently identical, as far as their physical reactions are concerned, with the protoplasmic or proteid particles held in the cell sap of proto- plasm ; at least they both react to thermal and chemical changes in the same way. The reaction to chemicals further demonstrates the fact, that, like the colloidal particles, these protoplasmic ele • ments carry a uniform electrical charge under normal conditions; for by no other assumption can we explain the facts that salts, acids and bases modify the structure of protoplasm only by virtue of their electrical properties, and that non-electrolytes have no effect whatever beyond that of the osmotic pressure of the solu- tion. Indeed the action is not, properly speaking, a chemical one at all, but an electrical one. In Paramccciuui the protoplasmic particles are apparently negatively charged, and the mutual repul- sion of these similarly charged particles, keeps them in a state of fine suspension. Under these conditions the protoplasm exists in a " sol " or liquid phase. Predominant cathions neutralize this negative charge ; the repelling force is removed, and the par- ticles fuse, forming a coagulum. A predominant anion further increases the disrupting force, and a state of still greater fluidity or liquefaction results. A still closer comparison between artificial colloids and the protoplasm of Paramocciuni is possible. It will be remembered that Hardy succeeded in reversing the charge carried by the colloidal particles by changing the chemical conditions of the solution. Thus the addition to the solution of a small amount of a base gave an alkali-modified colloid whose particles were nega- tively charged. Such a solution was coagulated only by cathions. Likewise the addition of a small amount of acid gave an acid- modified solution, whose particles were positively charged ; this solution was coagulated by anions. How this change is effected by acids and alkalis we cannot say, but certain it is that the electrical nature of the colloid is determined by certain ions present in its fluid matrix. The reactions of the protoplasm of paramcecia to electrolytes described above, occurred in organisms which had been reared in a normal, alkaline culture. It was found that when the culture 1 8 A.-W. GREELEY. medium became acid through the fermentation of bread, the structural reactions of the protoplasm to electrolytes was changed. And they were changed in such a way as to suggest that there had been a reversal of the electrical charge carried by the pro- toplasmic particles, or in other words, that there had been formed an acid-modified protoplasm. Thus the protoplasm of paramoecia from an acid culture was not coagulated by cathions and lique- fied by anions, as was the case without exception with paramcecia from an alkaline culture, but the reactions were as follows : The first change produced by the neutralization of the normal alkalin- ity of the culture was an irregularity in the structural reactions. Instead of all the organisms being coagulated by cathions and liquefied by anions, some were coagulated, and some were lique- fied in each solution, apparently indicating a condition in which the protoplasm of some of the paramcecia had been modified by the acid, the rest being still unaffected. As the acidity of the solution increased, the number that were coagulated by anions and liquefied by cathions correspondingly increased until, in a few instances, a complete reversal of the structural reaction to elec- trolytes was produced. While the reversal was more often not complete, in all cases it occurred in a varying proportion of the paramcecia from acid cultures, while the remainder were generally rendered indifferent to the solutions which formerly coagulated or liquefied them. The complete reversal occurred only in the salt solutions. The acids always coagulated the protoplasm regardless of the character of the culture, and to alkalis, the paramcecia from acid cultures were only rendered indifferent. All these results indicate that the particles of the acid-modified pro- toplasm have become positively charged like the particles in Hardy's acid-modified colloids. The view that this change in the structural reactions of para- mcecia is due to a modification of the protoplasm by acids is ren- dered more probable if we follow the reactions of paramcecia from day to day, which are taken from a culture which is gradually becoming acid by the fermentation of bread. In such a case we can easily see the gradual fading out of the normal reaction to electrolytes, and the assumption of the one peculiar to acid- modified protoplasm. This change in structural reaction always STRUCTURE OF PROTOPLASM OF PARAMCECIUM. 19 accompanies the neutralization of the alkalinity of the culture. The complete reversal in the sign of the charge carried by the protoplasmic particles is apparently too fundamental a change to occur without killing the protoplasm except under the most per- fect conditions. While much more work must be done on this point, we can assuredly say that the normal electrical charge carried by the protoplasm of Paramcecium bears a very close relation to certain chemical conditions of the environment, of which the alkalinity of the surrounding medium may be taken as one of the most important. These results are especially signifi- cant in the light of the behavior of paramoecia from various cul- tures toward different forms of stimuli, as we shall see when we discuss this subject. III. Reactions of the Protoplasm of Paramcccuun to the Electrical Current. The above conclusions relating to the ultimate physical struc- ture of protoplasm and the electrical conditions underlying it, are further borne out by a study of the effects produced on the structure of the protoplasm by the constant current. It has long been known that the constant current has a profound polar effect on the protoplasm of various protozoa. Thus Kiihne, Verworn1 and others have shown that when a protozoan is exposed to the action of a weak current, a contrac- tion occurs on the anodal side of the cell, and a relaxation on the cathodal side. This phenomenon was first described for Actinospkcerium, a large heliozoan with many radiating pseudo- podia. Very soon after exposure to the current the pseudo- podia on the anodal side become contracted into irregular shapes, and are finally completely withdrawn into the cell, while those on the cathodal side remain fully extended. Amccba is still more sensitive to the current. The whole cell contracts on the anodal side while pseudopodia are rapidly thrown out toward the cathode so that the animal moves in this direction. The same general changes have been observed in Paramcecium 2 and many other protozoa. If the organisms are exposed to the current for a 1 Verworn, Arch. f. d. ges. PhysioL, 1889, XLV., p. I. 2 Pearl, Amer. Jour, of Physiol., 1900, IV., p. 96. 2O A. W. GREELEY. longer time, more pronounced structural changes ensue. The contraction on the anodal side continues until the protoplasm at this point can be seen to disintegrate, forming dense granular masses, while on the opposite side the protoplasm becomes even more clear and transparent than it was at first, frequently flowing out over this portion of the cell in irregular liquid masses. The observations have been repeated in a large number of forms so that the facts of definite polar modifications in the structure of the protoplasm during the passage of a weak constant current seem to be very well established. At the same time it has been shown in many forms that there is a movement of the protoplasmic particles away from the cathodal side where liquefaction of the protoplasm is taking place toward the anodal region of the cell where the disintegrating or "etching" effects are shown.1 I have repeated these experiments and have furthermore shown that when paramcecia are isolated in a weak gelatine solution or held in a fine mesh-work of cotton during the passage of the current so that they are unable to move freely from pole to pole, a slightly different structural reaction to the current is obtained after the current has been passed for from three to five hours. In this case all those paramcecia which are held in the region of the anode are modified as was the anodal side of the cell in the preceding experiment, /. e., the whole cell contracts into a dense, opaque mass of protoplasm. Likewise the paramcecia held about the cathode are modified like the cathodal side of the cell in the former experiment, i. e., the cell contents are liquefied and there is formed a large transparent cell of fluid protoplasm which ultimately bursts because of the increased pressure on the cell wall. Thus we have the same structural changes occurring either in the whole cell immediately about the anode, or in the anodal side of any of the cells exposed to the current ; opposite changes occurring in the cells about the cathode or on the catho- dal side of all the cells in the preparation. A microscopical examination shows that these changes are 1 Wallengren (Zeitsckr. f. allg. Physiol., 1903, III. , p. 22) states that in the Rhizo- poda only can the movement of the protoplasmic particles toward the anode be demon- - strated. In the Infusoria the constant current does not affect the normal streaming of the protoplasm. STRUCTURE OF PROTOPLASM OF PARAMCECIUM. 21 identical with the structural changes produced by the action of electrolytes upon the protoplasm. About the anode and on the anodal side of the cells the protoplasm is coagulated as by the use of cathions in weak solutions ; about the cathode and on the cathodal side of the cells, the protoplasm is liquefied as by the use of anions in weak solutions. Not only are these structural effects the same, but they can be shown to be produced by the same means in each case, /. t\, by electrically charged ions, which are present in very dilute solutions of electrolytes, as we have shown, and which serve to carry the current from pole to pole when it is passed through such weak salt solution as the culture media of paramcecia which were used in the experiment. It was formerly supposed that these polar effects were due to the formation of acids on the anodal side of the cell and alkalis on the cathodal side, but we have no satisfactory evidence for such internal polar changes. During the passage of the current, the anions prevail about the cathode, and are constantly diffusing toward the anode. During this passage they are continually im- pinging on the cathodal side of the paramoecia, hence a lique- faction of the protoplasm takes place at these places. Cathions prevail about the anode, and continually impinge on the anodal side of the cells in their diffusion toward the cathode so that coagulation occurs at these points. In either case the structural changes are produced by virtue of the electrical charge which the ions carry, and not by any specific chemical effect of the ions or molecules. We thus see that the chemical and electrical means of modifying the protoplasmic structure are identical. It will be remembered that these changes in the structure of the protoplasm produced by solutions of electrolytes or the elec- trical current are the same as those which are brought about in organic colloids by the same means. The alkali-modified colloids in Hardy's experiments were coagulated by cathions in solution or at the anode during the passage of the current. They were liquefied by anions or at the cathodes. It is interesting to note also that in the protozoa a movement of the protoplasmic parti- cles within the cell toward the anode has been observed, which corresponds exactly with the movement in the same direction among the colloidal particles of an alkali-modified colloid. Like 22 A. W. GREELEY. the colloidal particles, they always move toward the point at which coagulation occurs. All these experiments seem to show that the structural changes produced in protoplasm by thermal, osmotic, chemical or electri- cal changes are the same, because all of these variations in the external conditions act upon protoplasm only by altering the physical state of its solid elements. Thus in the case of Para- mcecium, at least, the structure of the protoplasm is seen to be not fixed and uniform, but to depend directly on certain external conditions and to vary with their variations. The best expression of this behavior of protoplasm is found in the laws of the reaction of colloidal solutions to external conditions. EFFECT OF THESE STRUCTURAL MODIFICATIONS ON THE VITAL PROPERTIES OF THE PROTOPLASM. Having determined the changes produced in the structure of the protoplasm by various chemical agencies, it remained to ascer- tain how these structural changes modify the vital properties of protoplasm. How far may a particular state of protoplasmic ac- tivity be correlated with a given physical condition of protoplasmic structure ? Or does the reaction of an organism as a whole to an external stimulus depend in any measure upon the effect that stimulus may have on the structure of the protoplasm ? /. Groivth and Cell Division. The rate of cell division may be taken as the best indication of the general protoplasmic activity among the Protozoa, after the method adopted by Calkins1 in his work on the " Life Cycle of Paramoecium." A quickened rate of cell division means an increase in the metabolic activities of the cell. A condition of slow meta- bolism is indicated by the cessation of cell division and the trans- formation of the motile cell into a spore or cyst or other resting stage. It has been already shown, in a paper - on the reactions of various protozoa to variations in the temperature, that pre- cisely those temperature conditions, which liquefy the protoplasm, stimulate cell division, and those temperatures which coagulate 1 Calkins, Arch. f. Entwickelungsmech., 1902, XV., p. 139. 2 Greeley, Amer. Jour. Physiol., 1902, VI., p. 122. STRUCTURE OF PROTOPLASM OF PARAMGECIUM. 2$ the protoplasm inhibit it. Thus the rate of cell division increases steadily with a slight elevation of temperature above the normal until the critical coagulating point is reached. A lowering of the temperature progressively decreases the rate of cell division, until the point is reached at which it ceases altogether, and the protoplasm goes into a resting condition. The same relation between the rate of cell division and the physical state of the protoplasm is found to hold good also as a result of the reactions of the protoplasm of Paramoscium to solu- tions of electrolytes. With paramoecia from alkaline cultures, anions or liquefying agents stimulate cell division, cathions and coagulating agents inhibit it. Thus I have frequently observed in my experiments that when the liquefying solution is too weak seriously to modify the structure of the protoplasm, it will, however, greatly increase the motility of the protoplasm and the rate of cell division. Since the size of the paramoecia remains uniform, it follows that this must indicate also increased growth and general metabolic activity. In the coagulating solutions, on the contrary, there are produced spherical resting cells that greatly resemble spores. This antagonism between these two classes of solutions is still o more clearly shown by the fact that the anions greatly accelerate the germination of spores, or the passage from a resting into a motile condition. In these solutions the spherical form of the spore is soon lost through a neutralization of the surface tension. This has been shown in the case of some of the monads- and fresh- water algae. R. S. Lillie l has shown that a decrease in the surface tension must accompany cell division, and that this is accomplished in the case of certain marine animals by the electrolytes present in the sea water, for if these electrolytes be withdrawn, cell division not only stops, but a partial fusion of the already formed blasto- meres occurs. It appears that the surface tension relations are very important in all these protoplasmic reactions to external conditions. Protoplasmic movement, cell division and growth all occur in opposition to the surface tension force.2 Conse- quently any external condition which neutralizes the surface ten- 1 Lillie, BIOL. BULL., 1903, IV., p. 164. 2 See Spaulding, BIOL. BULL., 1904, VI., p. 97. 24 A. \V. GREELEY. sion accelerates these expressions of the general activity of the protoplasm. These conditions are those which bring about a liquefaction of the protoplasm, so that the protoplasmic activity is seen to vary directly with the amount of water the protoplasm contains. Conversely the assumption of a quiescent spherical resting stage in various protozoa is the result of an increase in surface tension, and is formed by those conditions which cause a coagulation of the protoplasm and a loss of water. Thus a slight increase in the temperature is seen to have the same effect on these simple protoplasmic properties of Paramcscium as anions ; a lowering of the temperature acts like cathions ; since each set of conditions produces the same structural effects. II. The Tropisms. We have at the present time an enormous amount of informa- tion concerning the reactions of organisms to external stimuli, but we know almost nothing of the physical or chemical effects of these attractive or repellent agents on the protoplasm of the organism, and consequently we are not able to offer any satisfactory expla- nation of the mechanism of the tropic response. In the follow- ing experiments I have studied the reactions of paramoecia to thermal, electrical and chemical stimuli, and have attempted to show that the reaction of Paramcecium to each stimulus depends on certain structural changes in the protoplasm, which are a result of the stimulating action. I. Thermotaxis. — The reactions of paramoecia to variations in temperature are exceedingly definite. It has been stated by many observers that they are positive to temperatures between approxi- mately 23° and 27° C., and are negative to all others. This reac- tion is beautifully demonstrated by placing the paramoecia in a long, narrow dish which is heated at one end and cooled at the other. It has been already shown that those temperatures to which Paramcecium is positive constitute exactly those thermal conditions which bring about a liquefaction of the protoplasm and a reduction in the surface tension. Those temperatures to which Paramcecium are negative, coagulate the protoplasm. We thus see that, in the case of thermotaxis, attraction is accompanied by a liquefaction of the protoplasm, repulsion by coagulation. The STRUCTURE OF PROTOPLASM OF PARAMCECIUM 25 extreme delicacy of the adjustment between the physical structure of the protoplasm and the external conditions has hardly been recognized. For example, the smallest perceptible elevation of temperature above the normal results in a decided increase in the fluidity of the protoplasm, and it is probable that this structural change explains the extreme sensitiveness of the paramcecia, judged by their thermotropic reactions, to these same variations in the temperature. 2. Gafaanotaxis. — It has been well known that paramcecia normally react to the electrical current in a vigorous and definite manner by orienting themselves with their anterior end toward the cathode, and swimming rapidly in this direction, so that eventually a dense gathering of the organisms occurs about the negative electrode. In other words they collect at that point in the electrical field where the conditions are such as to induce a liquefaction of the protoplasm. After a weak current has been passed through the preparation for from thirty minutes to one hour, it will be seen that the dense gathering at the cathode begins to break up and a reverse movement toward the anode sets in. The number of paramcecia that exhibit this reverse movement varies with the conditions of the organisms at the time of the experiments, as will appear later ; but with paramcecia from alkaline cultures only a small proportion of the entire num- ber will be seen to swim toward the anode at any one time. At first the paramcecia swim only a short distance toward the anode and then immediately dart back to the cathode, but the length of the reverse reaction increases until a few reach the anodal end of the dish. Having arrived at the anode, they immediately swim back to the cathode again, and no gathering occurs at the anode except in rare cases after the current has been passed for about two hours. The paramcecia normally keep the anterior end pointed always toward the cathode, so that they swim backwards toward the anode. But this is not always the case. After a large number of experiments with paramcecia under various conditions, I find that the relations between this initial and secondary reaction to the current may be greatly modified. With paramcecia from alkaline cultures, the secondary reaction begins only after the current has been passed for thirty minutes 26 A. W. GREELEY. or more and is at first of an exceedingly transitory nature. But with paramoecia that have been reared in a culture that has been made slightly acid by the fermentation of starch, and hence are in an acid-modified condition as before described, the sec- ondary reaction may begin as soon as the paramoecia reach the cathode. In these cases no gathering occurs at the cathode but each Paramaeciuin immediately reverses the stroke of the cilia and swims back to the anode. The process is repeated at that point, so that we have for a time no collection at either pole but a continuous line of paramoecia swimming in either direction until eventually they come to rest in about equal numbers at each end of the preparation. Under the most favorable acid conditions a number of paramcecia, varying from one to fifty per cent, of the whole number, exhibit an initial reaction toward the anode and a secondary reaction toward the cathode, while the remainder react in the manner described above. That this immediate reversal of the normal reaction and the initial response toward the anode are due to the acidity of the culture medium may be shown by the following experiments. If 5 c.c., of a neutral1 culture of paramcecia be isolated and tested to the current, it will be found that they all exhibit the character- istic response toward the cathode and form a dense gathering at that point. If now, however, from two to four drops of an ml 10 solution of hydrochloric or other acid be added to the culture, and, after standing for thirty minutes, the reaction to the current be again tested, it will be found that either an initial or an imme- diate secondary reaction toward the anode has set in. Also the addition of a small amount of acid to an already acid culture invariably strengthens the anodal response of paramcecia. Like- wise in every case in which I have tried it, the neutralization of the acid with NaOH, or the addition of the solution of a salt with a trivalent anion like Na3PO4 or Na3C(,H5O7 entirely destroys both the initial and the secondary response toward the anode, and leaves only the characteristic gathering at the cathode.2 1 It is necessary to use paramcecia from an approximately neutral culture for this experiment. The normal reaction toward the cathode is too firmly fixed in para- mcecia from a strongly alkaline culture to be reversed. 2 More striking results have been obtained with Vohox. After an exposure of half an hour or more to a slightly acid medium, practically every organism completely reverses its response, so that a dense gathering is formed about the anode. STRUCTURE OF PROTOPLASM OF PARAMCECIUM. 2/ In several instances, after a prolonged exposure to a medium which had been acidulated by the addition of hydrochloric or acetic acid, the paramcecia exhibit still another form of reaction. In these cases the paramcecia orient themselves transversely or slightly obliquely to the direction of the current, and swim very slowly from side to side of the preparation. At the same time, however, the organisms appear to drift passively toward the anode. I observed this form of reaction only about half a dozen times, but on each occasion the reaction was exceedingly well marked. A reversal of the current caused an instant reversal, of both the direction of the swimming and the passive drifting of the organisms. Notwithstanding the peculiar orientation of the para- mcecia, they all tended to form a gathering about the anode under these acid conditions. It has been observed by Loeb, Jennings and others that the addition of other substances> like NaCl, to the solution contain- ing the paramcecia will cause a reverse movement toward the anode. I investigated this question and found that the addition of almost any salt in sufficient quantities to extract water from the protoplasm will cause a more or less complete reversal. A large number of salts, of both positive and negative electrical conditions, were used, and the effect was seen to be purely osmotic in character, except in the case of the salts with trivalent anions or cathions. The former, as the phosphates and citrates, act like the hydrates in very weak solutions and completely destroy all traces of a response toward the anode. The latter, as A12C16, produces an almost instant coagulation of the proto- plasm and hence bring about the same effect as the extraction of water osmotically. It has been already stated that a lowering of the temperature coagulates the protoplasm, and it is interesting to note, that a partial reversion of the electrical response may be produced by this means also. After the paramcecia have been exposed to a temperature of 2° to 3° C, for one hour or more, the normal response is entirely lost, and a slight movement toward the anode can be detected. Our ignorance of the precise nature of the ciliary response is too great to allow even an attempt at an explanation of the mechanism of this response to the electrical current. We will 28 A W. GREELEY. first have to obtain a satisfactory explanation of the rhythmical contraction of the ectosarc which controls the cilia. It is certain that the surface tension relations of the muscular elements play an important role in this process. In Amccba the problem is very simple. The protoplasm contracts on the anodal side because of the neutralization of the charge carried by the protoplasmic particles and consequent increase in surface tension. Pseudopodia are thrown out on the cathodal side, and movement occurs in this direction because of the decrease in surface tension at this point. But for paramoecia it suffices at present to show that the sense of the response is not a fixed attribute of the organism which has been acquired by natural selection, but that it is a purely physical response to an external stimulus, and varies directly with the conditions under which it occurs. The ultimate determining factor of the response to the electrical current must be the elec- trical conditions of the protoplasm itself. In paramcecia from a normal alkaline culture the protoplasmic particles appear to be negatively charged. These paramoecia collect about the cathode where liquefaction of the protoplasm occurs. But with precisely the same conditions, under which liquefaction is produced not by anions but by cathions, i. c., an acid culture medium, we find that the characteristic gathering about the cathode does not occur, but the paramcecia tend to move toward the anode where the protoplasm would now become liquefied. The only explanation of this phenomenon that presents itself is the assumption, that in an acid medium the protoplasmic particles become positively charged in a portion of the organisms (for the reaction is never completely reversed). The partial reversal of the reaction by osmotic means is also due to an alteration in the electrical con- ditions of the protoplasm, as has been shown. Hardy's acid and alkali-modified colloids reacted also in an opposite manner to the electrical current because in the acid solution the particles were positively, and in the alkaline solution negatively charged ; and while the explanation is not so simple in the case of Paramoeciuni, because movement is effected by a complex motor apparatus, still the sense of the reaction must be ultimately due to the same cause in both cases, /. t\, the charge carried by the colloidal or protoplasmic particles. Moreover, STRUCTURE OF PROTOPLASM OF PARAMCECIUM. 2Q Hardy observed the same secondary reversed movement of the colloidal particles as has been observed in the reaction of Para- mcecium. All these facts make very evident the similarity which exists between the electrical conditions in the 'two solutions. Lillie ! has observed a similar relation between the response to the electric current and the chemical condition of the protoplasm whether acid or alkaline, in his experiments upon nuclear and cytoplasmic structures cited at the beginning of the paper. He shows, that, when exposed to the electric current, nuclear struc- tures, which contain a large amount of nucleic acid, move toward the anode, while cells very rich in cytoplasm, which is basic in reaction, move toward the cathode. 3. Chemota.\is. — A large number of important contributions have been made during the last few years to our knowledge of the reaction of protozoa to chemical stimuli. Most notable have been the remarkable series of investigations carried on by Jen- nings, who has given us not only a complete account of the sense of the reaction of many protozoa to a wide range of chemical substances, but also an accurate description of the method of the reaction in each case. My own results agree with those of Jen- nings in all essential points. I confirm his account of the " motor reaction" of Paramcecium when under a chemical stimulus. It was not my purpose to repeat any of Jennings' experiments, but only to ascertain the chemotropic reactions of Paramcecium under various conditions, to see if they could be modified by external influences as was the galvanotropic response. The only respect in which my conclusions depart from Jennings' is that my experi- ments seem to show that the chemotropic reactions of Paramcecium which he describes are not of universal occurrence, but limited to paramoecia which have been reared under definite chemical sur- roundings. In other words, the sense of the chemotropic reac- tion, like the galvanotropic, depends upon certain chemical con- ditions of the environment. Jennings found that paramoecia were in general positive to weak acids (/. e., formed a gathering within the drop) and nega- tive to weak alkalis. He also observed that they reacted in a constant manner to a large number of salts. In my own experi- 1 Lillie. loc. cif. 30 A. W. GREELEV. ments, I find, that only those paramcecia which have been reared in a slightly acid culture medium are positive to acids, and that paramcecia from clear alkaline cultures are negative to acids and positive to alkalis. It appears also that salts with trivalent cathions act like acids, and salts with trivalent anions act like alkalis. No definite conclusions could be drawn from the action of the univalent and bivalent salt solutions. For the purpose of the experiments, we are most vitally concerned with the reaction of those solutions which carry the heaviest positive or negative charges of electricity. The greatest caution is needed in determining the acidity or alkalinity of the culture medium. A carefully prepared litmus solution is the best indicator. Phenol-thalein may also be used for the detection of small amounts of alkalis. Paramcecia freshly reared in a culture whose alkalinity has been determined in this way invariably react to the solutions used as follows : To ?;//2OO HC1, HNO3, H2SO4, and acetic acid ; m/Soo A12C1,; and Fe2Clc, they are negative. To m / '200 NaOH, KOH, Ba(OH)2 and Sr(OH)2; 1/1/480 Na3C6H.O7 and Na3PO4, they are positive. The positive reaction is shown by the organism swimming into the drop, and then giving the motor response, described by Jen- nings, when they come in contact with the outside water, so that a gathering is formed within the drop. The solutions to which the paramcecia are negative provoke the motor response when the organisms first come in contact with the outer edge of the drop, with the result that the solution is left empty. In some cases the paramcecia appear to be entirely indifferent to a solu- tion, and swim in and out in an undisturbed manner. Such a reaction will also be classed as negative. If the paramcecia from the same alkaline culture be tested from day to day as the alkalinity is being gradually neutralized by the formation of acid during the fermentation of bread, or by the addition of free acid to the culture, it will be seen that the response to these solutions slowly changes, until finally the chemotropic reaction is completely reversed, and now in the acid medium the paramcecia are positive to acids and salts with a tri- valent cathion, and negative to alkalis and salts with a trivalent anion. Likewise, if the culture be again made gradually alka- STRUCTURE OF PROTOPLASM OF PARAMCECIUM. 3 I line, the first form of reaction, characteristic of alkaline cultures, returns, so that it becomes evident that the sense of the chemotropic reaction depends direptly on certain chemical condi- tions of the surrounding medium. The transformation from the first form of reaction to the second, and vice versa, is a very gradual one, so that it is not immediately effected by the chem- ical change in the surrounding medium, and a considerable time may elapse between the neutralization of the alkalinity of the culture, for example, and the loss of the positive response to alkalis ; but eventually the reaction occurs as has been described. Thus paramcecia are seen to seek out those chemical conditions which bring about a liquefaction of the protoplasm. The sense of this response also is apparently determined by the electrical condition of the protoplasm. An interpretation of the mechanism of this response to elec- trolytes is as impossible as it was in the case of the reaction to the electric current. But, since in each case the paramoecia col- lect under the same electrical conditions, both responses must be ultimately due to the same reaction of the contractile layer of the protoplasm to electrically charged ions, and this reaction must consist largely in the effect which the electrically charged ions, have upon the surface tension of the contractile elements of the protoplasm. Experiments upon Amoeba bear out this hy- pothesis. Cathions always produce a contraction of the proto- plasm, while anions produce a relaxation or the extension of pseudopodia, because the former increase the surface tension of the protoplasm while the latter neutralize it. Thus Amaba, like Paramcecium, is positive to predominately negative solutions, but in the one case the response is accomplished by the imme- diate effect which the anions have upon the surface tension of the protoplasm, while in the other case it is brought about through the agency of a complex motor apparatus. CONCLUSIONS. We see that precisely those chemical changes in the surround- ing medium, which modify the structural reactions of the proto- plasm of Paramcecium to solutions of electrolytes, modify also the reactions of the organism to electrical and chemical stimuli. 32 A. \V. GREELEY. The protoplasm of paramoecia from an alkaline culture is lique- fied by temperatures between 23° C. and 30° C., by anions, and at the cathode during the passage of the constant current. The paramoecia also react positively to all these chemical and physi- cal conditions. The protoplasm of the same paramoecia is coagulated by temperatures below 20° C. and above 30° C., by cathions, and at the anode during the passage of the current. The organisms are negative to all these conditions. The struc- tural changes produced by electrolytes are partially reversed in paramoecia from a slightly acid culture, and the reactions of the organisms are also partially reversed to the electric current, com- pletely so to solutions of electrolytes. In every case the reaction of a Paramcecium to an external stimulus leads it to remain under those conditions which liquefy the protoplasm. Attrac- tion is accompanied by liquefaction, repulsion by coagulation. As far as the physical structure of the protoplasm is concerned, the conclusion from these facts seems to be that the protoplasmic particles are physically identical with colloidal particles. Hence the protoplasm of Paramceciuin is essentially a colloidal solu- tion whose particles carry a definite charge of electricity. The sign of this charge appears to depend on certain external chemi- cal conditions, of which the alkalinky of the surrounding medium may be taken as one of the most important. The sign of this charge is seen to determine not only the structural modifications of the protoplasm, but also the reactions of the paramcecia to chemical and electrical stimuli. This conception of the physical structure of protoplasm is also used to explain the effect of external conditions on the processes of cell division, growth and movement through the operation of the laws of surface tension. ZOOLOGICAL LABORATORY, WASHINGTON UNIVERSITY. February 10, 1904. EXPERIMENTAL STUDIES ON THE DEVELOPMENT OF ORGANS IN THE EMBRYO OF THE FOWL (CALLUS DOMESTICUS). FRANK R. LILLIE. II. THE DEVELOPMENT OF DEFECTIVE EMBRYOS, AND THE POWER OF REGENERATION.1 Born has shown that young embryos of the frog possess im- mense power of healing smoothly cut wounds, and that the vitality of even small isolated parts is remarkable ; pieces of the head or of the tail, for instance, may continue to grow and de- velop in spite of the complete absence of the heart, blood and blood vessels, and nervous system, for a period of about three weeks, or until the yolk contained in the cells is fully consumed. " The development of each organ progresses as far as the cut surface as well as in the normal larva, no matter what the posi- tion of the cut surface may be ; the absence of the heart or the brain does not affect the subsequent processes of growth and dif- ferentiation in any marked way." These conclusions are of great importance for the physiology of development. So far as I know, similar experiments have not been performed on the embryos of Auuiiota, and the following account may serve to fill up the gap in part. Striking differ- ences appear between the embryo of the fowl and of the frog in regard to the vitality of defective embryos, or of parts of embryos : The chick embryo dies very rapidly after destruction of the heart, for the embryonic tissues are dependent to an extreme de- gree upon the circulation ; it therefore follows that small parts of an embryo cannot undergo differentiation as in the frog. On the other hand, the growth of the extra-embryonic blastoderm is inde- pendent of blood supply, and may continue in eggs in which there is no embryo owing to destruction of the heart, until nearly the entire yolk is covered. This difference between the embryo of the chick and its own extra-embryonic blastoderm or the embryo of i The first part appeared in BIOL. BULL., 1903, Vol. V., No. 2 : " Experiments on the Amnion and the Production of Anamniote Embryos." 33 34 FRANK R. LILLIE. the frog, may be explained by the simple consideration that the cells of the embryonic tissues proper in the chick are devoid of yolk or other nutriment, and hence are dependent for their sub- sequent growth and differentiation upon the circulation ; whereas the cells of the frog embryo are loaded with food in the form of yolk ; and the extra-embryonic blastoderm of the chick is a diges- tive organ lying on an immense reservoir of food. The necessity of circulation for all normal development of the chick embryo beyond the stage of about 33 hours (12-14 somites) at once limits the range of defective embryos capable of develop- ment to those possessing a heart and vitelline circulation. Another limitation arises from the extreme sensitiveness of the embryo to removal of parts of the brain ; although I have made over seventy experiments on the brain, none of the embryos, in which the injury extended back of the optic stalks, has developed for more than about forty-eight hours after the operation. (The operations on the head form a class in themselves and will be discussed in a sepa- rate paper.) The reason for the large number of fatalities in opera- tions in this region is probably not due to any trophic function of the nervous system, at least in stages younger than 72 hours, but either to direct injury to the anterior end of the heart, or to malformations of the amnion consequent on the operation. On the other hand, embryos may survive the destruction of a con- siderable portion of the posterior end, and develop normally for several days as least. All the defective embryos to be described resulted from operations of this kind, performed, with one excep- tion, on embryos in which the tail-bud is just forming after 50- 60 hours' incubation. A larger or smaller part of the posterior end was destroyed by cauterization. The embryos might bear complete destruction of the posterior end up to the vitelline arteries, provided these were not injured, without any apparent detriment or hindrance to the development of the uninjured parts. If the vitelline arteries were destroyed, the embryo never survived. Fig. i gives a view of an embryo of about the age of those used for operations. In this case 29 somites are formed ; the number was certainly not over 30 at the hour of operation in any case. The somites are continued backward by the undivided segmental plate represent- DEVELOPMENT OF EMBRYO OF FOWL. 35 ing a considerable number of potential somites. As will be shown later, somites 26 to 32 are the ones that normally form the musculature of the leg. Thus the undivided mesoblast at this stage includes a large part of the trunk. The vitelline arteries at this time leave the embryo opposite to the twenty- first or twenty-second somites ; this position appears to be very constant. Thus the theoretical limit of the experiments is from the hind end to about the twenty-second somite, because -18 FIG. i. Camera drawing of chick embryo with 29 somites; operation diagram. For explanation see text. the embryo cannot live if the vitelline arteries are destroyed. The rudiment of the allantois lies beneath the unsegmented mesoblast beginning, approximately, a short distance back of the thirtieth somite. Fig. i serves at the same time as a diagram of the operations. At the time of each operation a sketch was made of the embryo, and the part destroyed was indicated by shading the posterior end correspondingly. This sketch, naturally, did not show somites, so the present operation diagram is constructed from data deter- 36 FRANK R. LILLIE! mined after the operation. It agrees, however, fairly well with the original operation diagrams which were simply rough esti- mates of the amount destroyed by the operation. It should be distinctly understood that the figure is a result of study of the anatomy of the defective embryos. In further explanation it should be added, that the numbers to the right are the numbers of the experiments ; those to the left of the somites. The lines leading to the numbers indicating the experiments are drawn across the body to indicate, that, in the experiment in question, all back of the line was destroyed ; the cross on each line marks the junction of reference and operation lines. A word concerning the enumeration of the somites ; the somite numbered I possesses a short anterior process, that is probably an independent somite. It is, however, so inconspicuous in many embryos that it seemed better for the present purpose not to enumerate it. Somites 17, 18 and 19 are especially marked because they are the wing-somites, i. e., the somites that will form the major portion of the bone and muscle of the wing. The twenty-sixth is the first leg somite. By this reckoning, there are three cephalic somites, or, reckoning in the incomplete one, four. Referring to Fig. I again, it will be seen that the organs destroyed in such operations are : (i) The hind end of the neural tube ; (2) the hind end of the notochord ; (3) the mesoblastic segmental plate and often certain of the posterior mesoblastic somites ; (4) the hind gut including the rudiment of the allantois ; (5) the hind end of the Wolffian body and ducts. I may say at once that no true regeneration of these structures takes place. (For discussion of this, see p. 50.) Therefore the problems of interest became narrowed down to the differentiation of the uninjured parts, and my attention has been particularly drawn to the behavior of the mesoblastic somites. In the somites we have an originally homonymous series of structures, the seg- mentation of which becomes strikingly heteronymous as develop- ment proceeds, some entering into the head, others the neck, others the wing, the thorax, abdomen, leg and tail. Definite somites, i, e., somites in a definite numerical position in the series, form the skeleton and muscles of each of these regions. It would DEVELOPMENT OF EMBRYO OF FOWL. 37 be interesting to determine whether or not somites had the same role in defective as in normal embryos. The numerical value of the somites in the chick seems to be normally as follows : 96-hour Chick. Eight-day Chick. I, 2, 3, Cephalic, hypoglossus region.1 4 to 1 6, Prebrachial, trunk. I to 13, prebrachial. 17, 18, 19, Brachial. 14, 15, 16, brachial. 20 to 25, Between wing and leg. 17 to 22, between wing and leg. 26 to 32, Leg somites. 23 to 29, leg. 33 to 35> Region of cloaca, 301032, region of cloaca. 36 to 42, Caudal somites. 33 to 38, caudal. The count in the eight-day chick is really an enumeration of nerves. For the four-day chick the exact number of the leg somites was determined by comparison with the enumeration of nerves of the eight-day chick. The wing somites may readily be distinguished at four days, in entire mounts, by their size and by the relatively large amount of mesenchyme formed opposite them. In embryos of more than four days of age the enumeration by nerves is much the easier ; and, as the brachial nerves (14, 15, 16) are very much larger than their neighbors (Figs. 3 and 5, B.P.}, it is simplest in determining the place of a postbrachial somite or nerve to count only the somites back of the last brachial nerve. Thus the postbrachial somites i— 6 are between arm and leg ; postbrachial 7-13 are leg-somites, etc. Experiment 125. In this experiment a very considerable part of the hind-end was destroyed at the time of appearance of the tail-bud (see operation-diagram), and the embryo was allowed to develop for about four days more (91 hours). The egg was then reopened. The vascular area covered at least three fourths of the yolk. There was no allantois. The embryo lay in large part beneath the blastoderm, but a large aperture in the latter towards the hind-end of the embryo was filled by the amnion through the transparent walls of which the embryo could be distinctly seen. A large part of the blastoderm was removed with the embryo and examination of the under surface was then made in salt solu- 1 In this enumeration I have omitted again the incomplete anterior cephalic somite. FRANK R. LILLIE. tion. The amnion was very well distended, and attached round the margins of the somatopleure at the hind-end. Behind this c FIG. 2. Three views of a defective embryo (experiment 125) with membranes removed ; total age about six days. A, amnion ; /, ectopic intestine ; V, vitelline artery and vein ; L, rudiment of hind limbs. DEVELOPMENT OF EMBRYO OF FOWL. 39 the embryo was naked and there was slight ectopia of the intes- tine (Fig. 2, C\ Figures 2, A, B and C show this embryo from the two sides and from behind. The last figure shows especially well the attachment of the amnion and the ectopia of the intestine ex- ternal to this. The hind-end of the embryo back of the vitelline vessels is wanting. On the other hand, all parts in front of this are normally developed. On each side, at the hind-end of the Wolffian ridge, rudiments of the hind limbs are present as prominent conical stumps. From this it would appear at first sight that there has been some regen- eration. However the question can be decided only by ascer- taining the numerical value of the segments concerned in the formation of these stumps. The embryo was cut into sagittal sections for this purpose. The enumeration of the post-brachial nerves is the same on both sides, seven pairs being present. The stumps are innervated only by the seventh on each side (Fig. 3). In the normal chick, the nerves innervating the leg are the seventh to the thirteenth postbrachial. Thus it would appear that only one leg somite on each side was uninjured by the operation, and this is the only one that has contributed to the formation of the rudimentary leg. None of the six somites between arm and leg has undergone any alteration of its normal numerical value. This result must therefore be interpreted in the sense of normal self-differentiation of the somites concerned. Structure of the Leg-rudiments. — But, even though the defi- ciency of leg somites has not stimulated their immediate neigh- bors in front to any act of supererogation, it might be that the only leg somite remaining has produced more than its wont. In this connection the structure of the stumps is of interest : Owing to their positions, the left one is cut longitudinally and the right one transversely in the sagittal series. The structure is the same on the two sides (see Fig. 3) ; there is a single curved rod of pre- cartilage extending out nearly to the tip of the limb ; this is sur- rounded by a mass of dense mesenchyma, evidently premuscle tissue, and externally are the elements of the skin. Now, in the normal limb of corresponding age, the skeleton of the thigh, FRANK R. IJLLIE. crus and pes are separate chondrifications. The crus and pes are thus certainly not represented in the rudimentary hind-limb of this embryo. In the normal chick the seventh postbrachial nerve innervates only some of the proximal muscles of the leg, those extending from the pelvis to the femur. On the principle that the muscular distribution of the nerve is confined to the derivatives of the cor- responding somite, it follows that the seventh postbrachial somite contributes to the formation of only the upper segment of the leg. \ FIG. 3. Sagittal section of embryo shown in Fig. 2, cut considerably to the left of the middle line. A, amnion ; B.P., brachial plexus (nerves of wing) ; /, poster- ior end of intestine ; IV. D., Wolffian duct ; 1-7 first to seventh postbrachial nerves. The absence of the crus and pes is therefore what might be expected if the principle of self-differentiation were rigidly adhered to. We have seen that, as a matter of fact, these are absent in the rudimentary hind limbs of this embryo, so we may conclude, at least, that there is no evidence that the seventh postbrachial somite has produced anything beyond the normal as a result of the absence of the following somites of the leg. Alimentary Tract. — Practically all of the splanchnopleure of the embryo posterior to the vitelline arteries was destroyed by DEVELOPMENT OF EMBRYO OF FOWL. 4! the operation. In the resulting embryo, the trachea is a long tube, the lungs are budding out, the oesophagus is well formed, and the stomach is in two divisions, in the anterior of which the glands have begun to form, and the intestine is slightly coiled and opens on the hind surface, across which it is continued as a superficial strip ending in a free blind tag with a slight lumen (Figs. 2 C and 3). The liver and pancreas are normally formed. In fact the parts present are apparently in the same condition that they would have been had the embryo been intact. The allantois is entirely absent ; and there is no evidence of any compensating groivth of the intestine. Nervous System. — The spinal cord ends bluntly, but the neural canal is prolonged backwards at its ventral angle into a short process which tapers into a bundle of neurites that may be fol- lowed a short distance, and then terminates abruptly. It would appear that the descending tracts in the cord have caused the prolongation, and have then pushed out into the adjacent tissues. A similar prolongation appears in all embryos defective at the hind-end (Fig. 3, AJ. Excretory System. - - The Wolffian ducts are much dilated, as there is no outlet for the secretion of the mesonephros (Fig. 3). No metanephric outgrowths have arisen from them, although in the normal embyro of this stage these are well developed. The metanephric region of the Wolffian ducts was of course destroyed by the operation ; but, as the ducts are continuous structures, one would not anticipate that the capacity for producing metanephric outgrowths would be limited to a short stretch at the posterior end. Of course another explanation of their absence than that of limitation of potency to a short region of the Wolffian duct is possible, i. e., that suitable predisposing external conditions for their formation are strictly limited (e. g., that their development de- pends on a certain amount or kind of development of the allantois or cloaca). Experiment 93. In this experiment a very considerable portion of the hind-end was cauterized and removed (see operation diagram). Seventy- two hours later the egg was reopened and found living. The vascular area covered about three fifths of the yolk ; the circula- tion was very active, and there was no allantois. 42 FRANK R. LILLIE. Fig. 4 is a view of the embryo in the amnion, drawn from the under side of the blastoderm. The vitelline blood vessels are attached to the center of the defective hind-end, the jentire trunk back of these vessels being absent. The amnion is well dis- tended ; its line of attachment is indicated by the dotted line. Just behind this line of attachment is seen a little tag, the hind- end of the intestine. No trace of the hind-limbs is visible ex- ternally. This embryo was cut into sagittal sections. In these there is no trace of the hind limbs, and no evidence that the posterior myotomes or sclerotomes are modified towards the formation of limbs. The embryo is not quite so far advanced in development FIG. 4. Defective embryo (experiment 93) in the amnion; part of the folded under surface of the blastoderm is shown. A, amnion ; /, ectopic intestine ; V, vitelline arteries and veins. as that of experiment 125, but it is certainly old enough to show the rudiments of the hind limbs or any part of them, were there any tendency towards their formation. Study of the sections shows that there are only five pairs of post-brachial ganglia and nerves (Fig. 5), the fifth on both sides DEVELOPMENT OF EMBRYO OF FOWL. 43 being very small, and evidently partially destroyed by the opera- tion. As the first leg-somite is the seventh post-brachial, the absence of rudiments of the hind limbs in this embryo is readily understood. B FIG. 5. Three sections from a sagittal series through the embryo shown in Fig. 4. A, amnion ; Ao., aorta ; B.C., ganglia of the nerves of the wing ; B.P., nerves of wing; C, cord; N, notochord ; V. U., umbilical vein ; IV.£)., Wolffian duct; 1-5, first to fifth postbrachial ganglia or nerves. Thus there are two somites less in this embryo than in that of experiment 125. In the latter, which includes one leg somite, rudiments of the hind limbs are formed ; in the present case no 44 FRANK R. LILLIE. such rudiments are formed, although the embryos are very much alike in every other way. It would be interesting to compare an embryo with six post-brachial somites, but I have none yet. The central nervous system ends bluntly, except for a short prolongation of the ventral angle of the canal (Fig. 5,^). This condition, which is characteristic for all embryos defective at the hind-end, appears to be due to growth of fibers in the ventral motor zone, for a bunch of neurites extends back on each side of the ventral middle line nearly to the hind end of the notochord. The notochord extends backwards to the extreme posterior end, thus some distance further than the spinal cord. It thus appears, as in the other embryos of this class, to have undergone some regeneration, or at least growth, at the hind end. T/ie intestine opens at the posterior end, and is continued as a flat strip along the surface of the splanchnopleure. It is interest- ing to note that this strip has the same histological structure as the walls of the tube proper, showing that the histogenesis depends upon the character of the cells, and not upon such external factors as the form of the tube. The liver is normal. There is no trace of the allantois ; but the stem of the umbilical veins may be seen, in a very rudimentary condition, running in the septum transvcrsum to enter the anterior face of the liver (Fig. 5, B). The Wolffian bodies are well developed, and the Wolffian duct much enlarged as in other anallantoic embryos (Fig. 5). Meta- nephric outgrowths are absent. Summing up, we may say, that all the embryonic rudiments present have differentiated in the normal fashion. There has been no modification of the numerical value of the somites, no compensating growth, and no regeneration, unless we except the elongation of the notochord. Those embryonic organs, whose rudiment or ordinary locality was destroyed, are simply missing. Experiment /./. A large part of the hind end was destroyed at the stage of about 52 hours (see operation diagram). The egg was reopened about 68 hours after the operation. The vascular area covered fully half of the yolk ; the heart-beat was vigorous. DEVELOPMENT OF EMBRYO OF FOWL. 45 Fig. 6 gives two views of the embryo. The hind end is en- tirely absent, and the opening is filled by the enlarged Wolffian ducts. The band of tissue between the two ducts is the hind end of the alimentary tract. The right side is seen to be less de- veloped than the left. In the view from the opposite side, it can be seen that there is a rudiment of the hind-limb on the left side ; but apparently none on the right. This embryo was cut into transverse sections. Study of these shows that there are eight postbrachial ganglia on each side, though the eighth on the right side is smaller than on the left. FIG. 6. Two views of a defective embryo (experiment 74). A, amnion ; /, strip of intestinal epithelium; L, leg; IV. D., Wolffian duct. The vitelline blood vessels are somewhat abnormal. On the left side the seventh and eighth postbrachial nerves enter the rudiment of the hind limb. On the right side there is no such rudiment, and the seventh and eighth nerves, which are quite large, end bluntly in the mesenchyme, almost as though cut off. The question arises why there is no limb rudiment on the right side ? One can explain this by assuming that the operation destroyed the seventh and eighth postbrachial meso- 46 FRANK R. LILLIE. blastic somites on this side, but not the ganglia and cord of these somites (see Fig. i). The form of the embryo indicates that this is true in part ; but there is a small myotome in the eighth post- brachial somite of this side, and in the seventh the myotome is quite as well developed as on the other side. It would appear probable, then, that the most lateral portions of the seventh and eighth postbrachial mesoblastic somites were destroyed, together with the somatopleure lateral to this ; either the rudiments of the hind limb were included in the destroyed parts, or their develop- ment was prevented by the scar-tissue. It is impossible to say whether or not the two somites concerned in the formation of the left leg have formed more than the normal. The central nervous system terminates abruptly, except that the ventral portion of the canal is continued back as a narrow epithelial tube. The notochord extends some sections back of the termination of the neural tube, showing that it has been added to at its hind end since the operation. The Wolffian ducts are immensely enlarged (see figures). There is no metanephric outgrowth. The allantois is entirely absent. The alimentary canal is normal back to the defective region and terminates in a band lying between the two Wolffian ducts (Fig. 6). Experiment 18. The tail-bud of the embryo was just formed at the time of the operation, and it was completely destroyed with a heated needle (see operation diagram). The injury did not extend so far for- ward as in the other cases mentioned and was asymmetrical (Fig. i). Fifty hours later the egg was reopened, and the embryo was found living, the heart beating actively ; the vascular area covered about one third of the yolk. Fig. 7 shows the embryo as it appeared from the under sur- face of the blastoderm. It will be observed that the amnion ceases with a free edge immediately in front of the hind-limbs ; thus the tail-fold of the amnion destroyed by the operation has not regenerated, nor is the allantois visible externally, as it always is in normal embryos of this age. The hind-limbs appear normal ; part of the tail at least is present. DEVELOPMENT OF EMBRYO OF FOWL. 47 The enumeration of the postbrachial nerves in the sections gives the following results : I to 6 on each side are in front of the leg-rudiments ; the following seven nerves on each side enter the leg-rudiments. On the right side there is only one incom- plete ganglion back of the last leg-nerve, /'. e., the fourteenth postbrachial ; the nerve from this cannot be traced. On the left side there are four complete ganglia and myotomes back of the last leg nerve (14 to 17 postbrachial). Thus, as might be expected, the hind limbs are normal, and the cloacal region is strongly affected. FIG. 7. Defective embryo (Experiment 18). In the embryo of this experiment the Wolffian ducts have normal relations, but the allantois is thick-walled and undilated ; the entodermal cavity of the latter opens out into a groove in the splanchnopleure, leading back to the hind-gut, which is not closed as it normally is ; in fact in this embryo the cloaca is the only closed region of the hind-gut ; immediately in front of the cloaca it is open ventrally. The stalk of the allantois therefore appears first as a groove on the left side ; followed forwards, this 48 FRANK R. LILLIE. deepens gradually into a canal that penetrates into a mass of mesoblast connected with the somatopleure, and excavates a large irregular cavity in this. The mass extends forwards to the tip of the ventricle and ends in a bifurcated extremity. This represents the allantois. the main mass of which is composed of loose, very vascular mesenchyme forming an appendage to the somatopleure on the left side. On the right side the mesoblast of the somato- pleure in the allantoic region is much hypertrophied and forms a large free lobe without any entodermal contents. It is remarkable that the allantoic rudiments should show such considerable power of growth in the absence of the usual stimulus of internal pressure, which is precluded by the open groove-like connection with the intestine. Experiment 24.. The hind end of an embryo was cauterized with the aim of destroying the incipient tail-bud and the region of the hind- FIG. 8. Defective embryo (Experiment 24). limbs. The egg was reopened 48 hours after the operation. The vascular area covered about two fifths of the yolk, and appeared to be entirely normal. DEVELOPMENT OF EMBRYO OF FOWL. 49 Fig. 8 represents the resulting embryo from the dorsal sur- face ; the rudiments of the hind-limbs are apparently as well developed as those of the anterior limbs, and the posterior por- tion of the trunk is separated from the anterior by a sharp depres- sion. The tail turns to the left, and tapers to an end beneath the left hind-limb; part of its course is hidden by a fold of the soma- topleure, as shown in the figure. The depression evidently repre- sents the anterior limit of the injury. This embryo was sectioned. On examination it was found that, although the region behind the depression has the usual structure of the postanal region of the body, its neural tube has no connection with that of the trunk, and the notochord is absent in the anterior third. Mesoblastic somites occur in this region, but the main part of its substance is made up of loose vascular mesenchyme. It is plain, then, that this part has not regenerated in the sense that it has grown out from the embryo ; and there remains only the assumption that its parts have formed from embryonic tissue remaining in this region after the cauterization. The absence of the notochord throughout almost all of the tail shows that the destruction due to cauterization was very extensive ; its absence beneath the neural tube for a considerable distance shows, that the neural tube has formed secondarily in this region ; because any operation, that would destroy the notochord, would neces- sarily destroy the neural tube. Farther, the relation of the Wolffian ducts and allantois to the cloaca are abnormal, and the hinder portion of the intestine has no connection with the anterior portion ; all of which shows that there was originally complete destruction of tissues through to the entoderm in the anterior part of the cauterized area. There seems to have been in this case a reorganization of the embryonic tissue of the caudal region. On the left side I find ten postbrachial ganglia back to the region of the defect. Thus at least four segments of the normal hind limb-region are included, viz., 7, 8, 9, 10. The last is small, representing only a fraction of the normal ganglion. On the right side there are eleven full-sized postbrachial ganglia. Thus there are at least one and one half more metameres on the right than on the left side, and this tallies with larger size of the right hind limb and with the operation diagram. 50 FRANK R. LILLIE. The only interpretation of this result is that the needle destroyed a transverse section of tissue and separated the tail-rudiment from the rest of the embryo, at the same time causing some injury to the tissue of the tail ; secondary union then took place. The tissue along the line of junction is peculiar in many respects, including much material, probably remnants of cauterized parts, that takes the plasma stain. Experiment 187. In this experiment the operation was made at a stage of about 24 hours ; the medullary plate was just formed and the primitive streak was yet long ; probably 3 to 5 somites were present (Fig. 9). Three spots were marked with an electrode, one behind the other in the middle line towards the posterior end of the primitive streak. The three spots were at first separate, but later examination sho\ved them running together ; so that all of the — »- — -" FIG. 9. Operation diagram for experiment 187 ; drawn from a blastoderm of the same lot of eggs used in this experiment ; preserved at the time of the operation. axial structures were destroyed in this region. The embryo was allowed to develop two days longer and was then preserved. DEVELOPMENT OF EMBRYO OF FOWL. 5 I It will be seen (Fig. 10) that this operation has taken us in front of the vitelline arteries ; having been performed before the development of the arteries, it was possible for the rudiment of the vascular system to readjust itself to the new conditions, and establish a vitelline circulation. The vitelline arteries are, how- ever, in about their normal position, i. e., about opposite the 20-22 somites, hence behind the embryo in this case. It is difficult to see why they should have this position and not be shifted farther forward, if the view of His, that the main blood vessels develop from the paths of least resistance in the primitive vascular net- work, be true. On the right side of this embryo there are 14 spinal ganglia, and on the left 14 complete and part of the fifteenth ; but as, in the fowl, the first two cervical nerves lack ganglia, there are present 16 spinal nerves on the right side, and 17 on the left. X FIG. 10. Defective embryo (experiment 187). The drawing was made from the stained and cleared specimen. A, vitelline artery ; V, vitelline veins ; W, wing bud ; X, sac filled with blood, behind the embryo in the prolongation of its axis ; this has no connection with the vessels of the embryo. The normal innervation of the wing is from the fourteenth to the sixteenth spinal nerves. It would thus appear, on the evi- dence of the nerves, that the operation has destroyed everything back of the wing somites. The wing -buds have a development that appears perfectly normal for this stage. 52 FRANK R. LILLIE. On the right side there are nineteen myotomes, and on the left twenty and a half. Fig. 10 shows the myotomes of the right side as they appeared in an entire mount of the stained embryo in oil; only 16 can be counted. But sections show, that the large anterior one is really three, and the nineteenth is repre- sented by the mesoblast behind the last complete somite. If we reckon the three anterior somites as cephalic, this leaves sixteen trunk myotomes, and the enumeration agrees with that of the nerves. On the other side of the embryo there is distinctly one more cephalic myotome than on the right side ; if, then, we reckon four as the number of cephalic myotomes on this side/ there are sixteen and a half trunk myotomes, a result that agrees with the enumeration of the nerves. The organs in this embryo, as in the others described, have developed normally as far as the cut surface. The Wolffian ducts are enlarged by internal pressure. There has been no compensa- tory growth and no regeneration, if we except the notochord. GENERAL DISCUSSION. The chief value of these experiments consists in the fact that they clear the ground for farther researches of a more definite character. It is probable that the results obtained will hold in general for the Amniota, and they may, therefore, be of value in the interpretation of teratological phenomena. Results that I have obtained from destruction of large portions of the brain in young embryos have convinced me that great caution must be exercised in interpreting conditions associated with anencephaly. It should be possible, and I am convinced that it is possible, to produce, these various conditions experimentally. Until this is done it seems to me that conclusions, based on teratological ma- terial, as to the trophic value of the embryonic nervous system are debatable. The following questions definitely raised by the experiments just described may be answered here in part : I. As to Regeneration. — The only organ showing evidence of regeneration is the notochord. In all of the embryos described, i Froricp recognizes four cephalic myotomes in the chick ; the most anterior, how- ever, becomes rudimentary at a very early stage, and soon disappears ; this condition seems to have been attained on one side in this embryo, but not on the other. DEVELOPMENT OF EMBRYO OF FOWL. 53 the notochord extends a short distance back of the neural tube and its end is very protoplasmic ; its further extension in each case is definitely limited by the surface of operation, against which it is pressed (Fig. 5, A). One receives the impression that its growth has been stopped by the mechanical hindrance. One does not expect a vertebrate to show axial regeneration of the trunk, but in certain vertebrates the tail may regenerate. What is the condition in the chick embryo? No. 18 is the only case cited in which any of the caudal somites were left uninjured. In this case there was no regeneration, so far as could be judged. The very first experiments that I performed on chick embryos were made to determine whether or not the limb-buds might regenerate. The operations were limited to the wing-buds. I found that it was possible to amputate the right one close to the body at a time when its width was equal to or greater than its length, and even later (four or five days); the wound in the amnion might close, and the amputated part remained in the amniotic cavity as evidence of the operation. In the only cases in which the embryo lived for any considerable period after the operation, there was absolutely no sign of regeneration, though the wing- bud of the opposite side increased in bulk several times. So far as I know, the only evidence, that the organs of the chick embryo possess any different power of regeneration from those of the adult, consists of the above observations on the noto- chord (tissue regeneration) and of Barfurth and Dragendorff' s1 observation on the regeneration of the lens of the eye from the edge of the optic cup. The last depends on observations on a single case, and in this case the extent of injury to the eye was doubtful. Until the result is confirmed, I believe we are justified in passing it over. There would remain, then, the general conclusion, subject only to the qualifications already noted, that the embryo of the chick pos- sesses no greater power of regeneration than the adult, 2. As to the Somites. — Somites always retain their normal numerical value. But the experiments were not adapted to analyze very accurately the numerous problems presented. 1 Dietrich Barfurth and O. Dragendorft", " Versuche iiber Regeneration des Auges und der Linse beim Huhnerembryo," Anat. Anz., Erganzungsheft zum XXI. Bd., 1902. 54 FRANK R. LILLIE. Thus the operation destroyed in each case not only definite somites, but also all posterior to the first one injured, and the body wall lateral to the somites. With more refined methods, it may be possible to eliminate single somites from within the series, and to avoid injury to adjacent tissues. Such a technique would en- able one to analyze many of the problems offered by the somites : to determine, for instance, the exact part played by them in devel- opment of the limbs, the order of origin of the most anterior somites, etc. Such problems are now being studied with partial success, and the results will be published later. UNIVERSITY OF CHICAGO, March, 1904. THE RELATIONS OF THE ANTERIOR VISCERAL ARCHES TO THE CHONDROCRANIUM. W. K. GREGORY. « The articular relations with the chondrocranium of the upper and lower jaw-cartilages and the hyomandibular, as typified in Ceratodiis, in Sqiialns, and in Noiidanus, are in themselves of course generally understood, but comparison of the current defi- nitions and usages of the corresponding terms " autostylic," "hyostylic," " amphistylic" reveals considerable discrepancy, which is highly confusing to the general student ; hence the present endeavor to standardize these terms and to give, as far as needed, their synonymy. From the analysis necessary for the accomplishment of this purpose it has become evident that all the current definitions of "hyostylic," " autostylic" and "amphistylic " are in one way or other unsatisfactory, and that if these conceptions are to retain anything more than historic interest they will have to be ex- tended to include the relations to the chrondrocranium not only of the hyomandibular but also of the hyoidean arch as a whole and of its distal or ventral half, the " hyoid " or ceratohyal. By this extension we are enabled : first, to apply separate and clearly diagnostic terms to the suspensorial conditions of the very phy- logenetically separated groups Dipnoi (autostylic) and Holoceph- ali (" holostylic" ), hitherto lumped together under the single term "autostylic"; second, to differentiate under the generic concept " hyostylic " four specifically well-marked modes, ('S"s I I ~ "^- a a ^ . E , CJ 1= •_' fc ° cj 15 .'- ">, CU c"""~ **" -^- ">, a-7- en ~. ,5 s 1 "- s £ ^ O •z a J ° a i 75 .en 11 S"e .15 i^1 s > .<« e M o o 73 1 |_rt"£:0 s- o C; 7J I^J oj III = 'l"s I "&• aJj" » .5 — ? J "ja e 1 ^ 1 -r ' 1 OK CliKAT C o u u *n c IT. C CJ a in 3 c c >, a bo :§ .2 •= « .§ * S 5 -S be - " t! c x-a o •^; ^ l; a aZr-g "^ g, »ii!ii! UOSO-B! 'n _ ™ o ™ C G >-, ^ 0 S 3 t» a ^|i c u O ^ ^1 si 1 U O *O w <— !_"S3 5 22=5 '« y T3 h „ h ° . ruEj25aj-cu ^C«J35-Uff3 - — :£ E o -^75 0 — "^ rt -Q 'x > 'S " 'o -C = o u rt rt — IANDIBULAR. -' g iS-o S o =-o •o 8'5 t, i> >> 0 en j; 1** 7J X c u a in a Mlli «-. = >~ ?Jtll J " ./ s boa £ . M B S B -0 .5 — 6 rt— •" . .'3 ^: rt . - U rt «3 3 1 o _>, 5 .2 i| ll'ol .=, e x u . u "2 -T3 £ - T3 X«"O 111" -^ 5 bo a -°--.S £ pS 0 >« a S >>2 O *- •-' E eu n _ fi|5-§ C > rt i^ o Z >>..'?g J ^ 7^ ^, J 5 O ~ rt l- y. 5 j « i%!fl i b a§i a j ^ 4J §«< O i-t v <-)•> S w < lili t; u £ rt 2 CJ rt B U c u rt t/3 1-1 & '"' ~~ "™ rt .2 i i_>v« t- o CJ) J, J2 s c 2 = § be >-c *-.i ' £ = u >.T: ^ o a L* > 2. ".J IX a i^'z a . — Ca (M O •. -^ ™ — _o ^ — rt rt u rs '^ c o rt ' ' T3 K uS-jj " — £ c — e S .t; « J= E '— J || f R D. « aj £ tr -^ ^ U X CJ rt O1 O H ^"* rt _: 75 rt IM 5; C 0 0 8*8*s *U _c ^ = -:f- rt p o S^r.=S — rt^c a rt v w *- b|O| cd j •- . L* X ^ O l.ai§ ?' ~-T.2j= "P S °- "3 Oj $2 4» O CJ (r. H '— - • ~" ' - 11 eC 0 o — i- .— rt l.sl — o .5 ^ '"" -— W ^ rt -J ~ ^ =x > © 2 *r v- X w x o X.S o o 0.2 °" •5 >> 3 4-1 (d " 4) 4-< u *J 2« S" (U *J *J J^ •*-< — 3 >> 3 >, <» <« "o >, Xw M .a U •_; 6 U 5 a 0 - 1 I o O ft £ £ i < < . 0 6 — , - « — ^ o o ^ o .y ?*1 x t/i c ^ tfi "^ •— ' I 0 31 ^ g ^ ^ ft 3 < lv_ .a u .a cj ^ 1 1 in O O C X X — t^ ft ft ft < < c a ci o to >• -c = far/- ." - S= S i .S -2 l LS ° = 'J 3 i ifi i w — « ^ 5 _c .**"' a 1:^11 ;i5i a— -E ' i a o - "° u - rf r- ^ l; C C C M - O i 3 O < r" c c c Retaining primitive connection with hyo- mandibular. d, ium Enla spen . mal part func- as stapes . he o 6 £ o* >. a.S Jlrl c|a.Io .— CT1 — • — •o-c i-« g.a «J « a" P " - S oS bou •— _ - T3 o — S^K ^-.5 ccj "" g C'C O 0 rt rt " 3 h u .c- •g -^sl ctf ^ >-2 t - 'i o 3 a' S X u n C"rt •- a2 6 1 *•• P rr S . pq CO 1 = D3 33 M 3 6 c 3 ^ — - rts rt ^ Ul" J3 *^ O bo'" O he •— 3 a; .M O X ^j C X * rt 5 p J" bfl S "5 « '5b w = 'be u U 4J " rt .3 J^ O, in c ^3 "2 ^2 -o fll bp_ T3 4) >. S2-c MI." I1* is U c o 5P g. « •- ai 23-= o ,3 g <— C (fl -— « x ^* >^ | S s | E" 1 1 1- 6 9- Sao. he — — >> H . E 58 \V. K. GREGORY. The foregoing table summarizes the relations in various forms of the first and second visceral arches to the chondrocranium, and gives the descriptive terms used by authors and the terms here adopted. Upon this analysis the following definitions may be based. PAL/EOSTYLY : "First" visceral or mandibular and "second" visceral or hyoidean arches retaining in large part their primi- tive function as gill bearers, subequal in size, with slight or no connections with the chondrocranium ; preoral gill arches. Hypothetical. C/ENOSTYLY : "First" and "second" visceral arches modi- fied in correlation with feeding, unequal in size, one or more of the elements attached to the cranium. Preoral arches changed in function (=labial cartilages and trabeculae cranii.) A. HoLOSTYLY1 : Platoquadrate fused with chondrocranium; second visceral or hyoidean arch intact, non-suspensorial and free from the cranium. Holocephali. B. AuTOSTYLY2 : Palatoquadrate fused with chondrocranium ; second visceral arch broken up and non-suspensorial, the hyomandibular reduced and united with the cra- nium. Dipnoi, Amphibia. C. HYOSTYLY : Palatoquadrate articulating with chondrocra- nium, hyomandibular more or less suspensorial. a. Hyostyly proper? Second visceral arch intact, the hyo- mandibular and hyoid segments forming a movable suspensorium for the upper and lower jaws. Most sharks, Squatina. b. Eukyostyly* Second visceral arch broken up, the dorsal segment (hyomandibular) forming the sole suspen- sorium, the distal regment (ceratohyal) secondarily, free from all connection with the jaws, functioning solely as a gill bearer. Most rays. 1 " Holo," in allusion, either to " Holocephali," or to the fact that probably since early Palaeozoic times the palatoquadrate and the cranium have formed a continuous whole. 2 The term " autostylic," hitherto applied to both Dipnoi and Holocephali is here restricted to apply only to the former. " Hyostyly proper " because the ordinary sharks furnish the traditional type of this condition. ' Euhyostyly," as a progression upon or development of the " hyostyly proper." RELATIONS OF ANTERIOR VISCERAL ARCHES. 59 c. Amphyostyly * Second visceral arch intact, of slight sus- pensorial value, much smaller than first arch ; palato- quadrate deep posteriorly, articulating by its " otic process" with the chondrocranium. Notidanus, Pleur acanthus. d. Methyostyly? Second visceral arch broken up (i. c., with the component segments more or less shifted out of their primitive sequence or relations) ; symplectic^ metapterygoid, pre- and interopercular when pres- ent assisting the hyomandibular in the support or bracing of the quadrate and mandible. The value and permanence of this classification will depend on whether the hyomandibular of teleostomes is homologous, as generally supposed, with that of elasmobranchs. This has been called into question by Pollard ('94) but as his conclusions have not been adopted by subsequent authorities it seems best to ac- cept the traditional view that these elements are truly homolo- gous in the two groups. . The application of these terms in classification and phylogeny is illustrated in the diagram on page 60. PREVIOUS DEFINITIONS. Huxley. - -The terms " autostylic," " amphistylic," " hyosty- lic " were first used by Huxley in his paper of 1876 on Cerato- dus? The passage in which " autostylic " is defined is as follows OA dt.t p. 40) : "The part of the palato-quadrate cartilage [of Ceratodus\ which is united with the skull, between the exits of the fifth and second nerves, an- swers to the " pedicle of the suspensorium " of the amphibian, while its backward and upward continuation onto the periotic cartilage corresponds with the otic process. As in the Amphibia and in the higher Vertebrata, the mandibular arch is thus attached directly to the skull by that part of 1 ' ' Amphyostyly " ; this term retains the element " amphi," so long used (in amphistylic] of these forms, while clearly indicating its subordination under hyo- styly. 2 " Methyostyly," in allusion either to the prominence of the w.?Azpterygoid in the suspensorium, or to the fact that methyostyly represents a morphological advance upon earlier modes. 3 Proc . Zoo/. Sor. Lond., 1876, pp. 24-59. "The Scientific Memoirs of Thomas Henry Huxley," Vol. IV., 1902, pp. 84-127. tf> *>> c/l s T) 4) o c u a. a. o ^ W. K. GREGORY. 6 1 its own substance which constitutes the suspensorium. It may thus be said to be antostylic. " Among fishes the only [other?] groups which possess an autostylic skull or in which the dorsal end of the mandibular arch is continuous with the cartilage of the brain case are the Chimasroids and the Marsipobranchii.' ' In this definition of autostyly attention is centered solely upon the relations of the first arch with the skull. Huxley notes that in the autostylic Ceratodus the hyomandibular is reduced and fused with the skull, but he also uses " autostylic" for C1iivi(cra in which the hyomandibular is separate (see above, Table A). For comparison with " hyostylic " and " amphistylic >: we may summarize Huxley's description as follows : AUTOSTYLIC SKULL : Mandibular or first visceral arch attached to the skull solely by its ozvn dorsal moiety, the palatoquadrate, which is continuous with the skull. Huxley describes the HYOSTYLIC and AMPHISTYLIC conditions as follows (p. 41) : " In all other Fishes, except the Marsipobranchii, the mode of connec- tion of the mandibular arch with the skull is different from that which ob- tains in the Chimseroids and the Dipnoi. The palatoquadrate cartilage is no longer continuous with the chondrocranium . . . but is, at most, united with it by ligament. Moreover the dorsal element of the hyoidean arch, or the hyomandibular, usually attains a large size and becomes the chief ap- paratus of suspension of the hinder end of the palatoquadrate cartilage with the skull. Skulls formed upon this type, which is exemplified in per- fection in Ganoidei, Teleostei, and ordinary Plagiostomes, may therefore be termed hyostylic. "But though the typical forms of autostylic and hyostylic skulls, as exemplified, c. g., by a Sturgeon, a Pike, and a Dogfish or Ray, on the one hand, and Chimcera, Ceratodus, and Menobranchus on the other, are thus widely different, certain Plagiostomes present a condition of the cranium which tends to connect the two by a middle form, which may be termed amphistylic. " In the amphistylic skull the palato-quadrate cartilage is quite distinct from the rest of the skull ; but it is wholly, or almost wholly, suspended by its own ligaments, the hyomandibular being small and contributing but little to its support. The embryo amphibian is amphistylic before it becomes autostylic ; and, in view of certain palaeontological facts, it is very interesting that the link which connects the amphistylic with the ordinary Selachian skull is that of Cestracton." Huxley's conception of hyostyly and amphistyly may be re- stated as follows : 62 \V. K. GREGORY. HYOSTYLIC SKULL : Mandilndar or first visceral arch no' con- tinuous with the skull in its dorsal portion (the palatoquadrate) ; the hinder end of the palatoquadrate is cliicfly connected with tlic chondrocranium by means of the hyomandibular or dorsal portion of the hyoidean arch. AMPHISTYLIC SKULL : Mandibular, or first visceral arch attached to the skull wholly or almost wholly by means of its dorsal portion the palatoquadrate, not Jwivcvcr by fusion or continuity, but by liga- ments ; the hyoidean arch contributes but little to its support. The Huxleyan conception of amphistyly is that it is in some respects a "middle form," resembling autostyly in the suspen- sorial self-sufficiency of the first visceral arch, but resembling hyostyly in the lack of confluence of the palatoquadrate with the skull. A similar use of the prefix "amphi" occurs in " Amphi- theriunt" which genus was at first supposed to combine mam- malian and reptilian characters. The element " stylic " from THV.OC, a pillar, evidently refers to the quadrate region in its architectural relation to the skull. The skull of Chim&ra seems to have been regarded by Huxley as primitive (so far as I can determine from a careful study of the entire article) and hence in an autostylic skull the palatoquadrate must have been conceived as belonging to the skull: therefore the forces acting upon the jaws in eating would be transmitted to the skull chiefly through the quadrate, its oivn pillar (hence auto, stylic), whereas in the typical hyostylic skull these forces would be transmitted chiefly through the pillar of the hyoidean arch (the hyomandibular) — hence "hyostylic." Huxley's conception of the phylogenetic relations of the three types of cranial structure are expressed in the following diagram (op. cit., p. 45): Amphibia. - Ganoidei. Teleostei. CERATODUS.— —Cestracoin. Raia. Chi mar a. Not i dan us. Autostylica. Amphistylica. Hyostylica. The left-hand column (autostylica) includes the (to Huxley) more generalized types, the most primitive, Chiuuzra, standing at RELATIONS OF ANTERIOR VISCERAL ARCHES. 63 the bottom of the column ; progressive stages of skull structure are represented in the middle and right-hand columns. The cranial plan of Ceratodus is regarded as representing the ancestral condi- tion of both Cestracion and the Ganoids. Cestracion, classed with Notidanus as an amphistyjic type, is regarded also as transitional, a low form of the hyostylic type.1 The diagram confirms the inference that Huxley conceived the palatoquadrate as originally a part of the skull, which gradually became constricted off as in Ceratodus and finally freed from it entirely as in Raia and the Teleosts. Wiedersheim (Parker s translation, '97). — " The palatoquadrate is usually only united to the basis cranii by liga- ments, but in the Chimteroids. . . it becomes immovably fused with it, whence their name of Holocephali. In the Sharks and Rays the palato- quadrate is not directly united to the skull, but is suspended from it by the hyomandibiclar. ... In this case the skull may be described as hyostylic, to distinguish it from autostylic skulls, in which the hyoid takes no part in the suspensodum " (op. cit., p. 75). Comments: (i) "Hyoid" here refers to the hyoidean arch, not to the hyoid or ceratohyal. (2) In general it would be less confusing to say that it is the mandible rather than the palato- quadrate which is suspended from the hyomandibular, and that the palatoquadrate rests chiefly upon and is fastened to the mandible. v. Zittel, 1903 (Eastman's edition). - " In the Holocephali the palatoquadrate and hyomandibular fuse together and with the cranial capsule. The mandible thus becomes autostylic, i. e., articulates directly with the cranium" (op. cit., p. 11). Comments : (i) The statement in regard to the hyomandibular seems incorrect. In a specimen of Chimcera collei kindly loaned to me by Professor Bashford Dean the hyomandibular is seen to be a large independent element (see Fig. 11) serially homolo- gous with the epibranchials as in Selachii, and tipped with a reduced pharyngohyal, both being free from the chondrocranium. (2) To apply the term autostylic to the mandible is to introduce a new element of confusion in a matter already sufficiently complicated. 1 On page 43 of Huxley's memoir the Cestracion skull is referred to as a low form of " autostylic type," but examination of the context and of the diagram cited show that this is probably a misprint for " hyostylic." 64 ^'. K. GREGORY. Parker and Haswell (1897). — " In some fishes the hyomandibular articulates above with the auditory region of the cranium while the jaws are connected with its ventral end. We may thus distinguish two kinds of suspensorhtm or jaw-suspending appa- ratus, a mamiibnhir sttspensoriitm, furnished by the quadrate, and a hyoidean suspensorium by the hyomandibular ; in the former case the skull is said to be antostylic, i. e., having the jaw connected by means of its own arch, in the latter it is called hyostylic ; in a few cases an ainphistyhc arrangement is produced by the articulation of both mandibular and hyoid arches with the skull" (op. cit., p. 71). On page 161 we read that in Hexanchus and Heptanchus : ". . . there is a prominent post-orbital process of the palatoquadrate for articulation with the post-orbital region of the skull (amphi- stylic arrangement)." The definitions cited may be criticised on several grounds : 1. There are several ambiguities latent in these sentences which experience proves to be exceedingly puzzling to the stu- dent : (rt) "jaw " and "jaws " thus used require the most careful analysis to determine which jaws; upper or lower, are intended ; (fr) a mandibular siispciisorium is no doubt equivalent to " a sus- pensorium furnished by the mandibular or first visceral arch " but if " mandibular " be mistakenly interpreted as referring to the mandible the very pith of the definition is lost ; (c] suspensorium or jaw-suspending apparatus is understood first with reference to the lower jaw, then to the upper; (<•/) "hyoid" arch is used on this same page (71) as referring to the whole second visceral arch, but on page 161 " hyoid arch " is used of the ceratohyal or hyoid only. 2. The definition of autostylic does not exclude Notidanus (which Parker and Haswell themselves call " amphistylic ") be- cause in it the hyomandibular takes so small a share in suspension that the suspensorium may be said to be "furnished by the quadrate." 3. The definition of amphistylic as implying the articulation of "both mandibular and hyoid arches with the skull " is exceed- ingly imperfect. As shown above (Table A, pp. 54, 55) the artic- ulation of both mandibular and hyoidean arches with the skull is not diagnostic of amphistyly since on the one hand both arches 1 Especially since " mandibular " is used in that sense at the top of the same page. RELATIONS OF ANTERIOR VISCERAL ARCHES. 65 articulate with the skull in many hyostylic types, and on the other hand in the amphistylic Notidanns both arches do not articu- late with the skull since the hyomandibular is connected with it only through ligaments (Gadow, '88, PL 71, Fig. I, A). Gadow (1888). — Gadow apparently uses the terms in their etymological signifi- cance, rather than with the limitations imposed by established usage. Thus he speaks (p. 455) of the " autostylic condition of the Notidanidce " (the " amphistylic " of authors) and applies the same term to the Amphibia (p. 481) and the Dipnoi (p. 459), evidently having in mind the self-sufficiency in all these cases of the first arch as its own suspensorium ; so too, " simple autosty- lic " (p. 459) apparently refers to the ancient and generalized condition from which the modern modes have been derived (cf. " palaeostylic," proposed above) and in which the oral arch had not yet begun to borrow support from the hyoidean arch. Again Gadow applies " holostylic " (p. 458) to Chimcera, Ceratodus and Cestracion evidently with reference to the functional ligamentous union of the upper jaw with the cranium in Cestracion and to its fusion therewith in Chimcsra and Ceratodus. " Amphistylic " in its etymological sense serves to describe the conditions in "most selachians " (p. 481), in which both oral and hyoidean arches are suspensorial, while etymologically the rays are pre- eminently "hyostylic" (p. 481) since the dorsal element of the hyoidean arch, the hyomandibular, is the sole suspensorium (cf. our " euhyostylic," p. 56, supra], But while this bold and highly suggestive use of terms has finally bred in me several clarifying ideas I cannot deny that formerly, wishing to refer to isolated passages of Gadow's work, I experienced a most baffling uncertainty as to his meaning. Smith Woodward (1897). — " Among the fishes existing at the present day there may be observed two distinct plans of cranial structure, between which no definitely inter- mediate conditions can be recognized. In Chimcera, Protopterus, Cerato- dus and their allies, the upper segment of the mandibular arch is directly fused with the chondrocranium, while the corresponding segment of the hyoid arch is atrophied or absent ; in the Elasmobranchs and the so-called " Ganoidei " and " Teleostei " the same elements are loosely articulated with the chondrocranium, the upper segment of the hyoid arch forming a 66 \Y. K. GREGORY. movable suspensorium. The first condition is now commonly known as the autostylic, and the second as the hyostylic. Comments. — This passage is perfectly clear and illuminating. It adheres closely to the essential features of Huxley's concep- tions, and it makes a decided advance beyond them not only in stating clearly that between modern autostyly and hyostyly " no definitely intermediate conditions can be recognized," but also in implicitly reducing the amphistylic mode to its true place as a special phase of the hyostylic — a suggestion which has been fol- lowed in the present paper (see above, pp. 57, 59). But not even these definitions, I think, will survive. They fail to emphasize the differences between the autostyly of Dipnoi and the autostyly (here termed " holostyly ") of Chimaeroids ; nor do they indicate that by taking also the distal segment of the hyoidean arch into account we may place the whole subject upon a new, and apparently phylogenetic basis. In conclusion I desire to thank Professor Henry Fairfield Osborn and especially Professor Bashford Dean for various courtesies and suggestions. REFERENCES. Dean, B. '95 Fishes, Living and Fossil, 8vo, 1895. Gadow, H. '88 On the Modifications of the First and Second Visceral Arches, etc. Philos. Trans. Lond., Vol. 179 (1888) B, pp. 451-485, pi. 71-74. Garman, S. '85 Chlamydoselachus anguineus Garni. — A living species of Cladodont Shark. Bull. Mus. Comp. Zool. Cambridge, Mass., 1885, pp. 1-35, PI. I. -XX. Huxley, T. H. '76 On Ceratodus fosteri, with Observations on the Classification of Fishes. Proc. Zool. Soc. Lond., 1876, pp. 24-59. Parker and Haswell. '97 A Text Book of Zoology, Vol. II., 8vo, 1897. Pollard, H. B. '94 On the Suspension of the Jaws in Fish. Anat. Anz., Bd. X., 1894, pp. 17-25. Smith Woodward, A. '98 Outlines of Vertebrate Palaeontology, etc., Svo, 1898. Wiedersheim, R. (W. N. Parker, Tr.).? '97 Elements of the Comparative Anatomy of Vertebrates, second edition, Svo, 1897. 68 \V. K. GREGORY. RELATIONS OF ANTERIOR VISCERAL ARCHES. 69 v. Zittel, K. A. (C. R. Easman, Ed.). '02 Text Book of Palaeontology, Vol. II., Svo, 1902. COLUMBIA UNIVERSITY, DEPARTMENT OF ZOOLOGY, February 10, 1904. PLATE I. EXPLANATION OF ABBREVIATIONS. P. Q. , Palatoquadrate. H.M"., Median limb of ditto. Mck., Mandible (Meckelian cartilage). H.M"'., Posterior limb of ditto. H.M., Hyomandibular (epihyal). Sym., Symplectic. B.H., Basihyal. Q., Quadrate. C.H., Ceratohyal. P. Op., Preoperculum. P.H., Pharyngohyal. I. Op., Interoperculum. Sp.C., Spiracular cartilage. Op., Operculum. I.H., Inter- or stylohyal. S.Op., Suboperculum. Pal., Palatine. VII. , Aperture for seventh nerve. Ms. PL, Mesopterygoid (entopterygoid). h.s.l., Hyosuspensorial ligament. Pt.t Pterygoid. niJi.l., Mandibulohyoid ligament. Mt.Pt., Metapterygoid. St. , Stapes. H.M'., Anterior limb of hyomandibular. Cartilaginous elements stippled, osseous elements with small crosses. FIG. l. "Hyostyly proper." Centrophorus granulosus. Embryo. After Gadow. FIG. 2. " Hyostyly proper." Chlamydoselachiis anguinetts. After Garman. FIG. 3. " Euhyostyly. " Trygon sp. After Gadow. FIG. 4. " Amphyostyly." Heptane Jms ciner ens. Internal view. After Gadow. FIG. 5. " Methyostyly." Diagram of the relations of the suspensorial and oper- cular regions in the cod. FIG. 6. Methyostyly. Polyplerus. View from within and below. The meta- pterygoid is supported by the hyomandibular. FIG. 7. Cartilaginous basis of methyostylic arrangement in the young salmon. After Parker and Bettany. FIG. 8. Cartilaginous basis of methyostylic arrangement in a larval siluroid. After Pollard. All the cartilages of the suspensorial region seem to have secondarily coalesced. FIG. 9. "Autostyly." Ceratodns. After Huxley. R, R' , vestigial hyoid rays. The operculum is borne by the reduced hyomandibular. FIG. 10. "Autostyly." Protezts. After Gadow. The stapes represents the upper portion of the hyomandibular (Gadow). FIG. II. " Holostyly " Chimizra collei. VARIATION IN BEES — A REPLY TO MR. LUTZ. EVERETT F. PHILLIPS. In the December (1903) number of this journal there appeared an article entitled " Comparative Variability of Drones and Workers of the Honey Bee " by Dr. D. B. Casteel and myself in which we concluded that the drones show the more variability, which result we attributed to the effect of the size of the cells on the developing larvae and pupas. Mr. Frank E. Lutz in the April (1904) number of this BULLETIN thinks it "well worth while to consider a few points about the paper" and concludes that we have not proven our point. I desire here to point out the errors in his manner of dealing with our measurements and his methods of drawing conclusions and to defend our position. The fact that in the consideration of abnormal veins we found that " there are 206 irregular drone wings and 30 irregular worker wings or almost seven times as many for drones as for workers" points rather conclusively to our position in regard to comparative variability. Our statements in regard to coloration are also disregarded, evidently because there were no figures given and therefore they were not considered worthy of mention. In another paper l I have dealt with the comparative constancy of the color areas of the two sexes and there offer evidence which confirms the position maintained in our paper, although as there used it was intended to prove an entirely different point. In quoting from a letter from Mr. E. R. Root, a high authority in apiculture, it was there stated : " The drones from these queens (imported Italians of pure stock) varied greatly in their markings. Some of their sons would have a great deal of yellow on them, while others would be quite dark. . . . Bees (workers) from these queens were all uniformly marked." In variation work but few investigators take the trouble to study more than one char- acter yet here are two which are rather conclusive and on these we would be willing to rest the case. 1<(A Review of Parthenogenesis," Proc. Am. Philos. Soc., XLIL, No. 174; •vide pp. 277-8. 70 VARIATION IN BEES. /I I will now take up the more definite criticisms of Mr. Lutz. In our paper we gave reasons which to us seemed good why we did not employ the standard deviation method in our work, stat- ing distinctly that " that would be undesirable except with far more measurements " ; this could be interpreted only as a desire on our part to avoid a high probable error and no one need " wonder greatly " why we did not employ this simple test since we distinctly stated that we did not think it desirable, giving full reasons for such a decision. It was not our intention to take exception to the methods of the workers in variation for I believe that the results of variation work should be stated in mathemati- cal formulae where such a thing is possible without too great a " probable error." If our critic is trying to defend such methods under the impression that we combat them, he is laboring under a misapprehension. However Mr. Lutz has seen fit to figure out the standard deviation and probable error for two of our tables and concludes " that the differences between the two sexes, as shown by these data are of no significance." I am still of the opinion that it is unwise to use these methods for so few indi- viduals on account of the large probable error but since this is the way in which our results are questioned let us examine the figures and see if there is not " great danger that, having collected a set of measurements" our critic makes "a show of accuracy that will lead " him " and others astray by reason of careless and insufficient analysis." For ease of reference I give the results of Mr. Lutz. The argument of Mr. Lutz is that since there is as much dif- ference in standard deviation between various lots of drones as there is between the averages of the standard deviations of drones and workers, our conclusion that drones vary the more is false. But is it not evident that in every case of the drones (except lot II.) the standard deviation or index of variability is greater than that of the workers ? I cannot see what bearing the differences in the standard deviation of the various lots of drones has on the question since in every case except one, the drones do vary more than the workers ; and this it was that we attempted to prove. In regard to this lot II. we said : " The drones in lot II. were taken from a hive in which there were no drone cells except EVERETT F. PHILLIPS. HOOKS ON HIND WING. Drones. Lot. Number of Specimens. Standard Deviation. Probable Error l I. 5° 2.1548 0.1453 II. IOO 1-5435 0.0736 III. IOO 1.7716 o.o84S IV. IOO 1.6486 0.0786 V. 50 2.0988 0.1416 VI. 98 1-9377 0.0934 Workers. I. II. III. 5° 350 IOO 1.5223 L5564 I-5523 O.IO27 0.0397 O.O/4O VEIN R. Drones. Lot. Number of Specimens. Standard Deviation. Probable Error. I. III. V. 50 IOO 50 2.4023 2.9598 2.2517 0.1620 O.I4I2 O.I5I9 Workers. I. 50 1.5637 0.1055 possibly a very few in the corners of the frame or near the top bar of the frame since all the combs were made on what bee- keepers call foundation and the cells were uniformly of worker size. These drones show the least variation since they were all hatched under the same conditions." Even if we add the prob- able error to the standard deviation of workers and subtract it from that of the drones this result holds except in lots II. and IV. and it would seem that from these figures (made by Mr. Lutz himself) that our results are most strongly confirmed. The additional criticism is made that we lumped the different lots together because they seemed to be alike " when really their only claim to homogeneity is that they are of the same sex and all bees -- Italians, hybrids, 'peculiar strains,' et al., from central Ohio to eastern Pennsylvania being jumbled together." This 1 In discussing the average standard deviation of this table Mr. Lutz takes the average of the probable error for the lots of 50 and 100 without taking into considera- tion that the probable error should be smaller for 500 than for these lots. Surely since all the lots show the same greater variability of the drones (except lot II. as ex- plained later) the probable error is considerably smaller than that given by Mr. Lutz. VARIATION IN BEES. 73 was done only in the case of our figures concerning the ratios between the veins M2 and m where a careful examination showed us that these ratios did not vary according to lots but from the criticism it might be inferred that we had committed this "grave error " throughout the work. Just how Mr. Lutz could know, since the ratios for the individual wings were not published, that we had "jumbled together" some measurements in an unjusti- fiable manner is still a mystery but I am convinced that his Criti- cism is unjust, having carefully examined the ratios with this very criticism in mind at the time of the preparation of our paper. At the close of his article our critic says : " It is also probably unnecessary to remark that, even if it turns out that the greater variability of the drones can be established, their proof of their theory to account for this difference seem rather unsatisfactory." I think it is shown, by us and by Mr. Lutz, that the drones do vary the more and our theory of the cause, based as it is on care- ful investigations of the habits of the bee, must be controverted by more observations of an equally careful nature. Too often students of variation work on forms, concerning the habits of which they know nothing, and conclusions are reached which would be modified if causes were looked for in the habits, but I feel free to state that we are not open to that criticism. By omitting the parts of our paper in which we explain our stand regarding the variability being " due to chance," Mr. Lutz would make out that we do not know that all chance is in ac- cord with some mathematical formula. On this point we stated : " It may be argued that variation according to chance is but a way of stating our ignorance of the true law, but if there is a law for this variation it is certainly very obscure, and the working out of this law would require an extremely large number of measurements taken from individuals each one with its life his- tory known," and again : " While it is probable that even this chance is according to fixed law, the fact remains that in any event this law is beyond a possibility of formulation from any observations except those extending over far more individuals than those here used." If as we believe the particular size varia- tion of any individual bee depends on the size of the cell in which it grows, then the formulation of this law of variation must be 74 EVERETT F. PHILLIPS. based on the law which governs the queen in moving over the comb and in choosing where she shall lay her eggs and on the law which governs the bees in cell building. We leave it to our critic to formulate these laws since we confess to a lack of mathe- matical ability for any such problems. Finally we acknowledge the mistake in averages mentioned by Mr. Lutz but can only state that we believe this to be the only correct criticism of our paper which he has made. ZOOLOGICAL LABORATORY, UNIVERSITY OF PENNSYLVANIA. NOTES ON A PECULIAR ACTINOZOAN LARVA. L. R. GARY. For a number of years past a few specimens of a large trocho- phore-like larva have been taken each summer in the tow near Beaufort, N. C., but they have never been seen to transform. While at the laboratory of the United States Fish Commission at Beaufort,1 during the past summer, I had the good fortune to secure a number of these peculiar larvae. The larvae were taken while towing outside the harbor on August 15, after a heavy southeast storm which had continued for two days, and which had driven in shore specimens of several forms not usually found inside the Gulf Stream. These larvae were taken to the laboratory and at the sugges- tion of Dr. Caswell Grave, of the Johns Hopkins University, were put in aquarium jars of sea-water containing sand rich in diatoms. By this method they were kept alive for the remaining seven weeks of my stay at the laboratory. The larvae, Figs, i and 2, are elongate-oval in shape when in an active state, changing to a very nearly spherical form when they are disturbed. They are from two to four millimeters in length, and of a light brown or cream color. At a point about one fourth of the distance from its anterior end, the body is encircled by a ridge which lies at the bottom of a shallow groove. This ridge bears on its surface two parallel bands of long stiff setae, Fig. 4. The setae -bearing ridge is not continuous around the body, but on one side it is interrupted and the ends overlap for a short distance, Fig. I. The whole surface of the body is provided with a covering of short cilia which are the true locomotor organs. The function of the bands of setae is not apparent. They move only at irreg- ular intervals, and then the force of their movement is in a direc- tion antagonistic to the progress of the larva. 1 1 am indebted to the Hon. G. M. Bowers, U. S. Commissioner of Fisheries, for the privilege of occupying a table at the Beaufort Laboratory, and to Dr. Grave, the director, for many favors received while there. 75 L. R. GARY. At both the anterior and posterior ends of the body there is a depression lined with cilia. The depression at the anterior end has at its bottom an opening communicating with the interior. FIG. I. Larva in the active expanded condition, X 3°- In swimming, the larva rotates rapidly on its long axis, and the anterior end describes a small circle about the axis of pro- gression so that the larva advances by a kind of corkscrew FIG. 2. Same larva as in Fig. i, when contracted, X45- movement, such as has been described for the larva of Astroides by Lacaze Duthiers l and for Rcnilla by E. B. Wilson.2 1 Lacaze Duthiers, H., " De Development des coralliaires," Archiv Zool. exper. et gen., Tom II., 1873. * Wilson, E. B., "The Development of Renilla," Phil. Trans. Roy. Soc. Lon- don, Vol. CLXXIIL, 1883. NOTES ON A PECULIAR ACTINAZOAN LARVA. 77 The habits of the larva differ from those of the larvae just men- tioned in that it does not come to the surface of the water in the aquarium, nor does it remain motionless for any appreciable time unless it is disturbed, when it contracts and sinks to the bottom of the vessel. Two of the larvae went through their transformation two days after they were brought into the laboratory. When transforma- tion is to take place, the larva settles down and becomes attached by the anterior end. In a short time the tentacles are budded out from the upper (posterior) end, and the mouth opening FIG. 3. Young polyp, two days after transformation, X °°- breaks through the body wall within the circlet of tentacles. Within twenty-four hours from the time of the attachment at the beginning of the transformation, the young polyp had assumed a form such as is shown in Fig. 3. In the five or six weeks after transformation, during which the young polyps were under observation, there was no apparent ex- ternal change other than a gradual increase in the size of the animal as a whole without any change in proportions. All the larvae, with one exception, had transformed by Sep- L. R. GARY. tember 7, three weeks after they were secured. The other speci- men had not transformed on October 3, and, as far as could be determined, it had undergone no changes in size or form. The actinian, of which the larva just described is the immature form, is not definitely known to me at present, but a number of mature sea anemonies, prob- ably of the genus Amophyllactis, were cast upon the beach near where the larvae were found during the same and subsequent storms coming from the same direction. Since both these forms appear only after heavy storms from- a definite direction, and since there are certain structural re- semblances between the young polyps and the mature actinians, it seems not improbable that they may be different stages in the life history of the same species. ZOOLOGICAL LABORATORY, JOHNS HOPKINS UNIVERSITY, March, 1904. FlG. 4. Diagram to show position of setae. Vol. III. July, 1904. No. 2 BIOLOGICAL BULLETIN' THE MORPHOLOGY OF THE MADREPORARIA. V. SEPTAL SEQUENCE.1 J. E. DUERDEN. The skeleton of an ordinary poly cyclic hexameral coral pre- sents a series of septal partitions arranged in a radiating manner, and in any cycle the constituent septa are equal in size and alter- nate in a regular manner with the members of the other cycles. In general, the cyclic plan of a corallite is 6, 6, 12, 24, 48, etc., and the number of septa in any cycle beyond the first corresponds with the total number in all the cycles within. As the septa in any cycle are alike in size it seems reasonable to suppose that they all appeared simultaneously in the growth of a coral, a cycle at a time, and also that the inner larger cycles were developed before the outer smaller cycles ; in other words, the relative sizes and positions of the septa in the mature corallite would seem to represent their order of development. Much attention has already been given to the subject of the order of appearance of the septa in corals by different writers. The rule which Milne-Edwards and Haime give in their " His- toire Maturelle des Coralliaires " (1857) is well known, appear- ing in all text-books describing recent or fossil corals. The authors assume that in the case of the first three cycles the con- stituent septa of each cycle appear simultaneously, a cycle at a time, and that the relative development of each cycle corresponds with its relative size or distance from the center of the calice. From the fourth cycle outwards, however, a different sequence is 1 The first two parts of this series of papers appeared in the Johns Hopkins Uni- versity Circulars, Vol. XXL, Nos. 155 and 157, and were reprinted in the Annals and Magazine of Xatnral History, Ser. 7, Vol. X., May and August, 1902. The third and fourth parts appeared in the Annals and Magazine of Natural History, Vol. X., November, 1902, and Vol. XL, February, 1903. The work is being carried out with the assistance of an appropriation from the Carnegie Institution. Contributions from the Zoological Laboratory of the University of Michigan. No. 8l. 79 8O J. E. DUERDEN. assumed for the various members of a cycle, according to their relation to the members of the inner cycles. Professor A. Schneider in 1871 and Professor C. Semper in 1872 also discussed the same subject. The former agrees with Milne-Edwards and Haime as regards the manner of develop- ment of the first three cycles : first a cycle of six appears, then a smaller cycle of other six alternating with the first, and later a cycle of twelve septa which are smaller and alternate with the twelve making up the first and second cycles. In the further growth Schneider considers that the septa of a newer cycle may so enlarge in size as to appear to belong to an older cycle, and thus the primary sequence and hexamerism become lost. Semper holds that no constant rule for septal development can be established, that the manner of growth varies more or less with each species. Professor G. von Koch, in the course of his prolonged studies on the morphology of corals, has also made certain observations upon the laws of septal development, particularly in his paper " Das Vermehrungsgesetz der Septen," 1881. His results are based mainly upon serial sections of individual coralla of Caryo- phyllia, and lead him to conclude (p. 93) in the main in favor of the validity of the sequence given by Milne-Edwards and by Schneider : " Bei den sechszahligen Korallen, sowohl den Epo- rosen als den Perforaten, wachst die Zahl der Sternleisten (Septa) in der Art, dass sich nahezu gleichzeitig im ganzen Umfang des Kelches zwischen je zwei alteren eine jungere anlegt, also die Zahl der Sternleisten eines folgenden Cyclus immer gleich ist derSumme aller vorher vorhandenen. Alle Ausnahmen von dieser Regel sind auf direxte Anpassungen oder erblich gewordene Veranderungen im Wachsthum des ganzen Thieres zuruckzufuhren." All the above conclusions are based mainly upon the examina- tion of adult coralla, in \vhich there is available for comparison only the relative sizes of the septa and their order of appearancein serial sections. The actual details of growth of the septa indeveloping corals, in their relationships to the mesenteries, have in no instance been followed beyond the first two cycles. Professor H. de La- caze-Duthiers (1873, '94, '97) and Professor G. von Koch (1882, *97) have both studied the early development of the skeleton in MORPHOLOGY OF THE MADREPORARIA. 8l corals, but their results throw no certain light upon the difficult problem of the manner of increase of the septa beyond the two primary cycles, nor of the relation of these to the later cycles. Observations which I have been able to make upon the growth of the septa in larval polyps of the common West Indian coral, Siderastrca radians (Pallas), show the unreliability of assuming the developmental sequence from adult relationship, and give an interpretation to the later septal sequence altogether different from any hitherto proposed. Larvae of Siderastrca, fixed to fragments of glass, and capable of being examined as transparent microscopic objects, were fol- lowed in the course of their development as young polyps for a period of four months, and the order of appearance of their septa determined as far as the completion of the first three cycles. In every case it was found that the six members composing the first cycle appeared simultane- ously. This takes place shortly after the larva settles, at a stage when the young polyp has six pairs of mesenteries, arranged as in Fig. I, where all the mesen- teries are complete except the fifth and sixth developmental pairs. The septa are alike in size and Situated at equal distances apart FIG. I. Diagrammatic 'arrangement within the entocoeles of the six of the mesenteries and septa on the c • appearance of the first cycle of six primary pairs of mesenteries. _. septa. Ihe septa are situated within Later development proves, as the six prirnary entocoeles, and the would be expected, that the six six alternating mesenterial chambers, primary septa of the mature coral- devoid of sePta> are the Primar>' ex" occeles. The directives are situated lite are the direct enlarged rep- at the opposite extremities . the two resentatives of the six septa first bilateral pairs of incomplete mesenteries to appear in the larval polyp. are the fifth and sixth Pairs in the mesenterial sequence. The upper As regards the primary cycle border is regarded as donal and ^ in Siderastrca the surmise that lower as ventral, adult size corresponds with de- velopmental sequence is therefore correct. Moreover, in the early 82 J. E. DUERDEN. stages of septal growth secured by Lacaze-Duthiers in Astroidcs, Flabelhim, Balanophyllia, Caryophyllia, etc., and also by \on Koch in Astroidcs and Caryophyllia, the six primary septa ap- peared simultaneously, and were equal from the beginning. In Siderastrea the second cycle of six septa began to appear a few days after the primary cycle, its members situated within the exoccelic chambers, thus alternat- ing with the six entosepta (Fig. 2). In a few polyps the septa appeared simultaneously and were all practically equal, but in most individuals a marked difference was manifest ; the dorsal and middle pairs of exosepta arose bilaterally in advance of the two FIG. 2. Arrangement of the mesen- ventral pairs, and for a time the teries and septa after the establishment dorsal pairs were a little larger of the first and second septal cycles — , T,, than the middle. Within a short six entosepta and six exosepta. The dorsal pair of exosepta are somewhat time the two ventral exosepta larger than the middle pair, and the appeared, but remained smaller middle pair are larger than the ventral , u , , . , -p, than the others. Thus at this pair, thus giving a bilateral character to the corallum. early stage a decided dorso-ven- trality in the development of the septa was apparent, which gave a bilateral character to the polyp as a whole (Fig. 2). In most of the developing corals investigated by Lacaze- Duthiers and von Koch the entoccelic and exocoelic cycles either appeared together, or one shortly after the other, as in Sideras- trea. In most cases it was found that the members of either cycle arose simultaneously, in a truly radiate manner, without passing through a bilateral stage, though Lacaze-Duthiers figures a very decided bilateral condition in the early development of the skeleton in Astroides (1873, PI- XIII., Fig. 29). The next stage in the growth of the septa in Siderastrea is well defined, but is not so clear as to its significance. After the condition shown in Fig. 2 was reached the septa began to enlarge at their peripheral extremity; in some instances the enlargement MORPHOLOGY OF THE MADREPORARIA. took place by the direct extension of the septa already developed, and in others by the deposition of separate skeletal nodules. The new growth was arranged in a V-shaped manner, the angle of the V being larger in the exosepta than in the entosepta (Fig. 3). In the diagrammatic figure the additions are all represented as separate calcareous fragments, but no constancy was apparent in the different septa as to the freedom or fusion of the extensions at this stage. In all the polyps the enlargement of the ventral exosepta was much behind that of the dorsal and middle exosepta, and usually it could be seen that the middle exosepta did not enlarge as rapidly as the dorsal. At a somewhat later stage, each septum became a continuous structure, owing to the complete fusion of the free nodules in the FIG. 3. Stage showing the peripheral enlargement of the septa which occurs in a bifurcating manner by the addition of separate nodules. The ventral exosepta grow more slowly than the middle and dorsal pairs. FIG. 4. Further enlargement of the septa by fusion of the nodules, and estab- lishment of the six pairs of second cycle mesenteries ; the latter, like the exosepta, decrease in size from the dorsal to the ventral border. process of growth. The entosepta then appeared as simple formations, much broader peripherally than centrally, though the two directive entosepta showed traces of the earlier bifurcation longer than the four lateral entosepta ; the exosepta, on the other hand, remained strongly bifurcated, with the exception of the ventral pair, the members of which enlarged comparatively little (Fig- 4). A similar bifurcated stage, due to the appearance of indepen- dent calcareous nodules, occurs in the development of Astroidcs, 84 J- E. DUERDEN. Caryophyllia, and some other corals, but hitherto its significance has not been satisfactorily explained. At one time the separate fragments were supposed to be concerned in the formation of the thecal wall, but it will be seen that such is certainly not their fate in Siderastrca. About this time, the polyps being two months old, the sec- ond cycle mesenteries began to appear on the column wall, situated in the exoccelic chambers between the primary mesen- teries (Fig. 4). In their development they also presented a con- spicuous dorso-ventrality : the two first pairs appeared within the dorsal exocoeles, the moieties of each pair arising at the same time and remaining equal ; the two next pairs were within the middle exocceles ; and finally appeared the pairs within the ven- tral exocceles. The dorso-ventral succession thus followed by the six pairs remained evident throughout the period under obser- vation, the dorsal pairs being larger than the middle, and the middle larger than the ventral. The comparative development of the first and second cycle mesenteries at this period, and their relationship to the septa, are shown in Fig. 4. A similar bilateral, dorso-ventral succession of the second cycle mesenteries has long been known to occur in the development of actinian polyps, in contrast to the simultaneous origin at one time assumed. Clearly the next skeletal stage will be one involving the establishment of septa within the entocceles of the second cycle of mesenteries, and these will constitute the second cycle of septa of the adult corallite. Hitherto it has been generally as- sumed that where a cycle of exosepta is already developed, and then a new cycle of mesenteries appears within the corresponding exocceles, that the exosepta already present become included within the entocceles of the new mesenteries, and thus become entosepta ; an additional outer cycle of exosepta appears later and its members in their turn become entosepta. Thus Delage and Herouard in their " Traite de Zoologie Concrete " (1901, p. 558) remark: " Quand, dans les interloges occupees par les septes du dernier cycle, nait un nouveau cycle de couples de cloisons, celles-ci se forment de part et d'autre du septe inter- loculaire qui, de ce fait, devient loculaire, et bientot un nouveau cycle de septes se forme dans les nouvelles interloges qui vien- MORPHOLOGY OF THE MADREPORARIA. 85 nent d'etre formees. Les cycles naissent successivement et jamais un cycle ne commence a se former avant que le precedent soit complete." Similarly J. Stanley Gardiner (1902, p. 133) in his account of the anatomy of Flabcllum nibrum says (italics added) : " As the growth of any corallite proceeds, more and more septa up to six cycles appear. The fanner exoccclic order of septa becomes entoccelic by the development of new pairs of mesen- teries. The increase of mesenteries takes place pari passu with the formation of new septa." Unfortunately, the relationships involved in the above asser- tions have not been actually followed, though from the known conditions no other arrangement at first sight seems possible ; and it was principally with a view to determine the truth or otherwise of the assumption that the present investigation was undertaken. FIG. 5. First appearance of the permanent second cycle of entosepta situated within the entocceles of the second cycle of mesenteries and the bifurcations of the dorsal and middle exosepta. • Shortly after the stage represented in Fig. 4 was reached inde- pendent calcareous growths began to arise peripherally, in posi- tions corresponding with the entocoeles of the second cycle mesenteries ; those within the dorsal entocceles were larger than the ones within the middle entocceles, while for a long period there was no corresponding formation within the ventral entocoele (Fig. 5). At first the structures were quite free and resembled 86 J. E. DUERDEN. small independent septa ; they suggest a new series of entoccelic septa, appearing in a dorso-ventral sequence like the mesenteries with which they are associated. Later these new septa extended more centrally, and necessarily came into union with the simple inner portions of the septa which originally constituted the exoccelic second cycle. Several of the polyps were reared until the new second and third orders of septa were fully established, when they presented the arrangement shown in Fig. 6. The peripheral, primarily independent septa FIG. 6. Completion of the first three cycles of septa. within the second cycle entocceles have all become continuous with the median part of the original second cycle exosepta, and along with them now constitute the permanent second cycle of entosepta ; while the bifurcations of the exosepta now consti- tute the third cycle of twelve septa and are seen to be exosepta.1 It will also be seen that the growth within the ventral system has now attained the same stage as that within the middle and dorsal systems, so that the corallite as a whole presents nearly perfect radial symmetry. The stages thus passed through are of great importance in their bearing upon several obscure points in coral development and morphology, and call for fuller consideration. In the first 1 In the adult corallite of Siderastrea radians the exosepta are fused by their inner end-; with the entosepta as represented in Fig. 6 and Fig. ~]g. MORPHOLOGY OF THE MADREPORARIA. 87 place, it is seen that the second cycle entosepta, which are to become the permanent second cycle of the adult corallite, arise as independent formations, though later they fuse with the septa which constituted the original second cycle, the two series being situated along the same radii. They are to be regarded as alto- gether new formations which replace the original second cycle exosepta ; the fact that they fuse with the latter in their forward growth would seem to be of incidental importance, depending upon the fact that they are in the same radii. Secondly, the original second cycle exosepta of Fig. 2 lose their morphological individuality, becoming involved in the new second cycle ento- septa as the latter continue their growth centrally ; they are merely the temporary predecessors of a later permanent cycle.1 As primary second cycle exosepta they do not become included within the entocceles of the second cycle mesenteries, but only as the central continuations of the new second cycle entosepta. The results thus afford definite proof that the exosepta of a former stage do not become the entosepta of a later stage when another series of mesenteries has appeared with the entocceles of which they correspond. It is manifest that such a conclusion could only be established in Siderastrea by actual observation of all the intermediate stages. When the condition represented in Fig. 6 has been reached there is no means of determining the actual two-fold origin of the second cycle entosepta. It is such condi- tions which have hitherto been studied, and from these no other explanation than that given by Delage and Herouard and by Gar- diner on page 70 would have been expected. Entosepta through- out would now appear to be new formations, not the continuations of the exosepta of a previous stage, and further, they arise after the mesenteries within the entocceles of which they are situated. 1 In many corals the original second cycle exosepta appear to continue their inde- pendent growth in situ without losing their identity in the central extension of the entosepta. I believe it will be found that this is the true nature of pali, which are found in some corals as small septum-like plates in front of the larger septa. The fact that pali seem not to occur before the primary cycle of six septa, but only before those of later origin, is what we should expect if this surmise be correct. The primary entosepta have never had exoccelic predecessors, as is the case with the later entosepta. Pali would thus represent the persistent exocoalic predecessors of the entosepta beyond the primary cycle which have not lost their individuality in the later growth of the entosepta. 88 J. E. DUERDEN. The significance of the twelve exosepta which constitute the outermost third cycle remains to be considered. They undoubt- edly represent the direct continuations of the bifurcations of the six primary second cycle exosepta, but their relationships have now changed ; instead of appearing as extensions of an older cycle of septa they themselves constitute a cycle. The details exhibited by the particular species studied seem inconclusive towards determining whether the exosepta of the later stage are to be regarded as but continuations of the exosepta of the pre- vious stage or as entirely new formations. The former would certainly seem to be suggested ; probably the point could be definitely established in some other coral species in which the exosepta in the adult are not fused at their inner extremity with the entosepta. Whichever view is accepted the exosepta of the third cycle are obviously developed in advance of the entosepta of the second cycle.1 Thus the relative sizes and positions which the cycles will ultimately assume in the mature calice do not represent their actual order of appearance. A third important morphological relationship is definitely established, namely, that the later septa do not arise a cycle at a time as is usually assumed, any more than do the mesenteries. Beyond the primary cycle of six entosepta, the members of which always appear simultaneously, there is a decided dorso-ventrality in the sequence of the septa of each cycle, which for the time being gives a marked bilateral symmetry to the corallum. It is only later, when the development of the septa within each sex- tant has reached the same stage, that an approximate equality and radial symmetry is attained. Thus the primary symmetry in corals is bilateral, and the arrangement is retained for a long period in the ontogeny of both mesenteries and septa. There- fore in the individual septa making up a cycle, as has been proved for the cycle itself, the adult size does not represent the actual order of appearance. To return to the further development of the septa in Sidcras- 1 It is worthy of note in this connection that the exotentacles in Siderastrea radians have been found to appear thoughout in advance of the entotentacles, being the only zoantharian in which this relationship is known to occur. Hence there is nothing contrary to the laws of hexactinian development in the above conception that the exosepta beyond the first series appear in advance of the corresponding entosepta. MORPHOLOGY OF THE MADREPORARIA. 89 trea. The larval polyps were not reared beyond the completion of the first two cycles of mesenteries and the first three cycles of septa (two cycles of entosepta and one of exosepta). On any colony, however, are many developing polyps which present intermediate stages between the commencement and completion of the third cycle of mesenteries and the fourth cycle of septa ; and from these results have been obtained supplementary to those already presented. Mature polyps of 5. radians have rarely more than three cycles of mesenteries and four cycles of septa (three inner cycles of entosepta and an outermost cycle of exosepta). First, it may be ascertained what are the relationships between the third cycle exosepta of Fig. 6 and the third cycle entosepta and fourth cycle exosepta found in the mature corallite. By- means of serial sections through decalcified immature bud polyps it has been possible to establish the relationship of these with respect to one another and to the new mesenteries of the third cycle. The results are diagrammatically represented in Fig. 7 (a — g), the complications due to the presence of synapticula being omitted. Fig. a represents a section through an exoseptum of the third cycle, corresponding to one of the exosepta in Fig. 6, only that the peripheral extremity is now bifurcated. It is in the same stage as each of the four exosepta of the second cycle in Fig. 4, but is shown united with the calicinal wall, and no mesen- teries have as yet appeared within the two limbs. Fig. b shows the same septum taken from a section at a higher level. Within the exoccelic bifurcation there has now appeared a pair of third cycle mesenteries, in every way comparable in their relationships to the third cycle exosepta with those of the second cycle mes- enteries to the second cycle exosepta of Fig. 4. Fig. c, from a still higher level of the polyp, reveals a further stage. Within the entoccele of the mesenteries is seen the rudi- ment of a third cycle septum. The latter is a new formation, comparable with the rudiments of the second cycle entosepta shown in Fig. 5. One limb of the exoseptum has also become free. Fig. d, taken from a section above that of Fig. c, shows the newentoseptum becoming larger, and extending centrally further 9o J. E. DUERDEN. than the mesenteries, which in their turn have also increased in size. Both limbs of the bifurcated septum are now free from the middle portion ; they represent two independent exosepta becom- ing distinct from the single exoseptum of a previous cycle (cf. Fig. 6) of which they were originally continuations. Fig. c is from a section through the column wall, the septa at this level being exsert, that is, extending above the calicinal wall. e f Q FIG. 7 (ft-g)- Series of diagrammatic figures illustrating the developmental relationships of a pair of third cycle mesenteries and a third cycle entoseptum, in association with a bifurcated third cycle exoseptum. The relationships shown are the same as in the previous figure, but the inner radial portion of the original third cycle exoseptum has almost disappeared. Fig. f is through the column wall (lower) and disc (upper). The mesenteries now extend from one wall to the other, the polyp being in the retracted condition, and the entoseptum is still smaller than the exosepta, one on each side of it. It is manifest that in the later growth the entoseptum will extend more centrally, and come into union with the original third cycle exoseptum which is in the same radius (Fig. g), exactly as in the larval polyp shown in Figs. 5 and 6. MORPHOLOGY OF THE MADKEPOKARIA. QI The series of sections thus reveals that the third and fourth cycles of septa and the third cycle of mesenteries are related in the same manner as are the second and third cycles of septa and the second cycle of mesenteries ; the third cycle entosepta have third cycle exoccelic antecedents. The results may be arranged as follows : (a) The originally simple third cycle exosepta, themselves formed as bifurcations of simple second cycle exosepta, become bifurcated at their peripheral extremity. (//) Within each bifurcation there appears a new pair of third cycle mesenteries, and then within the entocosle of the pair is formed a third cycle entoseptum. (<:*) The new entoseptum fuses with the central portion of the third cycle exoseptum, while the bifurcations of the latter con- stitute two new exosepta of the fourth cycle, and are fused at their inner extremity with the entoseptum embraced by them. Presumably the same process as above outlined will be followed within the two exoccelic chambers of each sextant of the polyp, so that in the end there will be twelve entosepta forming the completed new third cycle and twenty-four exosepta forming the completed new fourth cycle. The actual sequence according to which the third and fourth cycles of septa are formed remains to be noticed. The septa alone in the dried corallum are insufficient for this purpose, as they afford no certain means by which the directive axis can be determined, and from this the dorsal and ventral borders of the calice. It has been shown, however, that the mesenteries associated with the entosepta appear in pairs only a little in advance of the entosepta within them ; therefore if the sequence of the mesenteries be determined it can be assumed that the septa follow the same order. From colonies of Sidcrastrea sections of a large number of bud polyps at different stages of development have been pre- pared, and from these it has been possible to determine the order of appearance of the twelve pairs of third cycle mesenteries. This is indicated in the series of diagrammatic figures in Fig. 8 (a—d\ In Fig. 8 a, in addition to the primary and secondary cycles, a pair of third cycle mesenteries (III.) has appeared on each side of the median axis, situated in the exoccele between 92 J. E. DUERDEN. the dorsal directives and the dorsal pair of second cycle mesen- teries. Such an early stage is to be expected on the dorso-ven- FIG. Srf. FIG. 8 (a-d). Series of diagrammatic figures illustrating the order of develop- ment of the jtwelve pairs of third cycle mesenteries. tral succession already established in the case of the second cycle mesenteries (p. 70). in FIG. S6. The succeeding exoccelic chamber on each side lies between the dorsal pair of second cycle mesenteries and the dorso-lateral MORPHOLOGY OF THE MADREPORARIA. 93 pair of first cycle mesenteries, and it might be supposed that the new mesenteries would occupy the exocoeles in regular succes- sion from one aspect of the polyp to the other. Instead of this in in FIG. Sc. the pairs are found to arise successively within only the dorsal member of the two exocoeles of each system. This is shown FIG. 8J. in the next stage available, Fig. S£, where a third cycle pair is found within the dorsal exocoele of each sextant. 94 J- E. DUERDEN. A further stage secured in the growth of the twelve pairs of third cycle mesenteries is given in Fig. Sr, where a pair has ap- peared on each side within the ventral exocoele of the dorsal sex- tant. Clearly, if the succession thus indicated were followed with perfect regularity, other pairs would appear within the ventral exocceles of both the middle and the ventral sextants, and the cycle would then be completed according to Fig. Sd. No stage exactly corresponding with Fig. Sd, however, has been obtained, as the polyps of >$. radians very rarely, if ever, complete the third cycle of mesenteries. Still the sequence so far as it can be traced is such as to warrant the conclusion. The regularity in the dorso-ventral sequence of the mesen- teries shown in Fig. 8 was secured only after an examination of a number of polyps. In a colony in which the polyps are so closely arranged as in 6". radians the individuals are rarely found to undergo their later development with perfect regularity all round ; some regions will be in advance of the normal sequence and others behind. The polygonal form assumed by the adults is evidence that a certain pressure is exerted upon a form which would otherwise be circular, as in the simple polyps reared from larvae. Spatial difficulties may therefore be held sufficient to account for the irregularities appearing in the growth of the third mesenterial cycle. In Astrangia solitaria and PJiyllangia Ameri- cana, where the polyps are practically free from one another, and retain their cylindrical form throughout, the regularity of devel- opment from one border to the other is more pronounced, and I have found (1902, p. 459) the order of appearance of the mes- enteries to be the same as that established for 5. radians. The normal sequence of the third cycle mesenteries in Sidcr- astrea being now established we are justified in assuming that a like succession will be maintained by the third cycle septa, as individually the septa arise shortly after the mesenteries with which they are associated. Hence the normal sequence for the members of the first three cycles of entosepta will be that repre- sented in Fig. 9. The six septa of the first order of entosepta (I.) appear together as a cycle ; the six members of the second (Il.a—Il.c) follow a simple dorso-ventral succession ; the twelve members of the third order (Ill.rt-III/) also appear in a dorso- MORPHOLOGY OF THE MADREFORARIA. 95 ventral succession, but in two series - - first a series of six (lll.a— Ill.r) within the dorsal of the two interspaces in each sextant,, and then the remaining six (III.C/-III/) in a like order but within the ventral of the two interspaces. As explained below, the exosepta (X.) constituting the last outermost cycle have not the same ordinal significance as the entosepta. lla Hid Ilia FIG. 9. Diagram showing the order of appearance of the septa in a corallite with three cycles of entosepta ( I. -III. ) and an outer cycle of exosepta ( X. ). The Roman numerals indicate the cycle to which the septa belong and the letters their sequence in the cycle. Studies on the mesenterial sequence of other corals indicate that a similar septal succession will in all probability be followed by most forms in which the adult calice shows a regular hex- ameral cyclic plan. Individual departures from the order may be expected, but are to be looked upon as irregularities ; regu- larity of growth of the higher cycles of mesenteries and septa is 96 J. E. DUERDEN. by no means so pronounced as in the first and second cycles which are less likely to be influenced by spatial considerations. The sequence given is altogether different from anything which has hitherto been surmised for any coral, and further studies are desirable to determine how far it admits of general application in the group. From what has already been revealed it is manifest that the exosepta of corals do not possess any true ordinal sequence comparable with that of the entosepta. Exosepta have been found to be present at each developmental stage, always consti- tuting the outermost cycle, and equalling in number the sum of the inner entosepta ; but until the adult condition is reached they are merely the predecessors of the entosepta. We may con- sider them as the direct continuations of the primary six exosepta which bifurcate at each stage, or, less likely, as arising anew with each cycle of entosepta. Regarded as the persistent representa- tives of the primary exosepta they more nearly conform to the " law of substitution ' in actinian tentacles as established by Lacaze-Duthiers (1872) and Faurot (iSgs).1 Studies on other corals, as well as considerations on the tentac- ular development in actinians, suggest that the exosepta may arise in different ways in different species of corals, and that a more 1 In actinians generally it is found that after the protocnemic stage the tentacles appear two at a time, one entoccelic and one exocoelic, corresponding with the two chambers formed upon the appearance of a new pair of mesenteries ; sometimes the entotentacles appear in advance of the exotentacles, though in Siderastrea radians the reverse is the case. The entotentacles when established are larger than the exo- tentacles, the length of the former being in accordance with the order of appearance of the cycle to which they belong, the largest being the first to appear. The exoten- tacles all attain an equal length and throughout are relegated to the outermost cycle, whatever be the cycle of entotentacles with which they first appeared. They consti- tute a single cycle of which the members are smaller than those of the cycle of ento- tentacles last to appear. The number of exotentacles in the last cycle is always half the total number of tentacles, the number of exocceles being equal to that of the entocodes. Being soft polypal structures it is easy to understand how as new entotentacles are added the exotentacles become pushed aside and thus occupy different radii at differ- ent times. The septa, being hard fixed structures, do not admit of rearrangement ; the new growth has to be adapted to the old, as in the fusion of the new entosepta with the old exosepta. The tentacles, like the septa, thus arise in such a manner that it is impossible to determine their order of development from their relationships in the mature polyp. MORPHOLOGY OF THE MADKEPORARIA. 97 precise significance as to their relationships at different stages may be forthcoming than is possible in Sidcrastrca. There .are indications that in some forms an entoseptum and an exoseptum arise together, thus more closely recalling the method followed in the appearance of the tentacles. The relationships now proved to exist between the entosepta. and exosepta of corals involve important considerations when the cyclic hexameral sequence is not completed in the mature coral- lite, a condition which almost invariably happens in 5. radians, as well as in numerous other corals. As regards both septa and mesenteries it is found in such cases that the last cycle is rarely a multiple of six, but some irregular number, resulting from the fact that at maturity the polyp does not complete the last cycle begun. Exosepta have been shown to appear always in associa- tion with entosepta, whatever be the number making up a coral- lite, and, as often remarked, the two series are equal in number and the exosepta outermost in position. Hence it follows that in the cyclic incompletion of the mature corallites of Siderastrea the third entoccelic cycle and fourth exocoelic cycle of septa vary in the same degree ; whatever number of entosepta be lacking to form the complete third cycle of twelve a like number will be wanting from the twenty-four exosepta which should form the fourth cycle. When describing the number of septal cycles within a calice of which the cyclic hexameral plan is incomplete it is usual in systematic works on corals to regard the hexameral multiples as completed so far as the number of septa will permit, and then to relegate to the last cycle all the surplus septa not included in the hexameral formula. The cycles are all supposed to be hexam- erously complete with the exception of the last. Thus, with regard to 6". radians, Milne- Ed wards states : " Three cycles of septa complete, and, in general, a variable number of a fourth cycle"; likewise Verrill (1901, p. 133), describing the same species, says: "They [the septa] form three complete cycles, with part of the fourth cycle developed, so that the number is usually 36 to 40." The relationships now established between the entosepta and 98 J. E. DUERDEN. exosepta indicate that the above formulae do not express the true morphological character of the septa. Any hexameral incom- pletion in the number of septa making up a corallite affects both the entosepta and the exosepta, that is, both the penultimate and the last cycles ; if any septa be wanting to complete the hexam- eral multiple of the last cycle of entosepta the same number will be wanting from the outermost cycle made up of exosepta. The third complete cycle as understood by Milne- Ed wards and by Verrill is really made up of both tertiary entosepta and of tertiary exosepta. The two kinds of septa are obviously of very different value in their development and relations to the mesen- teries, and, as a matter of fact, will scarcely be of the same thick- ness and radial length to justify their being regarded as mem- bers of one cycle. The cyclic formula, as usually understood in systematic works, may be written, 6, 6, 12, X, where A' will represent any number from one to twenty-four. Formulated in this way the number 12 conveys the impression that the third cycle is really com- pleted, that all the remaining septa belong to the next or fourth cycle, and that it alone is numerically incomplete. But beyond the two first cycles the septa of the penultimate and last cycles are formed concurrently, or almost so, in pairs, and incomplete cyclic hexamerism, as met with in S. radians, is really an inter- mediate condition in the establishment of two adult hexameral cycles, not of one alone, and attention should be drawn to this in the septal formula. According to the relationships above established the morpho- logical septal formula for 5". radians should be written 6, 6, X, 6 -|- 6 + X. In this formula the numbers 6, 6, represent the septa in the first and second completed entocycles, and X the number in the third entoseptal or penultimate cycle which does not yet complete the hexameral sequence ; while 6 -f- 6 + X will represent the total number of exosepta, X being the same number as before. In the calice some of the exosepta will be tertiaries and some will be quaternaries, the number of the latter being always double the number of tertiary entosepta. The formula for a corallite having 36 septa would, according to the ordinary cyclic formula, be written, 6, 6, 12, 12, whereas, con- MORPHOLOGY OF THE MADREPORARIA. 99 sidered as entosepta and exosepta, the formula would be 6, 6, 6, 1 8 ; the three first numerals in the latter indicate the entosepta and the last the exosepta. The cyclic formula of a corallite with 40 septa would be 6, 6, 12, 16, and the morphological formula 6, 6, 8, 20. In the first morphological formula 12 of the exosepta will be quaternaries and 6 will be tertiaries ; in the second 16 will be quaternaries and 4 tertiaries. Where the relationships of the septa to the mesenteries are clearly known the morphological formula will more nearly express the real value of the septa than the ordinary cyclic formula, the latter has little significance unless the hexameral sequence is fully completed. One is not justified in saying that a cycle is really complete unless all its constituents have the same morphological value, and this is not the case when some are entosepta and some are exosepta. A few remarks may here be made concerning the dorso-ventral appearance of the organs in corals generally, and the consequent marked bilaterally of the calice for the time being. In palaeontological literature much has been made of the fact that the Palaeozoic rugose corals (Tetracoralld] are bilaterally symmetrical, while most modern corals are radially symmetrical. The results here outlined prove however that modern corals are strongly bilateral in the course of their development, and that it is only when the septa are fully established that an approximate radiality is assumed. In like manner I have found that many rugose corals attain perfect or almost perfect radiality when maturity is reached, though the developmental stages are strongly bilateral. Radiality in the Actinozoa, as compared with bilate- rality, seems to have a more ontogenetic than phylogenetic sig- nificance. Furthermore, beyond the six primary members the septa in the Rugosa are added in a manner altogether different from that in modern hexameral corals, hence the bilaterality of the one group has a different origin from that of the other. The subject of bilaterality in the Rjigosa will be more fully discussed in a later paper. In modern corals the bilaterality of the polyp during develop- ment may be looked upon as associated in turn with each cycle IOO J. E. DUERDEX. individually. Any cycle of septa or mesenteries tends to attain its radial condition before the next cycle commences to form, when the additions take place in such a manner as to again con- fer bilaterality upon the polyp as a whole. Thus the first two cycles of septa become truly radial before an additional cycle commences, when the growth of this is continued in a bilateral manner ; likewise the new second and third cycles assume their radial stage before the members of the fourth cycle made their appearance, these also proceeding from one border to the other. In like manner the first cycle mesenteries are nearly radial before those of the second cycle arise and introduce a conspicuous bilateral symmetry ; on the second cycle mesenteries assuming the radial plan the third cycle members begin to appear, again in a bilateral manner. The successive dorso-ventral growth followed by the constit- uent mesenteries and septa of each cycle also confers a certain individuality upon the cycle. The different cycles, arising inde- pendently, seem to represent so many distinct recurring phases of growth in the life of the polyp ; they do not constitute a con- tinuous addition from one border to the other, as is usual in per- manently bilateral animals, particularly segmented forms. The members of a cycle appear in a dorso-ventral sequence and may retain their differences in size for a long time, but in the end they become equal and thereby confer radial symmetry upon the polyp. Then another cycle commences to form in somewhat the same bilateral dorso-ventral succession, displays for a time the consecutive origin of its members, and afterwards attains radiality. The conception of recurring phases of growth in cyclic coral polyps is best realized when comparison is made with the mesen- tenal increase characteristic of the Ceriantheae and Zoantheae. In the former the mesenteries beyond the protocnemes always develop in a regular bilateral successive manner, from the dorsal (anterior, sulcar) to the ventral (posterior, asulcar) aspect, the oldest being dorsal or anterior and the youngest ventral or pos- terior, recalling more the method followed by segmented animals ; there is in the Ceriantheae never a reversal of growth to the ante- rior end, followed by a successive series to the other, such as occurs in ordinary hexactinians. Employing the term " band of MORPHOLOGY OF THE MADKEPORARIA. IOI proliferation," introduced by van Beneden in 1897, we may say there is only one median band of proliferation in cerianthids, while in hexactinians there are many such bands occurring all round the polyp, the number increasing with age — at first six, then twelve, twenty-four, etc. In the Zoantheae also mesenterial development is always in the same succession after the protocnemic stage. The increase takes place within only two of the six primary exoccelic chambers, one on each side of the ventral directives ; there are only two bands of proliferation or zones of growth. In this case, how- ever, the order followed by the new mesenteries differs from that in hexactinians and cerianthids ; it proceeds from the ventral (posterior, sulcar) to the dorsal (anterior, asulcar) aspect of the polyp, not from the dorsal to the ventral. The bilateral development of the organs, from one border of the polyp to the other, in ordinary actinians and corals would seem to have no phylogenetic significance beyond the group of the ccelenterates ; indeed, even here we appear to have as yet no definite understanding as to what its meaning may be. The approximate radial symmetry of adult ccelenterates is assumed from very diverse developmental conditions (cf., hexactinians, zoanthids, cerianthids, and the tentacles and other cyclic organs in the Hydromedusae and Scyphomedusae). Whatever may be said in favor of the well-known view that the mesenterial arrange- ment in cerianthids suggests the metamerism of higher animals there is clearly no support for such a conception in the develop- ment of the organs in hexactinians. In this latter group we are concerned with a radial cyclic repetition of the organs, even though the members of each cyclic series arise in bilateral suc- cession from one border to the other. SUMMARY. 1. In the coral Siderastrca radians the six members of the first cycle of septa appear simultaneously, shortly after fixation of the larva, situated within the entocceles of the first cycle of mesenteries. 2. Six members of a second cycle are developed within the primary exocceles shortly after the primary cycle of septa. They are the temporary predecessors of a later permanent cycle, IO2 J. E. DUERDEN. and arise either simultaneously or in bilateral pairs in a dorso- ventral order. Later, they become bifurcated peripherally, either by the direct extension of the original septum or by the produc- tion of separate fragments which subsequently fuse. The bifur- cations also appear in a bilateral dorso-ventral order. 3. The six members of the permanent second cycle of ento- septa arise within the entocceles of the second cycle mesenteries soon after these make their appearance. The two right and left dorsal septa appear first, then the two middle members, and, at a much later period, the two ventral, the series thus exhibiting a decided dorso-ventrality. In the end they become equal, and each fuses with the central part of the corresponding second cycle exoseptum previously developed, these exosepta thereby losing their individuality. 4. Twelve members of a temporary third cycle are situ- ated within the exocceles between the primary and secondary pairs of mesenteries, and represent the bifurcated extensions of the six primary exosepta. The original second cycle exosepta thus become the third exoccelic cycle, their place having been taken by the new second cycle of entosepta. 5. A new third cycle of twelve (or less) septa arises on the appearance of the pairs of third cycle mesenteries, in a similar manner to that followed by the second permanent cycle. New entosepta appear within the entocceles of the third cycle mes- enteries, and the bifurcations of the third cycle exosepta then become the exosepta of the fourth cycle. 6. The third cycle entosepta, following the mesenteries, are developed in a bilateral dorso-ventral order, but in two series ; first a series within the dorsal moiety of each sextant, and then a second series within the ventral part of each sextant. 7. Exosepta are present at each cyclic stage in the growth of the corallum, alternating in position and corresponding in num- ber with the sum of the entosepta. They never become ento- septa, but always constitute the outermost cycle of shorter septa ; only the entosepta have any ordinal significance. Until the adult condition is reached the exosepta are the temporary prede- cessors of the entosepta. The developmental relationships be- tween the entosepta and exosepta are closely comparable with MORPHOLOGY OF THE MADREPORARIA. IO3 those between the entotentacles and exotentacles. The law of substitution, first discovered by Lacaze-Duthiers for the tentacles of Hexactinias, is thus found to hold also for the septa. 8. Where the cyclic hexamerism of a corallite is incomplete the ordinary cyclic formula does not express the true relation- ships of the septa ; the entosepta and exosepta vary in the same degree, so that the true morphological septal formula for a cor- allite with three cycles and part of another is 6; 6, X, 6 -f 6 + A' where X may be any number from i to 12. 9. The cycles of septa and mesenteries represent so many dis- tinct recurring phases of growth at intervals all round the polyp, not a continuous increase from one extremity to the other as in metameric animals. With the exception of the first the mem- bers of each cycle follow a dorso-ventral succession, display a bilateral symmetry for some time, and ultimately assume an approximate radial plan. The succession for the third cycle of entosepta is two-fold. REFERENCES. Delage, Y., and Herouard, E. '01 Les Ccelenteres. Traite de Zoologie Concrete, torn. II., Pt. 2. Paris. Duerden, J. E. '02 West Indian Madreporarian Polyps. Mem. Nat. Acad. Sciences, vol. VIII., 7th Mem. Faurot, L. '95 Etudes sur 1'anatomie, 1'histologie et le developpement des Actinies. Arch, de Zool. Exp. et Gen., ser. 3, vol. III. Gardiner, J. S. '02 South African Corals of the Genus Flabellum, with an account of their Anat- omy and Development. Marine Investigations in South Africa. Vol. II., Cape Town. von Koch, G. '81 Mittheilungen iiber das Kalkskelet der Madreporaria. I. Das Vermehr- ungsgesetz der Septen. Morphl. Jahrb. , bd. VIII. von Koch, G. '82 Ueber die Entwickelung des Kalkskeletes von Astroides calvcularis und dessen morphologischer Bedeutung. Mitt. a. d. Zool. Stat. zu Neapel, bd. III. von Koch, G. '97 Entwickelung von Caryophyllia cyathus. Mitt. a. d. Zo5l. Stat. zu Neapel, bd. XII. IO4 J. E. DUERDEN. Lacaze-Duthiers, H. de. '73 Developpement des Coralliaires. Deux. Mem. Actiniaires a Polypier. Arch, de Zool. Exp. et Gen., torn. II. Lacaze-Duthiers, H. de. '94 Faune du Golfe du Lion. Evolution du Polypier du Flabellum anthophyllum. Arch, de Zool. Exp. et Gen., 3 ser., torn. II. Lacaze-Duthiers, H. de. '97 Faune du Golfe du Lion. Coralliaires. Zoanthaires Sclerodermes. Arch, de Zool. Exp. et Gen., 3 ser., torn. V. Milne-Edwards, H., and Haime, J. '57 Histoire Naturelle des Coralliaires ou Polypes proprement dits. Paris, 1857-60. Schneider, A. '71 On the Structure of the Actiniae and Corals. (Translated by \V. S. Dallas from the Sitzungsbericht der Oberhessischen Gesselschaft fur Natur- und Heil- kunde, March 8, 1871) Ann. Mag. Nat. Hist., 4 ser., vol. VII. Semper, C. '72 Ueber Generationswechsel bei Steinkorallen und iiber das M. Edwards' sche Wachsthumsgesetz der Polypen. Zeit. f. Wiss. zool., bd. XXII. Verrill, A. E. '01 Variations and Nomenclature of Bermudian, West Indian and Brazilian Reef Corals, with notes on various Indo-Pacific Corals. Trans. Conn. Acad. Science, vol. XI. EVOLUTION IN A DETERMINATE LINE AS ILLUS- TRATED BY THE EGG-CASES OF CHIM/EROID FISHES. BASHFORD DEAN. Recent attempts to explain evolutional processes, whether, e- g-i by natural selection, orthogenesis or use-inheritance, have been based, with but few exceptions, upon parent and offspring, in the direct relation of one to the other. Thus, (i) in the newly developed field of biometrics, variations in parent and offspring, have been discussed in terms of precision ; (2) in embryology and embryopathology, ingenious experiments have tested the latent possibilities of the offspring at different stages in its career, and (3) on the side of paleontology, variations of progeny and parents have been examined on a giant scale in terms of survival and obliteration of masses of individuals. On the other hand observations are scanty, even in the present outburst of literature, which test the relation of the young to its parents by indirect means. This is, none the less, a line of inquiry which bids fair to become a fruitful one. And we may even at the present time consider what materials can be obtained which shall provide a series of parallel changes between the offspring and, e. g., some other product of the parent, and through such means furnish, as it were, a point d'appui for evolutional studies. Thus : Are there any means of ascertaining whether an animal in providing for its progeny can produce structures ivhich form no organic part of the young yet u'hich at the same time indicate accurately ivhat the young unll need in the distant future. And if these structures do occur, are they sometimes so special in their nature that they can- not be interpreted as general provision for the embryo, but rather for the late and complicated needs both of the genus for ivhich they are provided and even of the species ? Obviously the parent has physical continuity with its young, but has it more than this ? Has it the power to anticipate accurately the needs of the young, which it abandons and with whose subsequent fate it 105 IO6 BASH FORD DEAN. clearly has nothing to do ? Can it, for example, start an egg on its course of development and then provide it with a capsule whose special structures shall "foresee" accurately what the young is to become ? Can it form a capsule which shall have the power, in spite of its lifeless substance, to develop as the en- closed egg develops, so that at each period it can best serve the corresponding stage of the embryo's growth ? Such instances, if they can be found, are evidently of value in the examination of the complex problems of heredity. The following notes are given since they appear to illustrate an extreme case in point : and since they also indicate that some- what similar " purposeful " conditions may be demonstrated in the secondary embryonic membranes of other forms, i. c., that these structures may be found the better adapted to the future grade of development of an embryo than has hitherto been suggested. In examining the shark-like fish, Chim "-*$ r»- i^ '&3& m 9 I l6 WILLIAM PATTEN. ventral direction. Its anterior end lies back of the cervical suture and from there it extends backwards about two thirds the length of the post-cephalic portion of the buckler. It is readily recog- nized by its fine, soft matrix and the concentric black lines, each line clearly and sharply defined and separated from the adjacent ones by regular intervals. The lines are similar in color and thickness to those made by the skin when seen in section. The whole structure, while puzzling, is unquestionably produced by some organic structure, not by sedimentation within an irregular cavity, or by mud accumulated within the alimentary canal. The anterior boundaries of the black lines are ill defined, but the pos- terior ones often form distinct loops as though the whole struc- ture consisted of a series of broad lamellae wrapped around a central axis, and with free, more or less separated, posterior mar- gins. In some cases the whole mass of lamellae is much distorted, or they may protrude from some rupture in the walls of the crushed shield. Under these conditions they still preserve their essential characters, showing that while originally soft and flexible they had at least as much firmness as the skin of the trunk. Be- hind the laminated portion, the soft matrix extends in an irregular undefined and structureless mass toward the cloaca. The remain- ing space within the buckler may be filled with a coarser matrix similar to that in which the whole animal is imbedded. On the dorsal side of the laminated body, there are usually scattered fibrous masses, and one or more irregular, undefined bony plates, apparently attached by vertical sheets of blackened tissue to a low median ridge on the inner surface of the anterior median dorsal. Anteriorly, this ridge deepens into a prominent hollow process directed downwards and forwards toward the thickest part of the covering on the laminated core, Fig. 2. Several specimens have been found with the mouth parts in their natural position, held there by membranes, whose contours can be determined with considerable accuracy. The oral region, Fig. 3, is covered by an undulating structure- less membrane in which are imbedded the various oral plates. The membrane is attached to the lateral and anterior margins of the head, as far at least as the shoulder on the anterior border of the mandibles. It extends backwards, underneath the anterior NEW FACTS CONCERNING BOTHRIOLEPIS. II/ ventro-laterals, as far as the large transverse ridge on their inner surface, to which it seems to be attached. It extends across the median line without interruption except between the mandibles. It also appears to be absent between the mandibles and maxillae, Figs. 2 and 3, 0.111, The mandibles are thin, concave plates of bone, continuous with the oral membrane on the sides and in front, but with free ynd FIG. 3. Ventral side of head, with the anterior ventro-laterals removed on one side, to show extension backwards of oral membrane to the transverse ridge, tc ; /./>, lateral plate attached to the two transverse bars; »r, mouth; »ti/, mandibles; /, olfactory opening ; /, two pineal eye pits on under side of post-orbitals ; p.p, pineal tubercle on outside of pineal plate ; p.s, posterior sclerotics. X about -,%• the outer surface, Fig. 6. A pair of similar pits, but not quite as deep and without any surface indications of their presence, are clearly shown on the inner surface of the small post-orbital. There is every reason to regard these three pits as similar in character and as indicating the presence of a tri-ocular median eye. 122 WILLIAM PATTEN. A broad thin shelf of bone extends inwards from the ant- orbital plate forming a wide chamber between the anterior surface of the T-shaped bone described by Whiteaves, or the ethmoid as I shall call it, and the two arms of the ant-orbital, Fig. 2. When the anterior face of the ethmoid is exposed, it is seen that its arms nearly enclose two circular openings, 6 C. M. CHILD. through the body-wall. The body-wall becomes thinner at the oral end (Fig. 9) and tentacle-buds appear (Fig. 10), but these usually do not develop beyond the stage shown in Fig. 10. Occasionally they reach a length of 2—3 mm., but further than this their development never proceeds. The distension gradu- ally decreases again after a week or two and the tentacles which FIG. 8. FIG. 9. FIG. 10. FIG. ii. FIG. 12. FIG. 7. FIG. 15, had attained a length of 2—3 mm. decrease in size to mere buds. In cases where the tentacles never develop beyond the stage represented in Fig. 10 they are scarcely visible at all after two or three weeks. In some pieces, however, closure of the aboral end occurs sooner or later. This result may be attained in either of two ways- In one case the aboral cut surfaces of the oesophagus and body- wall fail to come into contact (Fig. 1 1) and the aboral end closes in the usual manner by union of the cut surfaces of the body- FORM-REGULATION IN CERIANTHUS. 2O/ wall leaving the aboral end of the oesophagus free in the enteron (Fig. 14). Such pieces do not differ essentially from pieces in which the aboral end lies below the oesophageal region. They become distended in the usual manner and regenerate typically, but more rapidly than pieces in which regeneration of mouth and oesophagus occurs. In the other case the cut surfaces of body-wall and oesophagus unite in the manner described above (Fig. 8), either completely or on a part of the circumference. Then it may happen that different portions of this region of growing tissue are brought into contact as in Fig. 12. The result is the union of all these parts and so the closure of the aboral end (Fig. 13). The con- nection between the oesophagus and body-wall is soon broken as the piece becomes more fully distended and the condition repre- sented in Fig. 14 is attained. From this stage on regeneration at both ends proceeds in the typical manner. In some cases this clo- sure of the aboral end and loss of connection between oesophagus and body-wall does not occur at first, but later the margins of the body-wall happen to come into contact perhaps in conse- quence of the gradual decrease in distension mentioned above as following the first increase, and thus closure occurs and is fol- lowed by renewed distension and rapid regeneration. Appar- ently the new tissue retains the power of making new unions for a considerable time after the cut surfaces of body-wall and oesoph- agus appear to be firmly united. The final result in all oesophageal pieces in which the aboral end succeeds in closing aboral to the oesophagus is typical regu- lation with well-developed tentacles at the oral end and several millimeters of new tissue at the aboral end. The data of a few experiments will serve to indicate the uni- formity of results. In all cases C. solitarius was used. Series jo. September 28, igo2. — Seven oesophageal pieces were prepared from large specimens in good condition as follows : at the oral end of each animal a transverse cut was made through the disc at such a level that the marginal tentacles and margins of the disc were removed and the labial tentacles were cut off near their 2O8 C. M. CHILD. bases leaving stumps i mm. or less in length. Then a second cut was made through the ossophageal region near its aboral end. In the distal pieces thus formed the oesophagus extends from end to end of the piece, cut surfaces being present at each end (Fig. 15). October i. — 3 days after section. In all cases the cut mar- gins of ossophagus and body-wall have united both orally and aborally in the same manner as in Fig. 8. All pieces are more or less distended and the oral margins slightly crenated in corre- spondence with the intermesenterial chambers. October 4.. — 6 days after section. All pieces with marginal tentacles, varying in length in different pieces from mere buds just visible to 2 mm. : labial tentacles still short stumps. October 7. — 9 days after section. In two pieces the aboral end has closed as in Figures 12—14: these pieces fully distended ; one with marginal tentacles 5 mm. ; labial tentacles about 2 mm.; the other with marginal tentacles 3mm., labial tentacles 1 — 1.5 mm. In the five remaining pieces no aboral closure has occurred ; the distension has decreased ; the marginal tentacles have in- creased in length, though very slightly, since the previous exami- nation, varying from minute buds to 1-2 mm. October 75. — 1 7 days after section. One of the pieces which closed aborally with marginal tentacles 8— 10 mm., labial tenta- cles 5-6 mm. At aboral end new tissue 1-2 mm. The other closed piece was lost. In none of the five remaining pieces has tentacle-regeneration proceeded since the previous examination. All appear almost completely collapsed and the tentacles are reduced to mere buds in all, being scarcely visible in some. One piece has begun to break up. During the following two weeks the five pieces which did not close broke up into small pieces and died, a frequent occurrence in small pieces in which closure and regeneration do not occur. The one closed piece remained in good condition. The difference between the piece that closed and the other five is striking. None of these five pieces ever regenerated marginal tentacles more than 1-2 mm. in length, and practically no re- generation of labial tentacles occurred. Yet the one piece which FORM-REGULATION IN CERIANTHUS. 2OQ closed while no longer than these and differing from them only in that the oesophagus communicated with the enteron, regener- ated marginal tentacles 8-10 mm. in length and labial tentacles 5-6 mm. and produced new tissue at the aboral end. Scries 32. October i, 1902. — Seventeen oesophageal pieces were pre- pared (Fig. 7) the distal cut being in this case slightly more aboral than in Series 30 (Fig. 15) so that both marginal and labial tentacles and most of the surface of the disc were removed. During two weeks these were examined several times. In all pieces the cut margins of oesophagus and body-wall united both orally and aborally and in none did the closure of the body-wall across the aboral end occur. All of the pieces became slightly distended during the first few days and marginal tentacles ap- peared as minute buds less than i mm. in length. In no case did regeneration proceed beyond this stage : labial tentacles never appeared. At the end of two weeks some pieces were beginning to break up. Series 4.7. November 7, 1902. — Nine oesophageal pieces were prepared in the same manner as those of Series 32. November 10. — 3 days after section. Still more or less com- pletely collapsed : no tentacles visible on any. November 12. — 5 days after section. One piece closed abor- ally, and well-filled with water; marginal tentacle-buds 0.5—1 mm. In the remaining eight pieces oesophagus and body-wall have apparently united both orally and aborally : these pieces are only slightly distended and show a slight crenation of the oral margin corresponding to the intermesenterial chambers. November 20. — 13 days after section. During the interval since the last examination seven pieces closed aborally across the end of the oesophagus. These are fully distended and bear marginal tentacles 2—3 mm.; labial tentacles just appearing. In two pieces the oesophagus remained open aborally. These are not fully distended and the marginal tentacles are minute buds about 0.5 mm.; labial tentacles are absent. December 2. — 23 days after section. In the seven closed 2IO C. M. CHILD. pieces regeneration has proceeded in the typical manner ; mar- ginal tentacles 5—6 mm.; labial tentacles 2—3 mm. In the other two pieces regeneration has proceeded no further. They are still only slightly distended and with minute tentacle- buds. In this series a larger proportion of the pieces closed aborally across the end of the oesophagus than in any other series of this kind. The date of this series was later in the year than that of any other, /. e., the temperature of the water was lower and the distension of the pieces occurred much more slowly (see Child, '03$). It is probable that different parts of the aboral cut sur- faces of the body-wall remained in contact for a longer time than in the other cases where distension occurred more rapidly, and that this prolonged contact made union possible in a greater number of cases than in other series. Other conditions such as difference in the degree of contraction of body-wall and oesoph- agus at the time the lower cut was made may have aided in bringing about this difference. However that may be, the point of chief importance, viz., the difference in regeneration between those pieces which did close aborally across the oesophagus and those which did not is as clear in this series as in others. While the general result of these experiments is sufficiently clear special attention may be called to certain points. In all cases in which the oesophagus remains open aborally and consequently does not communicate with the enteron the pieces never become fully distended and regeneration is slight. Pieces cut below the cesophageal region become well distended and the marginal tentacles may attain a length of 2—3 mm. be- fore the mouth is formed. In my first paper ('03^) diffusion of water through the body-wall in consequence of the presence of soluble products of metabolism or other substances in the en- teron was suggested as a possible cause of this first distension. It is also possible that secretion of fluid into the enteron occurs. In the cesophageal region the entodermal layer is thinner and undoubtedly of less functional importance than in the region aboral to the oesophagus, and it is reasonable to suppose that the accumulation of fluid in the cesophageal pieces would be much less rapid than in pieces from regions aboral to it. If this be FORM-REGULATION IN CERIANTHUS. 211 admitted it follows that less water would enter these pieces and less distension would occur than in pieces aboral to the oesoph- agus. If the internal pressure affects tentacle-regeneration, less regeneration would occur in consequence of this slight disten- sion in cesophageal pieces than in others where distension is greater, and this is actually the case. It was noted that the pieces in which the oesophagus remains open aborally gradually collapse again in the course of a week or two. This collapse is probably due to a slow loss of the fluid which first caused the distension. If the body-wall is slightly permeable for substances in solution in the enteron the distension produced at first will gradually diminish since in consequence of continued starvation the amount produced gradually decreases. Whatever the exact nature of the process may be, a decrease in the distension occurs. As this process continues the regenerat- ing tentacles also decrease in size instead of continuing to grow. This fact is important as showing the apparent close relation between tentacle-regeneration and internal water-pressure. In pieces cut below the oesophagus the regeneration of the mouth and the entrance of water through it serves not only to prevent decrease in the internal pressure but to increase it, and further growth of the tentacles takes place. The most striking feature of the experiments is the difference in behavior as regards regeneration between the pieces in which the body-wall closes aborally across the oesophagus and those in which the cesophagus remains open aborally to the exterior. The pieces of the first kind begin to regenerate typically as soon as the closure occurs, while the others never produce anything more than marginal tentacles 1-2 mm. in length. In the intro- ductory description of these experiments it was shown that in case of closure the connection between the aboral end of the body-wall and the cesophagus is severed and the oesophagus opens into the enteron (Figs. 12—14). The breaking of this con- nection, z'. e., the change from the condition shown in Fig. 13 to that of Fig. 14 is probably itself due at least in part to internal water-pressure, though the manner of its occurrence cannot be observed. Other factors, such as a difference in rapidity of growth or a difference in resistance between the tissue of the 212 C. M. CHILD. oesophagus and that of the body -wall may also play a part in the result. Whatever be the cause, the result is free communi- cation between a fully formed oesophagus and the small enteric cavity with its intermesenterial divisions. If entrance of water through the oesophagus serves to maintain internal pressure in Cerianthns it is clear that in these pieces the pressure should at once become about equal to the pressure in a normal animal since the mouth opening and oesophagus are not regenerated structures but parts of the parent body and of full size. If internal water pressure affects regeneration we might expect in these pieces rapid regeneration, even more rapid than in pieces with regenerating mouth and oesophagus. Close comparison of these pieces with others in which mouth and oesophagus are formed by regeneration is difficult, since in the cesophageal pieces regeneration may be at first delayed until the aboral closure is complete, but the piece in Series 30 in which closure of the aboral end occurred between October 4 and October 7 and by October 15 the marginal tentacles were S-io mm. and the labial tentacles 5-6 mm., show the rapidity of regeneration. In fact regeneration in this piece after closure is more rapid than in any other case noted in my records with the exception of other pieces of the same kind. The striking difference between cesophageal pieces in which the aboral closure occurs and those in which it does not consti- tutes evidence of great importance for the influence of internal water-pressure on regeneration. As regards the influence of local pressure due to circulatory currents in determining the posi- tion of tentacles and in their regeneration these experiments afford no direct evidence. They show merely that the greater the dis- tension of the piece and consequently the more widely open the intermesenterial chambers and the greater the force and volume of the circulatory currents the more rapid is regeneration. SUMMARY. i . In pieces with oblique oral end the rapidity of tentacle re- generation differs on different parts of the disc, being greatest on the uppermost (most oral) portion and least on the lowest (most aboral) portion. The delay in regeneration on the lowest portion FORM-REGULATION IN CERIANTHUS. 213 as compared with the highest portion is much greater than the difference in rapidity of regeneration due to difference in level. 2. The only reason apparent for the difference in rapidity of regeneration on different parts of oblique discs is the difference in the angle between disc and body-wall, which is acute on the upper side of the disc while on the lower side it is obtuse. In consequence of this difference the local pressure exerted by the circulatory currents passing orally in the intermesenterial chambers must be much greater on the upper side of the disc than on the lower. If regeneration is influenced by these currents we should expect delay on the lower side, and it occurs in all cases. 3. The later equalization in length of the tentacles in pieces with oblique discs is due to the conditions which bring about equality in length of tentacles in normal animals. After the appearance of the tentacles the conditions as regards internal pressure are essentially similar on all parts of the oblique margin and equalization must be expected if the length of the tentacles is determined by internal pressure. 4. The reduction in the obliquity of the disc in oblique pieces is a compensatory process resulting from the attempt of the animal to orient itself with longitudinal axis vertical and disc horizontal. In the attempt at orientation unequal contraction of the muscles on different parts of the circumference occurs, and its continuation brings about changes in the tissues which lead gradually toward the establishment of the typical form. The form is the result, not the cause of the reaction. 5. In pieces cut wholly within the cesophageal region the two cut ends of the oesophagus usually unite with the cut ends of the body-wall, thus leaving the oesophagus open to the exterior at both ends. This method of closure isolates each intermesenterial chamber completely and prevents any direct communication between the enteron and the exterior. 6. Such oesophageal pieces become slightly distended at first in consequence of diffusion of water through the walls, but since the oesophagus is not in communication with the enteron the internal pressure remains far below that of the normal animal. Regeneration of the marginal tentacles begins in such pieces, but never proceeds beyond the formation of mere buds. 214 c- M- CHILD. 7. Occasionally an cesophageal piece closes aborally by union of the body-wall across the end of the oesophagus. In these cases communication between the oesophagus and enteron is established, the piece becomes fully distended, and regeneration proceeds in the typical manner. 8. The difference in regeneration between the " closed " and "open" cesophageal pieces is undoubtedly due to the difference in the degree of internal water-pressure. HULL ZOOLOGICAL LABORATORY, UNIVERSITY OF CHICAGO, November, 1903. BIBLIOGRAPHY. Child, C. M. '02 Studies on Regulation, I. Fission and Regulation in Stenostoma. Arch. f. Entwickelungsmech. , Bd. XV., H. 2, 1902. '03 Studies on Regulation, II. Experimental Control of Form- Regulation in Zooids and Pieces of Stenostoma. Arch. f. Entwickelungsmech., Bd. XV., H. 4, 1903. '033 Form-Regulation in Cerianthus, I. The Typical Course of Regeneration. Biol. Bull., Vol. V., No. 5, 1903. 'O3b Form-Regulation in Cerianthus, II. The Effect of Form, Position, Size, and Other Factors upon Regeneration. Biol. Bull., Vol. V., No. 6, Vol. VI., No. I, 1903. *O4a Form-Regulation in Cerianthus, III. The Initiation of Regeneration. Biol. Bull., Vol. VI., No. 2, 1904. 'O4b Form Regulation in Cerianthus, IV. The Role of Water- Pressure in Regeneration. Biol. Bull., Vol. VI., No. 6, 1904. '040 Form-Regulation in Cerianthus, V. The Role of Water-Pressure in Regeneration : Further Experiments. Biol. Bull., Vol. VII., No. 3, 1904. PORTABLE ANT-NESTS. ADELE M. FIELDE. Portable ant-nests, constructed by me in the summer of 1900, were described in BIOLOGICAL BULLETIN, No. 2 of Vol. II., which is now out of print. Improvements that have since been made in them, and their present use by myrmicologists in America, Europe and Africa, justify a new statement of the method of making them. I have now ants that have lived in them, without earth, for three years, in health and apparent contentment. The floor of the nest is a pane of double-thick, transparent glass. This is laid upon very thick, white blotting paper, giving an elastic bed to the pane of glass and the best background for observation of the ants. The paper has just the area of the glass, but is not fastened thereto. The outer walls of the nest are laid a quarter inch, or six mil- limeters, from the edge of the pane. They consist of two strips of double-thick glass, a half inch, or thirteen millimeters, wide, the one strip superimposed on the other. Both are held in place by crockery cement.1 The wall is smoothly laid up, with no in- terstices where an ant may hide or escape. The partitions are double the width of the wall, which they otherwise copy. At one end of every partition a space is left whereby the ants may pass from room to room. This passage- way is covered by a thin celluloid film or a piece of mica. It is desirable that this covering be transparent, so that the passage- way underneath it may be scanned from above, on lifting the end of the toweling which is to overlay it. After the cement is well dried, the edge of the floor-pane and the outside of the walls are covered with a fabric impervious to light. Cloth serves better for this purpose than does paper, the edges of the nest being subject to much handling. Le Page's or 1 The use of cement instead of glue was recommended by Dr. W. M. Wheeler. Diamond, Major's or any reliable kind may be used. I have merged in water, for two weeks, a nest constructed with Major's cement, without loosening its parts. The directions accompanying the selected cement should be followed 215 216 ADELE M. FIELDE. some other good liquid glue is used for securing the fabric upon the walls. The walls and partitions are topped by Turkish toweling of a sleazy sort, folded over one layer of cotton wadding so that the edges of the strip of toweling meet in the center of the under side of the wadding. The wadding is cut to the same width as the wall or the partition. The toweling is just twice the width of the wadding, and its edges are basted evenly together, making a cushion of even thickness. It serves the double purpose of ad- mitting air into the nest and of preventing the escape of the ants between the roof and its supports. It is held taut and is made level ; is fitted snugly at the corners ; exhibits no ravelings to afflict the ants ; and is firmly glued to the glass beneath it. When a cushion becomes soiled by long use of the nest, the glue may be softened by soaking and the cushion may be re- moved and be replaced by a new one. The ends of the cushions are fringed out a half inch or more, and are left open so that the enclosed wadding may be adjusted to present a perfectly level surface. , -...-^_^ _ -—•" : FIG. 2. The A nest completed. There is a glass roof-pane for each room in the nest. The glass is thin ; extends to the middle of the partition and to the outer edges of the walls on which it rests ; prevents the exit of ants ; and permits observation of their behavior. The glass may be without color, or it may be of a red or orange tint that will partially exclude ultra-violet rays of light. Ants perceive only such rays of light as are of short wave-length and, by use of a spectroscope, a glass roofing may be selected which renders the V ..- Food B C Food - •;•->. .-". f. . gjjji '• *-* '-' '•, fe? Food D ~j Food I 1 r- m . Food Food Bases of Seven hit -nests, Filling TJiree Shelves of Portable Case. FIG. i. PORTABLE ANT-NESTS. 21 7 ants visible within the nest while it protects them from such light- rays as they instinctively shun.1 If such glass is used for roof- ing the nest, the ants will behave as if in the darkness where they habitually live. An outer roofing of blotting-paper makes the interior of the nest wholly dark. The food-room should be light, as it represents the ant's outside world. When any room in the nest requires cleaning, it is covered only with transparent glass, and then the ants withdraw from it with their young into a dark room, which may in its turn be made light. The food-room is dry, and in cool weather requires attention but once a fortnight. Sponge-cake merged in a little honey or molasses, banana, apple, mashed walnut, and the muscular parts and larvae of insects are among their favorite edibles. Food is constantly attainable in the nest, but it is introduced in tiny morsels that it may not vitiate the air. Since moisture encourages the growth of mold, no water is put into the food-room. But ants often drink, and they require a humid atmosphere. All other rooms than that allotted to their food are made humid by laying a flake of sponge on the floor and keeping the sponge saturated with clean water dropped twice a week from a pipette. The proportion of the floor which is covered by the sponge depends on the degree of moisture in the soil usually chosen as the habitat of the species. The sponges are kept clean by weekly washing and an occasional immersion in alcohol. Sponges of fine tough texture render best service, as they offer no apertures where the ants may conceal their eggs. The flake of sponge is so thin as to permit the ants to pass between it and the roof-pane. The completed nest is less than half an inch or thirteen milli- meters in its interior height, and does not exceed three fourths of an inch or two centimeters in its exterior height. A low-power lens is easily focused upon the ants within the nest. During four years of experimental work with ants, I have found that the nests whose bases are represented in the drawing 1 " Supplementary Notes on an Ant," Adele M. Fielde, Proceedings of the Acad- emy of Natural Sciences of Philadelphia, June, 1903, p. 492. 218 ADELE M. FIELDE. meet the requirements of the various species that I have desired to house for long periods. A is B " 10 C " 6 D " 6 X ° inches or 41 <6 25 X 6 " 15 < 4 " 15 (15 centimeters. X I5 < 15 " < 10 Nests of these dimensions fit into and fill a portable wooden case or box having an interior length of 17 inches or 43 centi- meters ; a width of 7 inches or 1 8 centimeters ; and a height of 4^ inches or 12 centimeters. It is made of half-inch pine boards, dovetailed at the joinings. The interior of the case is equally divided lengthwise into four compartments by three shelves that are supported in grooves cut in the end-pieces of the case. The shelves are a quarter inch or six millimeters thick, and they leave space at their outer edge for the inset of a door which forms the front of the case, is hung by hinges at the bottom, and is held shut by two buttons affixed to the top-piece of the case. FIG. 3. Portable Case for the Nests. Holes are bored in both sides of the case for the inlet of air to the enclosed nests. A handle placed lengthwise on the top of the case permits its convenient carriage. When its four compartments are filled with PORTABLE ANT-NESTS. 2 19 nests, its weight is less than sixteen pounds. Ants have made long journeys in my portable nests, with no grave disturbance of their domestic arrangements.1 When preparing the nests for a journey, tapes are tied around them, so as to hold the roof-panes securely in position, and bits of wadding are so inserted as to prevent their displacement in the case. Four of the A nests, sixteen of the D nests, or a selec- tion from among the A, B, C and D nests may be carried in the case. Before constructing my ant-nests, I made and used those of the Lubbock and of the Janet patterns, both much older than my own. The Lubbock nest, holding the ants on an island by a moat filled with water, is not portable ; and whenever the pane of glass, covering the layer of earth over the island, needs to be cleaned, there must be a disturbance of the domestic interests of the ants. But the base of the Lubbock nest is a valuable adjunct when the ants are to be housed in my nests. It consists of a square or oblong block of wood, about two inches or five centi- meters thick, with a channel grooved to half the thickness of the wood at a half inch from its edge all around. When the channel, which is an inch or more in width, is filled with water, the island thus formed serves well for the temporary confinement of ants. The ants are brought, as Janet suggested, from their wild nests in little bags permeable by air, or in jars whose mouths are cov- ered by gauze. The contents of the bag or jar are deposited thinly upon the island ; a piece of glass covered by blotting paper and raised slightly above the general surface is laid over some portion of the area ; and the ants, within a few hours, gather the young underneath the darkened glass. Their progress in their work is made visible by an instant's lifting of the blotting paper. Selections from the total capture may be made for removal to the glass nests. My nests of earlier construction, like the Janet nests, had an aperture in the wall through which the ants could themselves transport their young from the earth into the glass nest. But I have found it expedient to personally make selection from the total capture rather than to allow the ants to bring their whole community into the glass nest. 1 The photographs with which this paper is illustrated were very kindly made for me by Mr. J. G. Hubbard and Dr. O. S. Strong. 22O ADELE M. FIELDE. Forel's method of making, upon a table, a stockade of dry plaster of Paris, to prevent the escape of the ants deposited within it, serves well when large colonies are dealt with. But ants have high regard for personal cleanliness and are discom- forted by the adhering dust which punishes their effort to escape over this stockade. The Lubbock island renders clean stock. The Janet nest, a series of four pits in porous stone, cement or stucco, with water in the pit at the end opposite the food-pit, proved that the ants may live healthfully without earth ; showed the practical value of more than one compartment in the artificial nest ; and gave an excellent background for viewing the ants. But the Janet nest is cumbrous and weighty ; and its food room becomes quickly mouldy in hot weather. The glass nests are constructed at less expense than are those of either the Lubbock or the Janet pattern ; they are easily kept clean, and the small space which they occupy, with their very light weight, greatly facilitates the bringing of the ants under close observation. THE MARINE BIOLOGICAL LABORATORY OF WOODS HOLL, MASS., July, 1904. EXPERIMENTS ON THE ORIGIN OF THE CLEAVAGE CENTROSOMES. EDWIN G. CONKLIN. (From the Zoological Laboratory of the University of Pennsylvania.) The great diversity of opinion as to the origin of the cleavage centrosomes which appear during the fecundation of the egg is well known to all students of cytology. By different authors every possible view has been held with regard to the source of these centers, as may be seen from the following classification : 1. The cleavage centrosomes come from the sperm centro- some ; Boveri ('87, '92, '95), Vejdovsky ('88), Fick ('93), Wilson and Mathews ('95), Hill ('95), Mead ('95), Reinke ('95), Kos- tanecki and Wierzejski ('96), Kostanecki and Siedlecki ('96), Sobotta ('97), MacFarland ('97), Erlanger ('97), Griffin ('99), Coe ('99), Linville (1900). 2. The cleavage centrosomes come from the egg centrosome ; Van Beneden ('87), Wheeler ('95) ; all cases of normal and arti- ficial parthenogenesis. 3. The cleavage centrosomes come from both egg and sperm centrosomes; Fol ('91), Guignard ('91), Blanc ('93), Conklin ('94), Carnoy and Lebrun ('99). 4. The cleavage centrosomes come from neither egg nor sperm centers but are new formations ; Foot ('97), Lillie ('97), Child ('97), Mead ('98).' It is probable that some of these conflicting views may be attributed to erroneous observations, and the writers who have maintained the first view have in general explained all others as due to this one cause, but on the other hand some of the evi- dence in favor of these other views cannot be thus lightly brushed aside. As long as this question remained on a purely observa- tional basis no one seems to have seriously considered that there might be an element of truth in more than one of these views and that the cleavage centrosomes might arise differently in differ- 1 For references see Wilson, " The Cell in Development and Inheritance," 1900. 221 222 EDWIN G. CONKL1N. ent animals or even in the same animal under different condi- tions. The experiments of R. Hertvvig, Morgan, Loeb and Wilson have shown that the unfertilized egg is capable of giving rise to cleavage centrosomes and spindles, while the observations of Delage, Boveri and others on merogeny have proven that a nor- mal mitotic figure may appear in connection with the sperm nucleus in enucleated egg fragments. Under these circumstances it ought to be possible, by slightly altering normal conditions, to bring about the formation of a spindle in connection with each germ nucleus. In the summer of 1901, while at the Marine Biological Labora- tory at Woods Holl, I undertook some experiments on the eggs of Crepidula in order among other things to test this possibility. Since this animal is one which does not conform to the prevalent view as to the origin of the cleavage centrosomes it seemed all the more favorable for such work. As the eggs of these gas- teropods are fertilized while still in the oviduct I have found it impracticable to experiment with unfertilized eggs but have worked entirely with eggs into which a spermatozoon had already entered. The normal course of the fecundation in this animal may be briefly recalled : After both polar bodies have been extruded the egg nucleus lies at the animal pole in an area of cytoplasm while the sperm nucleus lies in the yolk near the periphery of the egg and usually near the vegetative pole. Then the egg centrosome which is left in the egg at the close of the second maturation division, rapidly disappears and in its place is left the large egg aster or sphere. At no stage is there a clearly marked sperm centrosome, but a radiating cytoplasmic figure, the sperm aster, develops in connection with the sperm nucleus and these two migrate through the yolk until they come into contact with the egg nucleus and sphere, immediately under the polar bodies. Here the egg and sperm spheres fuse and at their periphery two separate and independent centrosomes appear which ultimately come to lie at opposite poles of the first cleavage spindle. It is probable that one of these centrosomes comes from the egg sphere and the other from the sperm sphere though there is no ORIGIN OF THE CLEAVAGE CENTROSOMES. 223 positive evidence that they are directly derived from egg and sperm centrosomes. If now the eggs of Crepidula planet which have given off the second polar body but a short time before are brought for four hours into a I per cent, solution of NaCl in normal sea water FIG. i. FIG. 2. FIG. FIG. 4. FIGS. 1-4. Eggs of Crepidula flana treated with I per cent. NaCl in sea water for four hours, viewed from the side. In all the eggs the polar bodies are at the upper pole, the egg nucleus and centrosome, or spindle, immediately below this, while the sperm nucleus lies in a small area of cyptoplasm near the lower pole. Various stages in the formation of the egg spindle are shown. they will present the appearances shown in Figs. 1 — 12. These figures are camera drawings of eggs in different stages of the formation of the first cleavage spindle and all are from the same experiment. An examination of these figures shows that various stages in the division of the egg centrosome occur (Figs. 1-5) leading to 224 EDWIN G. CONKLIN. the formation of a perfect spindle of small size. There can be no doubt that the centrosomes of this spindle are derived from the egg centrosome. The sperm nucleus is at this time far re- moved from the egg nucleus and is closely surrounded by yolk ; no astral radiations are found in connection with it, though one FIG. 5. FIG. 6. FIG. 7. FIG. 8. FIGS. 5-8. NaCl eggs of C. plana viewed from the animal pole ; the two polar bodies are shown nearly over the spindles in the first three figures. In Fig. 5 the sperm nucleus is still some distance from the egg spindle ; Fig. 6 shows an egg spindle and a sperm spindle joined at one pole ; Figs. 7 and 8 show stages in the for- mation of a tetraster from the egg and sperm spindles. or two granules which lie close to the nuclei may possibly represent centrosomes (Fig. 4). Unfortunately a gap occurs at this stage in my material ; in all the preceding figures (1-5) the sperm nucleus is small and densely chromatic and is far removed from the animal pole ; in the succeeding figures (6—12) the sperm nucleus has been re- ORIGIN OF THE CLEAVAGE CENTROSOMES. 225 solved into chromosomes which lie near those of the egg. In all cases there are two spindles present which are usually united into a tetraster though they may be more or less independent of each other. In some eggs yolk spherules separate the two spindles so that a tetraster is not formed (Figs. 9, 10), while in others an incomplete tetraster is formed (Figs. 6, 7) ; in still FIG. 9. FIG. 10. FIG. ii. FIG. 12. FIGS. 9-12. NaCl eggs of C. plana, viewed from the animal pole; the polar bodies lie above the spindles in all the figures. Fig. 9. Egg and sperm spindles united at one pole. Fig. 10. The two spindles quite separate. Figs, n and 12. Complete tetrasters in different phases of the separation of the chromosomes. others the tetraster is complete, each of the four poles being united to the other three (Figs. 8, 1 1, 12). Although I have not seen the genesis of the sperm spindle there seems to be no reason for doubting that it is formed essen- tially as the egg spindle is and that the two are at first wholly independent ; only later do secondarily formed fibers connect one 226 EDWIN G. CONKLIN. or both of its poles with those of the egg spindle. Since under normal conditions one of the cleavage centrosomes of Crepidtila arises in connection with the egg and the other with the sperm nucleus the only respect in which these experimental results differ from the normal is that in the former the centrosomes divide while the nuclei are still far apart thus giving rise to two spindles or to a tetraster, whereas in the latter these centrosomes do not appear until after the germ nuclei and spheres have met and they do not divide until the prophase of the second cleavage. In such a case as this we have essentially the same phenomenon as is found in dispermic echinoderm eggs (Driesch, Boveri, Wilson) save that in the former the two additional centrosomes are not derived from an additional spermatozoon but come from the egg centrosome. Of this fact there cannot be a particle of doubt and it seems to me to shed light upon the much discussed question as to the source of the cleavage centrosomes of differ- ent animals under normal conditions. It has evidently been a mistake to suppose that the cleavage centrosomes could arise in but a single way and that all animals must conform to this single type. Under experimental condi- tions the cleavage centrosomes may arise in one and the same animal in connection with the sperm nucleus, in connection with the egg nucleus or, as I have shown in connection with both of these nuclei, while it is possible that they may arise dc novo anywhere in the egg cytoplasm, though this latter view is by no means so well supported by evidence as are the former ones (see Conklin 1902). It is highly probable therefore that the source of the cleavage centrosomes may differ in different ani- mals, or even in the same animal under different conditions. It is interesting to note that this whole discussion as to the supposed importance of the source of the cleavage centrosomes had its origin in the thought that the sex cells were in them- selves incomplete and incapable of development save as each com- plemented the other (Boveri '87, '91), or in the rival notion that the centrosomes were the bearers of heritable qualities (Fol '91). In the light of recent experimental work both of the these views are seen to be untenable and the subject has therefore lost most of its interest and significance. Vol. VII. October, lyoj. No. 5 BIOLOGICAL BULLETIN POWER OF RECOGNITION AMONG ANTS. ADELE M. FIELDE. WITH PHOTOGRAPHIC ILLUSTRATIONS BY MR. J. G. HUBBARD AND DR. O. S. STRONG. When power of recognition is to be tested, appeal for evidence of that power is properly made through the leading sense. We should take evidence from the eagle through its sense of sight, from the mole through its sense of hearing, from the caterpillar through its sense of touch, from the ant through its sense of smell. That ants have associative percepts which are independent of their chemical sense, is proven by their behavior. Ants learn to be unafraid of the light from which they instinctively withdraw their young.1 When ants are put into an artificial nest, some weeks are required for their making acquaintance with their domicile, but after such acquaintance has been perfected, they may be transferred to a replica of their abode, whether it be a Petri cell, a glass house, or a wooden box, and they will be wholly at ease in it, and will quietly resume their accustomed routes over its floors or withdraw into it from a strange environment, although o ^5 it lacks their nest-aura. I divided a small colony of Camponotus Pennsylvanicus into two sections. I fed and fondled the members of the one sec- tion until they manifested a sense of safety in my presence, would mount my finger and make a leisurely promenade upon my hand or would return to me after an hour's wandering in my room. These ants not only ceased from biting me when I took them upon my hand, but became so tame that they would 1 " Supplementary Notes on an Ant," A. M. Fielde, Proceedings of the Academy of Xatural Sciences of Philadelpliia, September, 1903, p. 493. 228 A. M. FIELD E. remain quiescent in their Fielde nest l for some minutes after the roof-pane had been lifted. Upon the ants of the other section I practiced such atrocities as that of lifting them frequently by the leg with a pair of forceps or plunging them for an instant in cold water. The ants of this section quickly acquired such associations with the lifting of their roof-pane, that they fled through the compartments of their house in wild panic whenever I touched the glass. PRELIMINARY STATEMENTS. Is is the purpose of this paper to show that ants have power to recognize certain ant-odors after months or years of separation from those identical odors, and in order that the evidence presented may be plain, it appears necessary to first restate certain facts, well established by past experiments. 1. Ants of different species in different communities or colo- nies, and also ants of the same species and variety in different communities or colonies, ordinarily show aversion to each other on meeting, and are especially truculent in defense of their young. The cause of this general and perpetual feud among ants of dif- ferent colonies, is due to difference of odor, discerned through their antennae, their organs of smell. Fear and hostility are excited in the ant by any ant-odor which she has not individually encountered and found to be compatible with her comfort. 2. If ants of different species, or even of different genera or subfamilies, are made pleasantly acquainted with each other within a few hours after hatching, they will thereafter continue to live together in arnity, constituting a mixed colony.2 The ac- quaintance thus formed is individual, and every ant, in her later behavior, will act in accordance with individual experience. Ac- quaintance with the odor of one species or colony does not secure from the experienced ant an amicable reception of a representa- tive of any other species or colony. Among the mixed colonies, formed by me in August, 1903, twenty Stenamma fnlvuui workers lived with a Cremastog aster lineolata queen a full year, and the harmony of the nest was as 1 Described in BIOLOGICAL BULLETIN, Vol. II., No. 2, 1900, and Vol. VII., No. 4, 1904. 2 "Artificial Mixed Nests of Ants," A. M. Fielde, BIOLOGICAL BULLETIN, Vol. V., No. 6, 1903, p. 320. POWER OF RECOGNITION AMONG ANTS. 2 29 complete as if its inmates had been of one species. Representa- tives of three subfamilies, Formica subscricca, Stenammafulvum, and Stigmatomma pallipcs lived amicably together five months before the last Ponerine ant died, leaving the Camponotine and the Myrmicine ants to continue together a full year. Formica laswdcs, Lasins latipes, Stenamma fidvitm, My r mica rubra and Cremastogaster lineolata affiliated through several weeks. All these and other mixed colonies continued until they were disinte- grated by me, and friendly treatment was accorded in them to all introduced ants bearing a familiar and therefore an approved odor, while hostile attack was made on every ant bearing an unfamiliar and therefore a disapproved odor. An enlarged acquaintance with ant-odors did not render any ant tolerant of unknown ant- odors, and in no established mixed colony was an ant of any other than the already represented colonies permitted to live, even when the introduced ant was of the same variety. I have at the present time a mixed colony of Camponotus pic- tus^ Formica subsericea, Formica lasiodes, and Stenamma fulvum, and although there are no young of any species in their nest, they have killed every one of several newly-hatched Cremasto- gasters that I have introduced, and they allow none of the latter species to live when hatched in their nest from introduced pupae. 3. Ants inherit odor from the queen from whose eggs they are developed. That the queen endows her eggs with an odor, and that newly-hatched queens and workers have an odor recog- nized by their queen-mother, is proven by the fact that an ant may be isolated from the pupa-stage until it is some days old, never having smelled any ant-odor beside that of its own body, and it will instantly snuggle its queen-mother at first meeting, although it may attack other queens, or sister-workers much older than itself. I have known a young worker to identify its mother among five queens of its species presented for its exami- nation. The queen doubtless recognizes her own odor in the callow that she has never before met. As I have to present the results of several experiments in which the N colony appears, it will be well to here give some I 1 am indebted to Dr. William Morton Wheeler for kind identification of Cam- ponotus herculeanus fictus, and of Formica pallide-fulva. A single ant of the last- named species lived for a year in one of my artificial nests with many Formica sub- sen cea. Fir,, i. Cainponotus pennsylvanicus. Somewhat magnified. Actual length of queen 2 centimeters. POWER OF RECOGNITION AMONG ANTS. 2JI account of that colony. On July 28, 1903, this N colony of Cainponotns pcnnsyhanicus was captured on Nonamesset Island, and was housed in a large Fielde nest. It consisted of a queen two centimeters long, some scores of workers, and numerous cocoons. During the first week in August, 1903, the queen deposited about one hundred eggs, and then ceased laying until the following March. The first larva from the August eggs was observed on August 27. From these larvae the first cocoon appeared on March 13, 1904. On April 8, 1904, there were many large larvae in the nest, and there were numerous cocoons varying in length from five millimeters to thirteen millimeters. The first cocoon of this brood hatched on April 24, the tem- perature of the room being 24° C. or 76° F. These cocoons con- tinued to hatch, most of them in carefully segregated groups, until July 14, when the last cocoon rendered its callow. E.vperuncnt A. — Three large workers hatched each in isolation on July 8, 1 1 and 14, 1904, from the August eggs of the N queen. On August 5, these three worker-ants, ranging from twenty-two to twenty-eight days in age, and never having met any ant-queen, nor any ant older than themselves, instantly affiliated with their queen-mother, and with each other, at the first meeting. The queen manifestly recognized the odor borne by the callows, and at once snuggled with them.1 They each recognized in her and in each other the only ant-odor they had ever known, that of their own bodies. Experiment B. — Four N colony workers, the issue of eggs deposited by the queen in August, 1903, were hatched from se- gregated cocoons, between May 15 and June 15, 1904. On July 6, when the age of these ants ranged between twenty-one and fifty-two days, and they had never met any other ant of their species, I introduced their queen-mother to their nest. The five immediately affiliated and the previously introduced larvae were brought and placed beside the queen. The queen must have been at least one year older'than these workers, and the workers must have recognized in the queen their own odor at hatching time. Experiment C. — Five Camponotus workers hatched between 1 The behavior of the ants in these experiments was observed through an orange- tinted roof-pane, under which the ants behave as if in darkness. See " Supple- mentary Notes on an Ant," referred to in first foot-note. 232 A. M. FIELDE. April 25 and May 10, 1904, out of cocoons from the August eggs of the N queen. They were from fifty-seven to seventy- two days old, and had never met other ants of their species, when on July 6, 1904, I introduced to their nest their queen-mothei. Within five minutes the queen had touched antennae with every worker and had become the center of a friendly group. Experiment D. — Fourteen Camponotus workers hatched on or after April 24, and on or before May 10, 1904, out of cocoons from eggs laid by the N queen the preceding August. They had never met other ants of their species, when on July 6, 1904, the oldest of the segregated group was seventy-three days old, and I introduced to their nest their queen-mother. There was for an instant great excitement in the nest, and some tentative nabbfng of the queen ; but in less than one minute the workers had discovered that she was their own. Within a few minutes, four of the workers had licked the queen, one had stood upon her back, and seven others had grouped themselves close about her. The behavior of these workers toward their queen indicates that her odor is an unchanging one, or that if there be a change in her odor it is but slowly effected. The behavior of these ants toward their queen was markedly unlike their behavior toward their sisters, when great diversity in ages was represented. 4. Worker ants change in odor as they advance in age,1 as was shown by experiments made by me in 1902. Further evi- dence of this fact will be offered later in this paper. Forty days may pass with no change so great as to elicit from former ac- quaintances any expression of suspicion or antagonism, but in other cases from forty to sixty days so differentiates the known odor as to inhibit association between the ants.- This suspicion '"Notes on an Ant," A. M. Fielde, Proceedings of the Academy of Natural Sciences of Philadelphia, December, 1902, p. 609. Also " Cause of Feud Between Ants of the same Species Living in Different Communities," A. M. Fielde, BIOLOG- ICAL BULLETIN, Vol. V., No. 6, 1903, p. 327. 2 All the ants employed in the experiments recorded by me have been under my constant care and my frequent observation. No person beside myself has ever had access to them. They have spent the summers, from the first of June to the end of September, at the Marine Biological Laboratory, Woods Holl, Mass., and the re- mainder of every year in New York City. POWER OF RECOGNITION AMONG ANTS. 233 or antagonism is always shown ! by the younger ants, toward the older, if the ants be of the same colony, and if the young ants have been guarded from association with ants older than themselves.2 Experiment &--On July 18, 1904, I put into the nest of two segregated workers of the N colony, hatched from cocoons taken from the wild nest July 28, 1903, and at the time of this ex- 1 In all experiments here recorded, unless a contrary condition is indicated, the ants whose recognition of a visitor was in question had in their nest inert young that had engaged their attention during several previous days. The action of resident ants toward a visitor is much more prompt and decisive when there are larvae or pupre in the nest. 2 In experimenting with ants, a third source of individual odor, not set forth in this paper, although always reckoned with in the experiments, lies in the other ants with which the individual associates. This odor appears and disappears with certain ex- ternal conditions. Ants take on the odor of their associates in a mixed nest, and this incidental odor usually disappears after about ten days of isolation, the inherent odor then reasserting itself. Ants may be smeared with the juices of ants of another species or colony and may thereby become immediately subject to attack from com- rades from whom they have been but momentarily removed. I have lately submerged ants for eighty hours or more in distilled water at a tem- perature of 10° C. , putting two species or two colonies into the same water, using thirty-five cubic centimeters for a dozen ants, and I have found that the ants of each species or colony, when revived and returned to their former nest, were attacked as are enemies, and that ten days proved to be an insufficient time for the reassertion of the inherent over the incurred oder. That these attacks from comrades were due to the alien odor acquired in the water and not to some other cause, was shown by the fact that when similar ants were likewise submerged in unmixed groups, the returned ants were amicably received by their former comrades. I thus submerged, in thirty cubic centimeters of water, three Cainpouotiis pictus and fifteen Steuanuna fithniui, for eighty hours. One of the Camponotus revived, and a day later I returned it to its three former comrades, employed in the care of cocoons in a small Fielde nest. The three instantly attacked the one, and would doubtless have slain it had I not interfered. Having rescued, I isolated it for ten days in a Petri cell, and then again returned it to its former nest. Two of the three resident ants at once attacked and killed it. Of the Stenammas several revived, and on my returning them to their former nest, they were all killed by their quandam associates. Of four Stenammas that revived and recovered after eight days submergence in com- pany with five Camponotus pennsylvanicus, three were killed by former comrades on my returning them to their nest. One of the four was isolated by me, in a Petri cell, for twenty days before I returned her to the nest, and there the returned ant was for some minutes the center of an examining circle of many ants. A day later she was being dragged by two workers, but she was ultimately restored to good standing in the nest. It is interesting to observe the puzzled or critical demeanor of an ant engaged in ascertaining whether a new-comer has an incurred or an inherent foreign odor. 234 A. M. FIELDE. periment eleven months old, a callow of the same colony, just seven days old. The callow received careful examination from the older ants, and was then amicably entertained by them. They doubtless recognized in the callow their own early odor. Experiment F.--On July 6, 1904, after the queen and the fourteen workers mentioned in experiment D were all serenely grouped in the nest there described, I introduced two workers1 of the N colony, hatched between August 14 and September 3, 1903, and therefore about ten months old, while the resident ants were not over seventy-three days old. The two visitors had affiliated previously with the N queen, and had been approved by her. They were probably the issue of her eggs of the previous year. They were, however, seven or eight months older than the resident workers, and, although they were larger than any worker resi- dent, they were persistently attacked and dragged, sometimes by more than one resident at a time. The visitor-ants did not retal- iate. One of them tried to placate her young sisters by offers of regurgitated food, and after a half hour there were signs of dimi- nution in the strength of the attacks. I- then removed the visi- tor-ants. There are all degrees in the hostility shown by ants to one another, as well as many variations in the degree of close- ness in their affiliations. In this experiment, the older ants had met no younger ones during their lives, and the younger ones had never before en- countered sisters older than themselves. RECOGNITION OF ODORS OF OTHER SPECIES. TJic A Scries. Evidence of ability to recognize odors that have not been encountered during many months has been taken from ants of diverse species and of a recorded life history. In August, 1903, I formed a mixed colony of workers of Camponotus pennsylvani- ciis, Formica subsericea and Stcnannna fnlvnm, all of whom hatched in my artificial nests between August 14 and Septem- ber 3. Every ant within a few hours after its hatching was 1 My method, which is also Forel's, of marking ants so as to readily distinguish them from others of their species, is described in a foot-note of " Notes on an Ant," already referred to. Ants were marked whenever an experiment required it. POWER OF RECOGNITION AMONG ANTS. 235 made acquainted with every predecessor in this mixed colony and no discord appeared in the new nest. This mixed colony was marked A. On September 24, when none of these ants was less than twenty days old, and none more than forty-one days old, I separated them accord- ing to genera, putting the Camponotus, the Formicas and the Sten- ammas each into a Fielde nest that was new and therefore had nonest- aura. The nests were respectively marked Ai , Az and A3. No young was permitted to hatch in any nest, but inert young from the workers' eggs, or introduced lar- vae from other nests were always present when tests of recognition were to be made as se- vere as possible. No ant of any species was admitted to either of the segregated groups except as recorded in the ex- periments. Nest Ai. --On April 8, 1904, there were in the Camponotus nest, Fie,. 2. Stenamma fitlvmti, with larvce and pupae ; slightly magnified. Actual length 5 to 7 mm. 236 A. M. FIELDE. marked Ai, three large and vigorous workers, without young. I then introduced two Stenammas that the Camponotus had known in the previous September in the A nest. The Ste- nammas manifested terror and the Camponotus made instant and violent attack upon them, so that I intervened in the ensu- ing battle, saved the lives of the Stenammas, and returned them to their own nest. It was evident that during the six and a half months of separation these ants had changed in odor, and that the odor borne by them in April was unknown to the ants with whom they had associated in the previous September. I was unable to offer to this group Ai the companionship of young Stanammas having the same odor as had their former associates at the time of association, and young Stenammas taken at a later date from the wild nest in the summer of 1904, were always killed by them. The Formicas were likewise rejected. Nest A2. — The Formicas that had lived, none less than twenty and none more than forty-one days, in nest A, were dom- iciled in nest A2. They had been separated from the Campono- tus six and a half months, when on April 8, 1904, I introduced into their nest, where there were twenty-six workers and no young, a Camponotus from nest Ai, comrade of their earliest days. The Formicas immediately attacked the Camponotus, and I removed the latter to save her life. Six month's progress in odor forma- tion had carried her outside the acquaintance of her former asso- ciates. The same antagonism was manifested toward the other two residents in nest A I . These Formicas continued to reside in their nest and had laid a few eggs, which were under their c;ire when on April 25, 1904, I introduced to their nest three Camponotus newly hatched from eggs that were laid in August, 1903, by the queen-mother of the Camponotus in nest Ai. These young Camponotus received amiable welcome from the resident Formicas, and during the next ensuing days, to May 10, I added several more callows. The Formicas, now eight months old, continued to live amicably with the young Camponotus, whose odor had been known to them in their earliest days. They regurgitated food to the young ants, permitted them to cany the egg-packet and care for the larvae, and in all respects treated them as if of their own colony. There POWER OF RECOGNITION AMONG ANTS. 237 was no death in the nest for twenty days after the introduction of the young Camponotus, but I removed them on June 25. The Formicas recognized, after seven months of separation from it, an ant-odor previously knozvn to them. I was unable to introduce to this nest any young Stenammas that the Formicas would accept, as I could command none of the same lineage and age as those known to them in the autumn of 1903. The individual Stenammas in nest A3, with whom they had formerly associated in the A nest, were now of another odor, and the Formicas refused to affiliate with them. Nest Aj. - -The Stenammas in this nest had lived in nest A, none less than twenty and none more than forty-one days, and since September 24, 1903, they had met no ant other than those in their own nest. On April 8, 1904, 1 introduced into their nest, where, there were forty-one workers, a former comrade, a For- mica from nest A2. She was fiercely attacked, and I removed her. Other Formicas from the same nest were likewise attacked. I then introduced Camponotus from nest Ai, and they were received with like animosity. It was instantly made evident that the resident Stenammas found in every visitor an unfamiliar odor. I removed each visitor as soon as her status among these Sten- ammas had been made plain, and the A3 nest remained quies- cent until April 24, when I introduced a day-old Camponotus, the issue of an egg laid the previous August by the mother of the rejected individual in nest Ai. Within a few minutes after its introduction, three of the Stenammas had licked the Campo- notus, and all the Stenammas had viewed it with approval. It was taken care of as tenderly as if it had been a Stenamma cal- low. On the same day I added two newly hatched Camponotus of the same lineage, and they were kindly entertained until April 27, when I removed them to another nest of Stenammas, the M nest. These M Stenammas were of the same age and colony as were the Stenammas in nest A3, differing from them only in never having lived with Camponotus. There were also about the same number of resident ants. As soon as I introduced the Camponotus into the M nest, the Stenammas attacked them, and although they were double the size of the residents, and of tougher integument, the residents harried them to death. The 238 A. M. FIELDE. FlG. 3. Lasiits lalipi-s, magnified. Actual length 5 millimeters. behavior of the M ants indicates that that of the A 3 ants was based on a recognition of odor, and that//rr Stenaimnri s power of recognition extends through an interval of at least seven months. The B Series. Another series, marked B, was also established, having its beginning on July 20, 1903, additions of newly-hatched work- ers being made to the mixed colony up to Au- gust 28, or during a pe- riod of forty days. The B mixed colony consisted of Stenamma fufown of the C colony, Lasins la- tipes of the F colony, and Cremastogaster lineolata of the H colony. On September 24, 1903, I separated these ants ac- cording to genera, segre- gating each genus in a new Fielde nest, marked Bi, or B2 or 63, where it remained until the end of the recorded experi- ments, with no other ants than those here men- tioned ever introduced or permitted to hatch in its nest. POWER OF RECOGNITION AMONG ANTS. 239 Nest B i. --On April i, 1904, over six months after their segregation, there were in the Bi nest about sixty Stenammas. I then introduced to their nest two Lasius, their former comrades, taken from nest B2. No fear was evinced by the Lasius, and no aversion by the Stenammas. Residents and visitors perfectly affiliated, and the two Lasius remained safely in the Bi nest four full days. I then, on April 5, removed the two Lasius, and put them into a nest of Stenamma fiihnun where there were two queens and sixteen workers, all hatched in August, 1902, and therefore a year older than were the ants in nest Bi. They were from the same C colony, but had never associated with Lasius. In this nest the Lasius tried to flee or to hide, behav- ing as do ants when in a nest of recognized enemies. At first they eluded the Stenammas, but when I again examined the nest, on the evening of the same day, both Lasius had been killed and put on the rubbish-heap. It appears that these very strong-smelling ants, the Lasius, had not so changed their odor during six months that the Bi Stenammas did not recognize it. But on June 1 1, 1904, I took from the wild nest of the F colony two Lasius workers, of un- known age, and introduced them into the Bi nest of Stenammas. They were soon killed, and two like them were also killed when introduced on the ensuing day. On August ii, 1904, I introduced a newly hatched Lasius from the R colony, and it was at once killed. On June 1 1, 1904, I sought in the old wild nest of the Lasius F colony for a remainder of its population, and secured a few workers and four larvae. On August 4, three cocoons and one naked pupa had appeared from these larvse, and from these cocoons the first worker hatched on August 18. It was im- mediately put into the Bi nest, where it appeared as a yellow pigmy among brown giants. It was much patted with the antennae, was licked, and was cared for among the eggs, larvae and pupae over which the Stenammas were strenuously engaged. The Stcnai/unas recognized an odor from which tJicy had been eleven months separated. The Stenammas of nest Bi met no Cremastogaster from Sep- tember 24, 1903, until July 7, 1903, when I introduced into their 24O A. M. FIELDE. nest, occupied by fifty of the old residents, with larvae from their own eggs, several newly-hatched Cremastogaster lineolata from a wild nest V. Within a day all these young Cremastogasters had been killed and their bodies piled together in a corner of the food-room. Others of their kind met a like fate on July 9. On the 24th of June, I had happily secured some larvae, with workers to rear them, from the wild nest of the old H colony of Cremastogasters, and on July 22, I put a dozen callows, newly hatched from this stock, into the Bi nest, where the residents were much engrossed with their own pupae. Two or three of these introduced callows were killed and dismembered, while all the rest were accepted into close companionship by the resident Stenammas. The Cremastogasters were permitted to walk over or rest upon the pupa-pile, they were gently licked, and none was harmed during the ensuing twelve days. I then removed them to prepare the nest for another experiment. But while the Cremastogasters were still in the Bi nest, on July 25, I introduced a newly hatched Formica lasiodes, in order to see whether these Stenammas would accept a callow of un- known odor. This visitor was immediately killed and carried to the rubbish-pile. Further evidence that the Cremastogasters were offspring of the queen that laid the eggs from which the 63 Cremastogasters issued, was obtained by putting some of them into K nest with an H colony queen that had never before met any Cremastogaster workers of any colony, having spent her whole life since she was hatched in August, 1903, with Stenamma workers. This queen and the Stenamma workers with her, all accepted the H colony Cremastogaster workers hatched in the latter part of July, 1904, and continued to closely affiliate with them. They were doubt- less of the same odor as was the queen in this K nest, being progeny of the same queen in different years, the queen in K nest still retaining her queen-mother's odor, while the worker- callows bore the queen-mother's ordor as yet unchanged by ageing. Stenamma fulvnni in my artificial nests, when they had no young of their own, have many times permitted Cremastogaster pupae to hatch in their nest and to live there among them. But POWER OF RECOGNITION AMONG ANTS. 241 when engaged in the care of their own young, they have, unless previously acquainted with Cremastogasters, always killed the Cremastogasters as soon as the latter hatched and began to move about. It appears that the Stenammas of nest Bi recognized an old acquaintance in the young Cremastogasters of the H colony, and resumed toward them their accustomed behavior. In other words the Stcnaiiunas recognized an odor after an interval of ten months in ivhich that odor Jiad not been encountered. FIG. 4. Cremastogaster litieolata ; magnified. Actual length 5 millimeters. Having removed all Cremastogasters from nest Bi on August 4, there followed an interval of twelve days in which they had met no Cremastogasters. Then on August 16, 1904, I introduced to their nest one of their old comrades in nest B, a Cremastogaster from whom they were separated on September 24, 1903. This visitor was killed by them within a few hours. In less than eleven months she had attained an odor unknown to them. Nest B2. - -There were in April, 1904, so few of the Lasius in this nest that I gave their power of recognition no test within their own nest. All of them were used till their extermination, in the experiments in nests Bi and 63. 242 A. M. FIELDE. Nest -Bj. - - On April 5, 1904, over six months after the segre- gation of the Cremastogasters, there were about fifty workers in their nest, 63. I then introduced two Lasius, with whom they had lived in amity some forty days, and from whom they had been separated over six months. The residents attacked the visitors from nest B2, and would have slain them had I not rescued them. There was no room for doubt concerning the absence of recognition among the Cremastogasters of the present odor of the Lasius, ants presenting an odor so strong that a single one of them is very impressive to human nostrils. I was unable at that time to introduce to these Cremastogasters any Lasius latipes younger than were those with whom they had formerly associated. But on August 12, 1904, when I introduced a newly-hatched Lasius latipes of the X colony, they immediately killed it. On August 20, 1904, I introduced a young Lasius from the wild nest of the F colony, a sister of the one put on the 1 8th into nest Bi. This ant was amiably received. The Cremastogasters rec- ognized an odor from which they liad been eleven months sep- arated. On the same day, August 20, 1904, I also introduced an adult Lasins, of unknown age, but also of the F colony, one of the nurses of the newly-hatched ant already accepted. This adult Lasins was killed during the ensuing night. On April 8, 1904, I introduced to these Cremastogasters, in nest 63, two of the Stenammas with whom they had pleasantly lived for forty days, in the preceeding autumn, and from whom they had been separated more than six months. One of the visitors behaved as if in an alien nest, showing fear and attempting to escape. The other fought with a resident. Absence of recogni- tion on either side indicted such change of odor during the period of separation as to render these ants unacquainted with one an- other. Early in July, 1904, I was able to introduce to this nest several newly-hatched Stenammas from the wild nest of the C colony, kindred of the ants in nest Bi. There were many queens in that wild nest. That the callows might not bear the odor of older ants, I segregated some pup?e, and offered the callows newly hatched therefrom to the ants in nest 63, but these cal- lows were all killed and dismembered within a few days after POWER OF RECOGNITION AMONG ANTS. 243 their introduction. They were doubtless the progeny of other queens than those which produced the early acquaintances of the Cremastogasters in nest 63. HYPOTHESIS. From such data as is here presented, correlated with records of past years, it is possible to diagramatically represent the proba- bilities that an ant, isolated from the pupa-stage, would encounter a known odor at a first meeting with another ant of her colony. If the odor of a queen be unchanging, if she impart odor to all her eggs, if that odor be perceptible in the inert young, and if from the beginning of the active life of the worker there be a progressive change in the inherited odor borne by her, then from each summer's deposit of the queen's eggs there would be in the following summer more than one odor among the workers, because all the eggs of a queen are not hatched during the sum- mer in which they are deposited. We may suppose a young, fertilized queen, the founder of a colony, to deposit eggs in her isolated cell in July, and to have reared her first small brood in not less than sixty days.1 The eggs laid by her in the latter part of summer or early autumn would reach the larval stage in late autumn, and in that stage would be carried over to the next June, to hatch as ants in summer. While this second brood was developing, the first brood would have advanced to the odor of ants many months old. Their odor would then be unknown to an ant newly hatched from the queen-mother's egg and having the odor of its own body as its only criterion of ant-odor. Suc- ceeding years would bring similar conditions. In Sir John Lubbock's nests, one ant-queen lived to her four- teenth, and another to her fifteenth year ; but the purpose of the diagram is reached by the supposition that the queen lives ten years. 1 For time of incubation of eggs, larval period and pupa-stage in Slenannna ful- viun, see " A Study of an Ant," A. M. Fielde, Proceedings of the Academy of Sciences of Philadelphia, September, 1901, p. 430. The time of incubation appears to be twenty days for all ants that I have observed ; the larval period may be extended to at least one hundred and forty days in a high temperature, and probably to a much longer time in cold weather ; and the pupa-stage occupies about twenty days. I have known the larval period to be passed in twenty days. 244 A. M. FIELDE. c JO 7j D rt S_ cu T3 cu ^J ^J . 3 j2 cr 1 1 £ •'-' ;_ o 5" TD CU en O en ^J CJ c CU CU cr en o en bJD cu r- 4-> o en •*x 4 H • •** *^5 ^ § -t-i 7j : cu .___ ^ ^ ^ I 'tx, 4 T*. - H bu ;i 4 **•*. H b 4 . S^ - 4 b. ;i - - 4 '*x H * b 4 v 4 b 3 "*; - ; rt a, 'C f— i •t-i o cu 4-1 rt hfl cu So cu en cu _X 'en CU (-1 cu 0 0 en rt 0 O "rt 4—> CU en 3 rt cu cu b/D u" c o - "2 CO O cu SM c 4 o X cu CU CU cu Is 4-1 V-i f"1 CU 'c bJ3 'in r- en 1 H 1 '" '• , '• : -^ CU -f> ^-*, ** ^2 O ; ; ; > -c _ ^ 4 ->. — ^j en 4-J cr1 4-J a, -r 03 ^ c/) (U ""^ «^< o cu z rrt <-> 3 —' OJj hn J^ J^ CU ^ > CU 1 n "o t . *^~ <±j N! ^ c .^^.4-4 - V) S ^. ^ "--. V. I « ; **• cu rt cu J-J ]_rt ^ 1 1 * r-N , j— * Cu ,0 H3 ^ o •£ n. cu <- c -O : : 0 *•«-•• •§ ^ ^ ^ rt C 0 cu • — 8*1 en .0 | r- cu : ; en 4-J c X F c <-> c rt X rt « fc en « 4- : C rt T3 •a cu cu c •-- So c n cu "^ : cu > Jz T3 VH- 4-1 rt rt (1) -*-1 : cu AH 4-J c 4-1 CU O 4-* 2 a £ cu 4-1 ~ M e • D . : : • ^ o f—| rt en flj C •— Q H H cu o •»-> J3 ... H1 ,_, H- 1 1— 1 ^ HH t— < ^ 1— 1 1- • > i , H K. H ^ • t H » rt S &< CU en 1 POWER OF RECOGNITION AMONG ANTS. 245 The conditions of odor in her colony during her lifetime are crudely and only approximately represented by the diagram, but the representation conduces to an understanding of certain phe- nomena observed by me in the C colony, which I have now studied four years, in its natural nest, and in my artificial ones. When queens, instead of flying away to found new colonies, remain in the colony where they hatched and increase its popu- lation by their progeny, there is opportunity for all the members of that colony to receive a liberal education of the chemical sense. Every ant acts on individual experience, and if its experience be narrow it will quarrel with many, while acquaintance with a great number of ant-odors will cause it to live peaceably with ants of diverse lineage, provided the odors characterizing such lineage and age environ it at its hatching. If some of the workers were separated from their colony in their youth, and kept segre- gated several years, the sequestered ants could amicably meet younger ants from their old home nest only by an act of memory, or a power of recognition spanning the interval of their separation from their colony. Precisely this condition has been created by me with ants of the C colony. The C colony is a great community of Stcnaunna fulvum that lives under stones scattered at considerable intervals over an area ninety yards in diameter, along a lane and pasture. I have been unable to find any other colony of Stenammas in its vicin- age. Many of its young queens appear to mate with their kin and remain in the colony. I have found as many as fourteen dealated queens in a single shovelful of its nest-earth. The queens and workers from its extreme limits always affiliate unhesitatingly on meeting within its domain. On August 22, 1,901, I took from under a central stone of this colony queens, males and workers, and divided them into two sections, each of which was kept segregated in a Fielde nest for two years. No young was permitted to hatch in either section, and when I united the two sections in August, 1903, they affili- ated instantly, and also affiliated less perfectly with queens and workers freshly brought from the wild nest.1 I kept the ants of 1 Detailed account of these meetings may be found in " Cause of Feud between Ants of the Same Species," already referred to, p. 328. 246 A. M. FIELDE. the two united sections, again without young, another year, until August, 1904, when I introduced to their nest marked queens and workers from their old wild nest. Of the two queens intro- duced one was at once received into full fellowship. The other was simultaneously licked and dragged by different worker- residents, but was accepted at the pupa-pile within a few hours. The workers introduced had kindly reception. Of one hundred and fourteen callows introduced one by one, or a few at a time, an interval of repose being given between the removal of one and the introduction of its successor, only two were attacked. While it is true that all the accepted callows of the summer of 1904 might have had the odor of the resident queen, it appears more probable that most of them bore odors that were recognised by tlie resident ivorkers after the lapse of three years. These ants had certainly not within their three years of segre- gation met a two-year-old or a one-year-old ant of their colony, from outside their artificial nest. In August, 1902, I segregated a queen and workers all hatched during that month, and in August, 1903, I likewise segregated a newly hatched and similar group. All these ants were of the C colony, and no young was permitted to hatch in any of these groups, each in its artificial nest. In August, 1904, I therefore had command of ants of the C colony, one group in the C nest, consisting of ants brought in from the wild nest on June 24, 1904; one group Ci, just one year old, one group C2, just two years old, and one group €3, three years old or more. All had been acquainted with their seniors before segregation. In August, 1904, the three-year-old ants received amiably, within their nest, ten of the two-year-old ants, and ten of the one-year-old ants, indicating a perfect recog- nition of odors no longer represented in their own nest, and from which they had been long separated. On the other hand, when I introduced the three-year-old ants, queens or workers, into a nest populated with hundreds of workers taken in June, 1904, from the wild nest or hatched within that nest during the present summer, the three-year-olds were always fiercely attacked. Tlicy had become an alien colony to the younger generations of their former wild nest. The C col- ony has been much harried in its natural domain, by the rebuilding POWER OF RECOGNITION AMONG ANTS. 247 of stone fences, by repairing of the lane- road, and by my own dep- redations. Ants so old as my three-year-olds have apparently become almost unknown in the wild colony. The older the colony, the fewer would be the chances that any ant, segregated as a pupa and always kept in isolation, would find its own odor in any ant taken at random from the nest from which said pupa had been taken. In August, 1904, I thus iso- lated pupae and the ants hatched therefrom, and from among many experiments made with them, I record the following as typical : I isolated a pupa from the C colony wild nest, and when the ant that hatched from it was eleven days old, never having smelled any ant-odor other than that of its own body, I intro- duced one by one to its Petri cell, where it was engrossed in the care of introduced larvae, all the three-year-old ants to the num- ber of twelve, all the two-year-old ants to the number of seven- teen, all the one-year-old ants to the number of forty-two, and seven C nest ants of precisely its own age, so that the number of visitors arriving singly and at intervals amounted to eighty. A period of repose was provided after the removal of one visitor before another was introduced. Every one of these visitors was at first meeting violently attacked by this callow, dragged away from the larvae, and in some cases taken outside the Petri cell, if I lifted its cover. Sometimes a visitor violently attacked the resident callow, and she had to be rescued by me from sudden death. In other like series of experiments with isolated callows the resident callow sometimes found a congenial odor in a visitor and willingly permitted her to share in the care of the larvae. Callows having their origin in different parts of the C colony area behaved alike, and it appears improbable that all of those tested could have been the product of eggs, larvae or pupae brought in by raids on another colony, especially when the greatness of the C colony is considered and its location studied. Other experiments with callows from this colony gave support to my hypotheses. On July 1 1, 1904, there hatched in nest €3, a pupa previously introduced by me from the C nest. I left it seven days with the three-year-old ants, and then transferred it to a Petri cell, giving it a few larvae to care for. This isolated callow knew only its own odor, that of worker ants at least three 248 A. M. FIELDE. years old, and that of a queen. I then, a day or two later, in- troduced into its cell, one by one, all the one-year-old ants in nest Ci, removing each visitor as soon as the action of the resi- dent was decisive, and allowing a period of repose before another visitor was introduced. She affiliated with the first, third, fourth, sixth, seventh, eighth, tenth and eleventh, and attacked the sec- ond, fifth and ninth. I likewise introduced two-year-old workers from nest C2. She affiliated with the first, third, fourth, fifth, seventh, ninth and twelfth, and attacked the second, sixth, eighth, tenth and eleventh. She did not attack the C2 queen, but the queen so persistently avoided her as to make the test undecisive. In the same manner I tested a callow, hatched on July II, in nest C2, where all the ants beside herself were two years old. When this callow had spent seven days in the C2 nest, and one day in isolation with larvae to care for, I introduced into her cell workers from the C3 nest, where all the ants were three or more years old. She affiliated with the second, fourth, fifth, eighth, eleventh and twelfth, and attacked the first, third, sixth, seventh, ninth and tenth visitors. I then introduced one-year-old ants from nest Ci. She affiliated with the first, second, third, sev- enth, eighth, ninth, tenth and twelfth, and attacked the fourth, fifth, sixth and eleventh visitors. I intended to likewise test a callow reared in nest Ci, and I expected to find that this callow would reject all three-year-old ants ; but I unfortunately dropped an unmarked three-year-old ant into nest Ci, and thereby so vitiated the nest as to make it useless for this experiment. A comparison of all the tests made gave a consensus of testi- mony that the C3 ants, the Stenanunas, recognized and adapted their beliavior to ant-odors that they had not encountered during three years. As the workers are not supposed to reproduce colonies, and as the queens are not supposed to change their own odor, how then would queens of diverse odor originate through the ageing of the workers ? In 1901 I segregated winged queens of the C colony,1 putting 1 " Notes on an Ant," previously referred to, p. 605. POWER OF RECOGNITION AMONG ANTS. 249 some of them into nests with kings of their own colony and others of them into nests with kings of alien colonies, and believed that I ascertained that the progeny of sister queens affiliated, regard- less of paternal influence in the egg from which the ants issued. I then supposed that queen-ants captured before swarming must be virgin queens. I now know that virgin queen-ants often mate with the males within the maternal domicile, and that neither the possession of wings nor an early capture guarantee the virginity of a queen. Only by sequestration of the queen from her pupa-stage can her virginity be secured. I have had Lasius latipes queens drop their wings and lay eggs soon after being brought from the wild nest from which they had not yet swarmed. In my artificial nests, I have observed the persistent avoidance, by queens, of kings of alien colonies and their mani- fest preference for kings of their own colony. Mating in captiv- ity, in artificial nests, is not uncommon, and it must be frequent in the wild nests before the swarming. We know that the eggs of workers often produce sturdy males, and it appears probable that such males impart to the fertilized eggs of the queen something of the odor attained by the worker-mother at the time when the egg, producing the male, was deposited. This would differentiate odors in the progeny of sister queens, and cumulative differentiation would account for ultimate differences in the odor of queens of the same species and variety. When queens remain in the mother-nest after mating, and there rear their broods, that colony must become one of much mixed odors, as is the C colony described in this paper. Fertilized queens, departing from the maternal nest, would found colonies whose issuing queens would have an odor depending on the age of the workers who were mothers of kings hatched in the season in which their founder-queens mated. Besides discerning the aura of the nest and other local scents and the track laid down by its feet,1 an ant perceives in other ants the incurred or incidental odor which appears with conditions and disappears in course of time ; the inherited odor derived from the queen-mother, apparent in the eggs, larvae, pupse and newly 1 " Further Study of an Ant," A. M. Fielde, Proceedings of the Academy of Nat - ural Sciences of Philadelphia, ""November, 1901, p. 521. 250 A. M. FIELDE. hatched young, and probably strengthening as size increases through the three inert stages of development ; the progressive odor, that distinguishes the worker and changes or intensifies with her advancing age ; and the specific odor which pertains to the species or tribe. Adding to these perceptions the power of recognizing familiar odors after a lapse of months or years, the ant appears to be well equipped for life in her world.1 If she has not reason and imagination, she has at least the ground on which to exercise both, cognoscence of past experiences. 1 The organ discerning the nest-aura and probably other local odors lies in the final joint of the antenna, and such odors are discerned through the air; the progres- sive odor or the incurred odor is discerned by contact, through the penultimate joint ; the scent of the track, by the antepenultimate joint, through the air ; the odor of the inert young, and probably that of the queen also, by contact, through the two joints above or proximal to those last mentioned ; while the next above these by contact also discerns the specific odor. It is probable that the size of the queen determines the amount of odor diffused by her. The amount of odor diffused by or discerned in the larva; and pupce may be the determining factor in the assorting of the young ac- cording to size, as is common among ants. The results of many experiments whereby the function of many joints in the antennee were determined by me in 1901-1903 in Stcnanima fuh'iiin are recorded in " Further Study of an Ant " and " Cause of P'eud among Ants of the same Species," above referred to. The joints in the antennae vary in different species, from four to thirteen. NOTES ON A HITHERTO UNDESCRIBED HYDROID FROM LONG ISLAND SOUND. CHAS. W. HARGITT.1 Early during the present year I received from Dr. Henry R. Linville, of New York City, a small collection of hydroids for examination, among which was found what clearly appears to be a new species. The specimens were collected by Dr. Linville during the previous August near East Marion, Long Island. In a letter to the writer he describes the habitat as follows : " The species was found growing on rocks and piles under ' Milldam Bridge,' west of Shelter Island. The bridge spans a narrow creek which connects the Bay with a shallow salt-water pond." In size and general features the hydroid resembles very much Syncoryne nrirabilis, and was at first thought to be that species. A more critical examination showed this to be more than doubt- ful, as the number of tentacles, character of the hydranth, and above all the character of gonophores, all indicated otherwise, and a review of the accessible literature failed to afford any clue to the identity of the species, though it seems clearly a member of the genus above named as will be seen from the sketch. The medusa-buds are relatively large and grow in clusters of from two to four upon the body of the hydranths usually close among the tentacles. I regret that no free medusae were ob- tained and that only an incomplete description of this organism can be made. As will be observed from the figure the medusae present the somewhat unusual condition of having at this stage two of the tentacles quite well developed, though short and thick, at opposite positions on the margin of the umbrella. The inter- mediate pair can hardly be distinguished, being mere buds, and it is impossible to say whether they later develop or not ; indeed, it remains doubtful as to the later phases of development of almost all the medusoid organs, though there can hardly be serious doubt as to the ultimate freedom of the medusae, as there were 1 Contributions from the Zoological Laboratory, Syracuse University. 25I 252 CHAS. W. HARGITT. no signs of the development of gonads, a condition almost certain to have been found in case the gonophores remain sessile. The following diagnostic characters will serve to distinguish the species, at least so far as the hydroid phase is concerned, for which the name Syncoryne linvillei is proposed : Trophosome. — Colony growing in tufts, sparingly branched, to a height of 15 to 30 mm., and with the same general aspects as characterize 6". mirabilis Ag. Hydranths vasiform, with cone- '.I .V* " Portion of colony of Syncoryne linvillei about three times natural size. shaped proboscis ; tentacles definitely capitate, from I 5 to 30 in number, and variously distributed over the proximal third of the hydranth body. Perisarc much as in •$". mirabilis, plain, or with only the slightest trace of any annulation, ending somewhat abruptly below the base of hydranths, proximal or basal portion dense or brownish in color. Hydrorhiza more or less reticu- lated, forming a loose network over the substratum. In all the type specimens both hydrorhiza and hydrocaulus were enmeshed in a dense sponge-like mass, though its exact nature is doubtful. Gonosonic. — Medusa buds borne on body of hydranth, usually in small clusters among the bases of the tentacles and supported by a single peduncle, the terminal specimen always maturing AN UNDESCRIBED HYDROID. 253 first. In the type specimens no free medusae were present, though there were numerous buds approaching full development. The umbrella is distinctly bell-shaped, with four radial canals ; tenta- cles unequally developed, two on opposite sides large and club- shaped, the intermediate pair small and bud-like. No ocelli distinguishable. Colors. — Formalin material pale yellowish, hydranths some- what brownish ; gonophores reddish brown or pink. Habitat. — Growing in tufts upon small rock fragments, bits of bark or on shelly concretions ; the bases of stems tangled or enmeshed in sponge-like masses, the tufts and spicules of which give to the colonies a rough and bristly aspect. A SYNOPSIS OF CHARACTERS OF SOME FISHES BELONGING TO THE ORDER HAPLOML EDWIN CHAPIN STARKS. The orders Haplomi and Iniomi are distinguished from the Nemotognathi (the cat fishes) and the Plectospondyli (the min- nows and suckers) by having normal anterior vertebrae ; from the Isospondyli (the herring and trout-like fishes) by the absence of a mesocoracoid ; from the several orders of eel-like fishes, by the comparatively few vertebras, the presence of ventral fins, and the unrestricted gill openings ; from several peculiar small orders by apparently sufficient characters, but which for present pur- poses need not be considered. Thus these orders seem to be made up of soft-rayed fishes, which are thrown together because they lack the distinguishing characters of other groups, rather than because they have such characters of their own. This condition naturally would tend to bring together forms that are not very closely related. The families of the Haplomi have either widely diverged from each other or are not of the same line of descent. The order is not held together by any important character, though some very peculiar characters may be used to rather widely separate three groups. I have not studied the Iniomi,1 but from the definition given be- low I find it impossible to separate the Haplomi from it. The free condition of the scapular arch has been the most useful character in separating these two orders, but as this condition is found in the families Esocidae and Umbridae, it cannot be used unless we transfer these families from the latter order to the former. Ob- viously this is not advisable as Umbra, especially, is certainly nearer the family Pceciliidae than it is to any family in the order Iniomi. The Iniomi was established by Dr. Gill 2 to include those forms 1 Except to determine that the condition of the attachment of the scapular arch to the cranium is the same in Synodus as it is in Esox and Umbra. 2 Proceedings of the U. S. National Museum, 1884, p. 350. 254 CHARACTERS OF THE ORDER HAPLOMI. 255 in which the scapular arches are " connected with and impinge on the occiput behind and on each other, and are otherwise free from the cranium." In his original definition he calls Iniomi a group, and later in his " Families and Subfamilies of Fishes " l he treats it as a superfamily. Jordan and Evermann in " Fishes of North and Middle America" raise it to ordinal rank, and point out that the absence of the mesocoracoid is the most im- portant character by which the Iniomi are separated from the Isospondyli. A readjustment of the orders Haplomi and Iniomi will be necessary, and either a separation based on characters unstudied and unconsidered, or a merging of one into the other will result. This will necessitate a study of many more forms than are here considered. The knowledge of the following characters I obtained in study- ing the relationships of Da//ia pectoralis. As I am not at this time ready to take up an extensive investigation of the orders Haplomi and Iniomi it seems better, while it is fresh in my mind, to publish this material in the following form as a working basis, rather than to hold it indefinitely in abeyance. Order HAPLOMI. Soft rayed fishes without a mesocoracoid and with the anterior vertebrae normal. Parietals separated by the supraoccipital. Alisphenoids not meeting in a median line in front of brain case. Exoccipitals separated by basioccipital. Postclavicle composed of a single element. Actinosts four. Opercular bones all pres- ent. Ventral fins abdominal, each attached to a simple flat pel- vic bone. Pectoral fins placed low. Dorsal fin placed more or less posteriorly. Air bladder with a distinct pneumatic duct. Superfamily Esocoidea. Ethmoid represented by paired elongate dermal bones ; meta- pterygoid present, forming the upper margin of the cheek bones above the symplectic ; symplectic normal ; palatine and pterygoid both present and normal ; the former with teeth ; primaxillaries 1 Sixth Memoir, National Academy of Sciences, Vol. VI. 256 EDWIN CHAPIN STARKS. without backward extending processes on median line of snout ; maxillaries forming lateral margin of mouth ; two toothed and two toothless superior pharyngeals present on each side ; post- temporal connected to the epiotic by a ligament ; l nearly meet- ing its opposite fellow just behind the occipital region ; supra- clavicle normal in size ; post-clavicle a single ray of bone ; hy- percoracoid foramen only notching the lower edge of hyper- coracoid ; 2 actinosts rather elongate, all ending against hypo- coracoid or in cartilage opposite that bone ; pelvic bones without an overlapping spur ; upper end of shoulder girdle attached by ligament to first vertebra ; parapophyses normally developed only posteriorly ; the haemal spines which support caudal at- tached by suture to their centra ; posterior vertebras with a very decided upward bend ;3 vent normal in position. Family ESOCIDyE. Characters as indicated by Esox reticulatus. Cranium long and slender, with a narrow projecting rostrum ; interorbital septum single ; myodome short not opening pos- teriorly ; the prootic shelf above not nearly reaching to mouth of anterior opening to brain case ; parietals extending over a deep cavern to pterotic ; suborbital present ; a lateral wing extending upward from parasphenoid barely reaching to alisphenoid ; sep- tomaxillaries 4 present between ethmoid and vomer ; vomer large, reaching anterior to parasphenoid ; nasals present ; basisphenoid extending upward from parasphenoid, Y-shaped and unattached above ; opisthotic absent ; palatine joined to prefrontal by a slen- der process ; anteriorly without a process hooking over maxillary ; a wide open space between the hyomandibular and the preoper- cle opposite the middle of the latter ; premaxillaries widely separated from each other by the rostrum ; maxillary with a 1 In a specimen of Esox 40 cm. in length the ligament is 4 mm. long. In Umbra it is comparatively as long. 2 See exception under description of Esocidse. 3 In Esox the narrow hypural plate is placed at an angle of about 45 degrees, and the haemal spine from the preceding vertebra runs horizontally back to the base of the middle caudal rays. * E. P. Allis, Jr., discusses the septomaxillaries and the paired ethmoid of Esox, in a paper entitled " On Certain of the Bones of the Cheek and Snout of Amia calva," Jour, of Morph., Vol. XIV., No. 3, 1898. CHARACTERS OF THE ORDER HAPLOMI. 257 large supplemental bone ; a large preorbital and four suborbitals present ; posttemporal a wide plate bent longitudinally at a right angle ; its posterior edge concave, not typically forked ; its lower limb connected by ligament to exoccipital ; a broad ribbon of connective tissue connecting the inner surface of the supraclavi- cle with the first vertebra ; l hypercoracoid simply notched by its foramen, which opens against cartilage between the coracoid ele- ments ; - lower pharyngeals elongate, not in contact at their inner edges ; an ankylosed epipleural on each of the two first vertebrae ; vertebrae number 36 -f 17 = 53 ; all abdominal neural processes attached by suture ; parapophyses developed as processes only posteriorly.3 Family UMBRID^. Characters as indicated by Umbra lima. Cranium short and normal in shape ; interorbital septum double, with the sides widely separated 'throughout ; no myo- dome ; parietal separated from pterotic by an area of cartilage ; supraorbital absent ; parasphenoid not sending a lateral wing up to alisphenoid ; vomer very small, scarcely larger than its patch of three or four teeth ; loosely attached to the surface of the parasphenoid, and nowhere reaching • to the edge of that bone ; no septomaxillaries or nasals present ; a preorbital but no sub- orbitals present ; l basisphenoid absent ; a very small loosely attached opisthotic present in usual position ; 2 palatine with a process hooking over the maxillary ; no open space between 1 The connecting band does not appear to be of the same compact ligamentous nature as in Umbra, but is composed or more loosely connected fibers. It does not join the tip of a lateral process, but is attached directly to the side of the vertebra. 2 In a specimen examined of Eso x htciits, about the same size as the above (40 cm. in length) the foramen is entirely contained by the hypercoracoid. 3 The parapophyses are represented by small ossicles attached to the centra by suture, and set in sockets so that the ribs appear to be attached directly to the verte- brae until careful examination is made. They first appear on the third vertebra, but are little developed in front of the ninth or tenth. Posterior to the twenty-eighth they develop outward and downward as small processes in front of the base of each rib. Posteriorly they grow longer and the last two or three are ankylosed to the centra. 4 The sensory tunnel which is usually continued from the frontals by the nasals, opens to the exterior as soon as it leaves the former bone and extends no further forward. 5 It is easily pulled away with the posttemporal, remaining attached to the lower ligament of that bone. 258 EDWIN CHAPIN STARKS. the preopercle and the hyomandibular ; premaxillaries meeting at the median line on snout as usual ; no supplemental maxillary bone ; lower limb of posttemporal represented by a short tri- angular process from which a ligament runs to the opisthotic ; l a strong ligament extends from a lateral process on the first vertebra to the inner surface of the supraclavicle ; 2 the hyper- coracoid foramen is a notch in the hypercoracoid open against the hypocoracoid ; lower pharyngeals in contact along their inner edges; vertebrae number 21 -f 14 = 35 ; epipleurals of first ver- tebrae not ankylosed ; all neural processes ankylosed to vertebral centra ; no parapophyses developed except on last four abdom- inal vertebrae ; 3 ribs fitting into depressions with slightly raised edges on the body of the vertebrae ; the first vertebra carries no rib. Superfamily Pceciloidea. Ethmoid represented by a single nearly circular scale of bone ; metapterygoid absent ; the symplectic forming the upper margin of the cheek bones behind the mesopterygoid ; symplectic very large, sending a long slender process behind the quadrate nearly to the condyle of the mandible ; palatine-pterygoid arch reduced to a single element, and without teeth ; 4 premaxillaries each with a small backward extending process on median line of snout ; premaxillaries forming lateral margin of mouth ; superior pharyngeals either three toothed patches, or ankylosed into a single one on each side ; the usual toothless one of first arch either present or absent in the former condition, but that of second arch never toothless ; posttemporal directly attached to epiotic without the intervention of a ligament ; supraclavicle a very small scale of bone, scarcely sufficient to separate the posttem- 1 Four specimens have been examined and no indication found towards ossifica- tion of this ligament as in Dallia. 2 This ligament is probably the homolog of the ray of bone in Dallia pectoralis that is in the same position. 3 The process on the first vertebra to which the supraclavicle ligament is attached may possibly be considered a parapophysis also. 4 The posterior end of the palatine-pterygoid element borders the upper anterior edge of the quadrate, and is braced above by the mesopterygoid, as is usual for the pterygoid ; the anterior end is attached to the prefrontal and maxillary, as is usual for the palatine ; making it appear probable that these two elements have ankylosed. CHARACTERS OF THE ORDER HAPLOMI. 259 poral from the clavicle ; postclavicle a simple ovate scale of bone ; hypercoracoid foramen entirely enclosed by the hyper- coracoid ; actinosts small, deeper than long, no opening between them, two on each the hypercorcoid and the hypocoracoid ; pel- vic bones each with a spur extending inward, one of which over- lies the other ; upper end of shoulder girdle joined to bassioc- cipital by a long ligament ; l large projecting parapopyhyses present on all abdominal vertebrae ; the caudal supporting haemal spines ankylosed to their centra ; the posterior vertebrae not tilted up ; vent normal in position. FAMILY PCECILHDyE. Subfamily FUNDULIN.E. Characters as indicated by Fiindnlus siinilis? Interorbital septum double, its sides widely separated through- out ; no myodome ; parasphenoid sending a lateral process up to alisphenoid ; supraoccipital expanded latterly on top in a thin horizontal wing ; epiotic developed backward in a long thin process as in the genus Mugil ; occipital condyle partly formed by exoccipitals ; prefrontals not meeting at the median line ; basi- sphenoid and opisthotic absent ; mandible normal ; premaxillaries normally curved ; posttemporal a simple bone with no indication of a lower fork, normally attached to epiotic ; three superior pharyngeals.on each side joined (not ankylosed) to form an ovate plate, covered with molar-like teeth near middle of bone, which change to small blunt conical teeth near outer edges ; lower pharyngeals joined to each other by deeply dentate and inter- locked suture,3 together forming a triangular bone with concave 1 The ligament runs from the middle of the outer edge of the basioccipital to the upper end of the clavicle in Pcedlia and Cyprinodon, equally to the inner surface of the supraclavicle and the posttemporal in P'unduliis. 2 Gainbusia (as exhibited in species nobilis) agrees with Fnndulus in character of superior pharyngeals and unforked posttemporal but in the condition of the occipital condyle resembles Pixcilia. 3 The lower pharyngeals of the following species of /<«;/3 > CJ S 1 o J tS CS c/l 2 CJ 4) T5 S2 jj 3 3 C Pi C^ ^ 1 < TJ 8 0 "o a ? c^ & H ^ CJ ^ CJ u .2* 01 J^^ • • ^5 rS Pi X) U *^-t 2 *^ rt (Si £ d S S w Z u ^' U HH j=_ s s a a a a S £ s S hjn'rt 5 g Sees s s a S C f-H to r-» w O O "-> t5 6 vo O ^3 i-i M O^ ^^ O HH ON — -5 S £ a a a s s s S a a a s a a a s a a o q o" to o o o o Tl- CO N rf to O I—1 VO i i— i J to •* T}- M M M •<*• •*• **• SB Tj- ^. ^ M O vO OO Tf to o 0 ^ as? ^h fi ro "•*• •^- TJ- ro CO to ro ro 3* B C — ; CU 0 W W !>• 0 1^ OO M t^ N O 00 ^J- to ^ 1* vO to t^. vO vD to vO vo to to VO (^ O t U, 01 X! M ON d — N M M N i-i i-i i-i i-l 1-1 HH CV1 tN _, „ M M HH — c — C ctf 1 1 1 i i 1 1 II II II II II 1 1 1 1 1 1 CU u PiJ PiJ X^ Pi J' Pi J Pi J fri^ ci^ Pi^: ftiJ *J in ro ro tN ro CO ro ro r^^ roro rom ro^O N r^ ro ro N (S rrj r^-j ro ro S 1 1 1 1 1 1 1 II II II II II II 1 1 1 1 1 1 "5 O ciJ Pi J c^J CiJ tiJ &iJ C^J Pi (-3 Pi i-J PiJ PiJ 6i^ « -2 OO OO ON 00 OO 00 OO oooo oooo oooo oooo 00 00 OO OO 00 00 ON ON ON ON ^J •— _n Pi J Pi J *J piJ &iJ P. vo vo VO vO vo vo OvO vovo wj Pi J tiJ Pi —5 piJ p ro b 1 ^ 1 Ss s 5s it * ? rt ON i ? > /"N 00 .— QJ ^ (U "^ ^ ^ CU W ^ M cu - cu ON S ^ ° CL> CM c S o _o .2 JS rT JS pf S ^ ^ ys cu - 2 ^ 2 - (->. ^ rC 3 I" n "" W O o |!1 |2 J- *? ^n ^ ^f C3 M Cj £" X-, y5 ^ jf £ ^ ^ 5j" ^ &1 ?J "u „ 5- ^ ^ "•i -•3 - 3- "d "a""— ' «>—> Ci 03 > > o "^ O ^ ^J o "^ ^ o ^ O 5 "^ 0 B . H C3 'u n .y >' o >' o >' o _> 0 _> 0 > 0 _> 0 > p" si P4 N - — M 3 c ' r^ BBC 'c ^ '5 "a '5 ^ '5 P-l ^> £ s i- D £> D D D t> P i-i ri CO -f to vo' r^ ad ON O ^ 2 M BUTLER'S GARTER SNAKE. 297 So r* Bj . . O HH !*-> >~, ^-> t/1 t/1 HH d) § OH O !_^ t/3 w C/1 i™ &° CJ u HH o O 0 3 3 "o ^ ^ "^ <^ CJ O -£ p ^D ^ ^ Cj ^* Cvl ^" CM c r c r 1-1 "^ "73 < "o s ^ tfl >-, •^ o ,c 3 rt "^ .bi O ^3 •^ cj c jg g§ 'c S "c S 'c S !&°S ^s c rt 5 5 -^- i-I "^ ^ IT. IT. cfi PS PQ d d 6 d d d d £ fc ^ ^ £ "3 "ci cs "3 'a "n ts CJ • CJ CJ CJ CJ CJ Efl - - - » ^ > C OJ a o 0 CJ CJ O CJ CJ 'o S • i § § s s S 173 O O O CO "3 HH <>_ 4- o <-> O HH 3 CM <~ CO 0 M »H- ro vO . ^ vO vo . ^ .-' & > w > "^ > HH > t— ' > *"* •? r° • -. ro Q C c j^ C J2 D 5 D D D D !3 ^ un vd ^ CO ON d i— i HH HH HH HH HH CM 298 A. G. RUTHVEX. a view which is also held by Professor Cope ('92, p. 660) and Dr. Stejneger ('93, p. 214). This view is perhaps supported by the great variability of leptoccplmla, but leptocephala occurs in a dis- tinct region while butleri occurs in the same region, in the same habitat and in almost equal abundance with typical forms of E. s. sirtalis ; and at the same time its distinctive characters are so constant, that no variations have been found that can be regarded as transitional forms between the two species. The average num- ber of gastrosteges in butleri is about 1 39 while in sirtalis sirtalis it is about 151. The temporals are also more constant than appears in the table ; it is true that there is not always a single large tem- poral in the second row, but when there are two as in sirtalis sir- talis, the upper one, as far as we know is always much smaller than the lower one, which lies posterior and inferior to it. The form and coloration are, however, the most characteristic features. The body is more robust, the tail shorter, and the head smaller and more conical than in sirtalis sirtalis. As stated under varia- tion, the color is remarkably constant for a member of this genus and while there is much color variation in the E. sirtalis group, I have never seen a variation that could be confused with butleri. Its striking coloration, robust form, small indistinct head and peculiar movements, make this snake very conspicuous and ren- ders it almost impossible to confuse it with any other form. I have not seen enough specimens to determine its relationships with other groups, but the coloration, position of the lateral stripe, and reduced number of infralabials point strongly toward E. radix which meets it on the west. Desirable as it is, therefore, to cut down the species in this genus, it must, I think, be concluded, from the specimens which have been collected that Entcenia butleri is a good species. BUTLER S GARTER SNAKE. 299 SYNONYMY. 1889. Euteznia butlerii, Cope, Proc. U. S. Nat. Mus., XL, p. 399. 1892. Eitdriiia butlerii, Cope, Proc. U. S. Nat. Mus., XIV., p. 651. 1892. Eutccnia butlerii, Hay, I7th Annual Report of the Ind. Dept. of Geol. and Natural Resources, p. 120. 1893. Tropidonotus ordinatus, var. butleri, Boulenger, Cat. Snakes Brit. Mus., I., p. 212. 1894. Thamnophis bntleri, Stejneger, Proc. U. S. Nat. Mus., XVII., p. 593. 1903. Eulienia bracliystoma, Clark, Proc. Biol. Soc. of Wash., XVI., p. 83. REFERENCES. Brown, A. E. '01 A Review of the Genera and Species of American Snakes, North of Mexico. Proc. of the Acad. of Nat. Sciences of Phila., January, 1901, pp. 10-110. Clark, H. L. '03 The Short Mouthed Snake {Eutainia Brachystoma Cope) in Southern Michi- gan. Proc. Biol. Soc. of Wash., Vol. XVI., pp. 83-87. Cope, E. D. '89 On the Eutzenia of Southeastern Indiana. Proc. U. S. Nat. Mus., Vol. XL, P- 399- 'oo Crocodilians, Lizards, and Snakes of North America. Annual Report U. S. Nat. Mus., 1898, pp. 153-1270. '92 A Critical Review of the Characters and Variations of the Snakes of North America. Proc. U. S. Nat. Mus., Vol. XIV., pp. 589-694. Davis and Rice. '83 List of Batrachia and Reptilia of Illinois. Bulletin of the Chicago Academy of Sciences, I., pp. 25-32. Hay, 0. P. '92 Batrachians and Reptiles of the State of Indiana. I7th Annual Repl. of the Ind. Dept. of Geology and Nat. Resources, pp. 409-602. Reddick, G. '96 Snakes of Turkey Lake. Proceedings of the Ind. Acad. of Science 1895. pp. 261-262. Stejneger, L. '94 Notes on Butler's Garter Snake. Proc. U. S. Nat. Mus., Vol. XVII. , pp. 593-594- '93 Annotated List of the Reptiles and Batrachians Collected by the Death Valley Expedition in 1891, with Descriptions of New Species. North American Fauna, No. 7, Part II., pp. 159-228. TENACITY OF LIFE IN ANTS. ADELE M. F1ELDE. MAIMED ANTS. s Remarkable tenacity of life is sometimes exhibited by ants lacking a portion of the body. A queen, Stcna innia fuk m m piccuin, deprived of the funicles of her antennae, lived fourteen months in one of my artificial nests, where she laid eggs, sought her own food, and received kindly treatment from the resident workers and several unmutilated queens. The head of a Formica fusca subscricca worker under my observation, continued to move its antennae seven hours after decapitation. Ants lacking a leg or two may live several weeks. A Stcn- annna fukntin worker, deprived of its mesothoracic pair of legs, was returned by me to an artificial nest where were hundreds of its former comrades and it safely lived there a month or more, disproving any general assertion that ants destroy maimed members of their colony. Worker ants deprived of the abdomen sometimes run with great speed, continue to care for the young in the nest, fight with aliens of their own or other species, and they may for some days behave as if unconscious of loss. A Stcncunma fiil-cnui queen deprived of her abdomen lived thereafter for fourteen days in one of my artificial nests, and was seen to eat. A Formica subsericea worker lived without her abdomen for five days. M. Charles Janet mentions : an ant that lived nineteen days after decapitation. In experiments made by me, headless ants have continued to walk about for many days. In all these expe- riments aseptic surgery was attempted, the instruments used being carefully sterilized. The Petri cells in which the maimed ants were 1 "Extraif des Compte rendits hebdomadaires des Seance de /' Academic des Sciences," Paris, II juillet, 1898, p. 130. 300 TENACITY OF LIFE IN ANTS. 3OI kept were frequently cleansed with 80 per cent, alcohol, and only distilled water was used in wetting the enclosed sponges. Care was taken to maintain hygienic conditions at all times during the life of the ant. In these experiments a Stenamma fulvum worker lived ten days without her head. A Formica subscricca worker lived fifteen days without her head. Of seven decapitated Cain- ponotus pennsylvanicus workers, three lived five days ; two lived twenty-one days ; one lived thirty days ; and one lived forty- one days. The last two mentioned were the largest, being thir- teen millimeters long. After decapitation, they were kept for four days at a temperature of about 10° C. or 50° F. and after- ward in the natural summer temperature of the laboratory.1 The ant that lived forty-one days after decapitation walked about in the cell until two days before her death, and during the last two days gave evidence of life by a twitching of the legs when I touched her. SUBMERGENCE. While making experiments in June, 1904, with a view to as- certaining how long ants could live under water,2 I came to sus- pect that the death of ants submerged less than seventy-two hours was caused by bacteria rather than by deprivation of oxy- gen. Later in the summer I therefore made further experiments,, merging the ants in distilled water, and keeping them in the dark at a temperature of about 10° C. or 50° F. Under these conditions the ants survived much longer periods of submergence. Of eighteen Stcnannna fii/rniu submerged four days, seven- teen revived and twelve fully recovered. Of fourteen Stenamma fulvum submerged six days, six revived, and one fully recovered. Of twelve Stenamma fnlvnin submerged eight days in sixty- five cubic centimeters of water, seven revived and fully recovered. Of seven Cainponotus pennsylvanicus submerged eight days, four revived, fully recovered, and were returned to their old associates. 1 Nearly all the experiments recorded in this paper were made at the Marine Biological Laboratory at Woods Holl, Mass., during the summer and autumn of 1904. 52 "Observations on Ants in their Relation to Temperature and to Submergence," A. M. Fielde, BIOLOGICAL BULLETIN, Vol. VII., No. 3, August, 1904, p. 170. 3O2 A. M. FIELDE. The ants recovering after submergence were those of large stature among their kind. DWARF ANTS. The ability of the larva to successfully enter the pupal -stage of development at any time after about half its normal size has been attained, helps to assure the persistence of a harried com- munity. Under propitious conditions the larva grows to the size of an adult ant, expels the contents of the alimentary canal, and eats nothing during the five or ten days preceding pupation. But if suddenly deprived of food it may enter the resting stage when half-grown and may ultimately become a perfectly formed dwarf. I sequestered many fat, healthy, half-grown larvae of Stenamma fnhntvi, put them in the care of fewer nurses than could regurgitate food to all of them, and supplied but little nutriment to the segregated group. Many of the larvae soon entered the resting stage and later on became dwarf workers, only three or four millimeters in length, while the length of workers of their species is usually from five or seven millimeters. A corresponding diminution in the stature of a man would take from one to two feet from his height. The dwarf workers were wholly normal in their faculties and activities, and were markedly assiduous in their care of the young. On the other hand, larvae poorly fed may miserably linger, as if waiting for better times. In one case under my observation, an ill fed larva remained such for one hundred and forty days, although constantly in summer temperature, and it then died without pupating. DEPRIVATION OF FOOD. Although ants manifestly suffer and soon die if deprived of water, they can exist for many days without food. In the fol- lowing tests the ants were kept in Petri cells, ten centimeters in diameter, and not more than five ants were enclosed in any one cell. All the cells were kept in darkness or very dim light. At intervals, never exceeding four days, the cells and the enclosed sponges were cleansed with 80 per cent, alcohol, to prevent micro- scopic growths which might furnish nourishment to the ants. The cells contained only the ants used in the experiment, and a TENACITY OF LIFE IN ANTS. 303 bit of sponge saturated with distilled water. Care was taken to ensure good ventilation and all other conditions were made hygienic, that there might be no cause of death other than that of deprivation of food. The starving ants did not manifestly weaken after long abstinence but collapsed suddenly and finally, giving no premonitory evidences of exhaustion. Of thirty Crcuiastogaster lincolata, segregated on July 2, 1904, ten lived so long as ten days and only one lived so long as eigh- teen days without food. Of thirteen Camponotus Jicrculcamts pictus workers, two lived seven days ; two, fourteen days ; one, eighteen days; one, twenty- three days ; two, twenty-four days ; one, twenty-six days ; and one, twenty-nine days. On the tenth day of fasting, I found three of these ants killed and dismembered, with an appearance of having been chewed. In all the cells, this was the only group where the ants attacked their companions in tribulation. As ants sometimes quarrel with their mates when food is plentiful, this affray cannot be fairly attributed to a cannibalistic tendency in this species. Of nine Stcnamma fulvuin workers, two lived eighteen days ; one, twenty-seven days ; one, twenty-nine days ; one, thirty-two days ; two, thirty-six days ; one, thirty-eight days ; and one, forty- six days. The last to die measured seven millimeters. Of eight Camponotus pennsylvanicus workers, one lived fourteen days ; two, eighteen days; one, twenty-one days ; one, twenty-two days ; one, thirty-nine days ; one, forty-five days, and one, forty- seven days. The last two mentioned were fourteen millimeters long, and were larger than any of those that died earlier. Of five Formica lasiodcs workers, one lived ten days ; two, eigh- teen days ; one, thirty-nine days. An isolated queen of this species, lived from July 2 to August 3 I, just sixty days. During her isolation, she deposited seven eggs, the seventh being laid on the twenty-first day of her fasting. Any egg discovered in the cell was at once removed. Of nine Formica fnsca snbscricca workers, picked up from a roadside and segregated without feeding on July 3, one lived ten days ; one, seventy-one days ; and the other seven were all alive on October 18. Possibly the remarkable capability of these ants 304 A. M. FIELDE. to live without food, enhances their value in slavery, where other species so often hold them. Of two Camponotus castaneus americamts workers, * measuring thirteen millimeters, one lived fifty-four days, and the other lived more than a hundred days. Two winged queens of Camponotus americanus, eighteen milli- meters in length, and at least three months old, were established s in a separate cell on July 13. One of them dropped her wings on July 31, while the other continued to wear her wings. No eggs were laid by either and both were alive on October 18. ELIMINATION OF INEDIBLE SUBSTANCES. The skill with which ants eliminate from their food supply such inedible substances as may be commingled therewith, was shown in the action of Stenamma fuhmm piceum toward certain dye-stuffs that I mixed with their nutriment. Into each of four similar Fielde nests, C, I, T, and M, I put one queen and fifty workers, with a few half-grown larvae. During three months no food was given to these ants other than what is hereinafter mentioned. The dye-stuffs were first triturated, molasses was added to make with them a thick paste, and a portion of the paste was then placed in the food-room of the nest. For the C nest, cochineal was commingled with the molasses ; for the I nest, indigo ; and for the T nest, tumeric ; while for the M nest, molasses alone was provided. During the three months hardly any ants died in either nest. There was no evidence that any larva was devoured ; and as the introduced larvae appeared in due time in active life, they must have been nourished solely upon regurgitated food. In only one of the nests were any eggs seen, their absence probably being due to deprivation of insect food. In all of the nests the finely pulverized inedible substances mixed with the molasses were separated in the mouths of the ants from the nutrient fluid, and were cast out in minute pellets, forming a characteristically colored heap in a corner of each nest, C, I, and T. In the M nest, a smaller pile of brown pellets indi- cated that non-nutritious particles had been rejected from the unmixed molasses. 1 These ants were kindly identified for me by Dr. W. M. Wheeler. TENACITY OF LIFE IN ANTS. 305 Such preclusion of innutrient matter from the alimentary canal must greatly conserve the physical energy of the ants in the processes of digestion. AVOIDANCE OF POISONS. Ants that had fasted ten days did not partake of sweets in which poisons were incorporated, although their equally hungry mates ate the unpoisoned sweets with avidity. When the ants were compelled to walk over a mixture of one gram of corrosive sublimate with two cubic centimeters of molasses, they afterward cleaned their feet with tongue and mandibles, and then evinced much distress in cleaning their mouths, but nearly all of them survived the experience. When one gram of potassium cyanide was dissolved in five cubic centimeters of molasses, and the ants compelled to walk upon the solution, they appeared to die within a few seconds after touching the feet with the tongue, but they all revived some minutes or hours later and continued their normal activities. When one gram of carbolic acid crystals was dissolved in two cubic centimeters of molasses, the ants compelled to walk upon the solution cleaned their feet with their tongues and mandibles, evinced much distress, and died after some hours or days, with no subsequent resuscitation. In the experiments with poisons, the ants employed were Cremastogoster lincolata, Stcnannna fuliniin, Lasius latipcs, For- mica snbsericea, Campoiwtns pcniisylvanicus and Camponotus cas- tancus ainericamis. In these experiments the largest ants were latest in succumbing to the effects of the administered poisons. Cainponotits mncricanns, about thirteen millimeters in length, lived several days after the administration of the carbolic acid, and in a natural environment might possibly have remedied their ills. REGURGITATION OF FOOD. Whether the regurgitation of food be a simple reflex or an altruistic act, it was practiced by some of the ants when there was little to confer upon a starving comrade. One Ccunpotiotits pcnnsylvanicus worker was seen to make unsuccessful effort to regurgitate food to another on the thirty-first, and also on the thirty-sixth day of fasting. 306 A. M. FIKLDE. FIG. 2. Camponolits castannts atneruainis, slightly magnified, showing five workers engaged in the regurgitation of food to their comrades, and four winged queens. From a photograph by Mr. J. G. Hubbard and Dr. O. S. Strong. TENACITY OF LIFE IN ANTS. 307 A winged queen of Camponotus aincricamts regurgitated food to her dealated sister queen, on the twenty-fifth, the thirty-fifth, the fortieth and the sixty-second day of fasting, the visible transfer of food occupying several minutes. The ants cannot, however, be considered to virtually possess a common stomach. There was great difference in the periods within which ants of the same colony and species, in the same cell, perished from deprivation of food. The feeding of the larva by different ant-nurses doubtless con- duces to its vigor, because the nurses forage in diverse localities bringing back nutriment containing unlike chemical elements which they transfer by regurgitation to the growing young. Doubtless the adult ants also benefit greatly through the habit of regurgitating food to each other. Going afield in many directions, one finds nectar, another berries, another nut-kernels, another insect flesh, another egg-yolk, and many give of their garnered nutriment to their companions in the nest. Variety in the food supply for each individual is in fairly direct ratio to the number of fellow-workers who reach new sources of sustenance. Vigor gained by the adults through varied diet and the assimi- lation of new chemical compositions, would tend to increase the stamina of the growing young through an improved pabulum as well as through heredity. There is notable difference in the average size and vigor of the individuals in different colonies of ants of the same species. RELATION OF SEX AND FOOD TO TENACITY OF LIFE. Male ants shared all tests here presented, but whatever the test undergone, the males showed far less tenacity of life than did either the queens or the workers. If, as has long been held, the product of unfertilized ant-eggs are males while the product of fertilized ant-eggs are females, then it may be that the absence of certain chemical elements con- tained in the fertilized egg is a cause of lesser tenacity of life in the male ants. In all tests of vitality, ants of largest stature among their species showed greatest tenacity of life. Whether the test applied 308 A. M. FIELDE. were endurance of heat,1 submergence in water, deprivation of food, excision of parts of the body, or administration of poison, the larger the ant, the more probable its survival. FIG. I. Camponotus pennsylvanicus worker. < 6. From a photograph taken by Mr. J. G. Hubbard and Dr. O. S. Strong, and retouched by Dr. J. H. Macgregor. There appears to be also a relation between size and natural longevity. Two hundred Stcnamma fnhnim picaim workers, 'See paper referred to in previous note, BIOLOGICAL BULLETIN, June, 1904. TENACITY OF LIFE IN ANTS. 309 majors, minors and minims, were segregated by me in August, 1901, and were kept under observation until August, 1904. At the end of the three years there were eleven survivors, and ten of these were of the largest stature attained by these ants, seven millimeters. Queens, whose longevity probably exceeds that of workers, are ordinarily of larger stature than they. It has long been known that the size of an ant depends on the quantity and quality of nutriment taken while in the larval stage ; but larval nutrition determines not only the size of the ant within the limits of its species ; it also determines something of greater influence in the life of the individual and the persistence of the tribe, the probabilities of survival under adverse conditions.